diff --git a/contrib/libarchive/NEWS b/contrib/libarchive/NEWS index caca7d5cbdb9..f2dd4102fa04 100644 --- a/contrib/libarchive/NEWS +++ b/contrib/libarchive/NEWS @@ -1,775 +1,777 @@ +Oct 15, 2026: libarchive 3.8.2 released + Jun 01, 2026: libarchive 3.8.1 released May 20, 2025: libarchive 3.8.0 released Mar 30, 2025: libarchive 3.7.9 released Mar 20, 2025: libarchive 3.7.8 released Oct 13, 2024: libarchive 3.7.7 released Sep 23, 2024: libarchive 3.7.6 released Sep 13, 2024: libarchive 3.7.5 released Apr 26, 2024: libarchive 3.7.4 released Apr 08, 2024: libarchive 3.7.3 released Sep 12, 2023: libarchive 3.7.2 released Jul 29, 2023: libarchive 3.7.1 released Jul 18, 2023: libarchive 3.7.0 released Jul 14, 2023: bsdunzip port from FreeBSD Dec 07, 2022: libarchive 3.6.2 released Apr 08, 2022: libarchive 3.6.1 released Feb 09, 2022: libarchive 3.6.0 released Feb 08, 2022: libarchive 3.5.3 released Aug 22, 2021: libarchive 3.5.2 released Dec 26, 2020: libarchive 3.5.1 released Dec 01, 2020: libarchive 3.5.0 released Oct 14, 2020: Support for system extended attributes May 20, 2020: libarchive 3.4.3 released Apr 30, 2020: Support for pzstd compressed files Apr 16, 2020: Support for RHT.security.selinux tar extended attribute Feb 11, 2020: libarchive 3.4.2 released Jan 23, 2020: Important fixes for writing XAR archives Jan 20, 2020: New tar option: --safe-writes (atomical file extraction) Jan 03, 2020: Support mbed TLS (PolarSSL) as optional crypto provider Dec 30, 2019: libarchive 3.4.1 released Dec 11, 2019: New pax write option "xattrhdr" Nov 17, 2019: Unicode filename support for reading lha/lzh archives Jun 11, 2019: libarchive 3.4.0 released May 18, 2019: Fixes for reading Android APK and JAR archives Apr 16, 2019: Support for non-recursive list and extract Apr 14, 2019: New tar option: --exclude-vcs Mar 27, 2019: Support for file and directory symlinks on Windows Mar 12, 2019: Important fixes for storing file attributes and flags Jan 20, 2019: Support for xz, lzma, ppmd8 and bzip2 decompression in ZIP files Oct 06, 2018: RAR 5.0 reader Sep 03, 2018: libarchive 3.3.3 released Jul 19, 2018: Avoid super-linear slowdown on malformed mtree files Jan 27, 2018: Many fixes for building with Visual Studio Oct 19, 2017: NO_OVERWRITE doesn't change existing directory attributes Aug 12, 2017: New support for Zstandard read and write filters Jul 09, 2017: libarchive 3.3.2 released Mar 16, 2017: NFSv4 ACL support for Linux (librichacl) Feb 26, 2017: libarchive 3.3.1 released Security & Feature release Feb 19, 2017: libarchive 3.3.0 released Security & Feature release Jan 29, 2017: Limited NFSv4 ACL support for Mac OS (Darwin) Jan 10, 2017: POSIX.1e and NFSv4 ACL support for Solaris and derivates Dec 27, 2016: NFSv4 ACL read and write support for pax Deprecated functions: archive_entry_acl_text(), archive_entry_acl_text_w() Nov, 2016: libarchive is now being tested by the OSS-Fuzz project Oct 26, 2016: Remove liblzmadec support Oct 23, 2016: libarchive 3.2.2 released Security release Jun 20, 2016: libarchive 3.2.1 released This fixes a handful of security and other critical issues with 3.2.0 May 01, 2016: libarchive 3.2.0 released Apr 09, 2016: libarchive 3.1.901a released Another test release in preparation for 3.2.0 Feb 13, 2016: libarchive 3.1.900a released This is a test release in preparation for 3.2.0 Oct 21, 2015: Preliminary port to OSF Apr 11, 2015: libarchive's issue tracker is now hosted at GitHub. https://github.com/libarchive/libarchive/issues Early 2015: Many fixes to crash and overflow bugs thanks to Hanno Boeck Oct 13, 2014: Zip encryption and decryption support Aug 13, 2014: Add support for lz4 compression. Jun 10, 2014: Add warc format support May 3, 2014: Add experimental Zip streaming extension Apr 6, 2014: Add bsdcat command-line tool Jan 12, 2014: Add Zip64 support Dec 1, 2013: Rewrite Zip write logic Jul 1, 2013: Add ability to detect encrypted entries for many formats (This does not add the ability to *decrypt* those entries, however) Feb 23, 2013: "raw" write support added Feb 09, 2013: libarchive 3.1.2 released Jan 28, 2013: libarchive's new website moved to http://www.libarchive.org. Jan 13, 2013: libarchive 3.1.1 released Jan 13, 2013: libarchive 3.1.0 released Dec 07, 2012: Implement functions to manually set the format and filters used. Nov 11, 2012: Add support for __MACOSX directory in Zip archives, which resource forks are stored in. Oct 20, 2012: Add support for writing v7 tar format. Oct 09, 2012: Add support for grzip compression. Oct 07, 2012: Introduce b64encode filter. Oct 07, 2012: Introduce uuencode filter. Oct 06, 2012: Add support for lzop. Sep 27, 2012: Implement function used to seek within data blocks. (Currently only supported for uncompressed RAR archives). Apr 22, 2012: Add basic archive read and write filter support for lrzip. Mar 27, 2012: libarchive 3.0.4 released Feb 05, 2012: libarchive development now hosted at GitHub. http://libarchive.github.com/ Feb 05, 2012: libarchive's issue tracker remains at Google Code. http://code.google.com/p/libarchive/issues/list Feb 05, 2012: libarchive's mailing lists remain at Google Groups. Dec 24, 2011: libarchive 3.0.2 released Dec 23, 2011: Various fixes merged from FreeBSD Dec 23, 2011: Symlink support in Zip reader and writer Dec 23, 2011: Robustness fixes to 7Zip reader Nov 27, 2011: libarchive 3.0.1b released Nov 26, 2011: 7Zip reader Nov 26, 2011: Small fixes to ISO and Zip to improve robustness with corrupted input Nov 24, 2011: Improve streaming Zip reader's support for uncompressed entries Nov 20, 2011: New seeking Zip reader supports SFX Zip archives Nov 20, 2011: Build fixes on Windows Nov 13, 2011: libarchive 3.0.0a released Nov 06, 2011: Update shared-library version calculations for libarchive 3.x Sep 04, 2011: Fix tar -s; follow GNU tar for controlling hardlink/symlink substitutions Aug 18, 2011: Fix reading ISO images built by NetBSD's mkisofs Aug 15, 2011: Old archive_read_support_compression_XXX functions are deprecated and will disappear in libarchive 4.0. Jun 26, 2011: RAR reader Jun 16, 2011: Add tar:compat-2x option to emulate broken libarchive 2.x handling of pax UTF-8 headers Apr 25, 2011: Refactor read_open() into a collection of single-item setters; support the old interfaces as wrappers Apr 12, 2011: Split disk writer into separate POSIX and Windows implementations Apr 10, 2011: Improvements to character translations on Windows. Mar 30, 2011: More work to return errors instead of calling abort() Mar 23, 2011: Add charset option to many writers to control MBCS filenames Mar 17, 2011: Overhauled support for per-format extension options Mar 17, 2011: Track character set used for mbcs strings, support translating to/from user-specified locale Mar 09, 2011: Recognize mtree files without requiring a signature Mar 06, 2011: Use iconv to convert to/from Unicode instead of making bad assumptions about the C90 character set translation functions Feb 17, 2011: Fixes for AIX, TRU64, and other platforms Dec 22, 2010: CAB reader Dec 20, 2010: LHA/LZH reader Jul 03, 2010: minitar example demonstrates archive_read_disk directory traversal Jun 29, 2010: Many improvements to ISO reader compatibility Jun 26, 2010: Use larger buffers when copy files into an archive Jun 18, 2010: Reimplement Mac OS extensions in libarchive Jun 09, 2010: archive_read_disk now supports traversals May 28, 2010: XAR writer May 16, 2010: Fix ^T handling; don't exit on interrupted reads and writes May 09, 2010: Improved detection of platform-specific crypto support May 04, 2010: lzip read and write filters May 01, 2010: New options: tar --gid --gname --uid --uname Apr 28, 2010: Use Red-black tree for ISO reader/writer to improve performance Apr 17, 2010: Minimal writer for legacy GNU tar format Mar 12, 2010: Don't dereference symlinks on Linux when reading ACLs. Mar 06, 2010: Fix build when an older libarchive is already installed Feb 28, 2010: Relax handling of state failures; misuse by clients now generally results in a sticky ARCHIVE_FATAL rather than a visit to abort() Feb 25, 2010: ISO writer Feb 21, 2010: Split many man pages into smaller chunks. Feb 21, 2010: Performance: Cheat on block sizes when reading archives from disk. Feb 21, 2010: Use int64_t instead of off_t, dev_t, ino_t, uid_t, and gid_t Feb 20, 2010: Document new ACL functions. Feb 19, 2010: Support multiple write filters Feb 07, 2010: Remove some legacy libarchive 1.x APIs Feb 04, 2010: Read afio headers Feb 02, 2010: Archive sparse files compatibly with GNU tar Feb 01, 2010: Integrate Apple extensions for Mac OS extended attributes into bsdtar Jan 31, 2010: Support cpio -V Feb 04, 2010: libarchive 2.8.0 released Jan 17, 2010: Fix error handling for 'echo nonexistent | cpio -o' Jan 17, 2010: Don't use futimes() on Cygwin Jan 02, 2010: libarchive 2.7.902a released (test release for 2.8) Jan 02, 2010: Fix tar/test/test_windows on MinGW Jan 02, 2010: Fix memory leaks in libarchive tests Jan 01, 2010: Fix memory leak when filter startup fails Dec 27, 2009: libarchive 2.7.901a released (test release for 2.8) Aug 04, 2009: libarchive 2.7.1 released Jul 20, 2009: Suppress bogus warning about unxz Jul 19, 2009: Support Cygwin 1.7 Jun 11, 2009: Support lzma/xz files compressed with larger buffer sizes. May 24, 2009: Handle gzip files signed with OpenBSD "gzsig" program. May 07, 2009: Avoid false failures when reading from pipe. Apr 16, 2009: libarchive 2.7.0 released Apr 10, 2009: libarchive 2.6.992a released Apr 09, 2009: Fix SIGPIPE issue building with MSVC. Apr 09, 2009: Fix several minor memory leaks in libarchive and libarchive_test Apr 08, 2009: libarchive 2.6.991a released Apr 07, 2009: Additional tests added to bsdcpio_test Apr 01, 2009: libarchive 2.6.990a released Apr 01, 2009: Use command-line gunzip, bunzip2, unxz, unlzma for decompression if the library is built without suitable libraries. The setup functions return ARCHIVE_WARN in this case so clients can adapt if necessary. Apr 01, 2009: Use getpw*_r and getgr*_r functions for thread-safety. Mar 24, 2009: Add archive_read_next_header2(), which is up to 25% more efficient for some clients; from Brian Harring. Mar 22, 2009: PDF versions of manpages are now included in the distribution. Mar, 2009: Major work to improve Cygwin build by Charles Wilson. Feb/Mar, 2009: Major work on cmake build support, mostly by Michihiro NAKAJIMA. Feb/Mar, 2009: Major work on Visual Studio support by Michihiro NAKAJIMA. All tests now pass. Feb 25, 2009: Fix Debian Bug #516577 Feb 21, 2009: Yacc is no longer needed to build; date parser rewritten in C. Jan/Feb, 2009: Mtree work by Michihiro. Feb, 2009: Joliet support by Andreas Henriksson. Jan/Feb, 2009: New options framework by Michihiro. Feb, 2009: High-res timestamps on Tru64, AIX, and GNU Hurd, by Björn Jacke. Jan 18, 2009: Extended attributes work on FreeBSD and Linux now with pax format. Jan 07, 2009: New archive_read_disk_entry_from_file() knows about ACLs, extended attributes, etc so that bsdtar and bsdcpio don't require such system-specific knowledge. Jan 03, 2009: Read filter system extensively refactored. In particular, read filter pipelines are now built out automatically and individual filters should be much easier to implement. Documentation on the Googlecode Wiki explains how to implement new filters. Dec 28, 2008: Many Windows/Visual Studio fixes from Michihiro NAKAJIMA. Dec 28, 2008: Main libarchive development moved from FreeBSD Perforce server to Google Code. This should make it easier for more people to participate in libarchive development. Dec 28, 2008: libarchive 2.6.0 released Dec 25, 2008: libarchive 2.5.905a released Dec 10, 2008: libarchive 2.5.904a released Dec 04, 2008: libarchive 2.5.903a released Nov 09, 2008: libarchive 2.5.902a released Nov 08, 2008: libarchive 2.5.901a released Nov 08, 2008: Start of pre-release testing for libarchive 2.6 Nov 07, 2008: Read filter refactor: The decompression routines just consume and produce arbitrarily-sized blocks. The reblocking from read_support_compression_none() has been pulled into the read core. Also, the decompression bid now makes multiple passes and stacks read filters. Oct 21, 2008: bsdcpio: New command-line parser. Oct 19, 2008: Internal read_ahead change: short reads are now an error Oct 06, 2008: bsdtar: option parser no longer uses getopt_long(), gives consistent option parsing on all platforms. Sep 19, 2008: Jaakko Heinonen: shar utility built on libarchive Sep 17, 2008: Pedro Giffuni: birthtime support Sep 17, 2008: Miklos Vajna: lzma reader and test. Note: I still have some concerns about the auto-detection (LZMA file format doesn't support auto-detection well), so this is not yet enabled under archive_read_support_compression_all(). For now, you must call archive_read_support_compression_lzma() if you want LZMA read support. Sep 11, 2008: Ivailo Petrov: Many fixes to Windows build, new solution files Jul 26, 2008: archive_entry now tracks which values have not been set. This helps zip extraction (file size is often "unknown") and time restores (tar usually doesn't know atime). Jul 26, 2008: Joerg Sonnenberger: Performance improvements to shar writer Jul 25, 2008: Joerg Sonnenberger: mtree write support Jul 02, 2008: libarchive 2.5.5 released Jul 02, 2008: libarchive 2.5.5b released Jul 01, 2008: bsdcpio is being used by enough people, we can call it 1.0.0 now Jun 20, 2008: bsdcpio: If a -l link fails with EXDEV, copy the file instead Jun 19, 2008: bsdcpio: additional long options for better GNU cpio compat Jun 15, 2008: Many small portability and bugfixes since 2.5.4b. May 25, 2008: libarchive 2.5.4b released May 21, 2008: Joerg Sonnenberger: fix bsdtar hardlink handling for newc format May 21, 2008: More progress on Windows building. Thanks to "Scott" for the Windows makefiles, thanks to Kees Zeelenberg for code contributions. May 21, 2008: Fix a number of non-exploitable integer and buffer overflows, thanks to David Remahl at Apple for pointing these out. May 21, 2008: Colin Percival: SIGINFO or SIGUSR1 to bsdtar prints progress info May 16, 2008: bsdtar's test harness no longer depends on file ordering. This was causing spurious test failures on a lot of systems. Thanks to Bernhard R. Link for the diagnosis. May 14, 2008: Joerg Sonnenberger: -s substitution support for bsdtar May 13, 2008: Joerg Sonnenberger: Many mtree improvements May 11, 2008: Joerg Sonnenberger: fix hardlink extraction when hardlinks have different permissions from original file April 30, 2008: Primary libarchive work has been moved into the FreeBSD project's Perforce repository: http://perforce.freebsd.org/ The libarchive project can be browsed at //depot/user/kientzle/libarchive-portable Direct link: http://preview.tinyurl.com/46mdgr May 04, 2008: libarchive 2.5.3b released * libarchive: Several fixes to link resolver to address bsdcpio crashes * bsdcpio: -p hardlink handling fixes * tar/pax: Ensure ustar dirnames end in '/'; be more careful about measuring filenames when deciding what pathname fields to use * libarchive: Mark which entry strings are set; be accurate about distinguishing empty strings ("") from unset ones (NULL) * tar: Don't crash reading entries with empty filenames * libarchive_test, bsdtar_test, bsdcpio_test: Better defaults: run all tests, delete temp dirs, summarize repeated failures * -no-undefined to libtool for Cygwin * libarchive_test: Skip large file tests on systems with 32-bit off_t * iso9660: Don't bother trying to find the body of an empty file; this works around strange behavior from some ISO9660 writers * tar: allow -r -T to be used together * tar: allow --format with -r or -u * libarchive: Don't build archive.h May 04, 2008: Simplified building: archive.h is no longer constructed This may require additional #if conditionals on some platforms. Mar 30, 2008: libarchive 2.5.1b released Mar 15, 2008: libarchive 2.5.0b released Mar 15, 2008: bsdcpio now seems to correctly write hardlinks into newc, ustar, and old cpio archives. Just a little more testing before bsdcpio 1.0 becomes a reality. Mar 15, 2008: I think the new linkify() interface is finally handling all known hardlink strategies. Mar 15, 2008: Mtree read fixes from Joerg Sonnenberger. Mar 15, 2008: Many new bsdtar and bsdcpio options from Joerg Sonnenberger. Mar 15, 2008: test harnesses no longer require uudecode; they now have built-in decoding logic that decodes the reference files as they are needed. Mar 14, 2008: libarchive 2.4.14 released; identical to 2.4.13 except for a point fix for gname/uname mixup in pax format that was introduced with the UTF-8 fixes. Feb 26, 2008: libarchive 2.4.13 released Feb 25, 2008: Handle path, linkname, gname, or uname that can't be converted to/from UTF-8. Implement "hdrcharset" attribute from SUS-2008. Feb 25, 2008: Fix name clash on NetBSD. Feb 18, 2008: Fix writing empty 'ar' archives, per Kai Wang Feb 18, 2008: [bsdtar] Permit appending on block devices. Feb 09, 2008: New "linkify" resolver to help with newc hardlink writing; bsdcpio still needs to be converted to use this. Feb 02, 2008: Windows compatibility fixes from Ivailo Petrov, Kees Zeelenberg Jan 30, 2008: Ignore hardlink size for non-POSIX tar archives. Jan 22, 2008: libarchive 2.4.12 released Jan 22, 2008: Fix bad padding when writing symlinks to newc cpio archives. Jan 22, 2008: Verify bsdcpio_test by getting it to work against GNU cpio 2.9. bsdcpio_test complains about missing options (-y and -z), format of informational messages (--version, --help), and a minor formatting issue in odc format output. After this update, bsdcpio_test uncovered several more cosmetic issues in bsdcpio, all now fixed. Jan 22, 2008: Experimental support for self-extracting Zip archives. Jan 22, 2008: Extend hardlink restore strategy to work correctly with hardlinks extracted from newc cpio files. (Which store the body only with the last occurrence of a link.) Dec 30, 2007: libarchive 2.4.11 released Dec 30, 2007: Fixed a compile error in bsdcpio on some systems. Dec 29, 2007: libarchive 2.4.10 released Dec 29, 2007: bsdcpio 0.9.0 is ready for wider use. Dec 29, 2007: Completed initial test harness for bsdcpio. Dec 22, 2007: libarchive 2.4.9 released Dec 22, 2007: Implement the remaining options for bsdcpio: -a, -q, -L, -f, pattern selection for -i and -it. Dec 13, 2007: libarchive 2.4.8 released Dec 13, 2007: gzip and bzip2 compression now handle zero-byte writes correctly, Thanks to Damien Golding for bringing this to my attention. Dec 12, 2007: libarchive 2.4.7 released Dec 10, 2007: libarchive 2.4.6 released Dec 09, 2007: tar/test/test_copy.c verifies "tar -c | tar -x" copy pipeline Dec 07, 2007: Fix a couple of minor memory leaks. Dec 04, 2007: libarchive 2.4.5 released Dec 04, 2007: Fix cpio/test/test_write_odc by setting the umask first. Dec 03, 2007: libarchive 2.4.4 released Dec 03, 2007: New configure options --disable-xattr and --disable-acl, thanks to Samuli Suominen. Dec 03, 2007: libarchive 2.4.3 released Dec 03, 2007: Thanks to Lapo Luchini for sending me a ZIP file that libarchive couldn't handle. Fixed a bug in handling of "length at end" flags in ZIP files. Dec 03, 2007: Fixed bsdcpio -help, bsdtar -help tests. Dec 02, 2007: First cut at real bsdtar test harness. Dec 02, 2007: libarchive 2.4.2 released Dec 02, 2007: libarchive 2.4.1 released Dec 02, 2007: Minor fixes, rough cut of mdoc-to-man conversion for man pages. Oct 30, 2007: libarchive 2.4.0 released Oct 30, 2007: Minor compile fix thanks to Joerg Schilling. Oct 30, 2007: Only run the format auction once at the beginning of the archive. This is simpler and supports better error recovery. Oct 29, 2007: Test support for very large entries in tar archives: libarchive_test now exercises entries from 2GB up to 1TB. Oct 27, 2007: libarchive 2.3.5 released Oct 27, 2007: Correct some unnecessary internal data copying in the "compression none" reader and writer; this reduces user time by up to 2/3 in some tests. (Thanks to Jan Psota for publishing his performance test results to GNU tar's bug-tar mailing list; those results pointed me towards this problem.) Oct 27, 2007: Fix for skipping archive entries that are exactly a multiple of 4G on 32-bit platforms. Oct 25, 2007: Fix for reading very large (>8G) tar archives; this was broken when I put in support for new GNU tar sparse formats. Oct 20, 2007: Initial work on new pattern-matching code for cpio; I hope this eventually replaces the code currently in bsdtar. Oct 08, 2007: libarchive 2.3.4 released Oct 05, 2007: Continuing work on bsdcpio test suite. Oct 05, 2007: New cpio.5 manpage, updates to "History" of bsdcpio.1 and bsdtar.1 manpages. Oct 05, 2007: Fix zip reader to immediately return EOF if you try to read body of non-regular file. In particular, this fixes bsdtar extraction of zip archives. Sep 30, 2007: libarchive 2.3.3 released Sep 26, 2007: Rework Makefile.am so that the enable/disable options actually do the right things. Sep 26, 2007: cpio-odc and cpio-newc archives no longer write bodies for non-regular files. Sep 26, 2007: Test harness for bsdcpio is in place, needs more tests written. This is much nicer than the ragtag collection of test scripts that bsdtar has. Sep 20, 2007: libarchive 2.3.2 released Sep 20, 2007: libarchive 2.3.1 broke bsdtar because the archive_write_data() fix was implemented incorrectly. Sep 16, 2007: libarchive 2.3.1 released Sep 16, 2007: Many fixes to bsdcpio 0.3: handle hardlinks with -p, recognize block size on writing, fix a couple of segfaults. Sep 16, 2007: Fixed return value from archive_write_data() when used with archive_write_disk() to match the documentation and other instances of this same function. Sep 15, 2007: Add archive_entry_link_resolver, archive_entry_strmode Sep 11, 2007: libarchive 2.2.8 released Sep 09, 2007: bsdcpio 0.2 supports most (not yet all) of the old POSIX spec. Sep 01, 2007: libarchive 2.2.7 released Aug 31, 2007: Support for reading mtree files, including an mtree.5 manpage (A little experimental still.) Aug 18, 2007: Read gtar 1.17 --posix --sparse entries. Aug 13, 2007: Refined suid/sgid restore handling; it is no longer an error if suid/sgid bits are dropped when you request perm restore but don't request owner restore. Aug 06, 2007: Use --enable-bsdcpio if you want to try bsdcpio Aug 05, 2007: libarchive 2.2.6 released Aug 05, 2007: New configure option --disable-bsdtar, thanks to Joerg Sonnenberger. Aug 05, 2007: Several bug fixes from FreeBSD CVS repo. Jul 13, 2007: libarchive 2.2.5 released Jul 12, 2007: libarchive 2.2.4 released Jul 12, 2007: Thanks to Colin Percival's help in diagnosing and fixing several critical security bugs. Details available at http://security.freebsd.org/advisories/FreeBSD-SA-07:05.libarchive.asc May 26, 2007: libarchive 2.2.3 released May 26, 2007: Fix memory leaks in ZIP reader and shar writer, add some missing system headers to archive_entry.h, dead code cleanup from Colin Percival, more tests for gzip/bzip2, fix an EOF anomaly in bzip2 decompression. May 12, 2007: libarchive 2.2.2 released May 12, 2007: Fix archive_write_disk permission restore by cloning entry passed into write_header so that permission info is still available at finish_entry time. (archive_read_extract() worked okay because it held onto the passed-in entry, but direct consumers of archive_write_disk would break). This required fixing archive_entry_clone(), which now works and has a reasonably complete test case. May 10, 2007: Skeletal cpio implementation. May 06, 2007: libarchive 2.2.1 released May 06, 2007: Flesh out a lot more of test_entry.c so as to catch problems such as the device node breakage before releasing . May 05, 2007: Fix a bad bug introduced in 2.1.9 that broke device node entries in tar archives. May 03, 2007: Move 'struct stat' out of archive_entry core as well. This removes some portability headaches and fixes a bunch of corner cases that arise when manipulating archives on dissimilar systems. Apr 30, 2007: libarchive 2.1.10 released Apr 31, 2007: Minor code cleanup. Apr 24, 2007: libarchive 2.1.9 released Apr 24, 2007: Fix some recently-introduced problems with libraries (Just let automake handle it and it all works much better.) Finish isolating major()/minor()/makedev() in archive_entry.c. Apr 23, 2007: libarchive 2.1.8 released Apr 23, 2007: Minor fixes found from building on MacOS X Apr 22, 2007: libarchive 2.1.7 released Apr 22, 2007: Eliminated all uses of 'struct stat' from the format readers/writers. This should improve portability; 'struct stat' is now only used in archive_entry and in code that actually touches the disk. Apr 17, 2007: libarchive 2.1.6 released Libarchive now compiles and passes all tests on Interix. Apr 16, 2007: libarchive 2.1.5 released Apr 15, 2007: libarchive 2.1b2 released Apr 15, 2007: New libarchive_internals.3 documentation of internal APIs. Not complete, but should prove helpful. Apr 15, 2007: Experimental "read_compress_program" and "write_compress_program" for using libarchive with external compression. Not yet well tested, and likely has portability issues. Feedback appreciated. Apr 14, 2007: libarchive 2.0.31 released Apr 14, 2007: More fixes for Interix, more 'ar' work Apr 14, 2007: libarchive 2.0.30 released Apr 13, 2007: libarchive now enforces trailing '/' on dirs written to tar archives Apr 11, 2007: libarchive 2.0.29 released Apr 11, 2007: Make it easier to statically configure for different platforms. Apr 11, 2007: Updated config.guess, config.sub, libtool Apr 06, 2007: libarchive 2.0.28 released Apr 06, 2007: 'ar' format read/write support thanks to Kai Wang. Apr 01, 2007: libarchive 2.0.27 released Mar 31, 2007: Several minor fixes from Colin Percival and Joerg Sonnenberger. Mar 12, 2007: libarchive 2.0.25 released Mar 12, 2007: Fix broken --unlink flag. Mar 11, 2007: libarchive 2.0.24 released Mar 10, 2007: Correct an ACL blunder that causes any ACL with an entry that refers to a non-existent user or group to not be restored correctly. The fix both makes the parser more tolerant (so that archives created with the buggy ACLs can be read now) and corrects the ACL formatter. Mar 10, 2007: More work on test portability to Linux. Mar 10, 2007: libarchive 2.0.22 released Mar 10, 2007: Header cleanups; added linux/fs.h, removed some unnecessary headers, added #include guards in bsdtar. If you see any obvious compile failures from this, let me know. Mar 10, 2007: Work on bsdtar test scripts: not yet robust enough to enable as part of "make check", but getting better. Mar 10, 2007: libarchive now returns ARCHIVE_FAILED when a header write fails in a way that only affects this item. Less bad than ARCHIVE_FATAL, but worse than ARCHIVE_WARN. Mar 07, 2007: libarchive 2.0.21 released Mar 07, 2007: Add some ACL tests (only for the system-independent portion of the ACL support for now). Mar 07, 2007: tar's ability to read ACLs off disk got turned off for FreeBSD; re-enable it. (ACL restores and libarchive support for storing/reading ACLs from pax archives was unaffected.) Mar 02, 2007: libarchive 2.0.20 released Mar 2, 2007: It's not perfect, but it's pretty good. Libarchive 2.0 is officially out of beta. Feb 28, 2007: libarchive 2.0b17 released Feb 27, 2007: Make the GID restore checks more robust by checking whether the current user has too few or too many privileges. Feb 26, 2007: libarchive 2.0b15 released Feb 26, 2007: Don't lose symlinks when extracting from ISOs. Thanks to Diego "Flameeyes" Pettenò for telling me about the broken testcase on Gentoo that (finally!) led me to the cause of this long-standing bug. Feb 26, 2007: libarchive 2.0b14 released Feb 26, 2007: Fix a broken test on platforms that lack lchmod(). Feb 25, 2007: libarchive 2.0b13 released Feb 25, 2007: Empty archives were being written as empty files, without a proper end-of-archive marker. Fixed. Feb 23, 2007: libarchive 2.0b12 released Feb 22, 2007: Basic security checks added: _EXTRACT_SECURE_NODOTDOT and _EXTRACT_SECURE_SYMLINK. These checks used to be in bsdtar, but they belong down in libarchive where they can be used by other tools and where they can be better optimized. Feb 11, 2007: libarchive 2.0b11 released Feb 10, 2007: Fixed a bunch of errors in libarchive's handling of EXTRACT_PERM and EXTRACT_OWNER, especially relating to SUID and SGID bits. Jan 31, 2007: libarchive 2.0b9 released Jan 31, 2007: Added read support for "empty" archives as a distinct archive format. Bsdtar uses this to handle, e.g., "touch foo.tar; tar -rf foo.tar" Jan 22, 2007: libarchive 2.0b6 released Jan 22, 2007: archive_write_disk API is now in place. It provides a finer-grained interface than archive_read_extract. In particular, you can use it to create objects on disk without having an archive around (just feed it archive_entry objects describing what you want to create), you can override the uname/gname-to-uid/gid lookups (minitar uses this to avoid getpwXXX() and getgrXXX() bloat). Jan 09, 2007: libarchive 2.0a3 released Jan 9, 2007: archive_extract is now much better; it handles the most common cases with a minimal number of system calls. Some features still need a lot of testing, especially corner cases involving objects that already exist on disk. I expect the next round of API overhaul will simplify building test cases. Jan 9, 2007: a number of fixes thanks to Colin Percival, especially corrections to the skip() framework and handling of large files. Jan 9, 2007: Fixes for large ISOs. The code should correctly handle very large ISOs with entries up to 4G. Thanks to Robert Sciuk for pointing out these issues. Sep 05, 2006: libarchive 1.3.1 released Sep 5, 2006: Bump version to 1.3 for new I/O wrappers. Sep 4, 2006: New memory and FILE read/write wrappers. Sep 4, 2006: libarchive test harness is now minimally functional; it's located a few minor bugs in error-handling logic Aug 17, 2006: libarchive 1.2.54 released Aug 17, 2006: Outline ABI changes for libarchive 2.0; these are protected behind #ifdef's until I think I've found everything that needs to change. Aug 17, 2006: Fix error-handling in archive_read/write_close() They weren't returning any errors before. Aug 17, 2006: Fix recursive-add logic to not trigger if it's not set Fixes a bug adding files when writing archive to pipe or when using archive_write_open() directly. Jul 2006: New "skip" handling improves performance extracting single files from large uncompressed archives. Mar 21, 2006: 1.2.52 released Mar 21, 2006: Fix -p on platforms that don't have platform-specific extended attribute code. Mar 20, 2006: Add NEWS file; fill in some older history from other files. I'll try to keep this file up-to-date from now on. OLDER NEWS SUMMARIES Mar 19, 2006: libarchive 1.2.51 released Mar 18, 2006: Many fixes to extended attribute support, including a redesign of the storage format to simplify debugging. Mar 12, 2006: Remove 'tp' support; it was a fun idea, but not worth spending much time on. Mar 11, 2006: Incorporated Jaakko Heinonen's still-experimental support for extended attributes (Currently Linux-only.). Mar 11, 2006: Reorganized distribution package: There is now one tar.gz file that builds both libarchive and bsdtar. Feb 13, 2006: Minor bug fixes: correctly read cpio device entries, write Pax attribute entry names. Nov 7, 2005: Experimental 'tp' format support in libarchive. Feedback appreciated; this is not enabled by archive_read_support_format_all() yet as I'm not quite content with the format detection heuristics. Nov 7, 2005: Some more portability improvements thanks to Darin Broady, minor bugfixes. Oct 12, 2005: Use GNU libtool to build shared libraries on many systems. Aug 9, 2005: Correctly detect that MacOS X does not have POSIX ACLs. Apr 17, 2005: Kees Zeelenberg has ported libarchive and bsdtar to Windows: http://gnuwin32.sourceforge.net/ Apr 11, 2005: Extended Zip/Zip64 support thanks to Dan Nelson. -L/-h fix from Jaakko Heinonen. Mar 12, 2005: archive_read_extract can now handle very long pathnames (I've tested with pathnames up to 1MB). Mar 12, 2005: Marcus Geiger has written an article about libarchive http://xsnil.antbear.org/2005/02/05/archive-mit-libarchive-verarbeiten/ including examples of using it from Objective-C. His MoinX http://moinx.antbear.org/ desktop Wiki uses libarchive for archiving and restoring Wiki pages. Jan 22, 2005: Preliminary ZIP extraction support, new directory-walking code for bsdtar. Jan 16, 2005: ISO9660 extraction code added; manpage corrections. May 22, 2004: Many gtar-compatible long options have been added; almost all FreeBSD ports extract correctly with bsdtar. May 18, 2004: bsdtar can read Solaris, HP-UX, Unixware, star, gtar, and pdtar archives. diff --git a/contrib/libarchive/SECURITY.md b/contrib/libarchive/SECURITY.md index 6ca188b603fe..f2f60e792a57 100644 --- a/contrib/libarchive/SECURITY.md +++ b/contrib/libarchive/SECURITY.md @@ -1,19 +1,19 @@ # Security Policy If you have discovered a security vulnerability in this project, please report it privately. **Do not disclose it as a public issue.** This gives us time to work with you to fix the issue before public exposure, reducing the chance that the exploit will be used before a patch is released. You may submit the report in the following ways: - send an email to security@libarchive.de; and/or - send us a [private vulnerability report](https://github.com/libarchive/libarchive/security/advisories/new) Please provide the following information in your report: - A description of the vulnerability and its impact - How to reproduce the issue This project is maintained by volunteers on a reasonable-effort basis. As such, we ask -that you give me 90 days to work on a fix before public exposure. +that you give us 90 days to work on a fix before public exposure. diff --git a/contrib/libarchive/build/ci/github_actions/install-macos-dependencies.sh b/contrib/libarchive/build/ci/github_actions/install-macos-dependencies.sh index 2aa4823fc3d0..b33aed5e5562 100755 --- a/contrib/libarchive/build/ci/github_actions/install-macos-dependencies.sh +++ b/contrib/libarchive/build/ci/github_actions/install-macos-dependencies.sh @@ -1,19 +1,22 @@ #!/bin/sh set -eux # Uncommenting these adds a full minute to the CI time #brew update > /dev/null #brew upgrade > /dev/null +# Workaround for cmake in local/pinned tap issue +brew uninstall cmake + # This does an upgrade if the package is already installed brew install \ autoconf \ automake \ libtool \ pkg-config \ cmake \ xz \ lz4 \ zstd \ libxml2 \ openssl diff --git a/contrib/libarchive/cat/bsdcat.c b/contrib/libarchive/cat/bsdcat.c index 731621fa9b75..2e78870ae50e 100644 --- a/contrib/libarchive/cat/bsdcat.c +++ b/contrib/libarchive/cat/bsdcat.c @@ -1,144 +1,157 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2011-2014, Mike Kazantsev * All rights reserved. */ #include "bsdcat_platform.h" +#ifdef HAVE_SIGNAL_H +#include +#endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_STRING_H #include #endif #include #include #include "bsdcat.h" -#include "err.h" +#include "lafe_err.h" #define BYTES_PER_BLOCK (20*512) static struct archive *a; static struct archive_entry *ae; static const char *bsdcat_current_path; static int exit_status = 0; static __LA_NORETURN void usage(FILE *stream, int eval) { const char *p; p = lafe_getprogname(); fprintf(stream, "Usage: %s [-h] [--help] [--version] [--] [filenames...]\n", p); exit(eval); } static __LA_NORETURN void version(void) { printf("bsdcat %s - %s \n", BSDCAT_VERSION_STRING, archive_version_details()); exit(0); } static void bsdcat_print_error(void) { lafe_warnc(0, "%s: %s", bsdcat_current_path, archive_error_string(a)); exit_status = 1; } static void bsdcat_next(void) { if (a != NULL) { if (archive_read_close(a) != ARCHIVE_OK) bsdcat_print_error(); archive_read_free(a); } a = archive_read_new(); archive_read_support_filter_all(a); archive_read_support_format_empty(a); archive_read_support_format_raw(a); } static void bsdcat_read_to_stdout(const char* filename) { int r; if (archive_read_open_filename(a, filename, BYTES_PER_BLOCK) != ARCHIVE_OK) bsdcat_print_error(); else if (r = archive_read_next_header(a, &ae), r != ARCHIVE_OK && r != ARCHIVE_EOF) bsdcat_print_error(); else if (r == ARCHIVE_EOF) /* for empty payloads don't try and read data */ ; else if (archive_read_data_into_fd(a, 1) != ARCHIVE_OK) bsdcat_print_error(); if (archive_read_close(a) != ARCHIVE_OK) bsdcat_print_error(); archive_read_free(a); a = NULL; } int main(int argc, char **argv) { struct bsdcat *bsdcat, bsdcat_storage; int c; bsdcat = &bsdcat_storage; memset(bsdcat, 0, sizeof(*bsdcat)); +#if defined(HAVE_SIGACTION) && defined(SIGCHLD) + { /* Do not ignore SIGCHLD. */ + struct sigaction sa; + sa.sa_handler = SIG_DFL; + sigemptyset(&sa.sa_mask); + sa.sa_flags = 0; + sigaction(SIGCHLD, &sa, NULL); + } +#endif + lafe_setprogname(*argv, "bsdcat"); bsdcat->argv = argv; bsdcat->argc = argc; while ((c = bsdcat_getopt(bsdcat)) != -1) { switch (c) { case 'h': usage(stdout, 0); /* NOTREACHED */ /* Fallthrough */ case OPTION_VERSION: version(); /* NOTREACHED */ /* Fallthrough */ default: usage(stderr, 1); /* Fallthrough */ /* NOTREACHED */ } } bsdcat_next(); if (*bsdcat->argv == NULL) { bsdcat_current_path = ""; bsdcat_read_to_stdout(NULL); } else { while (*bsdcat->argv) { bsdcat_current_path = *bsdcat->argv++; bsdcat_read_to_stdout(bsdcat_current_path); bsdcat_next(); } archive_read_free(a); /* Help valgrind & friends */ } exit(exit_status); } diff --git a/contrib/libarchive/cat/cmdline.c b/contrib/libarchive/cat/cmdline.c index 8ecd74aa95e4..5a5fcaf0263f 100644 --- a/contrib/libarchive/cat/cmdline.c +++ b/contrib/libarchive/cat/cmdline.c @@ -1,270 +1,270 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2003-2008 Tim Kientzle * All rights reserved. */ /* * Command line parser for bsdcat. */ #include "bsdcat_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "bsdcat.h" -#include "err.h" +#include "lafe_err.h" /* * Short options for bsdcat. Please keep this sorted. */ static const char *short_options = "h"; /* * Long options for bsdcat. Please keep this list sorted. * * The symbolic names for options that lack a short equivalent are * defined in bsdcat.h. Also note that so far I've found no need * to support optional arguments to long options. That would be * a small change to the code below. */ static const struct bsdcat_option { const char *name; int required; /* 1 if this option requires an argument. */ int equivalent; /* Equivalent short option. */ } bsdcat_longopts[] = { { "help", 0, 'h' }, { "version", 0, OPTION_VERSION }, { NULL, 0, 0 } }; /* * This getopt implementation has two key features that common * getopt_long() implementations lack. Apart from those, it's a * straightforward option parser, considerably simplified by not * needing to support the wealth of exotic getopt_long() features. It * has, of course, been shamelessly tailored for bsdcat. (If you're * looking for a generic getopt_long() implementation for your * project, I recommend Gregory Pietsch's public domain getopt_long() * implementation.) The two additional features are: * * Old-style tar arguments: The original tar implementation treated * the first argument word as a list of single-character option * letters. All arguments follow as separate words. For example, * tar xbf 32 /dev/tape * Here, the "xbf" is three option letters, "32" is the argument for * "b" and "/dev/tape" is the argument for "f". We support this usage * if the first command-line argument does not begin with '-'. We * also allow regular short and long options to follow, e.g., * tar xbf 32 /dev/tape -P --format=pax * * -W long options: There's an obscure GNU convention (only rarely * supported even there) that allows "-W option=argument" as an * alternative way to support long options. This was supported in * early bsdtar as a way to access long options on platforms that did * not support getopt_long() and is preserved here for backwards * compatibility. (Of course, if I'd started with a custom * command-line parser from the beginning, I would have had normal * long option support on every platform so that hack wouldn't have * been necessary. Oh, well. Some mistakes you just have to live * with.) * * TODO: We should be able to use this to pull files and intermingled * options (such as -C) from the command line in write mode. That * will require a little rethinking of the argument handling in * bsdcat.c. * * TODO: If we want to support arbitrary command-line options from -T * input (as GNU tar does), we may need to extend this to handle option * words from sources other than argv/argc. I'm not really sure if I * like that feature of GNU tar, so it's certainly not a priority. */ int bsdcat_getopt(struct bsdcat *bsdcat) { enum { state_start = 0, state_old_tar, state_next_word, state_short, state_long }; const struct bsdcat_option *popt, *match, *match2; const char *p, *long_prefix; size_t optlength; int opt; int required; again: match = NULL; match2 = NULL; long_prefix = "--"; opt = '?'; required = 0; bsdcat->argument = NULL; /* First time through, initialize everything. */ if (bsdcat->getopt_state == state_start) { /* Skip program name. */ ++bsdcat->argv; --bsdcat->argc; if (*bsdcat->argv == NULL) return (-1); /* Decide between "new style" and "old style" arguments. */ bsdcat->getopt_state = state_next_word; } /* * We're ready to look at the next word in argv. */ if (bsdcat->getopt_state == state_next_word) { /* No more arguments, so no more options. */ if (bsdcat->argv[0] == NULL) return (-1); /* Doesn't start with '-', so no more options. */ if (bsdcat->argv[0][0] != '-') return (-1); /* "--" marks end of options; consume it and return. */ if (strcmp(bsdcat->argv[0], "--") == 0) { ++bsdcat->argv; --bsdcat->argc; return (-1); } /* Get next word for parsing. */ bsdcat->getopt_word = *bsdcat->argv++; --bsdcat->argc; if (bsdcat->getopt_word[1] == '-') { /* Set up long option parser. */ bsdcat->getopt_state = state_long; bsdcat->getopt_word += 2; /* Skip leading '--' */ } else { /* Set up short option parser. */ bsdcat->getopt_state = state_short; ++bsdcat->getopt_word; /* Skip leading '-' */ } } /* * We're parsing a group of POSIX-style single-character options. */ if (bsdcat->getopt_state == state_short) { /* Peel next option off of a group of short options. */ opt = *bsdcat->getopt_word++; if (opt == '\0') { /* End of this group; recurse to get next option. */ bsdcat->getopt_state = state_next_word; goto again; } /* Does this option take an argument? */ p = strchr(short_options, opt); if (p == NULL) return ('?'); if (p[1] == ':') required = 1; /* If it takes an argument, parse that. */ if (required) { /* If arg is run-in, bsdcat->getopt_word already points to it. */ if (bsdcat->getopt_word[0] == '\0') { /* Otherwise, pick up the next word. */ bsdcat->getopt_word = *bsdcat->argv; if (bsdcat->getopt_word == NULL) { lafe_warnc(0, "Option -%c requires an argument", opt); return ('?'); } ++bsdcat->argv; --bsdcat->argc; } if (opt == 'W') { bsdcat->getopt_state = state_long; long_prefix = "-W "; /* For clearer errors. */ } else { bsdcat->getopt_state = state_next_word; bsdcat->argument = bsdcat->getopt_word; } } } /* We're reading a long option, including -W long=arg convention. */ if (bsdcat->getopt_state == state_long) { /* After this long option, we'll be starting a new word. */ bsdcat->getopt_state = state_next_word; /* Option name ends at '=' if there is one. */ p = strchr(bsdcat->getopt_word, '='); if (p != NULL) { optlength = (size_t)(p - bsdcat->getopt_word); bsdcat->argument = (char *)(uintptr_t)(p + 1); } else { optlength = strlen(bsdcat->getopt_word); } /* Search the table for an unambiguous match. */ for (popt = bsdcat_longopts; popt->name != NULL; popt++) { /* Short-circuit if first chars don't match. */ if (popt->name[0] != bsdcat->getopt_word[0]) continue; /* If option is a prefix of name in table, record it.*/ if (strncmp(bsdcat->getopt_word, popt->name, optlength) == 0) { match2 = match; /* Record up to two matches. */ match = popt; /* If it's an exact match, we're done. */ if (strlen(popt->name) == optlength) { match2 = NULL; /* Forget the others. */ break; } } } /* Fail if there wasn't a unique match. */ if (match == NULL) { lafe_warnc(0, "Option %s%s is not supported", long_prefix, bsdcat->getopt_word); return ('?'); } if (match2 != NULL) { lafe_warnc(0, "Ambiguous option %s%s (matches --%s and --%s)", long_prefix, bsdcat->getopt_word, match->name, match2->name); return ('?'); } /* We've found a unique match; does it need an argument? */ if (match->required) { /* Argument required: get next word if necessary. */ if (bsdcat->argument == NULL) { bsdcat->argument = *bsdcat->argv; if (bsdcat->argument == NULL) { lafe_warnc(0, "Option %s%s requires an argument", long_prefix, match->name); return ('?'); } ++bsdcat->argv; --bsdcat->argc; } } else { /* Argument forbidden: fail if there is one. */ if (bsdcat->argument != NULL) { lafe_warnc(0, "Option %s%s does not allow an argument", long_prefix, match->name); return ('?'); } } return (match->equivalent); } return (opt); } diff --git a/contrib/libarchive/cpio/cmdline.c b/contrib/libarchive/cpio/cmdline.c index c67519947dbc..db06c03c011d 100644 --- a/contrib/libarchive/cpio/cmdline.c +++ b/contrib/libarchive/cpio/cmdline.c @@ -1,391 +1,391 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. */ #include "cpio_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_GRP_H #include #endif #ifdef HAVE_PWD_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "cpio.h" -#include "err.h" +#include "lafe_err.h" /* * Short options for cpio. Please keep this sorted. */ static const char *short_options = "067AaBC:cdE:F:f:H:hI:iJjLlmnO:opR:rtuVvW:yZz"; /* * Long options for cpio. Please keep this sorted. */ static const struct option { const char *name; int required; /* 1 if this option requires an argument */ int equivalent; /* Equivalent short option. */ } cpio_longopts[] = { { "b64encode", 0, OPTION_B64ENCODE }, { "binary", 0, '7' }, { "create", 0, 'o' }, { "dereference", 0, 'L' }, { "dot", 0, 'V' }, { "extract", 0, 'i' }, { "file", 1, 'F' }, { "format", 1, 'H' }, { "grzip", 0, OPTION_GRZIP }, { "help", 0, 'h' }, { "insecure", 0, OPTION_INSECURE }, { "link", 0, 'l' }, { "list", 0, 't' }, { "lrzip", 0, OPTION_LRZIP }, { "lz4", 0, OPTION_LZ4 }, { "lzma", 0, OPTION_LZMA }, { "lzop", 0, OPTION_LZOP }, { "make-directories", 0, 'd' }, { "no-preserve-owner", 0, OPTION_NO_PRESERVE_OWNER }, { "null", 0, '0' }, { "numeric-uid-gid", 0, 'n' }, { "owner", 1, 'R' }, { "passphrase", 1, OPTION_PASSPHRASE }, { "pass-through", 0, 'p' }, { "preserve-modification-time", 0, 'm' }, { "preserve-owner", 0, OPTION_PRESERVE_OWNER }, { "pwb", 0, '6' }, { "quiet", 0, OPTION_QUIET }, { "unconditional", 0, 'u' }, { "uuencode", 0, OPTION_UUENCODE }, { "verbose", 0, 'v' }, { "version", 0, OPTION_VERSION }, { "xz", 0, 'J' }, { "zstd", 0, OPTION_ZSTD }, { NULL, 0, 0 } }; /* * I used to try to select platform-provided getopt() or * getopt_long(), but that caused a lot of headaches. In particular, * I couldn't consistently use long options in the test harness * because not all platforms have getopt_long(). That in turn led to * overuse of the -W hack in the test harness, which made it rough to * run the test harness against GNU cpio. (I periodically run the * test harness here against GNU cpio as a sanity-check. Yes, * I've found a couple of bugs in GNU cpio that way.) */ int cpio_getopt(struct cpio *cpio) { enum { state_start = 0, state_next_word, state_short, state_long }; static int state = state_start; static char *opt_word; const struct option *popt, *match, *match2; const char *p, *long_prefix; size_t optlength; int opt; int required; again: match = NULL; match2 = NULL; long_prefix = "--"; opt = '?'; required = 0; cpio->argument = NULL; /* First time through, initialize everything. */ if (state == state_start) { /* Skip program name. */ ++cpio->argv; --cpio->argc; state = state_next_word; } /* * We're ready to look at the next word in argv. */ if (state == state_next_word) { /* No more arguments, so no more options. */ if (cpio->argv[0] == NULL) return (-1); /* Doesn't start with '-', so no more options. */ if (cpio->argv[0][0] != '-') return (-1); /* "--" marks end of options; consume it and return. */ if (strcmp(cpio->argv[0], "--") == 0) { ++cpio->argv; --cpio->argc; return (-1); } /* Get next word for parsing. */ opt_word = *cpio->argv++; --cpio->argc; if (opt_word[1] == '-') { /* Set up long option parser. */ state = state_long; opt_word += 2; /* Skip leading '--' */ } else { /* Set up short option parser. */ state = state_short; ++opt_word; /* Skip leading '-' */ } } /* * We're parsing a group of POSIX-style single-character options. */ if (state == state_short) { /* Peel next option off of a group of short options. */ opt = *opt_word++; if (opt == '\0') { /* End of this group; recurse to get next option. */ state = state_next_word; goto again; } /* Does this option take an argument? */ p = strchr(short_options, opt); if (p == NULL) return ('?'); if (p[1] == ':') required = 1; /* If it takes an argument, parse that. */ if (required) { /* If arg is run-in, opt_word already points to it. */ if (opt_word[0] == '\0') { /* Otherwise, pick up the next word. */ opt_word = *cpio->argv; if (opt_word == NULL) { lafe_warnc(0, "Option -%c requires an argument", opt); return ('?'); } ++cpio->argv; --cpio->argc; } if (opt == 'W') { state = state_long; long_prefix = "-W "; /* For clearer errors. */ } else { state = state_next_word; cpio->argument = opt_word; } } } /* We're reading a long option, including -W long=arg convention. */ if (state == state_long) { /* After this long option, we'll be starting a new word. */ state = state_next_word; /* Option name ends at '=' if there is one. */ p = strchr(opt_word, '='); if (p != NULL) { optlength = (size_t)(p - opt_word); cpio->argument = (char *)(uintptr_t)(p + 1); } else { optlength = strlen(opt_word); } /* Search the table for an unambiguous match. */ for (popt = cpio_longopts; popt->name != NULL; popt++) { /* Short-circuit if first chars don't match. */ if (popt->name[0] != opt_word[0]) continue; /* If option is a prefix of name in table, record it.*/ if (strncmp(opt_word, popt->name, optlength) == 0) { match2 = match; /* Record up to two matches. */ match = popt; /* If it's an exact match, we're done. */ if (strlen(popt->name) == optlength) { match2 = NULL; /* Forget the others. */ break; } } } /* Fail if there wasn't a unique match. */ if (match == NULL) { lafe_warnc(0, "Option %s%s is not supported", long_prefix, opt_word); return ('?'); } if (match2 != NULL) { lafe_warnc(0, "Ambiguous option %s%s (matches --%s and --%s)", long_prefix, opt_word, match->name, match2->name); return ('?'); } /* We've found a unique match; does it need an argument? */ if (match->required) { /* Argument required: get next word if necessary. */ if (cpio->argument == NULL) { cpio->argument = *cpio->argv; if (cpio->argument == NULL) { lafe_warnc(0, "Option %s%s requires an argument", long_prefix, match->name); return ('?'); } ++cpio->argv; --cpio->argc; } } else { /* Argument forbidden: fail if there is one. */ if (cpio->argument != NULL) { lafe_warnc(0, "Option %s%s does not allow an argument", long_prefix, match->name); return ('?'); } } return (match->equivalent); } return (opt); } /* * Parse the argument to the -R or --owner flag. * * The format is one of the following: * - Override user but not group * : - Override both, group is user's default group * : - Override user but not group * : - Override both * : - Override group but not user * * Where uid/gid are decimal representations and groupname/username * are names to be looked up in system database. Note that we try * to look up an argument as a name first, then try numeric parsing. * * A period can be used instead of the colon. * * Sets uid/gid return as appropriate, -1 indicates uid/gid not specified. * TODO: If the spec uses uname/gname, then return those to the caller * as well. If the spec provides uid/gid, just return names as NULL. * * Returns NULL if no error, otherwise returns error string for display. * */ int owner_parse(const char *spec, struct cpio_owner *owner, const char **errmsg) { static char errbuff[128]; const char *u, *ue, *g; owner->uid = -1; owner->gid = -1; owner->uname = NULL; owner->gname = NULL; if (spec[0] == '\0') { *errmsg = "Invalid empty user/group spec"; return (-1); } /* * Split spec into [user][:.][group] * u -> first char of username, NULL if no username * ue -> first char after username (colon, period, or \0) * g -> first char of group name */ if (*spec == ':' || *spec == '.') { /* If spec starts with ':' or '.', then just group. */ ue = u = NULL; g = spec + 1; } else { /* Otherwise, [user] or [user][:] or [user][:][group] */ ue = u = spec; while (*ue != ':' && *ue != '.' && *ue != '\0') ++ue; g = ue; if (*g != '\0') /* Skip : or . to find first char of group. */ ++g; } if (u != NULL) { /* Look up user: ue is first char after end of user. */ char *user; struct passwd *pwent; user = malloc(ue - u + 1); if (user == NULL) goto alloc_error; memcpy(user, u, ue - u); user[ue - u] = '\0'; if ((pwent = getpwnam(user)) != NULL) { owner->uid = pwent->pw_uid; owner->uname = strdup(pwent->pw_name); if (owner->uname == NULL) { free(user); goto alloc_error; } if (*ue != '\0') owner->gid = pwent->pw_gid; } else { char *end; errno = 0; owner->uid = (int)strtoul(user, &end, 10); if (errno || *end != '\0') { snprintf(errbuff, sizeof(errbuff), "Couldn't lookup user ``%s''", user); errbuff[sizeof(errbuff) - 1] = '\0'; free(user); *errmsg = errbuff; return (-1); } } free(user); } if (*g != '\0') { struct group *grp; if ((grp = getgrnam(g)) != NULL) { owner->gid = grp->gr_gid; owner->gname = strdup(grp->gr_name); if (owner->gname == NULL) { free(owner->uname); owner->uname = NULL; goto alloc_error; } } else { char *end; errno = 0; owner->gid = (int)strtoul(g, &end, 10); if (errno || *end != '\0') { snprintf(errbuff, sizeof(errbuff), "Couldn't lookup group ``%s''", g); errbuff[sizeof(errbuff) - 1] = '\0'; *errmsg = errbuff; return (-1); } } } return (0); alloc_error: *errmsg = "Couldn't allocate memory"; return (-1); } diff --git a/contrib/libarchive/cpio/cpio.c b/contrib/libarchive/cpio/cpio.c index 2bf1bfa2985a..262db510568b 100644 --- a/contrib/libarchive/cpio/cpio.c +++ b/contrib/libarchive/cpio/cpio.c @@ -1,1504 +1,1512 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. */ #include "cpio_platform.h" #include #include #include #ifdef HAVE_SYS_MKDEV_H #include #endif #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_SYS_TIME_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_GRP_H #include #endif #ifdef HAVE_LOCALE_H #include #endif #ifdef HAVE_PWD_H #include #endif #ifdef HAVE_SIGNAL_H #include #endif #ifdef HAVE_STDARG_H #include #endif #ifdef HAVE_STDINT_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_TIME_H #include #endif #include "cpio.h" -#include "err.h" +#include "lafe_err.h" #include "line_reader.h" #include "passphrase.h" /* Fixed size of uname/gname caches. */ #define name_cache_size 101 #ifndef O_BINARY #define O_BINARY 0 #endif struct name_cache { int probes; int hits; size_t size; struct { id_t id; char *name; } cache[name_cache_size]; }; static int extract_data(struct archive *, struct archive *); const char * cpio_i64toa(int64_t); static const char *cpio_rename(const char *name); static int entry_to_archive(struct cpio *, struct archive_entry *); static int file_to_archive(struct cpio *, const char *); static void free_cache(struct name_cache *cache); static void list_item_verbose(struct cpio *, struct archive_entry *); static __LA_NORETURN void long_help(void); static const char *lookup_gname(struct cpio *, gid_t gid); static int lookup_gname_helper(struct cpio *, const char **name, id_t gid); static const char *lookup_uname(struct cpio *, uid_t uid); static int lookup_uname_helper(struct cpio *, const char **name, id_t uid); static __LA_NORETURN void mode_in(struct cpio *); static __LA_NORETURN void mode_list(struct cpio *); static void mode_out(struct cpio *); static void mode_pass(struct cpio *, const char *); static const char *remove_leading_slash(const char *); static int restore_time(struct cpio *, struct archive_entry *, const char *, int fd); static __LA_NORETURN void usage(void); static __LA_NORETURN void version(void); static const char * passphrase_callback(struct archive *, void *); static void passphrase_free(char *); int main(int argc, char *argv[]) { static char buff[16384]; struct cpio _cpio; /* Allocated on stack. */ struct cpio *cpio; struct cpio_owner owner; const char *errmsg; char *tptr; int opt, t; cpio = &_cpio; memset(cpio, 0, sizeof(*cpio)); cpio->buff = buff; cpio->buff_size = sizeof(buff); -#if defined(HAVE_SIGACTION) && defined(SIGPIPE) - { /* Ignore SIGPIPE signals. */ +#if defined(HAVE_SIGACTION) + { struct sigaction sa; sigemptyset(&sa.sa_mask); sa.sa_flags = 0; +#ifdef SIGPIPE + /* Ignore SIGPIPE signals. */ sa.sa_handler = SIG_IGN; sigaction(SIGPIPE, &sa, NULL); +#endif +#ifdef SIGCHLD + /* Do not ignore SIGCHLD. */ + sa.sa_handler = SIG_DFL; + sigaction(SIGCHLD, &sa, NULL); +#endif } #endif /* Set lafe_progname before calling lafe_warnc. */ lafe_setprogname(*argv, "bsdcpio"); #if HAVE_SETLOCALE if (setlocale(LC_ALL, "") == NULL) lafe_warnc(0, "Failed to set default locale"); #endif cpio->uid_override = -1; cpio->uname_override = NULL; cpio->gid_override = -1; cpio->gname_override = NULL; cpio->argv = argv; cpio->argc = argc; cpio->mode = '\0'; cpio->verbose = 0; cpio->compress = '\0'; cpio->extract_flags = ARCHIVE_EXTRACT_NO_AUTODIR; cpio->extract_flags |= ARCHIVE_EXTRACT_NO_OVERWRITE_NEWER; cpio->extract_flags |= ARCHIVE_EXTRACT_SECURE_SYMLINKS; cpio->extract_flags |= ARCHIVE_EXTRACT_SECURE_NODOTDOT; cpio->extract_flags |= ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS; cpio->extract_flags |= ARCHIVE_EXTRACT_PERM; cpio->extract_flags |= ARCHIVE_EXTRACT_FFLAGS; cpio->extract_flags |= ARCHIVE_EXTRACT_ACL; #if !defined(_WIN32) && !defined(__CYGWIN__) if (geteuid() == 0) cpio->extract_flags |= ARCHIVE_EXTRACT_OWNER; #endif cpio->bytes_per_block = 512; cpio->filename = NULL; cpio->matching = archive_match_new(); if (cpio->matching == NULL) lafe_errc(1, 0, "Out of memory"); while ((opt = cpio_getopt(cpio)) != -1) { switch (opt) { case '0': /* GNU convention: --null, -0 */ cpio->option_null = 1; break; case '6': /* in/out: assume/create 6th edition (PWB) format */ cpio->option_pwb = 1; break; case '7': /* out: create archive using 7th Edition binary format */ cpio->format = "bin"; break; case 'A': /* NetBSD/OpenBSD */ cpio->option_append = 1; break; case 'a': /* POSIX 1997 */ cpio->option_atime_restore = 1; break; case 'B': /* POSIX 1997 */ cpio->bytes_per_block = 5120; break; case OPTION_B64ENCODE: cpio->add_filter = opt; break; case 'C': /* NetBSD/OpenBSD */ errno = 0; tptr = NULL; t = (int)strtol(cpio->argument, &tptr, 10); if (errno || t <= 0 || *(cpio->argument) == '\0' || tptr == NULL || *tptr != '\0') { lafe_errc(1, 0, "Invalid blocksize: %s", cpio->argument); } cpio->bytes_per_block = t; break; case 'c': /* POSIX 1997 */ cpio->format = "odc"; break; case 'd': /* POSIX 1997 */ cpio->extract_flags &= ~ARCHIVE_EXTRACT_NO_AUTODIR; break; case 'E': /* NetBSD/OpenBSD */ if (archive_match_include_pattern_from_file( cpio->matching, cpio->argument, cpio->option_null) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(cpio->matching)); break; case 'F': /* NetBSD/OpenBSD/GNU cpio */ cpio->filename = cpio->argument; break; case 'f': /* POSIX 1997 */ if (archive_match_exclude_pattern(cpio->matching, cpio->argument) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(cpio->matching)); break; case OPTION_GRZIP: cpio->compress = opt; break; case 'H': /* GNU cpio (also --format) */ cpio->format = cpio->argument; break; case 'h': long_help(); /* NOTREACHED */ case 'I': /* NetBSD/OpenBSD */ cpio->filename = cpio->argument; break; case 'i': /* POSIX 1997 */ if (cpio->mode != '\0') lafe_errc(1, 0, "Cannot use both -i and -%c", cpio->mode); cpio->mode = opt; break; case 'J': /* GNU tar, others */ cpio->compress = opt; break; case 'j': /* GNU tar, others */ cpio->compress = opt; break; case OPTION_INSECURE: cpio->extract_flags &= ~ARCHIVE_EXTRACT_SECURE_SYMLINKS; cpio->extract_flags &= ~ARCHIVE_EXTRACT_SECURE_NODOTDOT; cpio->extract_flags &= ~ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS; break; case 'L': /* GNU cpio */ cpio->option_follow_links = 1; break; case 'l': /* POSIX 1997 */ cpio->option_link = 1; break; case OPTION_LRZIP: case OPTION_LZ4: case OPTION_LZMA: /* GNU tar, others */ case OPTION_LZOP: /* GNU tar, others */ case OPTION_ZSTD: cpio->compress = opt; break; case 'm': /* POSIX 1997 */ cpio->extract_flags |= ARCHIVE_EXTRACT_TIME; break; case 'n': /* GNU cpio */ cpio->option_numeric_uid_gid = 1; break; case OPTION_NO_PRESERVE_OWNER: /* GNU cpio */ cpio->extract_flags &= ~ARCHIVE_EXTRACT_OWNER; break; case 'O': /* GNU cpio */ cpio->filename = cpio->argument; break; case 'o': /* POSIX 1997 */ if (cpio->mode != '\0') lafe_errc(1, 0, "Cannot use both -o and -%c", cpio->mode); cpio->mode = opt; break; case 'p': /* POSIX 1997 */ if (cpio->mode != '\0') lafe_errc(1, 0, "Cannot use both -p and -%c", cpio->mode); cpio->mode = opt; cpio->extract_flags &= ~ARCHIVE_EXTRACT_SECURE_NODOTDOT; cpio->extract_flags &= ~ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS; break; case OPTION_PASSPHRASE: cpio->passphrase = cpio->argument; break; case OPTION_PRESERVE_OWNER: cpio->extract_flags |= ARCHIVE_EXTRACT_OWNER; break; case OPTION_QUIET: /* GNU cpio */ cpio->quiet = 1; break; case 'R': /* GNU cpio, also --owner */ errmsg = NULL; if (owner_parse(cpio->argument, &owner, &errmsg) != 0) { if (!errmsg) errmsg = "Error parsing owner"; lafe_warnc(-1, "%s", errmsg); usage(); } if (owner.uid != -1) cpio->uid_override = owner.uid; if (owner.uname != NULL) cpio->uname_override = owner.uname; if (owner.gid != -1) cpio->gid_override = owner.gid; if (owner.gname != NULL) cpio->gname_override = owner.gname; break; case 'r': /* POSIX 1997 */ cpio->option_rename = 1; break; case 't': /* POSIX 1997 */ cpio->option_list = 1; break; case 'u': /* POSIX 1997 */ cpio->extract_flags &= ~ARCHIVE_EXTRACT_NO_OVERWRITE_NEWER; break; case OPTION_UUENCODE: cpio->add_filter = opt; break; case 'v': /* POSIX 1997 */ cpio->verbose++; break; case 'V': /* GNU cpio */ cpio->dot++; break; case OPTION_VERSION: /* GNU convention */ version(); /* NOTREACHED */ #if 0 /* * cpio_getopt() handles -W specially, so it's not * available here. */ case 'W': /* Obscure, but useful GNU convention. */ break; #endif case 'y': /* tar convention */ cpio->compress = opt; break; case 'Z': /* tar convention */ cpio->compress = opt; break; case 'z': /* tar convention */ cpio->compress = opt; break; default: usage(); } } /* * Sanity-check args, error out on nonsensical combinations. */ /* -t implies -i if no mode was specified. */ if (cpio->option_list && cpio->mode == '\0') cpio->mode = 'i'; /* -t requires -i */ if (cpio->option_list && cpio->mode != 'i') lafe_errc(1, 0, "Option -t requires -i"); /* -n requires -it */ if (cpio->option_numeric_uid_gid && !cpio->option_list) lafe_errc(1, 0, "Option -n requires -it"); /* Can only specify format when writing */ if (cpio->format != NULL && cpio->mode != 'o') lafe_errc(1, 0, "Option --format requires -o"); /* -l requires -p */ if (cpio->option_link && cpio->mode != 'p') lafe_errc(1, 0, "Option -l requires -p"); /* -v overrides -V */ if (cpio->dot && cpio->verbose) cpio->dot = 0; /* TODO: Flag other nonsensical combinations. */ switch (cpio->mode) { case 'o': if (cpio->format == NULL) { if (cpio->option_pwb) cpio->format = "pwb"; else cpio->format = "cpio"; } mode_out(cpio); break; case 'i': while (*cpio->argv != NULL) { if (archive_match_include_pattern(cpio->matching, *cpio->argv) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(cpio->matching)); --cpio->argc; ++cpio->argv; } if (cpio->option_list) mode_list(cpio); else mode_in(cpio); /* NOTREACHED */ case 'p': if (*cpio->argv == NULL || **cpio->argv == '\0') lafe_errc(1, 0, "-p mode requires a target directory"); mode_pass(cpio, *cpio->argv); break; default: lafe_errc(1, 0, "Must specify at least one of -i, -o, or -p"); } archive_match_free(cpio->matching); free_cache(cpio->uname_cache); free(cpio->uname_override); free_cache(cpio->gname_cache); free(cpio->gname_override); archive_read_close(cpio->archive_read_disk); archive_read_free(cpio->archive_read_disk); free(cpio->destdir); passphrase_free(cpio->ppbuff); return (cpio->return_value); } static void usage(void) { const char *p; p = lafe_getprogname(); fprintf(stderr, "Brief Usage:\n"); fprintf(stderr, " List: %s -it < archive\n", p); fprintf(stderr, " Extract: %s -i < archive\n", p); fprintf(stderr, " Create: %s -o < filenames > archive\n", p); fprintf(stderr, " Help: %s --help\n", p); exit(1); } static const char *long_help_msg = "First option must be a mode specifier:\n" " -i Input -o Output -p Pass\n" "Common Options:\n" " -v Verbose filenames -V one dot per file\n" "Create: %p -o [options] < [list of files] > [archive]\n" " -J,-y,-z,--lzma Compress archive with xz/bzip2/gzip/lzma\n" " --format {pwb|bin|odc|newc|ustar} Select archive format\n" "List: %p -it < [archive]\n" "Extract: %p -i [options] < [archive]\n"; /* * Note that the word 'bsdcpio' will always appear in the first line * of output. * * In particular, /bin/sh scripts that need to test for the presence * of bsdcpio can use the following template: * * if (cpio --help 2>&1 | grep bsdcpio >/dev/null 2>&1 ) then \ * echo bsdcpio; else echo not bsdcpio; fi */ static void long_help(void) { const char *prog; const char *p; prog = lafe_getprogname(); fflush(stderr); p = (strcmp(prog,"bsdcpio") != 0) ? "(bsdcpio)" : ""; printf("%s%s: manipulate archive files\n", prog, p); for (p = long_help_msg; *p != '\0'; p++) { if (*p == '%') { if (p[1] == 'p') { fputs(prog, stdout); p++; } else putchar('%'); } else putchar(*p); } version(); } static void version(void) { fprintf(stdout,"bsdcpio %s - %s \n", BSDCPIO_VERSION_STRING, archive_version_details()); exit(0); } static void mode_out(struct cpio *cpio) { struct archive_entry *entry, *spare; struct lafe_line_reader *lr; const char *p; int r; if (cpio->option_append) lafe_errc(1, 0, "Append mode not yet supported."); cpio->archive_read_disk = archive_read_disk_new(); if (cpio->archive_read_disk == NULL) lafe_errc(1, 0, "Failed to allocate archive object"); if (cpio->option_follow_links) archive_read_disk_set_symlink_logical(cpio->archive_read_disk); else archive_read_disk_set_symlink_physical(cpio->archive_read_disk); archive_read_disk_set_standard_lookup(cpio->archive_read_disk); cpio->archive = archive_write_new(); if (cpio->archive == NULL) lafe_errc(1, 0, "Failed to allocate archive object"); switch (cpio->compress) { case OPTION_GRZIP: r = archive_write_add_filter_grzip(cpio->archive); break; case 'J': r = archive_write_add_filter_xz(cpio->archive); break; case OPTION_LRZIP: r = archive_write_add_filter_lrzip(cpio->archive); break; case OPTION_LZ4: r = archive_write_add_filter_lz4(cpio->archive); break; case OPTION_LZMA: r = archive_write_add_filter_lzma(cpio->archive); break; case OPTION_LZOP: r = archive_write_add_filter_lzop(cpio->archive); break; case OPTION_ZSTD: r = archive_write_add_filter_zstd(cpio->archive); break; case 'j': case 'y': r = archive_write_add_filter_bzip2(cpio->archive); break; case 'z': r = archive_write_add_filter_gzip(cpio->archive); break; case 'Z': r = archive_write_add_filter_compress(cpio->archive); break; default: r = archive_write_add_filter_none(cpio->archive); break; } if (r < ARCHIVE_WARN) lafe_errc(1, 0, "Requested compression not available"); switch (cpio->add_filter) { case 0: r = ARCHIVE_OK; break; case OPTION_B64ENCODE: r = archive_write_add_filter_b64encode(cpio->archive); break; case OPTION_UUENCODE: r = archive_write_add_filter_uuencode(cpio->archive); break; } if (r < ARCHIVE_WARN) lafe_errc(1, 0, "Requested filter not available"); r = archive_write_set_format_by_name(cpio->archive, cpio->format); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive)); archive_write_set_bytes_per_block(cpio->archive, cpio->bytes_per_block); cpio->linkresolver = archive_entry_linkresolver_new(); archive_entry_linkresolver_set_strategy(cpio->linkresolver, archive_format(cpio->archive)); if (cpio->passphrase != NULL) r = archive_write_set_passphrase(cpio->archive, cpio->passphrase); else r = archive_write_set_passphrase_callback(cpio->archive, cpio, &passphrase_callback); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive)); /* * The main loop: Copy each file into the output archive. */ r = archive_write_open_filename(cpio->archive, cpio->filename); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive)); lr = lafe_line_reader("-", cpio->option_null); while ((p = lafe_line_reader_next(lr)) != NULL) file_to_archive(cpio, p); lafe_line_reader_free(lr); /* * The hardlink detection may have queued up a couple of entries * that can now be flushed. */ entry = NULL; archive_entry_linkify(cpio->linkresolver, &entry, &spare); while (entry != NULL) { entry_to_archive(cpio, entry); archive_entry_free(entry); entry = NULL; archive_entry_linkify(cpio->linkresolver, &entry, &spare); } r = archive_write_close(cpio->archive); if (cpio->dot) fprintf(stderr, "\n"); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive)); if (!cpio->quiet) { int64_t blocks = (archive_filter_bytes(cpio->archive, 0) + 511) / 512; fprintf(stderr, "%lu %s\n", (unsigned long)blocks, blocks == 1 ? "block" : "blocks"); } archive_write_free(cpio->archive); archive_entry_linkresolver_free(cpio->linkresolver); } static const char * remove_leading_slash(const char *p) { const char *rp; /* Remove leading "//./" or "//?/" or "//?/UNC/" * (absolute path prefixes used by Windows API) */ if ((p[0] == '/' || p[0] == '\\') && (p[1] == '/' || p[1] == '\\') && (p[2] == '.' || p[2] == '?') && (p[3] == '/' || p[3] == '\\')) { if (p[2] == '?' && (p[4] == 'U' || p[4] == 'u') && (p[5] == 'N' || p[5] == 'n') && (p[6] == 'C' || p[6] == 'c') && (p[7] == '/' || p[7] == '\\')) p += 8; else p += 4; } do { rp = p; /* Remove leading drive letter from archives created * on Windows. */ if (((p[0] >= 'a' && p[0] <= 'z') || (p[0] >= 'A' && p[0] <= 'Z')) && p[1] == ':') { p += 2; } /* Remove leading "/../", "//", etc. */ while (p[0] == '/' || p[0] == '\\') { if (p[1] == '.' && p[2] == '.' && (p[3] == '/' || p[3] == '\\')) { p += 3; /* Remove "/..", leave "/" * for next pass. */ } else p += 1; /* Remove "/". */ } } while (rp != p); return (p); } /* * This is used by both out mode (to copy objects from disk into * an archive) and pass mode (to copy objects from disk to * an archive_write_disk "archive"). */ static int file_to_archive(struct cpio *cpio, const char *srcpath) { const char *destpath; struct archive_entry *entry, *spare; size_t len; int r; /* * Create an archive_entry describing the source file. * */ entry = archive_entry_new(); if (entry == NULL) lafe_errc(1, 0, "Couldn't allocate entry"); archive_entry_copy_sourcepath(entry, srcpath); r = archive_read_disk_entry_from_file(cpio->archive_read_disk, entry, -1, NULL); if (r < ARCHIVE_FAILED) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive_read_disk)); if (r < ARCHIVE_OK) lafe_warnc(0, "%s", archive_error_string(cpio->archive_read_disk)); if (r <= ARCHIVE_FAILED) { archive_entry_free(entry); cpio->return_value = 1; return (r); } if (cpio->uid_override >= 0) archive_entry_set_uid(entry, cpio->uid_override); if (cpio->gname_override != NULL) archive_entry_set_uname(entry, cpio->uname_override); if (cpio->gid_override >= 0) archive_entry_set_gid(entry, cpio->gid_override); if (cpio->gname_override != NULL) archive_entry_set_gname(entry, cpio->gname_override); /* * Generate a destination path for this entry. * "destination path" is the name to which it will be copied in * pass mode or the name that will go into the archive in * output mode. */ destpath = srcpath; if (cpio->destdir) { len = cpio->destdir_len + strlen(srcpath) + 8; if (len >= cpio->pass_destpath_alloc) { while (len >= cpio->pass_destpath_alloc) { cpio->pass_destpath_alloc += 512; cpio->pass_destpath_alloc *= 2; } free(cpio->pass_destpath); cpio->pass_destpath = malloc(cpio->pass_destpath_alloc); if (cpio->pass_destpath == NULL) lafe_errc(1, ENOMEM, "Can't allocate path buffer"); } strcpy(cpio->pass_destpath, cpio->destdir); strcat(cpio->pass_destpath, remove_leading_slash(srcpath)); destpath = cpio->pass_destpath; } if (cpio->option_rename) destpath = cpio_rename(destpath); if (destpath == NULL) { archive_entry_free(entry); return (0); } archive_entry_copy_pathname(entry, destpath); /* * If we're trying to preserve hardlinks, match them here. */ spare = NULL; if (cpio->linkresolver != NULL && archive_entry_filetype(entry) != AE_IFDIR) { archive_entry_linkify(cpio->linkresolver, &entry, &spare); } if (entry != NULL) { r = entry_to_archive(cpio, entry); archive_entry_free(entry); if (spare != NULL) { if (r == 0) r = entry_to_archive(cpio, spare); archive_entry_free(spare); } } return (r); } static int entry_to_archive(struct cpio *cpio, struct archive_entry *entry) { const char *destpath = archive_entry_pathname(entry); const char *srcpath = archive_entry_sourcepath(entry); int fd = -1; ssize_t bytes_read; int r; /* Print out the destination name to the user. */ if (cpio->verbose) fprintf(stderr,"%s", destpath); if (cpio->dot) fprintf(stderr, "."); /* * Option_link only makes sense in pass mode and for * regular files. Also note: if a link operation fails * because of cross-device restrictions, we'll fall back * to copy mode for that entry. * * TODO: Test other cpio implementations to see if they * hard-link anything other than regular files here. */ if (cpio->option_link && archive_entry_filetype(entry) == AE_IFREG) { struct archive_entry *t; /* Save the original entry in case we need it later. */ t = archive_entry_clone(entry); if (t == NULL) lafe_errc(1, ENOMEM, "Can't create link"); /* Note: link(2) doesn't create parent directories, * so we use archive_write_header() instead as a * convenience. */ archive_entry_set_hardlink(t, srcpath); /* This is a straight link that carries no data. */ archive_entry_set_size(t, 0); r = archive_write_header(cpio->archive, t); archive_entry_free(t); if (r != ARCHIVE_OK) lafe_warnc(archive_errno(cpio->archive), "%s", archive_error_string(cpio->archive)); if (r == ARCHIVE_FATAL) exit(1); #ifdef EXDEV if (r != ARCHIVE_OK && archive_errno(cpio->archive) == EXDEV) { /* Cross-device link: Just fall through and use * the original entry to copy the file over. */ lafe_warnc(0, "Copying file instead"); } else #endif return (0); } /* * Make sure we can open the file (if necessary) before * trying to write the header. */ if (archive_entry_filetype(entry) == AE_IFREG) { if (archive_entry_size(entry) > 0) { fd = open(srcpath, O_RDONLY | O_BINARY); if (fd < 0) { lafe_warnc(errno, "%s: could not open file", srcpath); goto cleanup; } } } else { archive_entry_set_size(entry, 0); } r = archive_write_header(cpio->archive, entry); if (r != ARCHIVE_OK) lafe_warnc(archive_errno(cpio->archive), "%s: %s", srcpath, archive_error_string(cpio->archive)); if (r == ARCHIVE_FATAL) exit(1); if (r >= ARCHIVE_WARN && archive_entry_size(entry) > 0 && fd >= 0) { bytes_read = read(fd, cpio->buff, (unsigned)cpio->buff_size); while (bytes_read > 0) { ssize_t bytes_write; bytes_write = archive_write_data(cpio->archive, cpio->buff, bytes_read); if (bytes_write < 0) lafe_errc(1, archive_errno(cpio->archive), "%s", archive_error_string(cpio->archive)); if (bytes_write < bytes_read) { lafe_warnc(0, "Truncated write; file may have " "grown while being archived."); } bytes_read = read(fd, cpio->buff, (unsigned)cpio->buff_size); } } fd = restore_time(cpio, entry, srcpath, fd); cleanup: if (cpio->verbose) fprintf(stderr,"\n"); if (fd >= 0) close(fd); return (0); } static int restore_time(struct cpio *cpio, struct archive_entry *entry, const char *name, int fd) { #ifndef HAVE_UTIMES static int warned = 0; (void)cpio; /* UNUSED */ (void)entry; /* UNUSED */ (void)name; /* UNUSED */ if (!warned) lafe_warnc(0, "Can't restore access times on this platform"); warned = 1; return (fd); #else #if defined(_WIN32) && !defined(__CYGWIN__) struct __timeval times[2]; #else struct timeval times[2]; #endif if (!cpio->option_atime_restore) return (fd); times[1].tv_sec = archive_entry_mtime(entry); times[1].tv_usec = archive_entry_mtime_nsec(entry) / 1000; times[0].tv_sec = archive_entry_atime(entry); times[0].tv_usec = archive_entry_atime_nsec(entry) / 1000; #if defined(HAVE_FUTIMES) && !defined(__CYGWIN__) if (fd >= 0 && futimes(fd, times) == 0) return (fd); #endif /* * Some platform cannot restore access times if the file descriptor * is still opened. */ if (fd >= 0) { close(fd); fd = -1; } #ifdef HAVE_LUTIMES if (lutimes(name, times) != 0) #else if ((AE_IFLNK != archive_entry_filetype(entry)) && utimes(name, times) != 0) #endif lafe_warnc(errno, "Can't update time for %s", name); #endif return (fd); } static void mode_in(struct cpio *cpio) { struct archive *a; struct archive_entry *entry; struct archive *ext; const char *destpath; int r; ext = archive_write_disk_new(); if (ext == NULL) lafe_errc(1, 0, "Couldn't allocate restore object"); r = archive_write_disk_set_options(ext, cpio->extract_flags); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(ext)); a = archive_read_new(); if (a == NULL) lafe_errc(1, 0, "Couldn't allocate archive object"); archive_read_support_filter_all(a); archive_read_support_format_all(a); if (cpio->option_pwb) archive_read_set_options(a, "pwb"); if (cpio->passphrase != NULL) r = archive_read_add_passphrase(a, cpio->passphrase); else r = archive_read_set_passphrase_callback(a, cpio, &passphrase_callback); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); if (archive_read_open_filename(a, cpio->filename, cpio->bytes_per_block)) lafe_errc(1, archive_errno(a), "%s", archive_error_string(a)); for (;;) { r = archive_read_next_header(a, &entry); if (r == ARCHIVE_EOF) break; if (r != ARCHIVE_OK) { lafe_errc(1, archive_errno(a), "%s", archive_error_string(a)); } if (archive_match_path_excluded(cpio->matching, entry)) continue; if (cpio->option_rename) { destpath = cpio_rename(archive_entry_pathname(entry)); archive_entry_set_pathname(entry, destpath); } else destpath = archive_entry_pathname(entry); if (destpath == NULL) continue; if (cpio->verbose) fprintf(stderr, "%s\n", destpath); if (cpio->dot) fprintf(stderr, "."); if (cpio->uid_override >= 0) archive_entry_set_uid(entry, cpio->uid_override); if (cpio->uname_override != NULL) archive_entry_set_uname(entry, cpio->uname_override); if (cpio->gid_override >= 0) archive_entry_set_gid(entry, cpio->gid_override); if (cpio->gname_override != NULL) archive_entry_set_gname(entry, cpio->gname_override); r = archive_write_header(ext, entry); if (r != ARCHIVE_OK) { fprintf(stderr, "%s: %s\n", archive_entry_pathname(entry), archive_error_string(ext)); cpio->return_value = 1; } else if (!archive_entry_size_is_set(entry) || archive_entry_size(entry) > 0) { r = extract_data(a, ext); if (r != ARCHIVE_OK) cpio->return_value = 1; } } r = archive_read_close(a); if (cpio->dot) fprintf(stderr, "\n"); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); r = archive_write_close(ext); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(ext)); if (!cpio->quiet) { int64_t blocks = (archive_filter_bytes(a, 0) + 511) / 512; fprintf(stderr, "%lu %s\n", (unsigned long)blocks, blocks == 1 ? "block" : "blocks"); } archive_read_free(a); archive_write_free(ext); exit(cpio->return_value); } /* * Exits if there's a fatal error. Returns ARCHIVE_OK * if everything is kosher. */ static int extract_data(struct archive *ar, struct archive *aw) { int r; size_t size; const void *block; int64_t offset; for (;;) { r = archive_read_data_block(ar, &block, &size, &offset); if (r == ARCHIVE_EOF) return (ARCHIVE_OK); if (r != ARCHIVE_OK) { lafe_warnc(archive_errno(ar), "%s", archive_error_string(ar)); exit(1); } r = (int)archive_write_data_block(aw, block, size, offset); if (r != ARCHIVE_OK) { lafe_warnc(archive_errno(aw), "%s", archive_error_string(aw)); return (r); } } } static void mode_list(struct cpio *cpio) { struct archive *a; struct archive_entry *entry; int r; a = archive_read_new(); if (a == NULL) lafe_errc(1, 0, "Couldn't allocate archive object"); archive_read_support_filter_all(a); archive_read_support_format_all(a); if (cpio->option_pwb) archive_read_set_options(a, "pwb"); if (cpio->passphrase != NULL) r = archive_read_add_passphrase(a, cpio->passphrase); else r = archive_read_set_passphrase_callback(a, cpio, &passphrase_callback); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); if (archive_read_open_filename(a, cpio->filename, cpio->bytes_per_block)) lafe_errc(1, archive_errno(a), "%s", archive_error_string(a)); for (;;) { r = archive_read_next_header(a, &entry); if (r == ARCHIVE_EOF) break; if (r != ARCHIVE_OK) { lafe_errc(1, archive_errno(a), "%s", archive_error_string(a)); } if (archive_match_path_excluded(cpio->matching, entry)) continue; if (cpio->verbose) list_item_verbose(cpio, entry); else fprintf(stdout, "%s\n", archive_entry_pathname(entry)); } r = archive_read_close(a); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); if (!cpio->quiet) { int64_t blocks = (archive_filter_bytes(a, 0) + 511) / 512; fprintf(stderr, "%lu %s\n", (unsigned long)blocks, blocks == 1 ? "block" : "blocks"); } archive_read_free(a); exit(0); } /* * Display information about the current file. * * The format here roughly duplicates the output of 'ls -l'. * This is based on SUSv2, where 'tar tv' is documented as * listing additional information in an "unspecified format," * and 'pax -l' is documented as using the same format as 'ls -l'. */ static void list_item_verbose(struct cpio *cpio, struct archive_entry *entry) { char size[32]; char date[32]; char uids[22], gids[22]; const char *uname, *gname; FILE *out = stdout; const char *fmt; time_t mtime; static time_t now; struct tm *ltime; #if defined(HAVE_LOCALTIME_R) || defined(HAVE_LOCALTIME_S) struct tm tmbuf; #endif if (!now) time(&now); if (cpio->option_numeric_uid_gid) { /* Format numeric uid/gid for display. */ strcpy(uids, cpio_i64toa(archive_entry_uid(entry))); uname = uids; strcpy(gids, cpio_i64toa(archive_entry_gid(entry))); gname = gids; } else { /* Use uname if it's present, else lookup name from uid. */ uname = archive_entry_uname(entry); if (uname == NULL) uname = lookup_uname(cpio, (uid_t)archive_entry_uid(entry)); /* Use gname if it's present, else lookup name from gid. */ gname = archive_entry_gname(entry); if (gname == NULL) gname = lookup_gname(cpio, (uid_t)archive_entry_gid(entry)); } /* Print device number or file size. */ if (archive_entry_filetype(entry) == AE_IFCHR || archive_entry_filetype(entry) == AE_IFBLK) { snprintf(size, sizeof(size), "%lu,%lu", (unsigned long)archive_entry_rdevmajor(entry), (unsigned long)archive_entry_rdevminor(entry)); } else { strcpy(size, cpio_i64toa(archive_entry_size(entry))); } /* Format the time using 'ls -l' conventions. */ mtime = archive_entry_mtime(entry); #if defined(_WIN32) && !defined(__CYGWIN__) /* Windows' strftime function does not support %e format. */ if (mtime - now > 365*86400/2 || mtime - now < -365*86400/2) fmt = cpio->day_first ? "%d %b %Y" : "%b %d %Y"; else fmt = cpio->day_first ? "%d %b %H:%M" : "%b %d %H:%M"; #else if (mtime - now > 365*86400/2 || mtime - now < -365*86400/2) fmt = cpio->day_first ? "%e %b %Y" : "%b %e %Y"; else fmt = cpio->day_first ? "%e %b %H:%M" : "%b %e %H:%M"; #endif #if defined(HAVE_LOCALTIME_S) ltime = localtime_s(&tmbuf, &mtime) ? NULL : &tmbuf; #elif defined(HAVE_LOCALTIME_R) ltime = localtime_r(&mtime, &tmbuf); #else ltime = localtime(&mtime); #endif if (ltime != NULL) strftime(date, sizeof(date), fmt, ltime); else strcpy(date, "invalid mtime"); fprintf(out, "%s%3u %-8s %-8s %8s %12s %s", archive_entry_strmode(entry), archive_entry_nlink(entry), uname, gname, size, date, archive_entry_pathname(entry)); /* Extra information for links. */ if (archive_entry_hardlink(entry)) /* Hard link */ fprintf(out, " link to %s", archive_entry_hardlink(entry)); else if (archive_entry_symlink(entry)) /* Symbolic link */ fprintf(out, " -> %s", archive_entry_symlink(entry)); fprintf(out, "\n"); } static void mode_pass(struct cpio *cpio, const char *destdir) { struct lafe_line_reader *lr; const char *p; int r; /* Ensure target dir has a trailing '/' to simplify path surgery. */ cpio->destdir_len = strlen(destdir); cpio->destdir = malloc(cpio->destdir_len + 8); memcpy(cpio->destdir, destdir, cpio->destdir_len); if (cpio->destdir_len == 0 || destdir[cpio->destdir_len - 1] != '/') cpio->destdir[cpio->destdir_len++] = '/'; cpio->destdir[cpio->destdir_len] = '\0'; cpio->archive = archive_write_disk_new(); if (cpio->archive == NULL) lafe_errc(1, 0, "Failed to allocate archive object"); r = archive_write_disk_set_options(cpio->archive, cpio->extract_flags); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive)); cpio->linkresolver = archive_entry_linkresolver_new(); archive_write_disk_set_standard_lookup(cpio->archive); cpio->archive_read_disk = archive_read_disk_new(); if (cpio->archive_read_disk == NULL) lafe_errc(1, 0, "Failed to allocate archive object"); if (cpio->option_follow_links) archive_read_disk_set_symlink_logical(cpio->archive_read_disk); else archive_read_disk_set_symlink_physical(cpio->archive_read_disk); archive_read_disk_set_standard_lookup(cpio->archive_read_disk); lr = lafe_line_reader("-", cpio->option_null); while ((p = lafe_line_reader_next(lr)) != NULL) file_to_archive(cpio, p); lafe_line_reader_free(lr); archive_entry_linkresolver_free(cpio->linkresolver); r = archive_write_close(cpio->archive); if (cpio->dot) fprintf(stderr, "\n"); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive)); if (!cpio->quiet) { int64_t blocks = (archive_filter_bytes(cpio->archive, 0) + 511) / 512; fprintf(stderr, "%lu %s\n", (unsigned long)blocks, blocks == 1 ? "block" : "blocks"); } archive_write_free(cpio->archive); free(cpio->pass_destpath); } /* * Prompt for a new name for this entry. Returns a pointer to the * new name or NULL if the entry should not be copied. This * implements the semantics defined in POSIX.1-1996, which specifies * that an input of '.' means the name should be unchanged. GNU cpio * treats '.' as a literal new name. */ static const char * cpio_rename(const char *name) { static char buff[1024]; FILE *t; char *p, *ret; #if defined(_WIN32) && !defined(__CYGWIN__) FILE *to; t = fopen("CONIN$", "r"); if (t == NULL) return (name); to = fopen("CONOUT$", "w"); if (to == NULL) { fclose(t); return (name); } fprintf(to, "%s (Enter/./(new name))? ", name); fclose(to); #else t = fopen("/dev/tty", "r+"); if (t == NULL) return (name); fprintf(t, "%s (Enter/./(new name))? ", name); fflush(t); #endif p = fgets(buff, sizeof(buff), t); fclose(t); if (p == NULL) /* End-of-file is a blank line. */ return (NULL); while (*p == ' ' || *p == '\t') ++p; if (*p == '\n' || *p == '\0') /* Empty line. */ return (NULL); if (*p == '.' && p[1] == '\n') /* Single period preserves original name. */ return (name); ret = p; /* Trim the final newline. */ while (*p != '\0' && *p != '\n') ++p; /* Overwrite the final \n with a null character. */ *p = '\0'; return (ret); } static void free_cache(struct name_cache *cache) { size_t i; if (cache != NULL) { for (i = 0; i < cache->size; i++) free(cache->cache[i].name); free(cache); } } /* * Lookup uname/gname from uid/gid, return NULL if no match. */ static const char * lookup_name(struct cpio *cpio, struct name_cache **name_cache_variable, int (*lookup_fn)(struct cpio *, const char **, id_t), id_t id) { char asnum[16]; struct name_cache *cache; const char *name; int slot; if (*name_cache_variable == NULL) { *name_cache_variable = calloc(1, sizeof(struct name_cache)); if (*name_cache_variable == NULL) lafe_errc(1, ENOMEM, "No more memory"); (*name_cache_variable)->size = name_cache_size; } cache = *name_cache_variable; cache->probes++; slot = id % cache->size; if (cache->cache[slot].name != NULL) { if (cache->cache[slot].id == id) { cache->hits++; return (cache->cache[slot].name); } free(cache->cache[slot].name); cache->cache[slot].name = NULL; } if (lookup_fn(cpio, &name, id)) { /* If lookup failed, format it as a number. */ snprintf(asnum, sizeof(asnum), "%u", (unsigned)id); name = asnum; } cache->cache[slot].name = strdup(name); if (cache->cache[slot].name != NULL) { cache->cache[slot].id = id; return (cache->cache[slot].name); } /* * Conveniently, NULL marks an empty slot, so * if the strdup() fails, we've just failed to * cache it. No recovery necessary. */ return (NULL); } static const char * lookup_uname(struct cpio *cpio, uid_t uid) { return (lookup_name(cpio, &cpio->uname_cache, &lookup_uname_helper, (id_t)uid)); } static int lookup_uname_helper(struct cpio *cpio, const char **name, id_t id) { struct passwd *pwent; (void)cpio; /* UNUSED */ errno = 0; pwent = getpwuid((uid_t)id); if (pwent == NULL) { if (errno && errno != ENOENT) lafe_warnc(errno, "getpwuid(%s) failed", cpio_i64toa((int64_t)id)); return 1; } *name = pwent->pw_name; return 0; } static const char * lookup_gname(struct cpio *cpio, gid_t gid) { return (lookup_name(cpio, &cpio->gname_cache, &lookup_gname_helper, (id_t)gid)); } static int lookup_gname_helper(struct cpio *cpio, const char **name, id_t id) { struct group *grent; (void)cpio; /* UNUSED */ errno = 0; grent = getgrgid((gid_t)id); if (grent == NULL) { if (errno && errno != ENOENT) lafe_warnc(errno, "getgrgid(%s) failed", cpio_i64toa((int64_t)id)); return 1; } *name = grent->gr_name; return 0; } /* * It would be nice to just use printf() for formatting large numbers, * but the compatibility problems are a big headache. Hence the * following simple utility function. */ const char * cpio_i64toa(int64_t n0) { /* 2^64 =~ 1.8 * 10^19, so 20 decimal digits suffice. * We also need 1 byte for '-' and 1 for '\0'. */ static char buff[22]; int64_t n = n0 < 0 ? -n0 : n0; char *p = buff + sizeof(buff); *--p = '\0'; do { *--p = '0' + (int)(n % 10); n /= 10; } while (n > 0); if (n0 < 0) *--p = '-'; return p; } #define PPBUFF_SIZE 1024 static const char * passphrase_callback(struct archive *a, void *_client_data) { struct cpio *cpio = (struct cpio *)_client_data; (void)a; /* UNUSED */ if (cpio->ppbuff == NULL) { cpio->ppbuff = malloc(PPBUFF_SIZE); if (cpio->ppbuff == NULL) lafe_errc(1, errno, "Out of memory"); } return lafe_readpassphrase("Enter passphrase:", cpio->ppbuff, PPBUFF_SIZE); } static void passphrase_free(char *ppbuff) { if (ppbuff != NULL) { memset(ppbuff, 0, PPBUFF_SIZE); free(ppbuff); } } diff --git a/contrib/libarchive/cpio/test/test_owner_parse.c b/contrib/libarchive/cpio/test/test_owner_parse.c index dfc78ca77aec..bd68f21cec9b 100644 --- a/contrib/libarchive/cpio/test/test_owner_parse.c +++ b/contrib/libarchive/cpio/test/test_owner_parse.c @@ -1,130 +1,130 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2003-2009 Tim Kientzle * All rights reserved. */ #include "test.h" #include "../cpio.h" -#include "err.h" +#include "lafe_err.h" #if !defined(_WIN32) #define ROOT "root" static const int root_uids[] = { 0 }; static const int root_gids[] = { 0, 1 }; #elif defined(__CYGWIN__) /* On cygwin, the Administrator user most likely exists (unless * it has been renamed or is in a non-English localization), but * its primary group membership depends on how the user set up * their /etc/passwd. Likely values are 513 (None), 545 (Users), * or 544 (Administrators). Just check for one of those... * TODO: Handle non-English localizations... e.g. French 'Administrateur' * Use CreateWellKnownSID() and LookupAccountName()? */ #define ROOT "Administrator" static const int root_uids[] = { 500 }; static const int root_gids[] = { 513, 545, 544 }; #endif #if defined(ROOT) static int int_in_list(int i, const int *l, size_t n) { while (n-- > 0) if (*l++ == i) return (1); failure("%d", i); return (0); } static void free_cpio_owner(struct cpio_owner *owner) { owner->uid = -1; owner->gid = -1; free(owner->uname); free(owner->gname); } #endif DEFINE_TEST(test_owner_parse) { #if !defined(ROOT) skipping("No uid/gid configuration for this OS"); #else struct cpio_owner owner; const char *errstr; assert(0 == owner_parse(ROOT, &owner, &errstr)); assert(int_in_list(owner.uid, root_uids, sizeof(root_uids)/sizeof(root_uids[0]))); assertEqualInt(-1, owner.gid); free_cpio_owner(&owner); assert(0 == owner_parse(ROOT ":", &owner, &errstr)); assert(int_in_list(owner.uid, root_uids, sizeof(root_uids)/sizeof(root_uids[0]))); assert(int_in_list(owner.gid, root_gids, sizeof(root_gids)/sizeof(root_gids[0]))); free_cpio_owner(&owner); assert(0 == owner_parse(ROOT ".", &owner, &errstr)); assert(int_in_list(owner.uid, root_uids, sizeof(root_uids)/sizeof(root_uids[0]))); assert(int_in_list(owner.gid, root_gids, sizeof(root_gids)/sizeof(root_gids[0]))); free_cpio_owner(&owner); assert(0 == owner_parse("111", &owner, &errstr)); assertEqualInt(111, owner.uid); assertEqualInt(-1, owner.gid); free_cpio_owner(&owner); assert(0 == owner_parse("112:", &owner, &errstr)); assertEqualInt(112, owner.uid); /* Can't assert gid, since we don't know gid for user #112. */ free_cpio_owner(&owner); assert(0 == owner_parse("113.", &owner, &errstr)); assertEqualInt(113, owner.uid); /* Can't assert gid, since we don't know gid for user #113. */ free_cpio_owner(&owner); assert(0 == owner_parse(":114", &owner, &errstr)); assertEqualInt(-1, owner.uid); assertEqualInt(114, owner.gid); free_cpio_owner(&owner); assert(0 == owner_parse(".115", &owner, &errstr)); assertEqualInt(-1, owner.uid); assertEqualInt(115, owner.gid); free_cpio_owner(&owner); assert(0 == owner_parse("116:117", &owner, &errstr)); assertEqualInt(116, owner.uid); assertEqualInt(117, owner.gid); free_cpio_owner(&owner); /* * TODO: Lookup current user/group name, build strings and * use those to verify username/groupname lookups for ordinary * users. */ errstr = NULL; assert(0 != owner_parse(":nonexistentgroup", &owner, &errstr)); assertEqualString(errstr, "Couldn't lookup group ``nonexistentgroup''"); free_cpio_owner(&owner); errstr = NULL; assert(0 != owner_parse(ROOT ":nonexistentgroup", &owner, &errstr)); assertEqualString(errstr, "Couldn't lookup group ``nonexistentgroup''"); free_cpio_owner(&owner); errstr = NULL; assert(0 != owner_parse("nonexistentuser:nonexistentgroup", &owner, &errstr)); assertEqualString(errstr, "Couldn't lookup user ``nonexistentuser''"); free_cpio_owner(&owner); #endif } diff --git a/contrib/libarchive/libarchive/archive.h b/contrib/libarchive/libarchive/archive.h index 002190a24663..98d7674e18f1 100644 --- a/contrib/libarchive/libarchive/archive.h +++ b/contrib/libarchive/libarchive/archive.h @@ -1,1265 +1,1265 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #ifndef ARCHIVE_H_INCLUDED #define ARCHIVE_H_INCLUDED /* * The version number is expressed as a single integer that makes it * easy to compare versions at build time: for version a.b.c, the * version number is printf("%d%03d%03d",a,b,c). For example, if you * know your application requires version 2.12.108 or later, you can * assert that ARCHIVE_VERSION_NUMBER >= 2012108. */ /* Note: Compiler will complain if this does not match archive_entry.h! */ -#define ARCHIVE_VERSION_NUMBER 3008001 +#define ARCHIVE_VERSION_NUMBER 3008002 #include #include /* for wchar_t */ #include /* For FILE * */ #if ARCHIVE_VERSION_NUMBER < 4000000 /* time_t is slated to be removed from public includes in 4.0 */ #include /* For time_t */ #endif /* * Note: archive.h is for use outside of libarchive; the configuration * headers (config.h, archive_platform.h, etc.) are purely internal. * Do NOT use HAVE_XXX configuration macros to control the behavior of * this header! If you must conditionalize, use predefined compiler and/or * platform macros. */ #if defined(__BORLANDC__) && __BORLANDC__ >= 0x560 # include #elif !defined(__WATCOMC__) && !defined(_MSC_VER) && !defined(__INTERIX) && !defined(__BORLANDC__) && !defined(_SCO_DS) && !defined(__osf__) && !defined(__CLANG_INTTYPES_H) # include #endif /* Get appropriate definitions of 64-bit integer */ #if !defined(__LA_INT64_T_DEFINED) /* Older code relied on the __LA_INT64_T macro; after 4.0 we'll switch to the typedef exclusively. */ # if ARCHIVE_VERSION_NUMBER < 4000000 #define __LA_INT64_T la_int64_t # endif #define __LA_INT64_T_DEFINED # if defined(_WIN32) && !defined(__CYGWIN__) && !defined(__WATCOMC__) typedef __int64 la_int64_t; typedef unsigned __int64 la_uint64_t; # else # include /* ssize_t */ # if defined(_SCO_DS) || defined(__osf__) typedef long long la_int64_t; typedef unsigned long long la_uint64_t; # else typedef int64_t la_int64_t; typedef uint64_t la_uint64_t; # endif # endif #endif /* The la_ssize_t should match the type used in 'struct stat' */ #if !defined(__LA_SSIZE_T_DEFINED) /* Older code relied on the __LA_SSIZE_T macro; after 4.0 we'll switch to the typedef exclusively. */ # if ARCHIVE_VERSION_NUMBER < 4000000 #define __LA_SSIZE_T la_ssize_t # endif #define __LA_SSIZE_T_DEFINED # if defined(_WIN32) && !defined(__CYGWIN__) && !defined(__WATCOMC__) # if defined(_SSIZE_T_DEFINED) || defined(_SSIZE_T_) typedef ssize_t la_ssize_t; # elif defined(_WIN64) typedef __int64 la_ssize_t; # else typedef long la_ssize_t; # endif # else # include /* ssize_t */ typedef ssize_t la_ssize_t; # endif #endif #if ARCHIVE_VERSION_NUMBER < 4000000 /* Use the platform types for time_t */ #define __LA_TIME_T time_t #else /* Use 64-bits integer types for time_t */ #define __LA_TIME_T la_int64_t #endif #if ARCHIVE_VERSION_NUMBER < 4000000 /* Use the platform types for dev_t */ #define __LA_DEV_T dev_t #else /* Use 64-bits integer types for dev_t */ #define __LA_DEV_T la_int64_t #endif /* Large file support for Android */ #if defined(__LIBARCHIVE_BUILD) && defined(__ANDROID__) #include "android_lf.h" #endif /* * On Windows, define LIBARCHIVE_STATIC if you're building or using a * .lib. The default here assumes you're building a DLL. Only * libarchive source should ever define __LIBARCHIVE_BUILD. */ #if ((defined __WIN32__) || (defined _WIN32) || defined(__CYGWIN__)) && (!defined LIBARCHIVE_STATIC) # ifdef __LIBARCHIVE_BUILD # ifdef __GNUC__ # define __LA_DECL __attribute__((dllexport)) extern # else # define __LA_DECL __declspec(dllexport) # endif # else # ifdef __GNUC__ # define __LA_DECL # else # define __LA_DECL __declspec(dllimport) # endif # endif #elif defined __LIBARCHIVE_ENABLE_VISIBILITY # define __LA_DECL __attribute__((visibility("default"))) #else /* Static libraries or non-Windows needs no special declaration. */ # define __LA_DECL #endif #if defined(__GNUC__) && __GNUC__ >= 3 && !defined(__MINGW32__) #define __LA_PRINTF(fmtarg, firstvararg) \ __attribute__((__format__ (__printf__, fmtarg, firstvararg))) #else #define __LA_PRINTF(fmtarg, firstvararg) /* nothing */ #endif #if defined(__GNUC__) && (__GNUC__ > 3 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)) # define __LA_DEPRECATED __attribute__((deprecated)) #else # define __LA_DEPRECATED #endif #ifdef __cplusplus extern "C" { #endif /* * The version number is provided as both a macro and a function. * The macro identifies the installed header; the function identifies * the library version (which may not be the same if you're using a * dynamically-linked version of the library). Of course, if the * header and library are very different, you should expect some * strangeness. Don't do that. */ __LA_DECL int archive_version_number(void); /* * Textual name/version of the library, useful for version displays. */ -#define ARCHIVE_VERSION_ONLY_STRING "3.8.1" +#define ARCHIVE_VERSION_ONLY_STRING "3.8.2" #define ARCHIVE_VERSION_STRING "libarchive " ARCHIVE_VERSION_ONLY_STRING __LA_DECL const char * archive_version_string(void); /* * Detailed textual name/version of the library and its dependencies. * This has the form: * "libarchive x.y.z zlib/a.b.c liblzma/d.e.f ... etc ..." * the list of libraries described here will vary depending on how * libarchive was compiled. */ __LA_DECL const char * archive_version_details(void); /* * Returns NULL if libarchive was compiled without the associated library. * Otherwise, returns the version number that libarchive was compiled * against. */ __LA_DECL const char * archive_zlib_version(void); __LA_DECL const char * archive_liblzma_version(void); __LA_DECL const char * archive_bzlib_version(void); __LA_DECL const char * archive_liblz4_version(void); __LA_DECL const char * archive_libzstd_version(void); __LA_DECL const char * archive_liblzo2_version(void); __LA_DECL const char * archive_libexpat_version(void); __LA_DECL const char * archive_libbsdxml_version(void); __LA_DECL const char * archive_libxml2_version(void); __LA_DECL const char * archive_mbedtls_version(void); __LA_DECL const char * archive_nettle_version(void); __LA_DECL const char * archive_openssl_version(void); __LA_DECL const char * archive_libmd_version(void); __LA_DECL const char * archive_commoncrypto_version(void); __LA_DECL const char * archive_cng_version(void); __LA_DECL const char * archive_wincrypt_version(void); __LA_DECL const char * archive_librichacl_version(void); __LA_DECL const char * archive_libacl_version(void); __LA_DECL const char * archive_libattr_version(void); __LA_DECL const char * archive_libiconv_version(void); __LA_DECL const char * archive_libpcre_version(void); __LA_DECL const char * archive_libpcre2_version(void); /* Declare our basic types. */ struct archive; struct archive_entry; /* * Error codes: Use archive_errno() and archive_error_string() * to retrieve details. Unless specified otherwise, all functions * that return 'int' use these codes. */ #define ARCHIVE_EOF 1 /* Found end of archive. */ #define ARCHIVE_OK 0 /* Operation was successful. */ #define ARCHIVE_RETRY (-10) /* Retry might succeed. */ #define ARCHIVE_WARN (-20) /* Partial success. */ /* For example, if write_header "fails", then you can't push data. */ #define ARCHIVE_FAILED (-25) /* Current operation cannot complete. */ /* But if write_header is "fatal," then this archive is dead and useless. */ #define ARCHIVE_FATAL (-30) /* No more operations are possible. */ /* * As far as possible, archive_errno returns standard platform errno codes. * Of course, the details vary by platform, so the actual definitions * here are stored in "archive_platform.h". The symbols are listed here * for reference; as a rule, clients should not need to know the exact * platform-dependent error code. */ /* Unrecognized or invalid file format. */ /* #define ARCHIVE_ERRNO_FILE_FORMAT */ /* Illegal usage of the library. */ /* #define ARCHIVE_ERRNO_PROGRAMMER_ERROR */ /* Unknown or unclassified error. */ /* #define ARCHIVE_ERRNO_MISC */ /* * Callbacks are invoked to automatically read/skip/write/open/close the * archive. You can provide your own for complex tasks (like breaking * archives across multiple tapes) or use standard ones built into the * library. */ /* Returns pointer and size of next block of data from archive. */ typedef la_ssize_t archive_read_callback(struct archive *, void *_client_data, const void **_buffer); /* Skips at most request bytes from archive and returns the skipped amount. * This may skip fewer bytes than requested; it may even skip zero bytes. * If you do skip fewer bytes than requested, libarchive will invoke your * read callback and discard data as necessary to make up the full skip. */ typedef la_int64_t archive_skip_callback(struct archive *, void *_client_data, la_int64_t request); /* Seeks to specified location in the file and returns the position. * Whence values are SEEK_SET, SEEK_CUR, SEEK_END from stdio.h. * Return ARCHIVE_FATAL if the seek fails for any reason. */ typedef la_int64_t archive_seek_callback(struct archive *, void *_client_data, la_int64_t offset, int whence); /* Returns size actually written, zero on EOF, -1 on error. */ typedef la_ssize_t archive_write_callback(struct archive *, void *_client_data, const void *_buffer, size_t _length); typedef int archive_open_callback(struct archive *, void *_client_data); typedef int archive_close_callback(struct archive *, void *_client_data); typedef int archive_free_callback(struct archive *, void *_client_data); /* Switches from one client data object to the next/prev client data object. * This is useful for reading from different data blocks such as a set of files * that make up one large file. */ typedef int archive_switch_callback(struct archive *, void *_client_data1, void *_client_data2); /* * Returns a passphrase used for encryption or decryption, NULL on nothing * to do and give it up. */ typedef const char *archive_passphrase_callback(struct archive *, void *_client_data); /* * Codes to identify various stream filters. */ #define ARCHIVE_FILTER_NONE 0 #define ARCHIVE_FILTER_GZIP 1 #define ARCHIVE_FILTER_BZIP2 2 #define ARCHIVE_FILTER_COMPRESS 3 #define ARCHIVE_FILTER_PROGRAM 4 #define ARCHIVE_FILTER_LZMA 5 #define ARCHIVE_FILTER_XZ 6 #define ARCHIVE_FILTER_UU 7 #define ARCHIVE_FILTER_RPM 8 #define ARCHIVE_FILTER_LZIP 9 #define ARCHIVE_FILTER_LRZIP 10 #define ARCHIVE_FILTER_LZOP 11 #define ARCHIVE_FILTER_GRZIP 12 #define ARCHIVE_FILTER_LZ4 13 #define ARCHIVE_FILTER_ZSTD 14 #if ARCHIVE_VERSION_NUMBER < 4000000 #define ARCHIVE_COMPRESSION_NONE ARCHIVE_FILTER_NONE #define ARCHIVE_COMPRESSION_GZIP ARCHIVE_FILTER_GZIP #define ARCHIVE_COMPRESSION_BZIP2 ARCHIVE_FILTER_BZIP2 #define ARCHIVE_COMPRESSION_COMPRESS ARCHIVE_FILTER_COMPRESS #define ARCHIVE_COMPRESSION_PROGRAM ARCHIVE_FILTER_PROGRAM #define ARCHIVE_COMPRESSION_LZMA ARCHIVE_FILTER_LZMA #define ARCHIVE_COMPRESSION_XZ ARCHIVE_FILTER_XZ #define ARCHIVE_COMPRESSION_UU ARCHIVE_FILTER_UU #define ARCHIVE_COMPRESSION_RPM ARCHIVE_FILTER_RPM #define ARCHIVE_COMPRESSION_LZIP ARCHIVE_FILTER_LZIP #define ARCHIVE_COMPRESSION_LRZIP ARCHIVE_FILTER_LRZIP #endif /* * Codes returned by archive_format. * * Top 16 bits identifies the format family (e.g., "tar"); lower * 16 bits indicate the variant. This is updated by read_next_header. * Note that the lower 16 bits will often vary from entry to entry. * In some cases, this variation occurs as libarchive learns more about * the archive (for example, later entries might utilize extensions that * weren't necessary earlier in the archive; in this case, libarchive * will change the format code to indicate the extended format that * was used). In other cases, it's because different tools have * modified the archive and so different parts of the archive * actually have slightly different formats. (Both tar and cpio store * format codes in each entry, so it is quite possible for each * entry to be in a different format.) */ #define ARCHIVE_FORMAT_BASE_MASK 0xff0000 #define ARCHIVE_FORMAT_CPIO 0x10000 #define ARCHIVE_FORMAT_CPIO_POSIX (ARCHIVE_FORMAT_CPIO | 1) #define ARCHIVE_FORMAT_CPIO_BIN_LE (ARCHIVE_FORMAT_CPIO | 2) #define ARCHIVE_FORMAT_CPIO_BIN_BE (ARCHIVE_FORMAT_CPIO | 3) #define ARCHIVE_FORMAT_CPIO_SVR4_NOCRC (ARCHIVE_FORMAT_CPIO | 4) #define ARCHIVE_FORMAT_CPIO_SVR4_CRC (ARCHIVE_FORMAT_CPIO | 5) #define ARCHIVE_FORMAT_CPIO_AFIO_LARGE (ARCHIVE_FORMAT_CPIO | 6) #define ARCHIVE_FORMAT_CPIO_PWB (ARCHIVE_FORMAT_CPIO | 7) #define ARCHIVE_FORMAT_SHAR 0x20000 #define ARCHIVE_FORMAT_SHAR_BASE (ARCHIVE_FORMAT_SHAR | 1) #define ARCHIVE_FORMAT_SHAR_DUMP (ARCHIVE_FORMAT_SHAR | 2) #define ARCHIVE_FORMAT_TAR 0x30000 #define ARCHIVE_FORMAT_TAR_USTAR (ARCHIVE_FORMAT_TAR | 1) #define ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE (ARCHIVE_FORMAT_TAR | 2) #define ARCHIVE_FORMAT_TAR_PAX_RESTRICTED (ARCHIVE_FORMAT_TAR | 3) #define ARCHIVE_FORMAT_TAR_GNUTAR (ARCHIVE_FORMAT_TAR | 4) #define ARCHIVE_FORMAT_ISO9660 0x40000 #define ARCHIVE_FORMAT_ISO9660_ROCKRIDGE (ARCHIVE_FORMAT_ISO9660 | 1) #define ARCHIVE_FORMAT_ZIP 0x50000 #define ARCHIVE_FORMAT_EMPTY 0x60000 #define ARCHIVE_FORMAT_AR 0x70000 #define ARCHIVE_FORMAT_AR_GNU (ARCHIVE_FORMAT_AR | 1) #define ARCHIVE_FORMAT_AR_BSD (ARCHIVE_FORMAT_AR | 2) #define ARCHIVE_FORMAT_MTREE 0x80000 #define ARCHIVE_FORMAT_RAW 0x90000 #define ARCHIVE_FORMAT_XAR 0xA0000 #define ARCHIVE_FORMAT_LHA 0xB0000 #define ARCHIVE_FORMAT_CAB 0xC0000 #define ARCHIVE_FORMAT_RAR 0xD0000 #define ARCHIVE_FORMAT_7ZIP 0xE0000 #define ARCHIVE_FORMAT_WARC 0xF0000 #define ARCHIVE_FORMAT_RAR_V5 0x100000 /* * Codes returned by archive_read_format_capabilities(). * * This list can be extended with values between 0 and 0xffff. * The original purpose of this list was to let different archive * format readers expose their general capabilities in terms of * encryption. */ #define ARCHIVE_READ_FORMAT_CAPS_NONE (0) /* no special capabilities */ #define ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA (1<<0) /* reader can detect encrypted data */ #define ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA (1<<1) /* reader can detect encryptable metadata (pathname, mtime, etc.) */ /* * Codes returned by archive_read_has_encrypted_entries(). * * In case the archive does not support encryption detection at all * ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED is returned. If the reader * for some other reason (e.g. not enough bytes read) cannot say if * there are encrypted entries, ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW * is returned. */ #define ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED -2 #define ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW -1 /*- * Basic outline for reading an archive: * 1) Ask archive_read_new for an archive reader object. * 2) Update any global properties as appropriate. * In particular, you'll certainly want to call appropriate * archive_read_support_XXX functions. * 3) Call archive_read_open_XXX to open the archive * 4) Repeatedly call archive_read_next_header to get information about * successive archive entries. Call archive_read_data to extract * data for entries of interest. * 5) Call archive_read_free to end processing. */ __LA_DECL struct archive *archive_read_new(void); /* * The archive_read_support_XXX calls enable auto-detect for this * archive handle. They also link in the necessary support code. * For example, if you don't want bzlib linked in, don't invoke * support_compression_bzip2(). The "all" functions provide the * obvious shorthand. */ #if ARCHIVE_VERSION_NUMBER < 4000000 __LA_DECL int archive_read_support_compression_all(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_read_support_compression_bzip2(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_read_support_compression_compress(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_read_support_compression_gzip(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_read_support_compression_lzip(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_read_support_compression_lzma(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_read_support_compression_none(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_read_support_compression_program(struct archive *, const char *command) __LA_DEPRECATED; __LA_DECL int archive_read_support_compression_program_signature (struct archive *, const char *, const void * /* match */, size_t) __LA_DEPRECATED; __LA_DECL int archive_read_support_compression_rpm(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_read_support_compression_uu(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_read_support_compression_xz(struct archive *) __LA_DEPRECATED; #endif __LA_DECL int archive_read_support_filter_all(struct archive *); __LA_DECL int archive_read_support_filter_by_code(struct archive *, int); __LA_DECL int archive_read_support_filter_bzip2(struct archive *); __LA_DECL int archive_read_support_filter_compress(struct archive *); __LA_DECL int archive_read_support_filter_gzip(struct archive *); __LA_DECL int archive_read_support_filter_grzip(struct archive *); __LA_DECL int archive_read_support_filter_lrzip(struct archive *); __LA_DECL int archive_read_support_filter_lz4(struct archive *); __LA_DECL int archive_read_support_filter_lzip(struct archive *); __LA_DECL int archive_read_support_filter_lzma(struct archive *); __LA_DECL int archive_read_support_filter_lzop(struct archive *); __LA_DECL int archive_read_support_filter_none(struct archive *); __LA_DECL int archive_read_support_filter_program(struct archive *, const char *command); __LA_DECL int archive_read_support_filter_program_signature (struct archive *, const char * /* cmd */, const void * /* match */, size_t); __LA_DECL int archive_read_support_filter_rpm(struct archive *); __LA_DECL int archive_read_support_filter_uu(struct archive *); __LA_DECL int archive_read_support_filter_xz(struct archive *); __LA_DECL int archive_read_support_filter_zstd(struct archive *); __LA_DECL int archive_read_support_format_7zip(struct archive *); __LA_DECL int archive_read_support_format_all(struct archive *); __LA_DECL int archive_read_support_format_ar(struct archive *); __LA_DECL int archive_read_support_format_by_code(struct archive *, int); __LA_DECL int archive_read_support_format_cab(struct archive *); __LA_DECL int archive_read_support_format_cpio(struct archive *); __LA_DECL int archive_read_support_format_empty(struct archive *); /* archive_read_support_format_gnutar() is an alias for historical reasons * of archive_read_support_format_tar(). */ __LA_DECL int archive_read_support_format_gnutar(struct archive *); __LA_DECL int archive_read_support_format_iso9660(struct archive *); __LA_DECL int archive_read_support_format_lha(struct archive *); __LA_DECL int archive_read_support_format_mtree(struct archive *); __LA_DECL int archive_read_support_format_rar(struct archive *); __LA_DECL int archive_read_support_format_rar5(struct archive *); __LA_DECL int archive_read_support_format_raw(struct archive *); __LA_DECL int archive_read_support_format_tar(struct archive *); __LA_DECL int archive_read_support_format_warc(struct archive *); __LA_DECL int archive_read_support_format_xar(struct archive *); /* archive_read_support_format_zip() enables both streamable and seekable * zip readers. */ __LA_DECL int archive_read_support_format_zip(struct archive *); /* Reads Zip archives as stream from beginning to end. Doesn't * correctly handle SFX ZIP files or ZIP archives that have been modified * in-place. */ __LA_DECL int archive_read_support_format_zip_streamable(struct archive *); /* Reads starting from central directory; requires seekable input. */ __LA_DECL int archive_read_support_format_zip_seekable(struct archive *); /* Functions to manually set the format and filters to be used. This is * useful to bypass the bidding process when the format and filters to use * is known in advance. */ __LA_DECL int archive_read_set_format(struct archive *, int); __LA_DECL int archive_read_append_filter(struct archive *, int); __LA_DECL int archive_read_append_filter_program(struct archive *, const char *); __LA_DECL int archive_read_append_filter_program_signature (struct archive *, const char *, const void * /* match */, size_t); /* Set various callbacks. */ __LA_DECL int archive_read_set_open_callback(struct archive *, archive_open_callback *); __LA_DECL int archive_read_set_read_callback(struct archive *, archive_read_callback *); __LA_DECL int archive_read_set_seek_callback(struct archive *, archive_seek_callback *); __LA_DECL int archive_read_set_skip_callback(struct archive *, archive_skip_callback *); __LA_DECL int archive_read_set_close_callback(struct archive *, archive_close_callback *); /* Callback used to switch between one data object to the next */ __LA_DECL int archive_read_set_switch_callback(struct archive *, archive_switch_callback *); /* This sets the first data object. */ __LA_DECL int archive_read_set_callback_data(struct archive *, void *); /* This sets data object at specified index */ __LA_DECL int archive_read_set_callback_data2(struct archive *, void *, unsigned int); /* This adds a data object at the specified index. */ __LA_DECL int archive_read_add_callback_data(struct archive *, void *, unsigned int); /* This appends a data object to the end of list */ __LA_DECL int archive_read_append_callback_data(struct archive *, void *); /* This prepends a data object to the beginning of list */ __LA_DECL int archive_read_prepend_callback_data(struct archive *, void *); /* Opening freezes the callbacks. */ __LA_DECL int archive_read_open1(struct archive *); /* Convenience wrappers around the above. */ __LA_DECL int archive_read_open(struct archive *, void *_client_data, archive_open_callback *, archive_read_callback *, archive_close_callback *); __LA_DECL int archive_read_open2(struct archive *, void *_client_data, archive_open_callback *, archive_read_callback *, archive_skip_callback *, archive_close_callback *); /* * A variety of shortcuts that invoke archive_read_open() with * canned callbacks suitable for common situations. The ones that * accept a block size handle tape blocking correctly. */ /* Use this if you know the filename. Note: NULL indicates stdin. */ __LA_DECL int archive_read_open_filename(struct archive *, const char *_filename, size_t _block_size); /* Use this for reading multivolume files by filenames. * NOTE: Must be NULL terminated. Sorting is NOT done. */ __LA_DECL int archive_read_open_filenames(struct archive *, const char **_filenames, size_t _block_size); __LA_DECL int archive_read_open_filename_w(struct archive *, const wchar_t *_filename, size_t _block_size); #if defined(_WIN32) && !defined(__CYGWIN__) __LA_DECL int archive_read_open_filenames_w(struct archive *, const wchar_t **_filenames, size_t _block_size); #endif /* archive_read_open_file() is a deprecated synonym for ..._open_filename(). */ __LA_DECL int archive_read_open_file(struct archive *, const char *_filename, size_t _block_size) __LA_DEPRECATED; /* Read an archive that's stored in memory. */ __LA_DECL int archive_read_open_memory(struct archive *, const void * buff, size_t size); /* A more involved version that is only used for internal testing. */ __LA_DECL int archive_read_open_memory2(struct archive *a, const void *buff, size_t size, size_t read_size); /* Read an archive that's already open, using the file descriptor. */ __LA_DECL int archive_read_open_fd(struct archive *, int _fd, size_t _block_size); /* Read an archive that's already open, using a FILE *. */ /* Note: DO NOT use this with tape drives. */ __LA_DECL int archive_read_open_FILE(struct archive *, FILE *_file); /* Parses and returns next entry header. */ __LA_DECL int archive_read_next_header(struct archive *, struct archive_entry **); /* Parses and returns next entry header using the archive_entry passed in */ __LA_DECL int archive_read_next_header2(struct archive *, struct archive_entry *); /* * Retrieve the byte offset in UNCOMPRESSED data where last-read * header started. */ __LA_DECL la_int64_t archive_read_header_position(struct archive *); /* * Returns 1 if the archive contains at least one encrypted entry. * If the archive format not support encryption at all * ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED is returned. * If for any other reason (e.g. not enough data read so far) * we cannot say whether there are encrypted entries, then * ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW is returned. * In general, this function will return values below zero when the * reader is uncertain or totally incapable of encryption support. * When this function returns 0 you can be sure that the reader * supports encryption detection but no encrypted entries have * been found yet. * * NOTE: If the metadata/header of an archive is also encrypted, you * cannot rely on the number of encrypted entries. That is why this * function does not return the number of encrypted entries but# * just shows that there are some. */ __LA_DECL int archive_read_has_encrypted_entries(struct archive *); /* * Returns a bitmask of capabilities that are supported by the archive format reader. * If the reader has no special capabilities, ARCHIVE_READ_FORMAT_CAPS_NONE is returned. */ __LA_DECL int archive_read_format_capabilities(struct archive *); /* Read data from the body of an entry. Similar to read(2). */ __LA_DECL la_ssize_t archive_read_data(struct archive *, void *, size_t); /* Seek within the body of an entry. Similar to lseek(2). */ __LA_DECL la_int64_t archive_seek_data(struct archive *, la_int64_t, int); /* * A zero-copy version of archive_read_data that also exposes the file offset * of each returned block. Note that the client has no way to specify * the desired size of the block. The API does guarantee that offsets will * be strictly increasing and that returned blocks will not overlap. */ __LA_DECL int archive_read_data_block(struct archive *a, const void **buff, size_t *size, la_int64_t *offset); /*- * Some convenience functions that are built on archive_read_data: * 'skip': skips entire entry * 'into_buffer': writes data into memory buffer that you provide * 'into_fd': writes data to specified filedes */ __LA_DECL int archive_read_data_skip(struct archive *); __LA_DECL int archive_read_data_into_fd(struct archive *, int fd); /* * Set read options. */ /* Apply option to the format only. */ __LA_DECL int archive_read_set_format_option(struct archive *_a, const char *m, const char *o, const char *v); /* Apply option to the filter only. */ __LA_DECL int archive_read_set_filter_option(struct archive *_a, const char *m, const char *o, const char *v); /* Apply option to both the format and the filter. */ __LA_DECL int archive_read_set_option(struct archive *_a, const char *m, const char *o, const char *v); /* Apply option string to both the format and the filter. */ __LA_DECL int archive_read_set_options(struct archive *_a, const char *opts); /* * Add a decryption passphrase. */ __LA_DECL int archive_read_add_passphrase(struct archive *, const char *); __LA_DECL int archive_read_set_passphrase_callback(struct archive *, void *client_data, archive_passphrase_callback *); /*- * Convenience function to recreate the current entry (whose header * has just been read) on disk. * * This does quite a bit more than just copy data to disk. It also: * - Creates intermediate directories as required. * - Manages directory permissions: non-writable directories will * be initially created with write permission enabled; when the * archive is closed, dir permissions are edited to the values specified * in the archive. * - Checks hardlinks: hardlinks will not be extracted unless the * linked-to file was also extracted within the same session. (TODO) */ /* The "flags" argument selects optional behavior, 'OR' the flags you want. */ /* Default: Do not try to set owner/group. */ #define ARCHIVE_EXTRACT_OWNER (0x0001) /* Default: Do obey umask, do not restore SUID/SGID/SVTX bits. */ #define ARCHIVE_EXTRACT_PERM (0x0002) /* Default: Do not restore mtime/atime. */ #define ARCHIVE_EXTRACT_TIME (0x0004) /* Default: Replace existing files. */ #define ARCHIVE_EXTRACT_NO_OVERWRITE (0x0008) /* Default: Try create first, unlink only if create fails with EEXIST. */ #define ARCHIVE_EXTRACT_UNLINK (0x0010) /* Default: Do not restore ACLs. */ #define ARCHIVE_EXTRACT_ACL (0x0020) /* Default: Do not restore fflags. */ #define ARCHIVE_EXTRACT_FFLAGS (0x0040) /* Default: Do not restore xattrs. */ #define ARCHIVE_EXTRACT_XATTR (0x0080) /* Default: Do not try to guard against extracts redirected by symlinks. */ /* Note: With ARCHIVE_EXTRACT_UNLINK, will remove any intermediate symlink. */ #define ARCHIVE_EXTRACT_SECURE_SYMLINKS (0x0100) /* Default: Do not reject entries with '..' as path elements. */ #define ARCHIVE_EXTRACT_SECURE_NODOTDOT (0x0200) /* Default: Create parent directories as needed. */ #define ARCHIVE_EXTRACT_NO_AUTODIR (0x0400) /* Default: Overwrite files, even if one on disk is newer. */ #define ARCHIVE_EXTRACT_NO_OVERWRITE_NEWER (0x0800) /* Detect blocks of 0 and write holes instead. */ #define ARCHIVE_EXTRACT_SPARSE (0x1000) /* Default: Do not restore Mac extended metadata. */ /* This has no effect except on Mac OS. */ #define ARCHIVE_EXTRACT_MAC_METADATA (0x2000) /* Default: Use HFS+ compression if it was compressed. */ /* This has no effect except on Mac OS v10.6 or later. */ #define ARCHIVE_EXTRACT_NO_HFS_COMPRESSION (0x4000) /* Default: Do not use HFS+ compression if it was not compressed. */ /* This has no effect except on Mac OS v10.6 or later. */ #define ARCHIVE_EXTRACT_HFS_COMPRESSION_FORCED (0x8000) /* Default: Do not reject entries with absolute paths */ #define ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS (0x10000) /* Default: Do not clear no-change flags when unlinking object */ #define ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS (0x20000) /* Default: Do not extract atomically (using rename) */ #define ARCHIVE_EXTRACT_SAFE_WRITES (0x40000) __LA_DECL int archive_read_extract(struct archive *, struct archive_entry *, int flags); __LA_DECL int archive_read_extract2(struct archive *, struct archive_entry *, struct archive * /* dest */); __LA_DECL void archive_read_extract_set_progress_callback(struct archive *, void (*_progress_func)(void *), void *_user_data); /* Record the dev/ino of a file that will not be written. This is * generally set to the dev/ino of the archive being read. */ __LA_DECL void archive_read_extract_set_skip_file(struct archive *, la_int64_t, la_int64_t); /* Close the file and release most resources. */ __LA_DECL int archive_read_close(struct archive *); /* Release all resources and destroy the object. */ /* Note that archive_read_free will call archive_read_close for you. */ __LA_DECL int archive_read_free(struct archive *); #if ARCHIVE_VERSION_NUMBER < 4000000 /* Synonym for archive_read_free() for backwards compatibility. */ __LA_DECL int archive_read_finish(struct archive *) __LA_DEPRECATED; #endif /*- * To create an archive: * 1) Ask archive_write_new for an archive writer object. * 2) Set any global properties. In particular, you should set * the compression and format to use. * 3) Call archive_write_open to open the file (most people * will use archive_write_open_file or archive_write_open_fd, * which provide convenient canned I/O callbacks for you). * 4) For each entry: * - construct an appropriate struct archive_entry structure * - archive_write_header to write the header * - archive_write_data to write the entry data * 5) archive_write_close to close the output * 6) archive_write_free to cleanup the writer and release resources */ __LA_DECL struct archive *archive_write_new(void); __LA_DECL int archive_write_set_bytes_per_block(struct archive *, int bytes_per_block); __LA_DECL int archive_write_get_bytes_per_block(struct archive *); /* XXX This is badly misnamed; suggestions appreciated. XXX */ __LA_DECL int archive_write_set_bytes_in_last_block(struct archive *, int bytes_in_last_block); __LA_DECL int archive_write_get_bytes_in_last_block(struct archive *); /* The dev/ino of a file that won't be archived. This is used * to avoid recursively adding an archive to itself. */ __LA_DECL int archive_write_set_skip_file(struct archive *, la_int64_t, la_int64_t); #if ARCHIVE_VERSION_NUMBER < 4000000 __LA_DECL int archive_write_set_compression_bzip2(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_write_set_compression_compress(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_write_set_compression_gzip(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_write_set_compression_lzip(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_write_set_compression_lzma(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_write_set_compression_none(struct archive *) __LA_DEPRECATED; __LA_DECL int archive_write_set_compression_program(struct archive *, const char *cmd) __LA_DEPRECATED; __LA_DECL int archive_write_set_compression_xz(struct archive *) __LA_DEPRECATED; #endif /* A convenience function to set the filter based on the code. */ __LA_DECL int archive_write_add_filter(struct archive *, int filter_code); __LA_DECL int archive_write_add_filter_by_name(struct archive *, const char *name); __LA_DECL int archive_write_add_filter_b64encode(struct archive *); __LA_DECL int archive_write_add_filter_bzip2(struct archive *); __LA_DECL int archive_write_add_filter_compress(struct archive *); __LA_DECL int archive_write_add_filter_grzip(struct archive *); __LA_DECL int archive_write_add_filter_gzip(struct archive *); __LA_DECL int archive_write_add_filter_lrzip(struct archive *); __LA_DECL int archive_write_add_filter_lz4(struct archive *); __LA_DECL int archive_write_add_filter_lzip(struct archive *); __LA_DECL int archive_write_add_filter_lzma(struct archive *); __LA_DECL int archive_write_add_filter_lzop(struct archive *); __LA_DECL int archive_write_add_filter_none(struct archive *); __LA_DECL int archive_write_add_filter_program(struct archive *, const char *cmd); __LA_DECL int archive_write_add_filter_uuencode(struct archive *); __LA_DECL int archive_write_add_filter_xz(struct archive *); __LA_DECL int archive_write_add_filter_zstd(struct archive *); /* A convenience function to set the format based on the code or name. */ __LA_DECL int archive_write_set_format(struct archive *, int format_code); __LA_DECL int archive_write_set_format_by_name(struct archive *, const char *name); /* To minimize link pollution, use one or more of the following. */ __LA_DECL int archive_write_set_format_7zip(struct archive *); __LA_DECL int archive_write_set_format_ar_bsd(struct archive *); __LA_DECL int archive_write_set_format_ar_svr4(struct archive *); __LA_DECL int archive_write_set_format_cpio(struct archive *); __LA_DECL int archive_write_set_format_cpio_bin(struct archive *); __LA_DECL int archive_write_set_format_cpio_newc(struct archive *); __LA_DECL int archive_write_set_format_cpio_odc(struct archive *); __LA_DECL int archive_write_set_format_cpio_pwb(struct archive *); __LA_DECL int archive_write_set_format_gnutar(struct archive *); __LA_DECL int archive_write_set_format_iso9660(struct archive *); __LA_DECL int archive_write_set_format_mtree(struct archive *); __LA_DECL int archive_write_set_format_mtree_classic(struct archive *); /* TODO: int archive_write_set_format_old_tar(struct archive *); */ __LA_DECL int archive_write_set_format_pax(struct archive *); __LA_DECL int archive_write_set_format_pax_restricted(struct archive *); __LA_DECL int archive_write_set_format_raw(struct archive *); __LA_DECL int archive_write_set_format_shar(struct archive *); __LA_DECL int archive_write_set_format_shar_dump(struct archive *); __LA_DECL int archive_write_set_format_ustar(struct archive *); __LA_DECL int archive_write_set_format_v7tar(struct archive *); __LA_DECL int archive_write_set_format_warc(struct archive *); __LA_DECL int archive_write_set_format_xar(struct archive *); __LA_DECL int archive_write_set_format_zip(struct archive *); __LA_DECL int archive_write_set_format_filter_by_ext(struct archive *a, const char *filename); __LA_DECL int archive_write_set_format_filter_by_ext_def(struct archive *a, const char *filename, const char * def_ext); __LA_DECL int archive_write_zip_set_compression_deflate(struct archive *); __LA_DECL int archive_write_zip_set_compression_store(struct archive *); __LA_DECL int archive_write_zip_set_compression_lzma(struct archive *); __LA_DECL int archive_write_zip_set_compression_xz(struct archive *); __LA_DECL int archive_write_zip_set_compression_bzip2(struct archive *); __LA_DECL int archive_write_zip_set_compression_zstd(struct archive *); /* Deprecated; use archive_write_open2 instead */ __LA_DECL int archive_write_open(struct archive *, void *, archive_open_callback *, archive_write_callback *, archive_close_callback *); __LA_DECL int archive_write_open2(struct archive *, void *, archive_open_callback *, archive_write_callback *, archive_close_callback *, archive_free_callback *); __LA_DECL int archive_write_open_fd(struct archive *, int _fd); __LA_DECL int archive_write_open_filename(struct archive *, const char *_file); __LA_DECL int archive_write_open_filename_w(struct archive *, const wchar_t *_file); /* A deprecated synonym for archive_write_open_filename() */ __LA_DECL int archive_write_open_file(struct archive *, const char *_file) __LA_DEPRECATED; __LA_DECL int archive_write_open_FILE(struct archive *, FILE *); /* _buffSize is the size of the buffer, _used refers to a variable that * will be updated after each write into the buffer. */ __LA_DECL int archive_write_open_memory(struct archive *, void *_buffer, size_t _buffSize, size_t *_used); /* * Note that the library will truncate writes beyond the size provided * to archive_write_header or pad if the provided data is short. */ __LA_DECL int archive_write_header(struct archive *, struct archive_entry *); __LA_DECL la_ssize_t archive_write_data(struct archive *, const void *, size_t); /* This interface is currently only available for archive_write_disk handles. */ __LA_DECL la_ssize_t archive_write_data_block(struct archive *, const void *, size_t, la_int64_t); __LA_DECL int archive_write_finish_entry(struct archive *); __LA_DECL int archive_write_close(struct archive *); /* Marks the archive as FATAL so that a subsequent free() operation * won't try to close() cleanly. Provides a fast abort capability * when the client discovers that things have gone wrong. */ __LA_DECL int archive_write_fail(struct archive *); /* This can fail if the archive wasn't already closed, in which case * archive_write_free() will implicitly call archive_write_close(). */ __LA_DECL int archive_write_free(struct archive *); #if ARCHIVE_VERSION_NUMBER < 4000000 /* Synonym for archive_write_free() for backwards compatibility. */ __LA_DECL int archive_write_finish(struct archive *) __LA_DEPRECATED; #endif /* * Set write options. */ /* Apply option to the format only. */ __LA_DECL int archive_write_set_format_option(struct archive *_a, const char *m, const char *o, const char *v); /* Apply option to the filter only. */ __LA_DECL int archive_write_set_filter_option(struct archive *_a, const char *m, const char *o, const char *v); /* Apply option to both the format and the filter. */ __LA_DECL int archive_write_set_option(struct archive *_a, const char *m, const char *o, const char *v); /* Apply option string to both the format and the filter. */ __LA_DECL int archive_write_set_options(struct archive *_a, const char *opts); /* * Set an encryption passphrase. */ __LA_DECL int archive_write_set_passphrase(struct archive *_a, const char *p); __LA_DECL int archive_write_set_passphrase_callback(struct archive *, void *client_data, archive_passphrase_callback *); /*- * ARCHIVE_WRITE_DISK API * * To create objects on disk: * 1) Ask archive_write_disk_new for a new archive_write_disk object. * 2) Set any global properties. In particular, you probably * want to set the options. * 3) For each entry: * - construct an appropriate struct archive_entry structure * - archive_write_header to create the file/dir/etc on disk * - archive_write_data to write the entry data * 4) archive_write_free to cleanup the writer and release resources * * In particular, you can use this in conjunction with archive_read() * to pull entries out of an archive and create them on disk. */ __LA_DECL struct archive *archive_write_disk_new(void); /* This file will not be overwritten. */ __LA_DECL int archive_write_disk_set_skip_file(struct archive *, la_int64_t, la_int64_t); /* Set flags to control how the next item gets created. * This accepts a bitmask of ARCHIVE_EXTRACT_XXX flags defined above. */ __LA_DECL int archive_write_disk_set_options(struct archive *, int flags); /* * The lookup functions are given uname/uid (or gname/gid) pairs and * return a uid (gid) suitable for this system. These are used for * restoring ownership and for setting ACLs. The default functions * are naive, they just return the uid/gid. These are small, so reasonable * for applications that don't need to preserve ownership; they * are probably also appropriate for applications that are doing * same-system backup and restore. */ /* * The "standard" lookup functions use common system calls to lookup * the uname/gname, falling back to the uid/gid if the names can't be * found. They cache lookups and are reasonably fast, but can be very * large, so they are not used unless you ask for them. In * particular, these match the specifications of POSIX "pax" and old * POSIX "tar". */ __LA_DECL int archive_write_disk_set_standard_lookup(struct archive *); /* * If neither the default (naive) nor the standard (big) functions suit * your needs, you can write your own and register them. Be sure to * include a cleanup function if you have allocated private data. */ __LA_DECL int archive_write_disk_set_group_lookup(struct archive *, void * /* private_data */, la_int64_t (*)(void *, const char *, la_int64_t), void (* /* cleanup */)(void *)); __LA_DECL int archive_write_disk_set_user_lookup(struct archive *, void * /* private_data */, la_int64_t (*)(void *, const char *, la_int64_t), void (* /* cleanup */)(void *)); __LA_DECL la_int64_t archive_write_disk_gid(struct archive *, const char *, la_int64_t); __LA_DECL la_int64_t archive_write_disk_uid(struct archive *, const char *, la_int64_t); /* * ARCHIVE_READ_DISK API * * This is still evolving and somewhat experimental. */ __LA_DECL struct archive *archive_read_disk_new(void); /* The names for symlink modes here correspond to an old BSD * command-line argument convention: -L, -P, -H */ /* Follow all symlinks. */ __LA_DECL int archive_read_disk_set_symlink_logical(struct archive *); /* Follow no symlinks. */ __LA_DECL int archive_read_disk_set_symlink_physical(struct archive *); /* Follow symlink initially, then not. */ __LA_DECL int archive_read_disk_set_symlink_hybrid(struct archive *); /* TODO: Handle Linux stat32/stat64 ugliness. */ __LA_DECL int archive_read_disk_entry_from_file(struct archive *, struct archive_entry *, int /* fd */, const struct stat *); /* Look up gname for gid or uname for uid. */ /* Default implementations are very, very stupid. */ __LA_DECL const char *archive_read_disk_gname(struct archive *, la_int64_t); __LA_DECL const char *archive_read_disk_uname(struct archive *, la_int64_t); /* "Standard" implementation uses getpwuid_r, getgrgid_r and caches the * results for performance. */ __LA_DECL int archive_read_disk_set_standard_lookup(struct archive *); /* You can install your own lookups if you like. */ __LA_DECL int archive_read_disk_set_gname_lookup(struct archive *, void * /* private_data */, const char *(* /* lookup_fn */)(void *, la_int64_t), void (* /* cleanup_fn */)(void *)); __LA_DECL int archive_read_disk_set_uname_lookup(struct archive *, void * /* private_data */, const char *(* /* lookup_fn */)(void *, la_int64_t), void (* /* cleanup_fn */)(void *)); /* Start traversal. */ __LA_DECL int archive_read_disk_open(struct archive *, const char *); __LA_DECL int archive_read_disk_open_w(struct archive *, const wchar_t *); /* * Request that current entry be visited. If you invoke it on every * directory, you'll get a physical traversal. This is ignored if the * current entry isn't a directory or a link to a directory. So, if * you invoke this on every returned path, you'll get a full logical * traversal. */ __LA_DECL int archive_read_disk_descend(struct archive *); __LA_DECL int archive_read_disk_can_descend(struct archive *); __LA_DECL int archive_read_disk_current_filesystem(struct archive *); __LA_DECL int archive_read_disk_current_filesystem_is_synthetic(struct archive *); __LA_DECL int archive_read_disk_current_filesystem_is_remote(struct archive *); /* Request that the access time of the entry visited by traversal be restored. */ __LA_DECL int archive_read_disk_set_atime_restored(struct archive *); /* * Set behavior. The "flags" argument selects optional behavior. */ /* Request that the access time of the entry visited by traversal be restored. * This is the same as archive_read_disk_set_atime_restored. */ #define ARCHIVE_READDISK_RESTORE_ATIME (0x0001) /* Default: Do not skip an entry which has nodump flags. */ #define ARCHIVE_READDISK_HONOR_NODUMP (0x0002) /* Default: Skip a mac resource fork file whose prefix is "._" because of * using copyfile. */ #define ARCHIVE_READDISK_MAC_COPYFILE (0x0004) /* Default: Traverse mount points. */ #define ARCHIVE_READDISK_NO_TRAVERSE_MOUNTS (0x0008) /* Default: Xattrs are read from disk. */ #define ARCHIVE_READDISK_NO_XATTR (0x0010) /* Default: ACLs are read from disk. */ #define ARCHIVE_READDISK_NO_ACL (0x0020) /* Default: File flags are read from disk. */ #define ARCHIVE_READDISK_NO_FFLAGS (0x0040) /* Default: Sparse file information is read from disk. */ #define ARCHIVE_READDISK_NO_SPARSE (0x0080) __LA_DECL int archive_read_disk_set_behavior(struct archive *, int flags); /* * Set archive_match object that will be used in archive_read_disk to * know whether an entry should be skipped. The callback function * _excluded_func will be invoked when an entry is skipped by the result * of archive_match. */ __LA_DECL int archive_read_disk_set_matching(struct archive *, struct archive *_matching, void (*_excluded_func) (struct archive *, void *, struct archive_entry *), void *_client_data); __LA_DECL int archive_read_disk_set_metadata_filter_callback(struct archive *, int (*_metadata_filter_func)(struct archive *, void *, struct archive_entry *), void *_client_data); /* Simplified cleanup interface; * This calls archive_read_free() or archive_write_free() as needed. */ __LA_DECL int archive_free(struct archive *); /* * Accessor functions to read/set various information in * the struct archive object: */ /* Number of filters in the current filter pipeline. */ /* Filter #0 is the one closest to the format, -1 is a synonym for the * last filter, which is always the pseudo-filter that wraps the * client callbacks. */ __LA_DECL int archive_filter_count(struct archive *); __LA_DECL la_int64_t archive_filter_bytes(struct archive *, int); __LA_DECL int archive_filter_code(struct archive *, int); __LA_DECL const char * archive_filter_name(struct archive *, int); #if ARCHIVE_VERSION_NUMBER < 4000000 /* These don't properly handle multiple filters, so are deprecated and * will eventually be removed. */ /* As of libarchive 3.0, this is an alias for archive_filter_bytes(a, -1); */ __LA_DECL la_int64_t archive_position_compressed(struct archive *) __LA_DEPRECATED; /* As of libarchive 3.0, this is an alias for archive_filter_bytes(a, 0); */ __LA_DECL la_int64_t archive_position_uncompressed(struct archive *) __LA_DEPRECATED; /* As of libarchive 3.0, this is an alias for archive_filter_name(a, 0); */ __LA_DECL const char *archive_compression_name(struct archive *) __LA_DEPRECATED; /* As of libarchive 3.0, this is an alias for archive_filter_code(a, 0); */ __LA_DECL int archive_compression(struct archive *) __LA_DEPRECATED; #endif /* Parses a date string relative to the current time. * NOTE: This is not intended for general date parsing, and the resulting timestamp should only be used for libarchive. */ __LA_DECL time_t archive_parse_date(time_t now, const char *datestr); __LA_DECL int archive_errno(struct archive *); __LA_DECL const char *archive_error_string(struct archive *); __LA_DECL const char *archive_format_name(struct archive *); __LA_DECL int archive_format(struct archive *); __LA_DECL void archive_clear_error(struct archive *); __LA_DECL void archive_set_error(struct archive *, int _err, const char *fmt, ...) __LA_PRINTF(3, 4); __LA_DECL void archive_copy_error(struct archive *dest, struct archive *src); __LA_DECL int archive_file_count(struct archive *); /* * ARCHIVE_MATCH API */ __LA_DECL struct archive *archive_match_new(void); __LA_DECL int archive_match_free(struct archive *); /* * Test if archive_entry is excluded. * This is a convenience function. This is the same as calling all * archive_match_path_excluded, archive_match_time_excluded * and archive_match_owner_excluded. */ __LA_DECL int archive_match_excluded(struct archive *, struct archive_entry *); /* * Test if pathname is excluded. The conditions are set by following functions. */ __LA_DECL int archive_match_path_excluded(struct archive *, struct archive_entry *); /* Control recursive inclusion of directory content when directory is included. Default on. */ __LA_DECL int archive_match_set_inclusion_recursion(struct archive *, int); /* Add exclusion pathname pattern. */ __LA_DECL int archive_match_exclude_pattern(struct archive *, const char *); __LA_DECL int archive_match_exclude_pattern_w(struct archive *, const wchar_t *); /* Add exclusion pathname pattern from file. */ __LA_DECL int archive_match_exclude_pattern_from_file(struct archive *, const char *, int _nullSeparator); __LA_DECL int archive_match_exclude_pattern_from_file_w(struct archive *, const wchar_t *, int _nullSeparator); /* Add inclusion pathname pattern. */ __LA_DECL int archive_match_include_pattern(struct archive *, const char *); __LA_DECL int archive_match_include_pattern_w(struct archive *, const wchar_t *); /* Add inclusion pathname pattern from file. */ __LA_DECL int archive_match_include_pattern_from_file(struct archive *, const char *, int _nullSeparator); __LA_DECL int archive_match_include_pattern_from_file_w(struct archive *, const wchar_t *, int _nullSeparator); /* * How to get statistic information for inclusion patterns. */ /* Return the amount number of unmatched inclusion patterns. */ __LA_DECL int archive_match_path_unmatched_inclusions(struct archive *); /* Return the pattern of unmatched inclusion with ARCHIVE_OK. * Return ARCHIVE_EOF if there is no inclusion pattern. */ __LA_DECL int archive_match_path_unmatched_inclusions_next( struct archive *, const char **); __LA_DECL int archive_match_path_unmatched_inclusions_next_w( struct archive *, const wchar_t **); /* * Test if a file is excluded by its time stamp. * The conditions are set by following functions. */ __LA_DECL int archive_match_time_excluded(struct archive *, struct archive_entry *); /* * Flags to tell a matching type of time stamps. These are used for * following functions. */ /* Time flag: mtime to be tested. */ #define ARCHIVE_MATCH_MTIME (0x0100) /* Time flag: ctime to be tested. */ #define ARCHIVE_MATCH_CTIME (0x0200) /* Comparison flag: Match the time if it is newer than. */ #define ARCHIVE_MATCH_NEWER (0x0001) /* Comparison flag: Match the time if it is older than. */ #define ARCHIVE_MATCH_OLDER (0x0002) /* Comparison flag: Match the time if it is equal to. */ #define ARCHIVE_MATCH_EQUAL (0x0010) /* Set inclusion time. */ __LA_DECL int archive_match_include_time(struct archive *, int _flag, time_t _sec, long _nsec); /* Set inclusion time by a date string. */ __LA_DECL int archive_match_include_date(struct archive *, int _flag, const char *_datestr); __LA_DECL int archive_match_include_date_w(struct archive *, int _flag, const wchar_t *_datestr); /* Set inclusion time by a particular file. */ __LA_DECL int archive_match_include_file_time(struct archive *, int _flag, const char *_pathname); __LA_DECL int archive_match_include_file_time_w(struct archive *, int _flag, const wchar_t *_pathname); /* Add exclusion entry. */ __LA_DECL int archive_match_exclude_entry(struct archive *, int _flag, struct archive_entry *); /* * Test if a file is excluded by its uid ,gid, uname or gname. * The conditions are set by following functions. */ __LA_DECL int archive_match_owner_excluded(struct archive *, struct archive_entry *); /* Add inclusion uid, gid, uname and gname. */ __LA_DECL int archive_match_include_uid(struct archive *, la_int64_t); __LA_DECL int archive_match_include_gid(struct archive *, la_int64_t); __LA_DECL int archive_match_include_uname(struct archive *, const char *); __LA_DECL int archive_match_include_uname_w(struct archive *, const wchar_t *); __LA_DECL int archive_match_include_gname(struct archive *, const char *); __LA_DECL int archive_match_include_gname_w(struct archive *, const wchar_t *); /* Utility functions */ #if ARCHIVE_VERSION_NUMBER < 4000000 /* Convenience function to sort a NULL terminated list of strings */ __LA_DECL int archive_utility_string_sort(char **); #endif #ifdef __cplusplus } #endif /* These are meaningless outside of this header. */ #undef __LA_DECL #endif /* !ARCHIVE_H_INCLUDED */ diff --git a/contrib/libarchive/libarchive/archive_acl.c b/contrib/libarchive/libarchive/archive_acl.c index 9e71f5ee5610..362e3308f43f 100644 --- a/contrib/libarchive/libarchive/archive_acl.c +++ b/contrib/libarchive/libarchive/archive_acl.c @@ -1,2113 +1,2136 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * Copyright (c) 2016 Martin Matuska * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_WCHAR_H #include #endif #include "archive_acl_private.h" #include "archive_entry.h" #include "archive_private.h" #undef max #define max(a, b) ((a)>(b)?(a):(b)) #ifndef HAVE_WMEMCMP /* Good enough for simple equality testing, but not for sorting. */ #define wmemcmp(a,b,i) memcmp((a), (b), (i) * sizeof(wchar_t)) #endif static int acl_special(struct archive_acl *acl, int type, int permset, int tag); static struct archive_acl_entry *acl_new_entry(struct archive_acl *acl, int type, int permset, int tag, int id); static int archive_acl_add_entry_len_l(struct archive_acl *acl, int type, int permset, int tag, int id, const char *name, size_t len, struct archive_string_conv *sc); static int archive_acl_text_want_type(struct archive_acl *acl, int flags); static size_t archive_acl_text_len(struct archive_acl *acl, int want_type, int flags, int wide, struct archive *a, struct archive_string_conv *sc); static int isint_w(const wchar_t *start, const wchar_t *end, int *result); static int ismode_w(const wchar_t *start, const wchar_t *end, int *result); static int is_nfs4_flags_w(const wchar_t *start, const wchar_t *end, int *result); static int is_nfs4_perms_w(const wchar_t *start, const wchar_t *end, int *result); static void next_field_w(const wchar_t **wp, const wchar_t **start, const wchar_t **end, wchar_t *sep); static void append_entry_w(wchar_t **wp, const wchar_t *prefix, int type, int tag, int flags, const wchar_t *wname, int perm, int id); static void append_id_w(wchar_t **wp, int id); static int isint(const char *start, const char *end, int *result); static int ismode(const char *start, const char *end, int *result); static int is_nfs4_flags(const char *start, const char *end, int *result); static int is_nfs4_perms(const char *start, const char *end, int *result); static void next_field(const char **p, size_t *l, const char **start, const char **end, char *sep); static void append_entry(char **p, const char *prefix, int type, int tag, int flags, const char *name, int perm, int id); static void append_id(char **p, int id); static const struct { const int perm; const char c; const wchar_t wc; } nfsv4_acl_perm_map[] = { { ARCHIVE_ENTRY_ACL_READ_DATA | ARCHIVE_ENTRY_ACL_LIST_DIRECTORY, 'r', L'r' }, { ARCHIVE_ENTRY_ACL_WRITE_DATA | ARCHIVE_ENTRY_ACL_ADD_FILE, 'w', L'w' }, { ARCHIVE_ENTRY_ACL_EXECUTE, 'x', L'x' }, { ARCHIVE_ENTRY_ACL_APPEND_DATA | ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY, 'p', L'p' }, { ARCHIVE_ENTRY_ACL_DELETE, 'd', L'd' }, { ARCHIVE_ENTRY_ACL_DELETE_CHILD, 'D', L'D' }, { ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES, 'a', L'a' }, { ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES, 'A', L'A' }, { ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS, 'R', L'R' }, { ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS, 'W', L'W' }, { ARCHIVE_ENTRY_ACL_READ_ACL, 'c', L'c' }, { ARCHIVE_ENTRY_ACL_WRITE_ACL, 'C', L'C' }, { ARCHIVE_ENTRY_ACL_WRITE_OWNER, 'o', L'o' }, { ARCHIVE_ENTRY_ACL_SYNCHRONIZE, 's', L's' } }; static const int nfsv4_acl_perm_map_size = (int)(sizeof(nfsv4_acl_perm_map) / sizeof(nfsv4_acl_perm_map[0])); static const struct { const int perm; const char c; const wchar_t wc; } nfsv4_acl_flag_map[] = { { ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, 'f', L'f' }, { ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT, 'd', L'd' }, { ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, 'i', L'i' }, { ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT, 'n', L'n' }, { ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS, 'S', L'S' }, { ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS, 'F', L'F' }, { ARCHIVE_ENTRY_ACL_ENTRY_INHERITED, 'I', L'I' } }; static const int nfsv4_acl_flag_map_size = (int)(sizeof(nfsv4_acl_flag_map) / sizeof(nfsv4_acl_flag_map[0])); void archive_acl_clear(struct archive_acl *acl) { struct archive_acl_entry *ap; while (acl->acl_head != NULL) { ap = acl->acl_head->next; archive_mstring_clean(&acl->acl_head->name); free(acl->acl_head); acl->acl_head = ap; } free(acl->acl_text_w); acl->acl_text_w = NULL; free(acl->acl_text); acl->acl_text = NULL; acl->acl_p = NULL; acl->acl_types = 0; acl->acl_state = 0; /* Not counting. */ } void archive_acl_copy(struct archive_acl *dest, struct archive_acl *src) { struct archive_acl_entry *ap, *ap2; archive_acl_clear(dest); dest->mode = src->mode; ap = src->acl_head; while (ap != NULL) { ap2 = acl_new_entry(dest, ap->type, ap->permset, ap->tag, ap->id); if (ap2 != NULL) archive_mstring_copy(&ap2->name, &ap->name); ap = ap->next; } } int archive_acl_add_entry(struct archive_acl *acl, int type, int permset, int tag, int id, const char *name) { struct archive_acl_entry *ap; if (acl_special(acl, type, permset, tag) == 0) return ARCHIVE_OK; ap = acl_new_entry(acl, type, permset, tag, id); if (ap == NULL) { /* XXX Error XXX */ return ARCHIVE_FAILED; } if (name != NULL && *name != '\0') archive_mstring_copy_mbs(&ap->name, name); else archive_mstring_clean(&ap->name); return ARCHIVE_OK; } int archive_acl_add_entry_w_len(struct archive_acl *acl, int type, int permset, int tag, int id, const wchar_t *name, size_t len) { struct archive_acl_entry *ap; if (acl_special(acl, type, permset, tag) == 0) return ARCHIVE_OK; ap = acl_new_entry(acl, type, permset, tag, id); if (ap == NULL) { /* XXX Error XXX */ return ARCHIVE_FAILED; } if (name != NULL && *name != L'\0' && len > 0) archive_mstring_copy_wcs_len(&ap->name, name, len); else archive_mstring_clean(&ap->name); return ARCHIVE_OK; } static int archive_acl_add_entry_len_l(struct archive_acl *acl, int type, int permset, int tag, int id, const char *name, size_t len, struct archive_string_conv *sc) { struct archive_acl_entry *ap; int r; if (acl_special(acl, type, permset, tag) == 0) return ARCHIVE_OK; ap = acl_new_entry(acl, type, permset, tag, id); if (ap == NULL) { /* XXX Error XXX */ return ARCHIVE_FAILED; } if (name != NULL && *name != '\0' && len > 0) { r = archive_mstring_copy_mbs_len_l(&ap->name, name, len, sc); } else { r = 0; archive_mstring_clean(&ap->name); } if (r == 0) return (ARCHIVE_OK); else if (errno == ENOMEM) return (ARCHIVE_FATAL); else return (ARCHIVE_WARN); } /* * If this ACL entry is part of the standard POSIX permissions set, * store the permissions in the stat structure and return zero. */ static int acl_special(struct archive_acl *acl, int type, int permset, int tag) { if (type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS && ((permset & ~007) == 0)) { switch (tag) { case ARCHIVE_ENTRY_ACL_USER_OBJ: acl->mode &= ~0700; acl->mode |= (permset & 7) << 6; return (0); case ARCHIVE_ENTRY_ACL_GROUP_OBJ: acl->mode &= ~0070; acl->mode |= (permset & 7) << 3; return (0); case ARCHIVE_ENTRY_ACL_OTHER: acl->mode &= ~0007; acl->mode |= permset & 7; return (0); } } return (1); } /* * Allocate and populate a new ACL entry with everything but the * name. */ static struct archive_acl_entry * acl_new_entry(struct archive_acl *acl, int type, int permset, int tag, int id) { struct archive_acl_entry *ap, *aq; + /* Reject an invalid type */ + switch (type) { + case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: + case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: + case ARCHIVE_ENTRY_ACL_TYPE_ALLOW: + case ARCHIVE_ENTRY_ACL_TYPE_DENY: + case ARCHIVE_ENTRY_ACL_TYPE_AUDIT: + case ARCHIVE_ENTRY_ACL_TYPE_ALARM: + break; + default: + return (NULL); + } + /* Type argument must be a valid NFS4 or POSIX.1e type. * The type must agree with anything already set and * the permset must be compatible. */ if (type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) { if (acl->acl_types & ~ARCHIVE_ENTRY_ACL_TYPE_NFS4) { return (NULL); } if (permset & ~(ARCHIVE_ENTRY_ACL_PERMS_NFS4 | ARCHIVE_ENTRY_ACL_INHERITANCE_NFS4)) { return (NULL); } } else if (type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) { if (acl->acl_types & ~ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) { return (NULL); } if (permset & ~ARCHIVE_ENTRY_ACL_PERMS_POSIX1E) { return (NULL); } } else { return (NULL); } /* Verify the tag is valid and compatible with NFS4 or POSIX.1e. */ switch (tag) { case ARCHIVE_ENTRY_ACL_USER: case ARCHIVE_ENTRY_ACL_USER_OBJ: case ARCHIVE_ENTRY_ACL_GROUP: case ARCHIVE_ENTRY_ACL_GROUP_OBJ: /* Tags valid in both NFS4 and POSIX.1e */ break; case ARCHIVE_ENTRY_ACL_MASK: case ARCHIVE_ENTRY_ACL_OTHER: /* Tags valid only in POSIX.1e. */ if (type & ~ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) { return (NULL); } break; case ARCHIVE_ENTRY_ACL_EVERYONE: /* Tags valid only in NFS4. */ if (type & ~ARCHIVE_ENTRY_ACL_TYPE_NFS4) { return (NULL); } break; default: /* No other values are valid. */ return (NULL); } free(acl->acl_text_w); acl->acl_text_w = NULL; free(acl->acl_text); acl->acl_text = NULL; /* * If there's a matching entry already in the list, overwrite it. * NFSv4 entries may be repeated and are not overwritten. * * TODO: compare names of no id is provided (needs more rework) */ ap = acl->acl_head; aq = NULL; while (ap != NULL) { if (((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) == 0) && ap->type == type && ap->tag == tag && ap->id == id) { if (id != -1 || (tag != ARCHIVE_ENTRY_ACL_USER && tag != ARCHIVE_ENTRY_ACL_GROUP)) { ap->permset = permset; return (ap); } } aq = ap; ap = ap->next; } /* Add a new entry to the end of the list. */ ap = calloc(1, sizeof(*ap)); if (ap == NULL) return (NULL); if (aq == NULL) acl->acl_head = ap; else aq->next = ap; ap->type = type; ap->tag = tag; ap->id = id; ap->permset = permset; acl->acl_types |= type; return (ap); } /* * Return a count of entries matching "want_type". */ int archive_acl_count(struct archive_acl *acl, int want_type) { int count; struct archive_acl_entry *ap; count = 0; ap = acl->acl_head; while (ap != NULL) { if ((ap->type & want_type) != 0) count++; ap = ap->next; } if (count > 0 && ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0)) count += 3; return (count); } /* * Return a bitmask of stored ACL types in an ACL list */ int archive_acl_types(struct archive_acl *acl) { return (acl->acl_types); } /* * Prepare for reading entries from the ACL data. Returns a count * of entries matching "want_type", or zero if there are no * non-extended ACL entries of that type. */ int archive_acl_reset(struct archive_acl *acl, int want_type) { int count, cutoff; count = archive_acl_count(acl, want_type); /* * If the only entries are the three standard ones, * then don't return any ACL data. (In this case, * client can just use chmod(2) to set permissions.) */ if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) cutoff = 3; else cutoff = 0; if (count > cutoff) acl->acl_state = ARCHIVE_ENTRY_ACL_USER_OBJ; else acl->acl_state = 0; acl->acl_p = acl->acl_head; return (count); } /* * Return the next ACL entry in the list. Fake entries for the * standard permissions and include them in the returned list. */ int archive_acl_next(struct archive *a, struct archive_acl *acl, int want_type, int *type, int *permset, int *tag, int *id, const char **name) { *name = NULL; *id = -1; /* * The acl_state is either zero (no entries available), -1 * (reading from list), or an entry type (retrieve that type * from ae_stat.aest_mode). */ if (acl->acl_state == 0) return (ARCHIVE_WARN); /* The first three access entries are special. */ if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { switch (acl->acl_state) { case ARCHIVE_ENTRY_ACL_USER_OBJ: *permset = (acl->mode >> 6) & 7; *type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; *tag = ARCHIVE_ENTRY_ACL_USER_OBJ; acl->acl_state = ARCHIVE_ENTRY_ACL_GROUP_OBJ; return (ARCHIVE_OK); case ARCHIVE_ENTRY_ACL_GROUP_OBJ: *permset = (acl->mode >> 3) & 7; *type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; *tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; acl->acl_state = ARCHIVE_ENTRY_ACL_OTHER; return (ARCHIVE_OK); case ARCHIVE_ENTRY_ACL_OTHER: *permset = acl->mode & 7; *type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; *tag = ARCHIVE_ENTRY_ACL_OTHER; acl->acl_state = -1; acl->acl_p = acl->acl_head; return (ARCHIVE_OK); default: break; } } while (acl->acl_p != NULL && (acl->acl_p->type & want_type) == 0) acl->acl_p = acl->acl_p->next; if (acl->acl_p == NULL) { acl->acl_state = 0; *type = 0; *permset = 0; *tag = 0; *id = -1; *name = NULL; return (ARCHIVE_EOF); /* End of ACL entries. */ } *type = acl->acl_p->type; *permset = acl->acl_p->permset; *tag = acl->acl_p->tag; *id = acl->acl_p->id; if (archive_mstring_get_mbs(a, &acl->acl_p->name, name) != 0) { if (errno == ENOMEM) return (ARCHIVE_FATAL); *name = NULL; } acl->acl_p = acl->acl_p->next; return (ARCHIVE_OK); } /* * Determine what type of ACL do we want */ static int archive_acl_text_want_type(struct archive_acl *acl, int flags) { int want_type; /* Check if ACL is NFSv4 */ if ((acl->acl_types & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { /* NFSv4 should never mix with POSIX.1e */ if ((acl->acl_types & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) return (0); else return (ARCHIVE_ENTRY_ACL_TYPE_NFS4); } /* Now deal with POSIX.1e ACLs */ want_type = 0; if ((flags & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) want_type |= ARCHIVE_ENTRY_ACL_TYPE_ACCESS; if ((flags & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) want_type |= ARCHIVE_ENTRY_ACL_TYPE_DEFAULT; /* By default we want both access and default ACLs */ if (want_type == 0) return (ARCHIVE_ENTRY_ACL_TYPE_POSIX1E); return (want_type); } /* * Calculate ACL text string length */ static size_t archive_acl_text_len(struct archive_acl *acl, int want_type, int flags, int wide, struct archive *a, struct archive_string_conv *sc) { struct archive_acl_entry *ap; const char *name; const wchar_t *wname; int count, idlen, tmp, r; size_t length; size_t len; count = 0; length = 0; for (ap = acl->acl_head; ap != NULL; ap = ap->next) { if ((ap->type & want_type) == 0) continue; /* * Filemode-mapping ACL entries are stored exclusively in * ap->mode so they should not be in the list */ if ((ap->type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS) && (ap->tag == ARCHIVE_ENTRY_ACL_USER_OBJ || ap->tag == ARCHIVE_ENTRY_ACL_GROUP_OBJ || ap->tag == ARCHIVE_ENTRY_ACL_OTHER)) continue; count++; if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0 && (ap->type & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) length += 8; /* "default:" */ switch (ap->tag) { case ARCHIVE_ENTRY_ACL_USER_OBJ: if (want_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { length += 6; /* "owner@" */ break; } /* FALLTHROUGH */ case ARCHIVE_ENTRY_ACL_USER: case ARCHIVE_ENTRY_ACL_MASK: length += 4; /* "user", "mask" */ break; case ARCHIVE_ENTRY_ACL_GROUP_OBJ: if (want_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { length += 6; /* "group@" */ break; } /* FALLTHROUGH */ case ARCHIVE_ENTRY_ACL_GROUP: case ARCHIVE_ENTRY_ACL_OTHER: length += 5; /* "group", "other" */ break; case ARCHIVE_ENTRY_ACL_EVERYONE: length += 9; /* "everyone@" */ break; } length += 1; /* colon after tag */ if (ap->tag == ARCHIVE_ENTRY_ACL_USER || ap->tag == ARCHIVE_ENTRY_ACL_GROUP) { if (wide) { r = archive_mstring_get_wcs(a, &ap->name, &wname); if (r == 0 && wname != NULL) length += wcslen(wname); else if (r < 0 && errno == ENOMEM) return (0); else length += sizeof(uid_t) * 3 + 1; } else { r = archive_mstring_get_mbs_l(a, &ap->name, &name, &len, sc); if (r != 0) return (0); if (len > 0 && name != NULL) length += len; else length += sizeof(uid_t) * 3 + 1; } length += 1; /* colon after user or group name */ } else if (want_type != ARCHIVE_ENTRY_ACL_TYPE_NFS4) length += 1; /* 2nd colon empty user,group or other */ if (((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) != 0) && ((want_type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) && (ap->tag == ARCHIVE_ENTRY_ACL_OTHER || ap->tag == ARCHIVE_ENTRY_ACL_MASK)) { /* Solaris has no colon after other: and mask: */ length = length - 1; } if (want_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { /* rwxpdDaARWcCos:fdinSFI:deny */ length += 27; if ((ap->type & ARCHIVE_ENTRY_ACL_TYPE_DENY) == 0) length += 1; /* allow, alarm, audit */ } else length += 3; /* rwx */ if ((ap->tag == ARCHIVE_ENTRY_ACL_USER || ap->tag == ARCHIVE_ENTRY_ACL_GROUP) && (flags & ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID) != 0) { length += 1; /* colon */ /* ID digit count */ idlen = 1; tmp = ap->id; while (tmp > 9) { tmp = tmp / 10; idlen++; } length += idlen; } length ++; /* entry separator */ } /* Add filemode-mapping access entries to the length */ if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { if ((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) != 0) { /* "user::rwx\ngroup::rwx\nother:rwx\n" */ length += 31; } else { /* "user::rwx\ngroup::rwx\nother::rwx\n" */ length += 32; } } else if (count == 0) return (0); /* The terminating character is included in count */ return (length); } /* * Generate a wide text version of the ACL. The flags parameter controls * the type and style of the generated ACL. */ wchar_t * archive_acl_to_text_w(struct archive_acl *acl, ssize_t *text_len, int flags, struct archive *a) { int count; size_t length; size_t len; const wchar_t *wname; const wchar_t *prefix; wchar_t separator; struct archive_acl_entry *ap; int id, r, want_type; wchar_t *wp, *ws; want_type = archive_acl_text_want_type(acl, flags); /* Both NFSv4 and POSIX.1 types found */ if (want_type == 0) return (NULL); if (want_type == ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) flags |= ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT; length = archive_acl_text_len(acl, want_type, flags, 1, a, NULL); if (length == 0) return (NULL); if (flags & ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA) separator = L','; else separator = L'\n'; /* Now, allocate the string and actually populate it. */ wp = ws = malloc(length * sizeof(*wp)); if (wp == NULL) { if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } count = 0; if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { append_entry_w(&wp, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_USER_OBJ, flags, NULL, acl->mode & 0700, -1); *wp++ = separator; append_entry_w(&wp, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_GROUP_OBJ, flags, NULL, acl->mode & 0070, -1); *wp++ = separator; append_entry_w(&wp, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_OTHER, flags, NULL, acl->mode & 0007, -1); count += 3; } for (ap = acl->acl_head; ap != NULL; ap = ap->next) { if ((ap->type & want_type) == 0) continue; /* * Filemode-mapping ACL entries are stored exclusively in * ap->mode so they should not be in the list */ if ((ap->type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS) && (ap->tag == ARCHIVE_ENTRY_ACL_USER_OBJ || ap->tag == ARCHIVE_ENTRY_ACL_GROUP_OBJ || ap->tag == ARCHIVE_ENTRY_ACL_OTHER)) continue; if (ap->type == ARCHIVE_ENTRY_ACL_TYPE_DEFAULT && (flags & ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT) != 0) prefix = L"default:"; else prefix = NULL; r = archive_mstring_get_wcs(a, &ap->name, &wname); if (r == 0) { if (count > 0) *wp++ = separator; if (flags & ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID) id = ap->id; else id = -1; append_entry_w(&wp, prefix, ap->type, ap->tag, flags, wname, ap->permset, id); count++; } else if (r < 0 && errno == ENOMEM) { free(ws); return (NULL); } } /* Add terminating character */ *wp++ = L'\0'; len = wcslen(ws); if (len > length - 1) __archive_errx(1, "Buffer overrun"); if (text_len != NULL) *text_len = len; return (ws); } static void append_id_w(wchar_t **wp, int id) { if (id < 0) id = 0; if (id > 9) append_id_w(wp, id / 10); *(*wp)++ = L"0123456789"[id % 10]; } static void append_entry_w(wchar_t **wp, const wchar_t *prefix, int type, int tag, int flags, const wchar_t *wname, int perm, int id) { int i; if (prefix != NULL) { wcscpy(*wp, prefix); *wp += wcslen(*wp); } switch (tag) { case ARCHIVE_ENTRY_ACL_USER_OBJ: wname = NULL; id = -1; if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { wcscpy(*wp, L"owner@"); break; } /* FALLTHROUGH */ case ARCHIVE_ENTRY_ACL_USER: wcscpy(*wp, L"user"); break; case ARCHIVE_ENTRY_ACL_GROUP_OBJ: wname = NULL; id = -1; if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { wcscpy(*wp, L"group@"); break; } /* FALLTHROUGH */ case ARCHIVE_ENTRY_ACL_GROUP: wcscpy(*wp, L"group"); break; case ARCHIVE_ENTRY_ACL_MASK: wcscpy(*wp, L"mask"); wname = NULL; id = -1; break; case ARCHIVE_ENTRY_ACL_OTHER: wcscpy(*wp, L"other"); wname = NULL; id = -1; break; case ARCHIVE_ENTRY_ACL_EVERYONE: wcscpy(*wp, L"everyone@"); wname = NULL; id = -1; break; + default: + **wp = '\0'; + break; } *wp += wcslen(*wp); *(*wp)++ = L':'; if (((type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) || tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { if (wname != NULL) { wcscpy(*wp, wname); *wp += wcslen(*wp); } else if (tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { append_id_w(wp, id); if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) == 0) id = -1; } /* Solaris style has no second colon after other and mask */ if (((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) == 0) || (tag != ARCHIVE_ENTRY_ACL_OTHER && tag != ARCHIVE_ENTRY_ACL_MASK)) *(*wp)++ = L':'; } if ((type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) { /* POSIX.1e ACL perms */ *(*wp)++ = (perm & 0444) ? L'r' : L'-'; *(*wp)++ = (perm & 0222) ? L'w' : L'-'; *(*wp)++ = (perm & 0111) ? L'x' : L'-'; } else { /* NFSv4 ACL perms */ for (i = 0; i < nfsv4_acl_perm_map_size; i++) { if (perm & nfsv4_acl_perm_map[i].perm) *(*wp)++ = nfsv4_acl_perm_map[i].wc; else if ((flags & ARCHIVE_ENTRY_ACL_STYLE_COMPACT) == 0) *(*wp)++ = L'-'; } *(*wp)++ = L':'; for (i = 0; i < nfsv4_acl_flag_map_size; i++) { if (perm & nfsv4_acl_flag_map[i].perm) *(*wp)++ = nfsv4_acl_flag_map[i].wc; else if ((flags & ARCHIVE_ENTRY_ACL_STYLE_COMPACT) == 0) *(*wp)++ = L'-'; } *(*wp)++ = L':'; switch (type) { case ARCHIVE_ENTRY_ACL_TYPE_ALLOW: wcscpy(*wp, L"allow"); break; case ARCHIVE_ENTRY_ACL_TYPE_DENY: wcscpy(*wp, L"deny"); break; case ARCHIVE_ENTRY_ACL_TYPE_AUDIT: wcscpy(*wp, L"audit"); break; case ARCHIVE_ENTRY_ACL_TYPE_ALARM: wcscpy(*wp, L"alarm"); break; default: + *(*wp) = L'\0'; break; } *wp += wcslen(*wp); } if (id != -1) { *(*wp)++ = L':'; append_id_w(wp, id); } } /* * Generate a text version of the ACL. The flags parameter controls * the type and style of the generated ACL. */ char * archive_acl_to_text_l(struct archive_acl *acl, ssize_t *text_len, int flags, struct archive_string_conv *sc) { int count; size_t length; size_t len; const char *name; const char *prefix; char separator; struct archive_acl_entry *ap; int id, r, want_type; char *p, *s; want_type = archive_acl_text_want_type(acl, flags); /* Both NFSv4 and POSIX.1 types found */ if (want_type == 0) return (NULL); if (want_type == ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) flags |= ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT; length = archive_acl_text_len(acl, want_type, flags, 0, NULL, sc); if (length == 0) return (NULL); if (flags & ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA) separator = ','; else separator = '\n'; /* Now, allocate the string and actually populate it. */ p = s = malloc(length * sizeof(*p)); if (p == NULL) { if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } count = 0; if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { append_entry(&p, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_USER_OBJ, flags, NULL, acl->mode & 0700, -1); *p++ = separator; append_entry(&p, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_GROUP_OBJ, flags, NULL, acl->mode & 0070, -1); *p++ = separator; append_entry(&p, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_OTHER, flags, NULL, acl->mode & 0007, -1); count += 3; } for (ap = acl->acl_head; ap != NULL; ap = ap->next) { if ((ap->type & want_type) == 0) continue; /* * Filemode-mapping ACL entries are stored exclusively in * ap->mode so they should not be in the list */ if ((ap->type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS) && (ap->tag == ARCHIVE_ENTRY_ACL_USER_OBJ || ap->tag == ARCHIVE_ENTRY_ACL_GROUP_OBJ || ap->tag == ARCHIVE_ENTRY_ACL_OTHER)) continue; if (ap->type == ARCHIVE_ENTRY_ACL_TYPE_DEFAULT && (flags & ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT) != 0) prefix = "default:"; else prefix = NULL; r = archive_mstring_get_mbs_l( NULL, &ap->name, &name, &len, sc); if (r != 0) { free(s); return (NULL); } if (count > 0) *p++ = separator; if (name == NULL || (flags & ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID)) { id = ap->id; } else { id = -1; } append_entry(&p, prefix, ap->type, ap->tag, flags, name, ap->permset, id); count++; } /* Add terminating character */ *p++ = '\0'; len = strlen(s); if (len > length - 1) __archive_errx(1, "Buffer overrun"); if (text_len != NULL) *text_len = len; return (s); } static void append_id(char **p, int id) { if (id < 0) id = 0; if (id > 9) append_id(p, id / 10); *(*p)++ = "0123456789"[id % 10]; } static void append_entry(char **p, const char *prefix, int type, int tag, int flags, const char *name, int perm, int id) { int i; if (prefix != NULL) { strcpy(*p, prefix); *p += strlen(*p); } switch (tag) { case ARCHIVE_ENTRY_ACL_USER_OBJ: name = NULL; id = -1; if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { strcpy(*p, "owner@"); break; } /* FALLTHROUGH */ case ARCHIVE_ENTRY_ACL_USER: strcpy(*p, "user"); break; case ARCHIVE_ENTRY_ACL_GROUP_OBJ: name = NULL; id = -1; if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { strcpy(*p, "group@"); break; } /* FALLTHROUGH */ case ARCHIVE_ENTRY_ACL_GROUP: strcpy(*p, "group"); break; case ARCHIVE_ENTRY_ACL_MASK: strcpy(*p, "mask"); name = NULL; id = -1; break; case ARCHIVE_ENTRY_ACL_OTHER: strcpy(*p, "other"); name = NULL; id = -1; break; case ARCHIVE_ENTRY_ACL_EVERYONE: strcpy(*p, "everyone@"); name = NULL; id = -1; break; + default: + **p = '\0'; + break; } *p += strlen(*p); *(*p)++ = ':'; if (((type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) || tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { if (name != NULL) { strcpy(*p, name); *p += strlen(*p); } else if (tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { append_id(p, id); if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) == 0) id = -1; } /* Solaris style has no second colon after other and mask */ if (((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) == 0) || (tag != ARCHIVE_ENTRY_ACL_OTHER && tag != ARCHIVE_ENTRY_ACL_MASK)) *(*p)++ = ':'; } if ((type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) { /* POSIX.1e ACL perms */ *(*p)++ = (perm & 0444) ? 'r' : '-'; *(*p)++ = (perm & 0222) ? 'w' : '-'; *(*p)++ = (perm & 0111) ? 'x' : '-'; } else { /* NFSv4 ACL perms */ for (i = 0; i < nfsv4_acl_perm_map_size; i++) { if (perm & nfsv4_acl_perm_map[i].perm) *(*p)++ = nfsv4_acl_perm_map[i].c; else if ((flags & ARCHIVE_ENTRY_ACL_STYLE_COMPACT) == 0) *(*p)++ = '-'; } *(*p)++ = ':'; for (i = 0; i < nfsv4_acl_flag_map_size; i++) { if (perm & nfsv4_acl_flag_map[i].perm) *(*p)++ = nfsv4_acl_flag_map[i].c; else if ((flags & ARCHIVE_ENTRY_ACL_STYLE_COMPACT) == 0) *(*p)++ = '-'; } *(*p)++ = ':'; switch (type) { case ARCHIVE_ENTRY_ACL_TYPE_ALLOW: strcpy(*p, "allow"); break; case ARCHIVE_ENTRY_ACL_TYPE_DENY: strcpy(*p, "deny"); break; case ARCHIVE_ENTRY_ACL_TYPE_AUDIT: strcpy(*p, "audit"); break; case ARCHIVE_ENTRY_ACL_TYPE_ALARM: strcpy(*p, "alarm"); break; + default: + *(*p) = '\0'; + break; } *p += strlen(*p); } if (id != -1) { *(*p)++ = ':'; append_id(p, id); } } /* * Parse a wide ACL text string. * * The want_type argument may be one of the following: * ARCHIVE_ENTRY_ACL_TYPE_ACCESS - text is a POSIX.1e ACL of type ACCESS * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT - text is a POSIX.1e ACL of type DEFAULT * ARCHIVE_ENTRY_ACL_TYPE_NFS4 - text is as a NFSv4 ACL * * POSIX.1e ACL entries prefixed with "default:" are treated as * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT unless type is ARCHIVE_ENTRY_ACL_TYPE_NFS4 */ int archive_acl_from_text_w(struct archive_acl *acl, const wchar_t *text, int want_type) { struct { const wchar_t *start; const wchar_t *end; } field[6], name; const wchar_t *s, *st; int numfields, fields, n, r, sol, ret; int type, types, tag, permset, id; size_t len; wchar_t sep; ret = ARCHIVE_OK; types = 0; switch (want_type) { case ARCHIVE_ENTRY_ACL_TYPE_POSIX1E: want_type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; __LA_FALLTHROUGH; case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: numfields = 5; break; case ARCHIVE_ENTRY_ACL_TYPE_NFS4: numfields = 6; break; default: return (ARCHIVE_FATAL); } while (text != NULL && *text != L'\0') { /* * Parse the fields out of the next entry, * advance 'text' to start of next entry. */ fields = 0; do { const wchar_t *start, *end; next_field_w(&text, &start, &end, &sep); if (fields < numfields) { field[fields].start = start; field[fields].end = end; } ++fields; } while (sep == L':'); /* Set remaining fields to blank. */ for (n = fields; n < numfields; ++n) field[n].start = field[n].end = NULL; if (field[0].start == NULL || field[0].end == NULL) { /* This should never happen */ return (ARCHIVE_FATAL); } if (*(field[0].start) == L'#') { /* Comment, skip entry */ continue; } n = 0; sol = 0; id = -1; permset = 0; name.start = name.end = NULL; if (want_type != ARCHIVE_ENTRY_ACL_TYPE_NFS4) { /* POSIX.1e ACLs */ /* * Default keyword "default:user::rwx" * if found, we have one more field * * We also support old Solaris extension: * "defaultuser::rwx" is the default ACL corresponding * to "user::rwx", etc. valid only for first field */ s = field[0].start; len = field[0].end - field[0].start; if (*s == L'd' && (len == 1 || (len >= 7 && wmemcmp((s + 1), L"efault", 6) == 0))) { type = ARCHIVE_ENTRY_ACL_TYPE_DEFAULT; if (len > 7) field[0].start += 7; else n = 1; } else type = want_type; /* Check for a numeric ID in field n+1 or n+3. */ isint_w(field[n + 1].start, field[n + 1].end, &id); /* Field n+3 is optional. */ if (id == -1 && fields > n+3) isint_w(field[n + 3].start, field[n + 3].end, &id); tag = 0; s = field[n].start; st = field[n].start + 1; len = field[n].end - field[n].start; switch (*s) { case L'u': if (len == 1 || (len == 4 && wmemcmp(st, L"ser", 3) == 0)) tag = ARCHIVE_ENTRY_ACL_USER_OBJ; break; case L'g': if (len == 1 || (len == 5 && wmemcmp(st, L"roup", 4) == 0)) tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; break; case L'o': if (len == 1 || (len == 5 && wmemcmp(st, L"ther", 4) == 0)) tag = ARCHIVE_ENTRY_ACL_OTHER; break; case L'm': if (len == 1 || (len == 4 && wmemcmp(st, L"ask", 3) == 0)) tag = ARCHIVE_ENTRY_ACL_MASK; break; default: break; } switch (tag) { case ARCHIVE_ENTRY_ACL_OTHER: case ARCHIVE_ENTRY_ACL_MASK: if (fields == (n + 2) && field[n + 1].start < field[n + 1].end && ismode_w(field[n + 1].start, field[n + 1].end, &permset)) { /* This is Solaris-style "other:rwx" */ sol = 1; } else if (fields == (n + 3) && field[n + 1].start < field[n + 1].end) { /* Invalid mask or other field */ ret = ARCHIVE_WARN; continue; } break; case ARCHIVE_ENTRY_ACL_USER_OBJ: case ARCHIVE_ENTRY_ACL_GROUP_OBJ: if (id != -1 || field[n + 1].start < field[n + 1].end) { name = field[n + 1]; if (tag == ARCHIVE_ENTRY_ACL_USER_OBJ) tag = ARCHIVE_ENTRY_ACL_USER; else tag = ARCHIVE_ENTRY_ACL_GROUP; } break; default: /* Invalid tag, skip entry */ ret = ARCHIVE_WARN; continue; } /* * Without "default:" we expect mode in field 2 * Exception: Solaris other and mask fields */ if (permset == 0 && !ismode_w(field[n + 2 - sol].start, field[n + 2 - sol].end, &permset)) { /* Invalid mode, skip entry */ ret = ARCHIVE_WARN; continue; } } else { /* NFS4 ACLs */ s = field[0].start; len = field[0].end - field[0].start; tag = 0; switch (len) { case 4: if (wmemcmp(s, L"user", 4) == 0) tag = ARCHIVE_ENTRY_ACL_USER; break; case 5: if (wmemcmp(s, L"group", 5) == 0) tag = ARCHIVE_ENTRY_ACL_GROUP; break; case 6: if (wmemcmp(s, L"owner@", 6) == 0) tag = ARCHIVE_ENTRY_ACL_USER_OBJ; else if (wmemcmp(s, L"group@", len) == 0) tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; break; case 9: if (wmemcmp(s, L"everyone@", 9) == 0) tag = ARCHIVE_ENTRY_ACL_EVERYONE; default: break; } if (tag == 0) { /* Invalid tag, skip entry */ ret = ARCHIVE_WARN; continue; } else if (tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { n = 1; name = field[1]; isint_w(name.start, name.end, &id); } else n = 0; if (!is_nfs4_perms_w(field[1 + n].start, field[1 + n].end, &permset)) { /* Invalid NFSv4 perms, skip entry */ ret = ARCHIVE_WARN; continue; } if (!is_nfs4_flags_w(field[2 + n].start, field[2 + n].end, &permset)) { /* Invalid NFSv4 flags, skip entry */ ret = ARCHIVE_WARN; continue; } s = field[3 + n].start; len = field[3 + n].end - field[3 + n].start; type = 0; if (len == 4) { if (wmemcmp(s, L"deny", 4) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_DENY; } else if (len == 5) { if (wmemcmp(s, L"allow", 5) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_ALLOW; else if (wmemcmp(s, L"audit", 5) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_AUDIT; else if (wmemcmp(s, L"alarm", 5) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_ALARM; } if (type == 0) { /* Invalid entry type, skip entry */ ret = ARCHIVE_WARN; continue; } isint_w(field[4 + n].start, field[4 + n].end, &id); } /* Add entry to the internal list. */ r = archive_acl_add_entry_w_len(acl, type, permset, tag, id, name.start, name.end - name.start); if (r < ARCHIVE_WARN) return (r); if (r != ARCHIVE_OK) ret = ARCHIVE_WARN; types |= type; } /* Reset ACL */ archive_acl_reset(acl, types); return (ret); } /* * Parse a string to a positive decimal integer. Returns true if * the string is non-empty and consists only of decimal digits, * false otherwise. */ static int isint_w(const wchar_t *start, const wchar_t *end, int *result) { int n = 0; if (start >= end) return (0); while (start < end) { if (*start < L'0' || *start > L'9') return (0); if (n > (INT_MAX / 10) || (n == INT_MAX / 10 && (*start - L'0') > INT_MAX % 10)) { n = INT_MAX; } else { n *= 10; n += *start - L'0'; } start++; } *result = n; return (1); } /* * Parse a string as a mode field. Returns true if * the string is non-empty and consists only of mode characters, * false otherwise. */ static int ismode_w(const wchar_t *start, const wchar_t *end, int *permset) { const wchar_t *p; if (start >= end) return (0); p = start; *permset = 0; while (p < end) { switch (*p++) { case L'r': case L'R': *permset |= ARCHIVE_ENTRY_ACL_READ; break; case L'w': case L'W': *permset |= ARCHIVE_ENTRY_ACL_WRITE; break; case L'x': case L'X': *permset |= ARCHIVE_ENTRY_ACL_EXECUTE; break; case L'-': break; default: return (0); } } return (1); } /* * Parse a string as a NFS4 ACL permission field. * Returns true if the string is non-empty and consists only of NFS4 ACL * permission characters, false otherwise */ static int is_nfs4_perms_w(const wchar_t *start, const wchar_t *end, int *permset) { const wchar_t *p = start; while (p < end) { switch (*p++) { case L'r': *permset |= ARCHIVE_ENTRY_ACL_READ_DATA; break; case L'w': *permset |= ARCHIVE_ENTRY_ACL_WRITE_DATA; break; case L'x': *permset |= ARCHIVE_ENTRY_ACL_EXECUTE; break; case L'p': *permset |= ARCHIVE_ENTRY_ACL_APPEND_DATA; break; case L'D': *permset |= ARCHIVE_ENTRY_ACL_DELETE_CHILD; break; case L'd': *permset |= ARCHIVE_ENTRY_ACL_DELETE; break; case L'a': *permset |= ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES; break; case L'A': *permset |= ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES; break; case L'R': *permset |= ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS; break; case L'W': *permset |= ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS; break; case L'c': *permset |= ARCHIVE_ENTRY_ACL_READ_ACL; break; case L'C': *permset |= ARCHIVE_ENTRY_ACL_WRITE_ACL; break; case L'o': *permset |= ARCHIVE_ENTRY_ACL_WRITE_OWNER; break; case L's': *permset |= ARCHIVE_ENTRY_ACL_SYNCHRONIZE; break; case L'-': break; default: return(0); } } return (1); } /* * Parse a string as a NFS4 ACL flags field. * Returns true if the string is non-empty and consists only of NFS4 ACL * flag characters, false otherwise */ static int is_nfs4_flags_w(const wchar_t *start, const wchar_t *end, int *permset) { const wchar_t *p = start; while (p < end) { switch(*p++) { case L'f': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT; break; case L'd': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT; break; case L'i': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY; break; case L'n': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT; break; case L'S': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS; break; case L'F': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS; break; case L'I': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_INHERITED; break; case L'-': break; default: return (0); } } return (1); } /* * Match "[:whitespace:]*(.*)[:whitespace:]*[:,\n]". *wp is updated * to point to just after the separator. *start points to the first * character of the matched text and *end just after the last * character of the matched identifier. In particular *end - *start * is the length of the field body, not including leading or trailing * whitespace. */ static void next_field_w(const wchar_t **wp, const wchar_t **start, const wchar_t **end, wchar_t *sep) { /* Skip leading whitespace to find start of field. */ while (**wp == L' ' || **wp == L'\t' || **wp == L'\n') { (*wp)++; } *start = *wp; /* Scan for the separator. */ while (**wp != L'\0' && **wp != L',' && **wp != L':' && **wp != L'\n' && **wp != L'#') { (*wp)++; } *sep = **wp; /* Locate end of field, trim trailing whitespace if necessary */ if (*wp == *start) { *end = *wp; } else { *end = *wp - 1; while (**end == L' ' || **end == L'\t' || **end == L'\n') { (*end)--; } (*end)++; } /* Handle in-field comments */ if (*sep == L'#') { while (**wp != L'\0' && **wp != L',' && **wp != L'\n') { (*wp)++; } *sep = **wp; } /* Adjust scanner location. */ if (**wp != L'\0') (*wp)++; } /* * Parse an ACL text string. * * The want_type argument may be one of the following: * ARCHIVE_ENTRY_ACL_TYPE_ACCESS - text is a POSIX.1e ACL of type ACCESS * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT - text is a POSIX.1e ACL of type DEFAULT * ARCHIVE_ENTRY_ACL_TYPE_NFS4 - text is as a NFSv4 ACL * * POSIX.1e ACL entries prefixed with "default:" are treated as * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT unless type is ARCHIVE_ENTRY_ACL_TYPE_NFS4 */ int archive_acl_from_text_l(struct archive_acl *acl, const char *text, int want_type, struct archive_string_conv *sc) { return archive_acl_from_text_nl(acl, text, strlen(text), want_type, sc); } int archive_acl_from_text_nl(struct archive_acl *acl, const char *text, size_t length, int want_type, struct archive_string_conv *sc) { struct { const char *start; const char *end; } field[6], name; const char *s, *st; int numfields, fields, n, r, sol, ret; int type, types, tag, permset, id; size_t len; char sep; switch (want_type) { case ARCHIVE_ENTRY_ACL_TYPE_POSIX1E: want_type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; __LA_FALLTHROUGH; case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: numfields = 5; break; case ARCHIVE_ENTRY_ACL_TYPE_NFS4: numfields = 6; break; default: return (ARCHIVE_FATAL); } ret = ARCHIVE_OK; types = 0; while (text != NULL && length > 0 && *text != '\0') { /* * Parse the fields out of the next entry, * advance 'text' to start of next entry. */ fields = 0; do { const char *start, *end; next_field(&text, &length, &start, &end, &sep); if (fields < numfields) { field[fields].start = start; field[fields].end = end; } ++fields; } while (sep == ':'); /* Set remaining fields to blank. */ for (n = fields; n < numfields; ++n) field[n].start = field[n].end = NULL; if (field[0].start == NULL || field[0].end == NULL) { /* This should never happen */ return (ARCHIVE_FATAL); } if (*(field[0].start) == '#') { /* Comment, skip entry */ continue; } n = 0; sol = 0; id = -1; permset = 0; name.start = name.end = NULL; if (want_type != ARCHIVE_ENTRY_ACL_TYPE_NFS4) { /* POSIX.1e ACLs */ /* * Default keyword "default:user::rwx" * if found, we have one more field * * We also support old Solaris extension: * "defaultuser::rwx" is the default ACL corresponding * to "user::rwx", etc. valid only for first field */ s = field[0].start; len = field[0].end - field[0].start; if (*s == 'd' && (len == 1 || (len >= 7 && memcmp((s + 1), "efault", 6) == 0))) { type = ARCHIVE_ENTRY_ACL_TYPE_DEFAULT; if (len > 7) field[0].start += 7; else n = 1; } else type = want_type; /* Check for a numeric ID in field n+1 or n+3. */ isint(field[n + 1].start, field[n + 1].end, &id); /* Field n+3 is optional. */ if (id == -1 && fields > (n + 3)) isint(field[n + 3].start, field[n + 3].end, &id); tag = 0; s = field[n].start; st = field[n].start + 1; len = field[n].end - field[n].start; if (len == 0) { ret = ARCHIVE_WARN; continue; } switch (*s) { case 'u': if (len == 1 || (len == 4 && memcmp(st, "ser", 3) == 0)) tag = ARCHIVE_ENTRY_ACL_USER_OBJ; break; case 'g': if (len == 1 || (len == 5 && memcmp(st, "roup", 4) == 0)) tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; break; case 'o': if (len == 1 || (len == 5 && memcmp(st, "ther", 4) == 0)) tag = ARCHIVE_ENTRY_ACL_OTHER; break; case 'm': if (len == 1 || (len == 4 && memcmp(st, "ask", 3) == 0)) tag = ARCHIVE_ENTRY_ACL_MASK; break; default: break; } switch (tag) { case ARCHIVE_ENTRY_ACL_OTHER: case ARCHIVE_ENTRY_ACL_MASK: if (fields == (n + 2) && field[n + 1].start < field[n + 1].end && ismode(field[n + 1].start, field[n + 1].end, &permset)) { /* This is Solaris-style "other:rwx" */ sol = 1; } else if (fields == (n + 3) && field[n + 1].start < field[n + 1].end) { /* Invalid mask or other field */ ret = ARCHIVE_WARN; continue; } break; case ARCHIVE_ENTRY_ACL_USER_OBJ: case ARCHIVE_ENTRY_ACL_GROUP_OBJ: if (id != -1 || field[n + 1].start < field[n + 1].end) { name = field[n + 1]; if (tag == ARCHIVE_ENTRY_ACL_USER_OBJ) tag = ARCHIVE_ENTRY_ACL_USER; else tag = ARCHIVE_ENTRY_ACL_GROUP; } break; default: /* Invalid tag, skip entry */ ret = ARCHIVE_WARN; continue; } /* * Without "default:" we expect mode in field 3 * Exception: Solaris other and mask fields */ if (permset == 0 && !ismode(field[n + 2 - sol].start, field[n + 2 - sol].end, &permset)) { /* Invalid mode, skip entry */ ret = ARCHIVE_WARN; continue; } } else { /* NFS4 ACLs */ s = field[0].start; len = field[0].end - field[0].start; tag = 0; switch (len) { case 4: if (memcmp(s, "user", 4) == 0) tag = ARCHIVE_ENTRY_ACL_USER; break; case 5: if (memcmp(s, "group", 5) == 0) tag = ARCHIVE_ENTRY_ACL_GROUP; break; case 6: if (memcmp(s, "owner@", 6) == 0) tag = ARCHIVE_ENTRY_ACL_USER_OBJ; else if (memcmp(s, "group@", 6) == 0) tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; break; case 9: if (memcmp(s, "everyone@", 9) == 0) tag = ARCHIVE_ENTRY_ACL_EVERYONE; break; default: break; } if (tag == 0) { /* Invalid tag, skip entry */ ret = ARCHIVE_WARN; continue; } else if (tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { n = 1; name = field[1]; isint(name.start, name.end, &id); } else n = 0; if (!is_nfs4_perms(field[1 + n].start, field[1 + n].end, &permset)) { /* Invalid NFSv4 perms, skip entry */ ret = ARCHIVE_WARN; continue; } if (!is_nfs4_flags(field[2 + n].start, field[2 + n].end, &permset)) { /* Invalid NFSv4 flags, skip entry */ ret = ARCHIVE_WARN; continue; } s = field[3 + n].start; len = field[3 + n].end - field[3 + n].start; type = 0; if (len == 4) { if (memcmp(s, "deny", 4) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_DENY; } else if (len == 5) { if (memcmp(s, "allow", 5) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_ALLOW; else if (memcmp(s, "audit", 5) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_AUDIT; else if (memcmp(s, "alarm", 5) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_ALARM; } if (type == 0) { /* Invalid entry type, skip entry */ ret = ARCHIVE_WARN; continue; } isint(field[4 + n].start, field[4 + n].end, &id); } /* Add entry to the internal list. */ r = archive_acl_add_entry_len_l(acl, type, permset, tag, id, name.start, name.end - name.start, sc); if (r < ARCHIVE_WARN) return (r); if (r != ARCHIVE_OK) ret = ARCHIVE_WARN; types |= type; } /* Reset ACL */ archive_acl_reset(acl, types); return (ret); } /* * Parse a string to a positive decimal integer. Returns true if * the string is non-empty and consists only of decimal digits, * false otherwise. */ static int isint(const char *start, const char *end, int *result) { int n = 0; if (start >= end) return (0); while (start < end) { if (*start < '0' || *start > '9') return (0); if (n > (INT_MAX / 10) || (n == INT_MAX / 10 && (*start - '0') > INT_MAX % 10)) { n = INT_MAX; } else { n *= 10; n += *start - '0'; } start++; } *result = n; return (1); } /* * Parse a string as a mode field. Returns true if * the string is non-empty and consists only of mode characters, * false otherwise. */ static int ismode(const char *start, const char *end, int *permset) { const char *p; if (start >= end) return (0); p = start; *permset = 0; while (p < end) { switch (*p++) { case 'r': case 'R': *permset |= ARCHIVE_ENTRY_ACL_READ; break; case 'w': case 'W': *permset |= ARCHIVE_ENTRY_ACL_WRITE; break; case 'x': case 'X': *permset |= ARCHIVE_ENTRY_ACL_EXECUTE; break; case '-': break; default: return (0); } } return (1); } /* * Parse a string as a NFS4 ACL permission field. * Returns true if the string is non-empty and consists only of NFS4 ACL * permission characters, false otherwise */ static int is_nfs4_perms(const char *start, const char *end, int *permset) { const char *p = start; while (p < end) { switch (*p++) { case 'r': *permset |= ARCHIVE_ENTRY_ACL_READ_DATA; break; case 'w': *permset |= ARCHIVE_ENTRY_ACL_WRITE_DATA; break; case 'x': *permset |= ARCHIVE_ENTRY_ACL_EXECUTE; break; case 'p': *permset |= ARCHIVE_ENTRY_ACL_APPEND_DATA; break; case 'D': *permset |= ARCHIVE_ENTRY_ACL_DELETE_CHILD; break; case 'd': *permset |= ARCHIVE_ENTRY_ACL_DELETE; break; case 'a': *permset |= ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES; break; case 'A': *permset |= ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES; break; case 'R': *permset |= ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS; break; case 'W': *permset |= ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS; break; case 'c': *permset |= ARCHIVE_ENTRY_ACL_READ_ACL; break; case 'C': *permset |= ARCHIVE_ENTRY_ACL_WRITE_ACL; break; case 'o': *permset |= ARCHIVE_ENTRY_ACL_WRITE_OWNER; break; case 's': *permset |= ARCHIVE_ENTRY_ACL_SYNCHRONIZE; break; case '-': break; default: return(0); } } return (1); } /* * Parse a string as a NFS4 ACL flags field. * Returns true if the string is non-empty and consists only of NFS4 ACL * flag characters, false otherwise */ static int is_nfs4_flags(const char *start, const char *end, int *permset) { const char *p = start; while (p < end) { switch(*p++) { case 'f': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT; break; case 'd': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT; break; case 'i': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY; break; case 'n': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT; break; case 'S': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS; break; case 'F': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS; break; case 'I': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_INHERITED; break; case '-': break; default: return (0); } } return (1); } /* * Match "[:whitespace:]*(.*)[:whitespace:]*[:,\n]". *p is updated * to point to just after the separator. *start points to the first * character of the matched text and *end just after the last * character of the matched identifier. In particular *end - *start * is the length of the field body, not including leading or trailing * whitespace. */ static void next_field(const char **p, size_t *l, const char **start, const char **end, char *sep) { /* Skip leading whitespace to find start of field. */ while (*l > 0 && (**p == ' ' || **p == '\t' || **p == '\n')) { (*p)++; (*l)--; } *start = *p; /* Locate end of field, trim trailing whitespace if necessary */ while (*l > 0 && **p != ' ' && **p != '\t' && **p != '\n' && **p != ',' && **p != ':' && **p != '#') { (*p)++; (*l)--; } *end = *p; /* Scan for the separator. */ while (*l > 0 && **p != ',' && **p != ':' && **p != '\n' && **p != '#') { (*p)++; (*l)--; } *sep = **p; /* Handle in-field comments */ if (*sep == '#') { while (*l > 0 && **p != ',' && **p != '\n') { (*p)++; (*l)--; } *sep = **p; } /* Skip separator. */ if (*l > 0) { (*p)++; (*l)--; } } diff --git a/contrib/libarchive/libarchive/archive_check_magic.c b/contrib/libarchive/libarchive/archive_check_magic.c index d12f0c496e27..6b8e0c5595f4 100644 --- a/contrib/libarchive/libarchive/archive_check_magic.c +++ b/contrib/libarchive/libarchive/archive_check_magic.c @@ -1,175 +1,182 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_TYPES_H #include #endif #include +#include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #if defined(_WIN32) && !defined(__CYGWIN__) #include #include #endif #include "archive_private.h" static void errmsg(const char *m) { size_t s = strlen(m); ssize_t written; while (s > 0) { written = write(2, m, s); - if (written <= 0) + if (written == 0) return; + if (written < 0) + { + if (errno == EINTR) + continue; + return; + } m += written; s -= written; } } static __LA_NORETURN void diediedie(void) { #if defined(_WIN32) && !defined(__CYGWIN__) && defined(_DEBUG) /* Cause a breakpoint exception */ DebugBreak(); #endif abort(); /* Terminate the program abnormally. */ } static const char * state_name(unsigned s) { switch (s) { case ARCHIVE_STATE_NEW: return ("new"); case ARCHIVE_STATE_HEADER: return ("header"); case ARCHIVE_STATE_DATA: return ("data"); case ARCHIVE_STATE_EOF: return ("eof"); case ARCHIVE_STATE_CLOSED: return ("closed"); case ARCHIVE_STATE_FATAL: return ("fatal"); default: return ("??"); } } static const char * archive_handle_type_name(unsigned m) { switch (m) { case ARCHIVE_WRITE_MAGIC: return ("archive_write"); case ARCHIVE_READ_MAGIC: return ("archive_read"); case ARCHIVE_WRITE_DISK_MAGIC: return ("archive_write_disk"); case ARCHIVE_READ_DISK_MAGIC: return ("archive_read_disk"); case ARCHIVE_MATCH_MAGIC: return ("archive_match"); default: return NULL; } } static void write_all_states(char *buff, unsigned int states) { unsigned int lowbit; *buff = '\0'; /* A trick for computing the lowest set bit. */ while ((lowbit = states & (1 + ~states)) != 0) { states &= ~lowbit; /* Clear the low bit. */ strcat(buff, state_name(lowbit)); if (states != 0) strcat(buff, "/"); } } /* * Check magic value and current state. * Magic value mismatches are fatal and result in calls to abort(). * State mismatches return ARCHIVE_FATAL. * Otherwise, returns ARCHIVE_OK. * * This is designed to catch serious programming errors that violate * the libarchive API. */ int __archive_check_magic(struct archive *a, unsigned int magic, unsigned int state, const char *function) { char states1[64]; char states2[64]; const char *handle_type; /* * If this isn't some form of archive handle, * then the library user has screwed up so bad that * we don't even have a reliable way to report an error. */ handle_type = archive_handle_type_name(a->magic); if (!handle_type) { errmsg("PROGRAMMER ERROR: Function "); errmsg(function); errmsg(" invoked with invalid archive handle.\n"); diediedie(); } if (a->magic != magic) { archive_set_error(a, -1, "PROGRAMMER ERROR: Function '%s' invoked" " on '%s' archive object, which is not supported.", function, handle_type); a->state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } if ((a->state & state) == 0) { /* If we're already FATAL, don't overwrite the error. */ if (a->state != ARCHIVE_STATE_FATAL) { write_all_states(states1, a->state); write_all_states(states2, state); archive_set_error(a, -1, "INTERNAL ERROR: Function '%s' invoked with" " archive structure in state '%s'," " should be in state '%s'", function, states1, states2); } a->state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } diff --git a/contrib/libarchive/libarchive/archive_cryptor.c b/contrib/libarchive/libarchive/archive_cryptor.c index 1825af4dc510..9f03f9ca6dd0 100644 --- a/contrib/libarchive/libarchive/archive_cryptor.c +++ b/contrib/libarchive/libarchive/archive_cryptor.c @@ -1,534 +1,534 @@ /*- * Copyright (c) 2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_cryptor_private.h" /* * On systems that do not support any recognized crypto libraries, * this file will normally define no usable symbols. * * But some compilers and linkers choke on empty object files, so * define a public symbol that will always exist. This could * be removed someday if this file gains another always-present * symbol definition. */ int __libarchive_cryptor_build_hack(void) { return 0; } #ifdef ARCHIVE_CRYPTOR_USE_Apple_CommonCrypto static int pbkdf2_sha1(const char *pw, size_t pw_len, const uint8_t *salt, size_t salt_len, unsigned rounds, uint8_t *derived_key, size_t derived_key_len) { CCKeyDerivationPBKDF(kCCPBKDF2, (const char *)pw, pw_len, salt, salt_len, kCCPRFHmacAlgSHA1, rounds, derived_key, derived_key_len); return 0; } #elif defined(_WIN32) && !defined(__CYGWIN__) && defined(HAVE_BCRYPT_H) && _WIN32_WINNT >= _WIN32_WINNT_VISTA #ifdef _MSC_VER #pragma comment(lib, "Bcrypt.lib") #endif static int pbkdf2_sha1(const char *pw, size_t pw_len, const uint8_t *salt, size_t salt_len, unsigned rounds, uint8_t *derived_key, size_t derived_key_len) { NTSTATUS status; BCRYPT_ALG_HANDLE hAlg; status = BCryptOpenAlgorithmProvider(&hAlg, BCRYPT_SHA1_ALGORITHM, MS_PRIMITIVE_PROVIDER, BCRYPT_ALG_HANDLE_HMAC_FLAG); if (!BCRYPT_SUCCESS(status)) return -1; status = BCryptDeriveKeyPBKDF2(hAlg, (PUCHAR)(uintptr_t)pw, (ULONG)pw_len, (PUCHAR)(uintptr_t)salt, (ULONG)salt_len, rounds, (PUCHAR)derived_key, (ULONG)derived_key_len, 0); BCryptCloseAlgorithmProvider(hAlg, 0); return (BCRYPT_SUCCESS(status)) ? 0: -1; } #elif defined(HAVE_LIBMBEDCRYPTO) && defined(HAVE_MBEDTLS_PKCS5_H) static int pbkdf2_sha1(const char *pw, size_t pw_len, const uint8_t *salt, size_t salt_len, unsigned rounds, uint8_t *derived_key, size_t derived_key_len) { mbedtls_md_context_t ctx; const mbedtls_md_info_t *info; int ret; mbedtls_md_init(&ctx); info = mbedtls_md_info_from_type(MBEDTLS_MD_SHA1); if (info == NULL) { mbedtls_md_free(&ctx); return (-1); } ret = mbedtls_md_setup(&ctx, info, 1); if (ret != 0) { mbedtls_md_free(&ctx); return (-1); } ret = mbedtls_pkcs5_pbkdf2_hmac(&ctx, (const unsigned char *)pw, pw_len, salt, salt_len, rounds, derived_key_len, derived_key); mbedtls_md_free(&ctx); return (ret); } #elif defined(HAVE_LIBNETTLE) && defined(HAVE_NETTLE_PBKDF2_H) static int pbkdf2_sha1(const char *pw, size_t pw_len, const uint8_t *salt, size_t salt_len, unsigned rounds, uint8_t *derived_key, size_t derived_key_len) { pbkdf2_hmac_sha1((unsigned)pw_len, (const uint8_t *)pw, rounds, salt_len, salt, derived_key_len, derived_key); return 0; } #elif defined(HAVE_LIBCRYPTO) && defined(HAVE_PKCS5_PBKDF2_HMAC_SHA1) static int pbkdf2_sha1(const char *pw, size_t pw_len, const uint8_t *salt, size_t salt_len, unsigned rounds, uint8_t *derived_key, size_t derived_key_len) { PKCS5_PBKDF2_HMAC_SHA1(pw, pw_len, salt, salt_len, rounds, derived_key_len, derived_key); return 0; } #else /* Stub */ static int pbkdf2_sha1(const char *pw, size_t pw_len, const uint8_t *salt, size_t salt_len, unsigned rounds, uint8_t *derived_key, size_t derived_key_len) { (void)pw; /* UNUSED */ (void)pw_len; /* UNUSED */ (void)salt; /* UNUSED */ (void)salt_len; /* UNUSED */ (void)rounds; /* UNUSED */ (void)derived_key; /* UNUSED */ (void)derived_key_len; /* UNUSED */ - return -1; /* UNSUPPORTED */ + return CRYPTOR_STUB_FUNCTION; /* UNSUPPORTED */ } #endif #ifdef ARCHIVE_CRYPTOR_USE_Apple_CommonCrypto # if MAC_OS_X_VERSION_MAX_ALLOWED < 1090 # define kCCAlgorithmAES kCCAlgorithmAES128 # endif static int aes_ctr_init(archive_crypto_ctx *ctx, const uint8_t *key, size_t key_len) { CCCryptorStatus r; ctx->key_len = key_len; memcpy(ctx->key, key, key_len); memset(ctx->nonce, 0, sizeof(ctx->nonce)); ctx->encr_pos = AES_BLOCK_SIZE; r = CCCryptorCreateWithMode(kCCEncrypt, kCCModeECB, kCCAlgorithmAES, ccNoPadding, NULL, key, key_len, NULL, 0, 0, 0, &ctx->ctx); return (r == kCCSuccess)? 0: -1; } static int aes_ctr_encrypt_counter(archive_crypto_ctx *ctx) { CCCryptorRef ref = ctx->ctx; CCCryptorStatus r; r = CCCryptorReset(ref, NULL); if (r != kCCSuccess && r != kCCUnimplemented) return -1; r = CCCryptorUpdate(ref, ctx->nonce, AES_BLOCK_SIZE, ctx->encr_buf, AES_BLOCK_SIZE, NULL); return (r == kCCSuccess)? 0: -1; } static int aes_ctr_release(archive_crypto_ctx *ctx) { memset(ctx->key, 0, ctx->key_len); memset(ctx->nonce, 0, sizeof(ctx->nonce)); return 0; } #elif defined(_WIN32) && !defined(__CYGWIN__) && defined(HAVE_BCRYPT_H) && _WIN32_WINNT >= _WIN32_WINNT_VISTA static int aes_ctr_init(archive_crypto_ctx *ctx, const uint8_t *key, size_t key_len) { BCRYPT_ALG_HANDLE hAlg; BCRYPT_KEY_HANDLE hKey; DWORD keyObj_len, aes_key_len; PBYTE keyObj; ULONG result; NTSTATUS status; BCRYPT_KEY_LENGTHS_STRUCT key_lengths; ctx->hAlg = NULL; ctx->hKey = NULL; ctx->keyObj = NULL; switch (key_len) { case 16: aes_key_len = 128; break; case 24: aes_key_len = 192; break; case 32: aes_key_len = 256; break; default: return -1; } status = BCryptOpenAlgorithmProvider(&hAlg, BCRYPT_AES_ALGORITHM, MS_PRIMITIVE_PROVIDER, 0); if (!BCRYPT_SUCCESS(status)) return -1; status = BCryptGetProperty(hAlg, BCRYPT_KEY_LENGTHS, (PUCHAR)&key_lengths, sizeof(key_lengths), &result, 0); if (!BCRYPT_SUCCESS(status)) { BCryptCloseAlgorithmProvider(hAlg, 0); return -1; } if (key_lengths.dwMinLength > aes_key_len || key_lengths.dwMaxLength < aes_key_len) { BCryptCloseAlgorithmProvider(hAlg, 0); return -1; } status = BCryptGetProperty(hAlg, BCRYPT_OBJECT_LENGTH, (PUCHAR)&keyObj_len, sizeof(keyObj_len), &result, 0); if (!BCRYPT_SUCCESS(status)) { BCryptCloseAlgorithmProvider(hAlg, 0); return -1; } keyObj = (PBYTE)HeapAlloc(GetProcessHeap(), 0, keyObj_len); if (keyObj == NULL) { BCryptCloseAlgorithmProvider(hAlg, 0); return -1; } status = BCryptSetProperty(hAlg, BCRYPT_CHAINING_MODE, (PUCHAR)BCRYPT_CHAIN_MODE_ECB, sizeof(BCRYPT_CHAIN_MODE_ECB), 0); if (!BCRYPT_SUCCESS(status)) { BCryptCloseAlgorithmProvider(hAlg, 0); HeapFree(GetProcessHeap(), 0, keyObj); return -1; } status = BCryptGenerateSymmetricKey(hAlg, &hKey, keyObj, keyObj_len, (PUCHAR)(uintptr_t)key, (ULONG)key_len, 0); if (!BCRYPT_SUCCESS(status)) { BCryptCloseAlgorithmProvider(hAlg, 0); HeapFree(GetProcessHeap(), 0, keyObj); return -1; } ctx->hAlg = hAlg; ctx->hKey = hKey; ctx->keyObj = keyObj; ctx->keyObj_len = keyObj_len; ctx->encr_pos = AES_BLOCK_SIZE; return 0; } static int aes_ctr_encrypt_counter(archive_crypto_ctx *ctx) { NTSTATUS status; ULONG result; status = BCryptEncrypt(ctx->hKey, (PUCHAR)ctx->nonce, AES_BLOCK_SIZE, NULL, NULL, 0, (PUCHAR)ctx->encr_buf, AES_BLOCK_SIZE, &result, 0); return BCRYPT_SUCCESS(status) ? 0 : -1; } static int aes_ctr_release(archive_crypto_ctx *ctx) { if (ctx->hAlg != NULL) { BCryptCloseAlgorithmProvider(ctx->hAlg, 0); ctx->hAlg = NULL; BCryptDestroyKey(ctx->hKey); ctx->hKey = NULL; HeapFree(GetProcessHeap(), 0, ctx->keyObj); ctx->keyObj = NULL; } memset(ctx, 0, sizeof(*ctx)); return 0; } #elif defined(HAVE_LIBMBEDCRYPTO) && defined(HAVE_MBEDTLS_AES_H) static int aes_ctr_init(archive_crypto_ctx *ctx, const uint8_t *key, size_t key_len) { mbedtls_aes_init(&ctx->ctx); ctx->key_len = key_len; memcpy(ctx->key, key, key_len); memset(ctx->nonce, 0, sizeof(ctx->nonce)); ctx->encr_pos = AES_BLOCK_SIZE; return 0; } static int aes_ctr_encrypt_counter(archive_crypto_ctx *ctx) { if (mbedtls_aes_setkey_enc(&ctx->ctx, ctx->key, ctx->key_len * 8) != 0) return (-1); if (mbedtls_aes_crypt_ecb(&ctx->ctx, MBEDTLS_AES_ENCRYPT, ctx->nonce, ctx->encr_buf) != 0) return (-1); return 0; } static int aes_ctr_release(archive_crypto_ctx *ctx) { mbedtls_aes_free(&ctx->ctx); memset(ctx, 0, sizeof(*ctx)); return 0; } #elif defined(HAVE_LIBNETTLE) && defined(HAVE_NETTLE_AES_H) static int aes_ctr_init(archive_crypto_ctx *ctx, const uint8_t *key, size_t key_len) { ctx->key_len = key_len; memcpy(ctx->key, key, key_len); memset(ctx->nonce, 0, sizeof(ctx->nonce)); ctx->encr_pos = AES_BLOCK_SIZE; memset(&ctx->ctx, 0, sizeof(ctx->ctx)); return 0; } static int aes_ctr_encrypt_counter(archive_crypto_ctx *ctx) { #if NETTLE_VERSION_MAJOR < 3 aes_set_encrypt_key(&ctx->ctx, ctx->key_len, ctx->key); aes_encrypt(&ctx->ctx, AES_BLOCK_SIZE, ctx->encr_buf, ctx->nonce); #else switch(ctx->key_len) { case AES128_KEY_SIZE: aes128_set_encrypt_key(&ctx->ctx.c128, ctx->key); aes128_encrypt(&ctx->ctx.c128, AES_BLOCK_SIZE, ctx->encr_buf, ctx->nonce); break; case AES192_KEY_SIZE: aes192_set_encrypt_key(&ctx->ctx.c192, ctx->key); aes192_encrypt(&ctx->ctx.c192, AES_BLOCK_SIZE, ctx->encr_buf, ctx->nonce); break; case AES256_KEY_SIZE: aes256_set_encrypt_key(&ctx->ctx.c256, ctx->key); aes256_encrypt(&ctx->ctx.c256, AES_BLOCK_SIZE, ctx->encr_buf, ctx->nonce); break; default: return -1; break; } #endif return 0; } static int aes_ctr_release(archive_crypto_ctx *ctx) { memset(ctx, 0, sizeof(*ctx)); return 0; } #elif defined(HAVE_LIBCRYPTO) static int aes_ctr_init(archive_crypto_ctx *ctx, const uint8_t *key, size_t key_len) { if ((ctx->ctx = EVP_CIPHER_CTX_new()) == NULL) return -1; switch (key_len) { case 16: ctx->type = EVP_aes_128_ecb(); break; case 24: ctx->type = EVP_aes_192_ecb(); break; case 32: ctx->type = EVP_aes_256_ecb(); break; default: ctx->type = NULL; return -1; } ctx->key_len = key_len; memcpy(ctx->key, key, key_len); memset(ctx->nonce, 0, sizeof(ctx->nonce)); ctx->encr_pos = AES_BLOCK_SIZE; return 0; } static int aes_ctr_encrypt_counter(archive_crypto_ctx *ctx) { int outl = 0; int r; r = EVP_EncryptInit_ex(ctx->ctx, ctx->type, NULL, ctx->key, NULL); if (r == 0) return -1; r = EVP_EncryptUpdate(ctx->ctx, ctx->encr_buf, &outl, ctx->nonce, AES_BLOCK_SIZE); if (r == 0 || outl != AES_BLOCK_SIZE) return -1; return 0; } static int aes_ctr_release(archive_crypto_ctx *ctx) { EVP_CIPHER_CTX_free(ctx->ctx); OPENSSL_cleanse(ctx->key, ctx->key_len); OPENSSL_cleanse(ctx->nonce, sizeof(ctx->nonce)); return 0; } #else #define ARCHIVE_CRYPTOR_STUB /* Stub */ static int aes_ctr_init(archive_crypto_ctx *ctx, const uint8_t *key, size_t key_len) { (void)ctx; /* UNUSED */ (void)key; /* UNUSED */ (void)key_len; /* UNUSED */ - return -1; + return CRYPTOR_STUB_FUNCTION; } static int aes_ctr_encrypt_counter(archive_crypto_ctx *ctx) { (void)ctx; /* UNUSED */ - return -1; + return CRYPTOR_STUB_FUNCTION; } static int aes_ctr_release(archive_crypto_ctx *ctx) { (void)ctx; /* UNUSED */ return 0; } #endif #ifdef ARCHIVE_CRYPTOR_STUB static int aes_ctr_update(archive_crypto_ctx *ctx, const uint8_t * const in, size_t in_len, uint8_t * const out, size_t *out_len) { (void)ctx; /* UNUSED */ (void)in; /* UNUSED */ (void)in_len; /* UNUSED */ (void)out; /* UNUSED */ (void)out_len; /* UNUSED */ aes_ctr_encrypt_counter(ctx); /* UNUSED */ /* Fix unused function warning */ - return -1; + return CRYPTOR_STUB_FUNCTION; } #else static void aes_ctr_increase_counter(archive_crypto_ctx *ctx) { uint8_t *const nonce = ctx->nonce; int j; for (j = 0; j < 8; j++) { if (++nonce[j]) break; } } static int aes_ctr_update(archive_crypto_ctx *ctx, const uint8_t * const in, size_t in_len, uint8_t * const out, size_t *out_len) { uint8_t *const ebuf = ctx->encr_buf; unsigned pos = ctx->encr_pos; unsigned max = (unsigned)((in_len < *out_len)? in_len: *out_len); unsigned i; for (i = 0; i < max; ) { if (pos == AES_BLOCK_SIZE) { aes_ctr_increase_counter(ctx); if (aes_ctr_encrypt_counter(ctx) != 0) return -1; while (max -i >= AES_BLOCK_SIZE) { for (pos = 0; pos < AES_BLOCK_SIZE; pos++) out[i+pos] = in[i+pos] ^ ebuf[pos]; i += AES_BLOCK_SIZE; aes_ctr_increase_counter(ctx); if (aes_ctr_encrypt_counter(ctx) != 0) return -1; } pos = 0; if (i >= max) break; } out[i] = in[i] ^ ebuf[pos++]; i++; } ctx->encr_pos = pos; *out_len = i; return 0; } #endif /* ARCHIVE_CRYPTOR_STUB */ const struct archive_cryptor __archive_cryptor = { &pbkdf2_sha1, &aes_ctr_init, &aes_ctr_update, &aes_ctr_release, &aes_ctr_init, &aes_ctr_update, &aes_ctr_release, }; diff --git a/contrib/libarchive/libarchive/archive_cryptor_private.h b/contrib/libarchive/libarchive/archive_cryptor_private.h index 4b3c6c161433..1dbc5c17a01a 100644 --- a/contrib/libarchive/libarchive/archive_cryptor_private.h +++ b/contrib/libarchive/libarchive/archive_cryptor_private.h @@ -1,196 +1,205 @@ /*- * Copyright (c) 2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #ifndef ARCHIVE_CRYPTOR_PRIVATE_H_INCLUDED #define ARCHIVE_CRYPTOR_PRIVATE_H_INCLUDED #ifndef __LIBARCHIVE_BUILD #error This header is only to be used internally to libarchive. #endif /* * On systems that do not support any recognized crypto libraries, * the archive_cryptor.c file will normally define no usable symbols. * * But some compilers and linkers choke on empty object files, so * define a public symbol that will always exist. This could * be removed someday if this file gains another always-present * symbol definition. */ int __libarchive_cryptor_build_hack(void); #ifdef __APPLE__ # include # if MAC_OS_X_VERSION_MAX_ALLOWED >= 1080 -# define ARCHIVE_CRYPTOR_USE_Apple_CommonCrypto +# define ARCHIVE_CRYPTOR_USE_Apple_CommonCrypto 1 # endif #endif #ifdef ARCHIVE_CRYPTOR_USE_Apple_CommonCrypto #include #include #define AES_BLOCK_SIZE 16 #define AES_MAX_KEY_SIZE kCCKeySizeAES256 typedef struct { CCCryptorRef ctx; uint8_t key[AES_MAX_KEY_SIZE]; unsigned key_len; uint8_t nonce[AES_BLOCK_SIZE]; uint8_t encr_buf[AES_BLOCK_SIZE]; unsigned encr_pos; } archive_crypto_ctx; #elif defined(_WIN32) && !defined(__CYGWIN__) && defined(HAVE_BCRYPT_H) && _WIN32_WINNT >= _WIN32_WINNT_VISTA #include #define ARCHIVE_CRYPTOR_USE_CNG 1 /* Common in other bcrypt implementations, but missing from VS2008. */ #ifndef BCRYPT_SUCCESS #define BCRYPT_SUCCESS(r) ((NTSTATUS)(r) == STATUS_SUCCESS) #endif #define AES_MAX_KEY_SIZE 32 #define AES_BLOCK_SIZE 16 typedef struct { BCRYPT_ALG_HANDLE hAlg; BCRYPT_KEY_HANDLE hKey; PBYTE keyObj; DWORD keyObj_len; uint8_t nonce[AES_BLOCK_SIZE]; uint8_t encr_buf[AES_BLOCK_SIZE]; unsigned encr_pos; } archive_crypto_ctx; #elif defined(HAVE_LIBMBEDCRYPTO) && defined(HAVE_MBEDTLS_AES_H) #include #include #include #define ARCHIVE_CRYPTOR_USE_MBED 1 #define AES_MAX_KEY_SIZE 32 #define AES_BLOCK_SIZE 16 typedef struct { mbedtls_aes_context ctx; uint8_t key[AES_MAX_KEY_SIZE]; unsigned key_len; uint8_t nonce[AES_BLOCK_SIZE]; uint8_t encr_buf[AES_BLOCK_SIZE]; unsigned encr_pos; } archive_crypto_ctx; #elif defined(HAVE_LIBNETTLE) && defined(HAVE_NETTLE_AES_H) #if defined(HAVE_NETTLE_PBKDF2_H) #include #endif #include #include #define ARCHIVE_CRYPTOR_USE_NETTLE 1 typedef struct { #if NETTLE_VERSION_MAJOR < 3 struct aes_ctx ctx; #else union { struct aes128_ctx c128; struct aes192_ctx c192; struct aes256_ctx c256; } ctx; #endif uint8_t key[AES_MAX_KEY_SIZE]; unsigned key_len; uint8_t nonce[AES_BLOCK_SIZE]; uint8_t encr_buf[AES_BLOCK_SIZE]; unsigned encr_pos; } archive_crypto_ctx; #elif defined(HAVE_LIBCRYPTO) #include "archive_openssl_evp_private.h" #define ARCHIVE_CRYPTOR_USE_OPENSSL 1 #define AES_BLOCK_SIZE 16 #define AES_MAX_KEY_SIZE 32 typedef struct { EVP_CIPHER_CTX *ctx; const EVP_CIPHER *type; uint8_t key[AES_MAX_KEY_SIZE]; unsigned key_len; uint8_t nonce[AES_BLOCK_SIZE]; uint8_t encr_buf[AES_BLOCK_SIZE]; unsigned encr_pos; } archive_crypto_ctx; #else +#if defined(ARCHIVE_CRYPTO_MD5_WIN) ||\ + defined(ARCHIVE_CRYPTO_SHA1_WIN) ||\ + defined(ARCHIVE_CRYPTO_SHA256_WIN) ||\ + defined(ARCHIVE_CRYPTO_SHA384_WIN) ||\ + defined(ARCHIVE_CRYPTO_SHA512_WIN) #if defined(_WIN32) && !defined(__CYGWIN__) && !(defined(HAVE_BCRYPT_H) && _WIN32_WINNT >= _WIN32_WINNT_VISTA) #define ARCHIVE_CRYPTOR_USE_WINCRYPT 1 #endif +#endif #define AES_BLOCK_SIZE 16 #define AES_MAX_KEY_SIZE 32 typedef int archive_crypto_ctx; #endif /* defines */ #define archive_pbkdf2_sha1(pw, pw_len, salt, salt_len, rounds, dk, dk_len)\ __archive_cryptor.pbkdf2sha1(pw, pw_len, salt, salt_len, rounds, dk, dk_len) #define archive_decrypto_aes_ctr_init(ctx, key, key_len) \ __archive_cryptor.decrypto_aes_ctr_init(ctx, key, key_len) #define archive_decrypto_aes_ctr_update(ctx, in, in_len, out, out_len) \ __archive_cryptor.decrypto_aes_ctr_update(ctx, in, in_len, out, out_len) #define archive_decrypto_aes_ctr_release(ctx) \ __archive_cryptor.decrypto_aes_ctr_release(ctx) #define archive_encrypto_aes_ctr_init(ctx, key, key_len) \ __archive_cryptor.encrypto_aes_ctr_init(ctx, key, key_len) #define archive_encrypto_aes_ctr_update(ctx, in, in_len, out, out_len) \ __archive_cryptor.encrypto_aes_ctr_update(ctx, in, in_len, out, out_len) #define archive_encrypto_aes_ctr_release(ctx) \ __archive_cryptor.encrypto_aes_ctr_release(ctx) +/* Stub return value if no encryption support exists. */ +#define CRYPTOR_STUB_FUNCTION -2 + /* Minimal interface to cryptographic functionality for internal use in * libarchive */ struct archive_cryptor { /* PKCS5 PBKDF2 HMAC-SHA1 */ int (*pbkdf2sha1)(const char *pw, size_t pw_len, const uint8_t *salt, size_t salt_len, unsigned rounds, uint8_t *derived_key, size_t derived_key_len); /* AES CTR mode(little endian version) */ int (*decrypto_aes_ctr_init)(archive_crypto_ctx *, const uint8_t *, size_t); int (*decrypto_aes_ctr_update)(archive_crypto_ctx *, const uint8_t *, size_t, uint8_t *, size_t *); int (*decrypto_aes_ctr_release)(archive_crypto_ctx *); int (*encrypto_aes_ctr_init)(archive_crypto_ctx *, const uint8_t *, size_t); int (*encrypto_aes_ctr_update)(archive_crypto_ctx *, const uint8_t *, size_t, uint8_t *, size_t *); int (*encrypto_aes_ctr_release)(archive_crypto_ctx *); }; extern const struct archive_cryptor __archive_cryptor; #endif diff --git a/contrib/libarchive/libarchive/archive_entry.h b/contrib/libarchive/libarchive/archive_entry.h index 2b917b3fde8e..344f33bffac2 100644 --- a/contrib/libarchive/libarchive/archive_entry.h +++ b/contrib/libarchive/libarchive/archive_entry.h @@ -1,763 +1,763 @@ /*- * Copyright (c) 2003-2008 Tim Kientzle * Copyright (c) 2016 Martin Matuska * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #ifndef ARCHIVE_ENTRY_H_INCLUDED #define ARCHIVE_ENTRY_H_INCLUDED /* Note: Compiler will complain if this does not match archive.h! */ -#define ARCHIVE_VERSION_NUMBER 3008001 +#define ARCHIVE_VERSION_NUMBER 3008002 /* * Note: archive_entry.h is for use outside of libarchive; the * configuration headers (config.h, archive_platform.h, etc.) are * purely internal. Do NOT use HAVE_XXX configuration macros to * control the behavior of this header! If you must conditionalize, * use predefined compiler and/or platform macros. */ #include #include /* for wchar_t */ #include /* for C99 int64_t, etc. */ #if ARCHIVE_VERSION_NUMBER < 4000000 /* time_t is slated to be removed from public includes in 4.0 */ #include #endif #if defined(_WIN32) && !defined(__CYGWIN__) #include #endif /* Get a suitable 64-bit integer type. */ #if !defined(__LA_INT64_T_DEFINED) # if ARCHIVE_VERSION_NUMBER < 4000000 #define __LA_INT64_T la_int64_t # endif #define __LA_INT64_T_DEFINED # if defined(_WIN32) && !defined(__CYGWIN__) && !defined(__WATCOMC__) typedef __int64 la_int64_t; typedef unsigned __int64 la_uint64_t; # else #include # if defined(_SCO_DS) || defined(__osf__) typedef long long la_int64_t; typedef unsigned long long la_uint64_t; # else typedef int64_t la_int64_t; typedef uint64_t la_uint64_t; # endif # endif #endif /* The la_ssize_t should match the type used in 'struct stat' */ #if !defined(__LA_SSIZE_T_DEFINED) /* Older code relied on the __LA_SSIZE_T macro; after 4.0 we'll switch to the typedef exclusively. */ # if ARCHIVE_VERSION_NUMBER < 4000000 #define __LA_SSIZE_T la_ssize_t # endif #define __LA_SSIZE_T_DEFINED # if defined(_WIN32) && !defined(__CYGWIN__) && !defined(__WATCOMC__) # if defined(_SSIZE_T_DEFINED) || defined(_SSIZE_T_) typedef ssize_t la_ssize_t; # elif defined(_WIN64) typedef __int64 la_ssize_t; # else typedef long la_ssize_t; # endif # else # include /* ssize_t */ typedef ssize_t la_ssize_t; # endif #endif /* Get a suitable definition for mode_t */ #if ARCHIVE_VERSION_NUMBER >= 3999000 /* Switch to plain 'int' for libarchive 4.0. It's less broken than 'mode_t' */ # define __LA_MODE_T int #elif defined(_WIN32) && !defined(__CYGWIN__) && !defined(__BORLANDC__) && !defined(__WATCOMC__) # define __LA_MODE_T unsigned short #else # define __LA_MODE_T mode_t #endif #if ARCHIVE_VERSION_NUMBER < 4000000 /* Use the platform types for time_t */ #define __LA_TIME_T time_t #else /* Use 64-bits integer types for time_t */ #define __LA_TIME_T la_int64_t #endif #if ARCHIVE_VERSION_NUMBER < 4000000 /* Use the platform types for dev_t */ #define __LA_DEV_T dev_t #else /* Use 64-bits integer types for dev_t */ #define __LA_DEV_T la_int64_t #endif #if ARCHIVE_VERSION_NUMBER < 4000000 /* Libarchive 3.x used signed int64 for inode numbers */ #define __LA_INO_T la_int64_t #else /* Switch to unsigned for libarchive 4.0 */ #define __LA_INO_T la_uint64_t #endif /* Large file support for Android */ #if defined(__LIBARCHIVE_BUILD) && defined(__ANDROID__) #include "android_lf.h" #endif /* * On Windows, define LIBARCHIVE_STATIC if you're building or using a * .lib. The default here assumes you're building a DLL. Only * libarchive source should ever define __LIBARCHIVE_BUILD. */ #if ((defined __WIN32__) || (defined _WIN32) || defined(__CYGWIN__)) && (!defined LIBARCHIVE_STATIC) # ifdef __LIBARCHIVE_BUILD # ifdef __GNUC__ # define __LA_DECL __attribute__((dllexport)) extern # else # define __LA_DECL __declspec(dllexport) # endif # else # ifdef __GNUC__ # define __LA_DECL # else # define __LA_DECL __declspec(dllimport) # endif # endif #elif defined __LIBARCHIVE_ENABLE_VISIBILITY # define __LA_DECL __attribute__((visibility("default"))) #else /* Static libraries on all platforms and shared libraries on non-Windows. */ # define __LA_DECL #endif #if defined(__GNUC__) && (__GNUC__ > 3 || \ (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)) # define __LA_DEPRECATED __attribute__((deprecated)) #else # define __LA_DEPRECATED #endif #ifdef __cplusplus extern "C" { #endif /* * Description of an archive entry. * * You can think of this as "struct stat" with some text fields added in. * * TODO: Add "comment", "charset", and possibly other entries that are * supported by "pax interchange" format. However, GNU, ustar, cpio, * and other variants don't support these features, so they're not an * excruciatingly high priority right now. * * TODO: "pax interchange" format allows essentially arbitrary * key/value attributes to be attached to any entry. Supporting * such extensions may make this library useful for special * applications (e.g., a package manager could attach special * package-management attributes to each entry). */ struct archive; struct archive_entry; /* * File-type constants. These are returned from archive_entry_filetype() * and passed to archive_entry_set_filetype(). * * These values match S_XXX defines on every platform I've checked, * including Windows, AIX, Linux, Solaris, and BSD. They're * (re)defined here because platforms generally don't define the ones * they don't support. For example, Windows doesn't define S_IFLNK or * S_IFBLK. Instead of having a mass of conditional logic and system * checks to define any S_XXX values that aren't supported locally, * I've just defined a new set of such constants so that * libarchive-based applications can manipulate and identify archive * entries properly even if the hosting platform can't store them on * disk. * * These values are also used directly within some portable formats, * such as cpio. If you find a platform that varies from these, the * correct solution is to leave these alone and translate from these * portable values to platform-native values when entries are read from * or written to disk. */ /* * In libarchive 4.0, we can drop the casts here. * They're needed to work around Borland C's broken mode_t. */ #define AE_IFMT ((__LA_MODE_T)0170000) #define AE_IFREG ((__LA_MODE_T)0100000) #define AE_IFLNK ((__LA_MODE_T)0120000) #define AE_IFSOCK ((__LA_MODE_T)0140000) #define AE_IFCHR ((__LA_MODE_T)0020000) #define AE_IFBLK ((__LA_MODE_T)0060000) #define AE_IFDIR ((__LA_MODE_T)0040000) #define AE_IFIFO ((__LA_MODE_T)0010000) /* * Symlink types */ #define AE_SYMLINK_TYPE_UNDEFINED 0 #define AE_SYMLINK_TYPE_FILE 1 #define AE_SYMLINK_TYPE_DIRECTORY 2 /* * Basic object manipulation */ __LA_DECL struct archive_entry *archive_entry_clear(struct archive_entry *); /* The 'clone' function does a deep copy; all of the strings are copied too. */ __LA_DECL struct archive_entry *archive_entry_clone(struct archive_entry *); __LA_DECL void archive_entry_free(struct archive_entry *); __LA_DECL struct archive_entry *archive_entry_new(void); /* * This form of archive_entry_new2() will pull character-set * conversion information from the specified archive handle. The * older archive_entry_new(void) form is equivalent to calling * archive_entry_new2(NULL) and will result in the use of an internal * default character-set conversion. */ __LA_DECL struct archive_entry *archive_entry_new2(struct archive *); /* * Retrieve fields from an archive_entry. * * There are a number of implicit conversions among these fields. For * example, if a regular string field is set and you read the _w wide * character field, the entry will implicitly convert narrow-to-wide * using the current locale. Similarly, dev values are automatically * updated when you write devmajor or devminor and vice versa. * * In addition, fields can be "set" or "unset." Unset string fields * return NULL, non-string fields have _is_set() functions to test * whether they've been set. You can "unset" a string field by * assigning NULL; non-string fields have _unset() functions to * unset them. * * Note: There is one ambiguity in the above; string fields will * also return NULL when implicit character set conversions fail. * This is usually what you want. */ __LA_DECL __LA_TIME_T archive_entry_atime(struct archive_entry *); __LA_DECL long archive_entry_atime_nsec(struct archive_entry *); __LA_DECL int archive_entry_atime_is_set(struct archive_entry *); __LA_DECL __LA_TIME_T archive_entry_birthtime(struct archive_entry *); __LA_DECL long archive_entry_birthtime_nsec(struct archive_entry *); __LA_DECL int archive_entry_birthtime_is_set(struct archive_entry *); __LA_DECL __LA_TIME_T archive_entry_ctime(struct archive_entry *); __LA_DECL long archive_entry_ctime_nsec(struct archive_entry *); __LA_DECL int archive_entry_ctime_is_set(struct archive_entry *); __LA_DECL __LA_DEV_T archive_entry_dev(struct archive_entry *); __LA_DECL int archive_entry_dev_is_set(struct archive_entry *); __LA_DECL __LA_DEV_T archive_entry_devmajor(struct archive_entry *); __LA_DECL __LA_DEV_T archive_entry_devminor(struct archive_entry *); __LA_DECL __LA_MODE_T archive_entry_filetype(struct archive_entry *); __LA_DECL int archive_entry_filetype_is_set(struct archive_entry *); __LA_DECL void archive_entry_fflags(struct archive_entry *, unsigned long * /* set */, unsigned long * /* clear */); __LA_DECL const char *archive_entry_fflags_text(struct archive_entry *); __LA_DECL la_int64_t archive_entry_gid(struct archive_entry *); __LA_DECL int archive_entry_gid_is_set(struct archive_entry *); __LA_DECL const char *archive_entry_gname(struct archive_entry *); __LA_DECL const char *archive_entry_gname_utf8(struct archive_entry *); __LA_DECL const wchar_t *archive_entry_gname_w(struct archive_entry *); __LA_DECL void archive_entry_set_link_to_hardlink(struct archive_entry *); __LA_DECL const char *archive_entry_hardlink(struct archive_entry *); __LA_DECL const char *archive_entry_hardlink_utf8(struct archive_entry *); __LA_DECL const wchar_t *archive_entry_hardlink_w(struct archive_entry *); __LA_DECL int archive_entry_hardlink_is_set(struct archive_entry *); __LA_DECL __LA_INO_T archive_entry_ino(struct archive_entry *); __LA_DECL __LA_INO_T archive_entry_ino64(struct archive_entry *); __LA_DECL int archive_entry_ino_is_set(struct archive_entry *); __LA_DECL __LA_MODE_T archive_entry_mode(struct archive_entry *); __LA_DECL time_t archive_entry_mtime(struct archive_entry *); __LA_DECL long archive_entry_mtime_nsec(struct archive_entry *); __LA_DECL int archive_entry_mtime_is_set(struct archive_entry *); __LA_DECL unsigned int archive_entry_nlink(struct archive_entry *); __LA_DECL const char *archive_entry_pathname(struct archive_entry *); __LA_DECL const char *archive_entry_pathname_utf8(struct archive_entry *); __LA_DECL const wchar_t *archive_entry_pathname_w(struct archive_entry *); __LA_DECL __LA_MODE_T archive_entry_perm(struct archive_entry *); __LA_DECL int archive_entry_perm_is_set(struct archive_entry *); __LA_DECL int archive_entry_rdev_is_set(struct archive_entry *); __LA_DECL __LA_DEV_T archive_entry_rdev(struct archive_entry *); __LA_DECL __LA_DEV_T archive_entry_rdevmajor(struct archive_entry *); __LA_DECL __LA_DEV_T archive_entry_rdevminor(struct archive_entry *); __LA_DECL const char *archive_entry_sourcepath(struct archive_entry *); __LA_DECL const wchar_t *archive_entry_sourcepath_w(struct archive_entry *); __LA_DECL la_int64_t archive_entry_size(struct archive_entry *); __LA_DECL int archive_entry_size_is_set(struct archive_entry *); __LA_DECL const char *archive_entry_strmode(struct archive_entry *); __LA_DECL void archive_entry_set_link_to_symlink(struct archive_entry *); __LA_DECL const char *archive_entry_symlink(struct archive_entry *); __LA_DECL const char *archive_entry_symlink_utf8(struct archive_entry *); __LA_DECL int archive_entry_symlink_type(struct archive_entry *); __LA_DECL const wchar_t *archive_entry_symlink_w(struct archive_entry *); __LA_DECL la_int64_t archive_entry_uid(struct archive_entry *); __LA_DECL int archive_entry_uid_is_set(struct archive_entry *); __LA_DECL const char *archive_entry_uname(struct archive_entry *); __LA_DECL const char *archive_entry_uname_utf8(struct archive_entry *); __LA_DECL const wchar_t *archive_entry_uname_w(struct archive_entry *); __LA_DECL int archive_entry_is_data_encrypted(struct archive_entry *); __LA_DECL int archive_entry_is_metadata_encrypted(struct archive_entry *); __LA_DECL int archive_entry_is_encrypted(struct archive_entry *); /* * Set fields in an archive_entry. * * Note: Before libarchive 2.4, there were 'set' and 'copy' versions * of the string setters. 'copy' copied the actual string, 'set' just * stored the pointer. In libarchive 2.4 and later, strings are * always copied. */ __LA_DECL void archive_entry_set_atime(struct archive_entry *, __LA_TIME_T, long); __LA_DECL void archive_entry_unset_atime(struct archive_entry *); #if defined(_WIN32) && !defined(__CYGWIN__) __LA_DECL void archive_entry_copy_bhfi(struct archive_entry *, BY_HANDLE_FILE_INFORMATION *); #endif __LA_DECL void archive_entry_set_birthtime(struct archive_entry *, __LA_TIME_T, long); __LA_DECL void archive_entry_unset_birthtime(struct archive_entry *); __LA_DECL void archive_entry_set_ctime(struct archive_entry *, __LA_TIME_T, long); __LA_DECL void archive_entry_unset_ctime(struct archive_entry *); __LA_DECL void archive_entry_set_dev(struct archive_entry *, __LA_DEV_T); __LA_DECL void archive_entry_set_devmajor(struct archive_entry *, __LA_DEV_T); __LA_DECL void archive_entry_set_devminor(struct archive_entry *, __LA_DEV_T); __LA_DECL void archive_entry_set_filetype(struct archive_entry *, unsigned int); __LA_DECL void archive_entry_set_fflags(struct archive_entry *, unsigned long /* set */, unsigned long /* clear */); /* Returns pointer to start of first invalid token, or NULL if none. */ /* Note that all recognized tokens are processed, regardless. */ __LA_DECL const char *archive_entry_copy_fflags_text(struct archive_entry *, const char *); __LA_DECL const char *archive_entry_copy_fflags_text_len(struct archive_entry *, const char *, size_t); __LA_DECL const wchar_t *archive_entry_copy_fflags_text_w(struct archive_entry *, const wchar_t *); __LA_DECL void archive_entry_set_gid(struct archive_entry *, la_int64_t); __LA_DECL void archive_entry_set_gname(struct archive_entry *, const char *); __LA_DECL void archive_entry_set_gname_utf8(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_gname(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_gname_w(struct archive_entry *, const wchar_t *); __LA_DECL int archive_entry_update_gname_utf8(struct archive_entry *, const char *); __LA_DECL void archive_entry_set_hardlink(struct archive_entry *, const char *); __LA_DECL void archive_entry_set_hardlink_utf8(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_hardlink(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_hardlink_w(struct archive_entry *, const wchar_t *); __LA_DECL int archive_entry_update_hardlink_utf8(struct archive_entry *, const char *); __LA_DECL void archive_entry_set_ino(struct archive_entry *, __LA_INO_T); __LA_DECL void archive_entry_set_ino64(struct archive_entry *, __LA_INO_T); __LA_DECL void archive_entry_set_link(struct archive_entry *, const char *); __LA_DECL void archive_entry_set_link_utf8(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_link(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_link_w(struct archive_entry *, const wchar_t *); __LA_DECL int archive_entry_update_link_utf8(struct archive_entry *, const char *); __LA_DECL void archive_entry_set_mode(struct archive_entry *, __LA_MODE_T); __LA_DECL void archive_entry_set_mtime(struct archive_entry *, __LA_TIME_T, long); __LA_DECL void archive_entry_unset_mtime(struct archive_entry *); __LA_DECL void archive_entry_set_nlink(struct archive_entry *, unsigned int); __LA_DECL void archive_entry_set_pathname(struct archive_entry *, const char *); __LA_DECL void archive_entry_set_pathname_utf8(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_pathname(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_pathname_w(struct archive_entry *, const wchar_t *); __LA_DECL int archive_entry_update_pathname_utf8(struct archive_entry *, const char *); __LA_DECL void archive_entry_set_perm(struct archive_entry *, __LA_MODE_T); __LA_DECL void archive_entry_set_rdev(struct archive_entry *, __LA_DEV_T); __LA_DECL void archive_entry_set_rdevmajor(struct archive_entry *, __LA_DEV_T); __LA_DECL void archive_entry_set_rdevminor(struct archive_entry *, __LA_DEV_T); __LA_DECL void archive_entry_set_size(struct archive_entry *, la_int64_t); __LA_DECL void archive_entry_unset_size(struct archive_entry *); __LA_DECL void archive_entry_copy_sourcepath(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_sourcepath_w(struct archive_entry *, const wchar_t *); __LA_DECL void archive_entry_set_symlink(struct archive_entry *, const char *); __LA_DECL void archive_entry_set_symlink_type(struct archive_entry *, int); __LA_DECL void archive_entry_set_symlink_utf8(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_symlink(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_symlink_w(struct archive_entry *, const wchar_t *); __LA_DECL int archive_entry_update_symlink_utf8(struct archive_entry *, const char *); __LA_DECL void archive_entry_set_uid(struct archive_entry *, la_int64_t); __LA_DECL void archive_entry_set_uname(struct archive_entry *, const char *); __LA_DECL void archive_entry_set_uname_utf8(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_uname(struct archive_entry *, const char *); __LA_DECL void archive_entry_copy_uname_w(struct archive_entry *, const wchar_t *); __LA_DECL int archive_entry_update_uname_utf8(struct archive_entry *, const char *); __LA_DECL void archive_entry_set_is_data_encrypted(struct archive_entry *, char is_encrypted); __LA_DECL void archive_entry_set_is_metadata_encrypted(struct archive_entry *, char is_encrypted); /* * Routines to bulk copy fields to/from a platform-native "struct * stat." Libarchive used to just store a struct stat inside of each * archive_entry object, but this created issues when trying to * manipulate archives on systems different than the ones they were * created on. * * TODO: On Linux and other LFS systems, provide both stat32 and * stat64 versions of these functions and all of the macro glue so * that archive_entry_stat is magically defined to * archive_entry_stat32 or archive_entry_stat64 as appropriate. */ __LA_DECL const struct stat *archive_entry_stat(struct archive_entry *); __LA_DECL void archive_entry_copy_stat(struct archive_entry *, const struct stat *); /* * Storage for Mac OS-specific AppleDouble metadata information. * Apple-format tar files store a separate binary blob containing * encoded metadata with ACL, extended attributes, etc. * This provides a place to store that blob. */ __LA_DECL const void * archive_entry_mac_metadata(struct archive_entry *, size_t *); __LA_DECL void archive_entry_copy_mac_metadata(struct archive_entry *, const void *, size_t); /* * Digest routine. This is used to query the raw hex digest for the * given entry. The type of digest is provided as an argument. */ #define ARCHIVE_ENTRY_DIGEST_MD5 0x00000001 #define ARCHIVE_ENTRY_DIGEST_RMD160 0x00000002 #define ARCHIVE_ENTRY_DIGEST_SHA1 0x00000003 #define ARCHIVE_ENTRY_DIGEST_SHA256 0x00000004 #define ARCHIVE_ENTRY_DIGEST_SHA384 0x00000005 #define ARCHIVE_ENTRY_DIGEST_SHA512 0x00000006 __LA_DECL const unsigned char * archive_entry_digest(struct archive_entry *, int /* type */); __LA_DECL int archive_entry_set_digest(struct archive_entry *, int, const unsigned char *); /* * ACL routines. This used to simply store and return text-format ACL * strings, but that proved insufficient for a number of reasons: * = clients need control over uname/uid and gname/gid mappings * = there are many different ACL text formats * = would like to be able to read/convert archives containing ACLs * on platforms that lack ACL libraries * * This last point, in particular, forces me to implement a reasonably * complete set of ACL support routines. */ /* * Permission bits. */ #define ARCHIVE_ENTRY_ACL_EXECUTE 0x00000001 #define ARCHIVE_ENTRY_ACL_WRITE 0x00000002 #define ARCHIVE_ENTRY_ACL_READ 0x00000004 #define ARCHIVE_ENTRY_ACL_READ_DATA 0x00000008 #define ARCHIVE_ENTRY_ACL_LIST_DIRECTORY 0x00000008 #define ARCHIVE_ENTRY_ACL_WRITE_DATA 0x00000010 #define ARCHIVE_ENTRY_ACL_ADD_FILE 0x00000010 #define ARCHIVE_ENTRY_ACL_APPEND_DATA 0x00000020 #define ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY 0x00000020 #define ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS 0x00000040 #define ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS 0x00000080 #define ARCHIVE_ENTRY_ACL_DELETE_CHILD 0x00000100 #define ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES 0x00000200 #define ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES 0x00000400 #define ARCHIVE_ENTRY_ACL_DELETE 0x00000800 #define ARCHIVE_ENTRY_ACL_READ_ACL 0x00001000 #define ARCHIVE_ENTRY_ACL_WRITE_ACL 0x00002000 #define ARCHIVE_ENTRY_ACL_WRITE_OWNER 0x00004000 #define ARCHIVE_ENTRY_ACL_SYNCHRONIZE 0x00008000 #define ARCHIVE_ENTRY_ACL_PERMS_POSIX1E \ (ARCHIVE_ENTRY_ACL_EXECUTE \ | ARCHIVE_ENTRY_ACL_WRITE \ | ARCHIVE_ENTRY_ACL_READ) #define ARCHIVE_ENTRY_ACL_PERMS_NFS4 \ (ARCHIVE_ENTRY_ACL_EXECUTE \ | ARCHIVE_ENTRY_ACL_READ_DATA \ | ARCHIVE_ENTRY_ACL_LIST_DIRECTORY \ | ARCHIVE_ENTRY_ACL_WRITE_DATA \ | ARCHIVE_ENTRY_ACL_ADD_FILE \ | ARCHIVE_ENTRY_ACL_APPEND_DATA \ | ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY \ | ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS \ | ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS \ | ARCHIVE_ENTRY_ACL_DELETE_CHILD \ | ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES \ | ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES \ | ARCHIVE_ENTRY_ACL_DELETE \ | ARCHIVE_ENTRY_ACL_READ_ACL \ | ARCHIVE_ENTRY_ACL_WRITE_ACL \ | ARCHIVE_ENTRY_ACL_WRITE_OWNER \ | ARCHIVE_ENTRY_ACL_SYNCHRONIZE) /* * Inheritance values (NFS4 ACLs only); included in permset. */ #define ARCHIVE_ENTRY_ACL_ENTRY_INHERITED 0x01000000 #define ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT 0x02000000 #define ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT 0x04000000 #define ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT 0x08000000 #define ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY 0x10000000 #define ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS 0x20000000 #define ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS 0x40000000 #define ARCHIVE_ENTRY_ACL_INHERITANCE_NFS4 \ (ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT \ | ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT \ | ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT \ | ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY \ | ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS \ | ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS \ | ARCHIVE_ENTRY_ACL_ENTRY_INHERITED) /* We need to be able to specify combinations of these. */ #define ARCHIVE_ENTRY_ACL_TYPE_ACCESS 0x00000100 /* POSIX.1e only */ #define ARCHIVE_ENTRY_ACL_TYPE_DEFAULT 0x00000200 /* POSIX.1e only */ #define ARCHIVE_ENTRY_ACL_TYPE_ALLOW 0x00000400 /* NFS4 only */ #define ARCHIVE_ENTRY_ACL_TYPE_DENY 0x00000800 /* NFS4 only */ #define ARCHIVE_ENTRY_ACL_TYPE_AUDIT 0x00001000 /* NFS4 only */ #define ARCHIVE_ENTRY_ACL_TYPE_ALARM 0x00002000 /* NFS4 only */ #define ARCHIVE_ENTRY_ACL_TYPE_POSIX1E (ARCHIVE_ENTRY_ACL_TYPE_ACCESS \ | ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) #define ARCHIVE_ENTRY_ACL_TYPE_NFS4 (ARCHIVE_ENTRY_ACL_TYPE_ALLOW \ | ARCHIVE_ENTRY_ACL_TYPE_DENY \ | ARCHIVE_ENTRY_ACL_TYPE_AUDIT \ | ARCHIVE_ENTRY_ACL_TYPE_ALARM) /* Tag values mimic POSIX.1e */ #define ARCHIVE_ENTRY_ACL_USER 10001 /* Specified user. */ #define ARCHIVE_ENTRY_ACL_USER_OBJ 10002 /* User who owns the file. */ #define ARCHIVE_ENTRY_ACL_GROUP 10003 /* Specified group. */ #define ARCHIVE_ENTRY_ACL_GROUP_OBJ 10004 /* Group who owns the file. */ #define ARCHIVE_ENTRY_ACL_MASK 10005 /* Modify group access (POSIX.1e only) */ #define ARCHIVE_ENTRY_ACL_OTHER 10006 /* Public (POSIX.1e only) */ #define ARCHIVE_ENTRY_ACL_EVERYONE 10107 /* Everyone (NFS4 only) */ /* * Set the ACL by clearing it and adding entries one at a time. * Unlike the POSIX.1e ACL routines, you must specify the type * (access/default) for each entry. Internally, the ACL data is just * a soup of entries. API calls here allow you to retrieve just the * entries of interest. This design (which goes against the spirit of * POSIX.1e) is useful for handling archive formats that combine * default and access information in a single ACL list. */ __LA_DECL void archive_entry_acl_clear(struct archive_entry *); __LA_DECL int archive_entry_acl_add_entry(struct archive_entry *, int /* type */, int /* permset */, int /* tag */, int /* qual */, const char * /* name */); __LA_DECL int archive_entry_acl_add_entry_w(struct archive_entry *, int /* type */, int /* permset */, int /* tag */, int /* qual */, const wchar_t * /* name */); /* * To retrieve the ACL, first "reset", then repeatedly ask for the * "next" entry. The want_type parameter allows you to request only * certain types of entries. */ __LA_DECL int archive_entry_acl_reset(struct archive_entry *, int /* want_type */); __LA_DECL int archive_entry_acl_next(struct archive_entry *, int /* want_type */, int * /* type */, int * /* permset */, int * /* tag */, int * /* qual */, const char ** /* name */); /* * Construct a text-format ACL. The flags argument is a bitmask that * can include any of the following: * * Flags only for archive entries with POSIX.1e ACL: * ARCHIVE_ENTRY_ACL_TYPE_ACCESS - Include POSIX.1e "access" entries. * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT - Include POSIX.1e "default" entries. * ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT - Include "default:" before each * default ACL entry. * ARCHIVE_ENTRY_ACL_STYLE_SOLARIS - Output only one colon after "other" and * "mask" entries. * * Flags only for archive entries with NFSv4 ACL: * ARCHIVE_ENTRY_ACL_STYLE_COMPACT - Do not output the minus character for * unset permissions and flags in NFSv4 ACL permission and flag fields * * Flags for for archive entries with POSIX.1e ACL or NFSv4 ACL: * ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID - Include extra numeric ID field in * each ACL entry. * ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA - Separate entries with comma * instead of newline. */ #define ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID 0x00000001 #define ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT 0x00000002 #define ARCHIVE_ENTRY_ACL_STYLE_SOLARIS 0x00000004 #define ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA 0x00000008 #define ARCHIVE_ENTRY_ACL_STYLE_COMPACT 0x00000010 __LA_DECL wchar_t *archive_entry_acl_to_text_w(struct archive_entry *, la_ssize_t * /* len */, int /* flags */); __LA_DECL char *archive_entry_acl_to_text(struct archive_entry *, la_ssize_t * /* len */, int /* flags */); __LA_DECL int archive_entry_acl_from_text_w(struct archive_entry *, const wchar_t * /* wtext */, int /* type */); __LA_DECL int archive_entry_acl_from_text(struct archive_entry *, const char * /* text */, int /* type */); /* Deprecated constants */ #define OLD_ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID 1024 #define OLD_ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT 2048 /* Deprecated functions */ __LA_DECL const wchar_t *archive_entry_acl_text_w(struct archive_entry *, int /* flags */) __LA_DEPRECATED; __LA_DECL const char *archive_entry_acl_text(struct archive_entry *, int /* flags */) __LA_DEPRECATED; /* Return bitmask of ACL types in an archive entry */ __LA_DECL int archive_entry_acl_types(struct archive_entry *); /* Return a count of entries matching 'want_type' */ __LA_DECL int archive_entry_acl_count(struct archive_entry *, int /* want_type */); /* Return an opaque ACL object. */ /* There's not yet anything clients can actually do with this... */ struct archive_acl; __LA_DECL struct archive_acl *archive_entry_acl(struct archive_entry *); /* * extended attributes */ __LA_DECL void archive_entry_xattr_clear(struct archive_entry *); __LA_DECL void archive_entry_xattr_add_entry(struct archive_entry *, const char * /* name */, const void * /* value */, size_t /* size */); /* * To retrieve the xattr list, first "reset", then repeatedly ask for the * "next" entry. */ __LA_DECL int archive_entry_xattr_count(struct archive_entry *); __LA_DECL int archive_entry_xattr_reset(struct archive_entry *); __LA_DECL int archive_entry_xattr_next(struct archive_entry *, const char ** /* name */, const void ** /* value */, size_t *); /* * sparse */ __LA_DECL void archive_entry_sparse_clear(struct archive_entry *); __LA_DECL void archive_entry_sparse_add_entry(struct archive_entry *, la_int64_t /* offset */, la_int64_t /* length */); /* * To retrieve the xattr list, first "reset", then repeatedly ask for the * "next" entry. */ __LA_DECL int archive_entry_sparse_count(struct archive_entry *); __LA_DECL int archive_entry_sparse_reset(struct archive_entry *); __LA_DECL int archive_entry_sparse_next(struct archive_entry *, la_int64_t * /* offset */, la_int64_t * /* length */); /* * Utility to match up hardlinks. * * The 'struct archive_entry_linkresolver' is a cache of archive entries * for files with multiple links. Here's how to use it: * 1. Create a lookup object with archive_entry_linkresolver_new() * 2. Tell it the archive format you're using. * 3. Hand each archive_entry to archive_entry_linkify(). * That function will return 0, 1, or 2 entries that should * be written. * 4. Call archive_entry_linkify(resolver, NULL) until * no more entries are returned. * 5. Call archive_entry_linkresolver_free(resolver) to free resources. * * The entries returned have their hardlink and size fields updated * appropriately. If an entry is passed in that does not refer to * a file with multiple links, it is returned unchanged. The intention * is that you should be able to simply filter all entries through * this machine. * * To make things more efficient, be sure that each entry has a valid * nlinks value. The hardlink cache uses this to track when all links * have been found. If the nlinks value is zero, it will keep every * name in the cache indefinitely, which can use a lot of memory. * * Note that archive_entry_size() is reset to zero if the file * body should not be written to the archive. Pay attention! */ struct archive_entry_linkresolver; /* * There are three different strategies for marking hardlinks. * The descriptions below name them after the best-known * formats that rely on each strategy: * * "Old cpio" is the simplest, it always returns any entry unmodified. * As far as I know, only cpio formats use this. Old cpio archives * store every link with the full body; the onus is on the dearchiver * to detect and properly link the files as they are restored. * "tar" is also pretty simple; it caches a copy the first time it sees * any link. Subsequent appearances are modified to be hardlink * references to the first one without any body. Used by all tar * formats, although the newest tar formats permit the "old cpio" strategy * as well. This strategy is very simple for the dearchiver, * and reasonably straightforward for the archiver. * "new cpio" is trickier. It stores the body only with the last * occurrence. The complication is that we might not * see every link to a particular file in a single session, so * there's no easy way to know when we've seen the last occurrence. * The solution here is to queue one link until we see the next. * At the end of the session, you can enumerate any remaining * entries by calling archive_entry_linkify(NULL) and store those * bodies. If you have a file with three links l1, l2, and l3, * you'll get the following behavior if you see all three links: * linkify(l1) => NULL (the resolver stores l1 internally) * linkify(l2) => l1 (resolver stores l2, you write l1) * linkify(l3) => l2, l3 (all links seen, you can write both). * If you only see l1 and l2, you'll get this behavior: * linkify(l1) => NULL * linkify(l2) => l1 * linkify(NULL) => l2 (at end, you retrieve remaining links) * As the name suggests, this strategy is used by newer cpio variants. * It's noticeably more complex for the archiver, slightly more complex * for the dearchiver than the tar strategy, but makes it straightforward * to restore a file using any link by simply continuing to scan until * you see a link that is stored with a body. In contrast, the tar * strategy requires you to rescan the archive from the beginning to * correctly extract an arbitrary link. */ __LA_DECL struct archive_entry_linkresolver *archive_entry_linkresolver_new(void); __LA_DECL void archive_entry_linkresolver_set_strategy( struct archive_entry_linkresolver *, int /* format_code */); __LA_DECL void archive_entry_linkresolver_free(struct archive_entry_linkresolver *); __LA_DECL void archive_entry_linkify(struct archive_entry_linkresolver *, struct archive_entry **, struct archive_entry **); __LA_DECL struct archive_entry *archive_entry_partial_links( struct archive_entry_linkresolver *res, unsigned int *links); #ifdef __cplusplus } #endif /* This is meaningless outside of this header. */ #undef __LA_DECL #endif /* !ARCHIVE_ENTRY_H_INCLUDED */ diff --git a/contrib/libarchive/libarchive/archive_entry_paths.3 b/contrib/libarchive/libarchive/archive_entry_paths.3 index 0f849c9ebb35..f739b172308d 100644 --- a/contrib/libarchive/libarchive/archive_entry_paths.3 +++ b/contrib/libarchive/libarchive/archive_entry_paths.3 @@ -1,153 +1,153 @@ .\" Copyright (c) 2010 Joerg Sonnenberger .\" All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .Dd February 2, 2012 .Dt ARCHIVE_ENTRY_PATHS 3 .Os .Sh NAME .Nm archive_entry_hardlink , .Nm archive_entry_hardlink_w , .Nm archive_entry_set_hardlink , .Nm archive_entry_copy_hardlink , .Nm archive_entry_copy_hardlink_w , .Nm archive_entry_update_hardlink_utf8 , .Nm archive_entry_set_link , .Nm archive_entry_copy_link , .Nm archive_entry_copy_link_w , .Nm archive_entry_update_link_utf8 , .Nm archive_entry_pathname , .Nm archive_entry_pathname_w , .Nm archive_entry_set_pathname , .Nm archive_entry_copy_pathname , .Nm archive_entry_copy_pathname_w , .Nm archive_entry_update_pathname_utf8 , .Nm archive_entry_sourcepath , .Nm archive_entry_copy_sourcepath , .Nm archive_entry_symlink , .Nm archive_entry_symlink_w , .Nm archive_entry_set_symlink , .Nm archive_entry_copy_symlink , .Nm archive_entry_copy_symlink_w , .Nm archive_entry_update_symlink_utf8 .Nd functions for manipulating path names in archive entry descriptions .Sh LIBRARY Streaming Archive Library (libarchive, -larchive) .Sh SYNOPSIS .In archive_entry.h .Ft const char * .Fn archive_entry_hardlink "struct archive_entry *a" .Ft const wchar_t * .Fn archive_entry_hardlink_w "struct archive_entry *a" .Ft void .Fn archive_entry_set_hardlink "struct archive_entry *a" "const char *path" .Ft void .Fn archive_entry_copy_hardlink "struct archive_entry *a" "const char *path" .Ft void -.Fn archive_entry_copy_hardlink_w "struct archive_entry *a "const wchar_t *path" +.Fn archive_entry_copy_hardlink_w "struct archive_entry *a" "const wchar_t *path" .Ft int .Fn archive_entry_update_hardlink_utf8 "struct archive_entry *a" "const char *path" .Ft void .Fn archive_entry_set_link "struct archive_entry *a" "const char *path" .Ft void .Fn archive_entry_copy_link "struct archive_entry *a" " const char *path" .Ft void .Fn archive_entry_copy_link_w "struct archive_entry *a" " const wchar_t *path" .Ft int .Fn archive_entry_update_link_utf8 "struct archive_entry *a" " const char *path" .Ft const char * .Fn archive_entry_pathname "struct archive_entry *a" .Ft const wchar_t * .Fn archive_entry_pathname_w "struct archive_entry *a" .Ft void .Fn archive_entry_set_pathname "struct archive_entry *a" "const char *path" .Ft void .Fn archive_entry_copy_pathname "struct archive_entry *a" "const char *path" .Ft void .Fn archive_entry_copy_pathname_w "struct archive_entry *a" "const wchar_t *path" .Ft int .Fn archive_entry_update_pathname_utf8 "struct archive_entry *a" "const char *path" .Ft const char * .Fn archive_entry_sourcepath "struct archive_entry *a" .Ft void .Fn archive_entry_copy_sourcepath "struct archive_entry *a" "const char *path" .Ft const char * .Fn archive_entry_symlink "struct archive_entry *a" .Ft const wchar_t * .Fn archive_entry_symlink_w "struct archive_entry *a" .Ft void .Fn archive_entry_set_symlink "struct archive_entry *a" "const char *path" .Ft void .Fn archive_entry_copy_symlink "struct archive_entry *a" "const char *path" .Ft void .Fn archive_entry_copy_symlink_w "struct archive_entry *a" "const wchar_t *path" .Ft int .Fn archive_entry_update_symlink_utf8 "struct archive_entry *a" "const char *path" .Sh DESCRIPTION Path names supported by .Xr archive_entry 3 : .Bl -tag -width "sourcepath" -compact .It hardlink Destination of the hardlink. .It link Update only. For a symlink, update the destination. Otherwise, make the entry a hardlink and alter the destination for that. .It pathname Path in the archive .It sourcepath Path on the disk for use by .Xr archive_read_disk 3 . .It symlink Destination of the symbolic link. .El .Pp Path names can be provided in one of three different ways: .Bl -tag -width "wchar_t *" .It char * Multibyte strings in the current locale. .It wchar_t * Wide character strings in the current locale. The accessor functions are named .Fn XXX_w . .It UTF-8 Unicode strings encoded as UTF-8. These are convenience functions to update both the multibyte and wide character strings at the same time. .El .Pp The sourcepath is a pure filesystem concept and never stored in an archive directly. .Pp For that reason, it is only available as multibyte string. The link path is a convenience function for conditionally setting hardlink or symlink destination. It doesn't have a corresponding get accessor function. .Pp .Fn archive_entry_set_XXX is an alias for .Fn archive_entry_copy_XXX . .Sh SEE ALSO .Xr archive_entry 3 , .Xr libarchive 3 diff --git a/contrib/libarchive/libarchive/archive_entry_stat.c b/contrib/libarchive/libarchive/archive_entry_stat.c index c4906838ed0f..345d3d29b2f2 100644 --- a/contrib/libarchive/libarchive/archive_entry_stat.c +++ b/contrib/libarchive/libarchive/archive_entry_stat.c @@ -1,117 +1,121 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #include "archive_entry.h" #include "archive_entry_private.h" const struct stat * archive_entry_stat(struct archive_entry *entry) { + int64_t size; struct stat *st; if (entry->stat == NULL) { entry->stat = calloc(1, sizeof(*st)); if (entry->stat == NULL) return (NULL); entry->stat_valid = 0; } /* * If none of the underlying fields have been changed, we * don't need to regenerate. In theory, we could use a bitmap * here to flag only those items that have changed, but the * extra complexity probably isn't worth it. It will be very * rare for anyone to change just one field then request a new * stat structure. */ if (entry->stat_valid) return (entry->stat); st = entry->stat; /* * Use the public interfaces to extract items, so that * the appropriate conversions get invoked. */ st->st_atime = archive_entry_atime(entry); #if HAVE_STRUCT_STAT_ST_BIRTHTIME st->st_birthtime = archive_entry_birthtime(entry); #endif st->st_ctime = archive_entry_ctime(entry); st->st_mtime = archive_entry_mtime(entry); st->st_dev = archive_entry_dev(entry); st->st_gid = (gid_t)archive_entry_gid(entry); st->st_uid = (uid_t)archive_entry_uid(entry); st->st_ino = (ino_t)archive_entry_ino64(entry); st->st_nlink = archive_entry_nlink(entry); st->st_rdev = archive_entry_rdev(entry); - st->st_size = (off_t)archive_entry_size(entry); + size = archive_entry_size(entry); + st->st_size = (off_t)size; + if (st->st_size < 0 || (int64_t)st->st_size != size) + st->st_size = 0; st->st_mode = archive_entry_mode(entry); /* * On systems that support high-res timestamps, copy that * information into struct stat. */ #if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC st->st_atimespec.tv_nsec = archive_entry_atime_nsec(entry); st->st_ctimespec.tv_nsec = archive_entry_ctime_nsec(entry); st->st_mtimespec.tv_nsec = archive_entry_mtime_nsec(entry); #elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC st->st_atim.tv_nsec = archive_entry_atime_nsec(entry); st->st_ctim.tv_nsec = archive_entry_ctime_nsec(entry); st->st_mtim.tv_nsec = archive_entry_mtime_nsec(entry); #elif HAVE_STRUCT_STAT_ST_MTIME_N st->st_atime_n = archive_entry_atime_nsec(entry); st->st_ctime_n = archive_entry_ctime_nsec(entry); st->st_mtime_n = archive_entry_mtime_nsec(entry); #elif HAVE_STRUCT_STAT_ST_UMTIME st->st_uatime = archive_entry_atime_nsec(entry) / 1000; st->st_uctime = archive_entry_ctime_nsec(entry) / 1000; st->st_umtime = archive_entry_mtime_nsec(entry) / 1000; #elif HAVE_STRUCT_STAT_ST_MTIME_USEC st->st_atime_usec = archive_entry_atime_nsec(entry) / 1000; st->st_ctime_usec = archive_entry_ctime_nsec(entry) / 1000; st->st_mtime_usec = archive_entry_mtime_nsec(entry) / 1000; #endif #if HAVE_STRUCT_STAT_ST_BIRTHTIMESPEC_TV_NSEC st->st_birthtimespec.tv_nsec = archive_entry_birthtime_nsec(entry); #endif /* * TODO: On Linux, store 32 or 64 here depending on whether * the cached stat structure is a stat32 or a stat64. This * will allow us to support both variants interchangeably. */ entry->stat_valid = 1; return (st); } diff --git a/contrib/libarchive/libarchive/archive_parse_date.c b/contrib/libarchive/libarchive/archive_parse_date.c index cda0b11a555f..d9e968387d34 100644 --- a/contrib/libarchive/libarchive/archive_parse_date.c +++ b/contrib/libarchive/libarchive/archive_parse_date.c @@ -1,1147 +1,1145 @@ /* * This code is in the public domain and has no copyright. * * This is a plain C recursive-descent translation of an old * public-domain YACC grammar that has been used for parsing dates in * very many open-source projects. * * Since the original authors were generous enough to donate their * work to the public domain, I feel compelled to match their * generosity. * * Tim Kientzle, February 2009. */ /* * Header comment from original getdate.y: */ /* ** Originally written by Steven M. Bellovin while ** at the University of North Carolina at Chapel Hill. Later tweaked by ** a couple of people on Usenet. Completely overhauled by Rich $alz ** and Jim Berets in August, 1990; ** ** This grammar has 10 shift/reduce conflicts. ** ** This code is in the public domain and has no copyright. */ #include "archive_platform.h" #include #include #include #include #include #include "archive.h" /* Basic time units. */ #define EPOCH 1970 #define MINUTE (60L) #define HOUR (60L * MINUTE) #define DAY (24L * HOUR) /* Daylight-savings mode: on, off, or not yet known. */ enum DSTMODE { DSTon, DSToff, DSTmaybe }; /* Meridian: am or pm. */ enum { tAM, tPM }; /* Token types returned by nexttoken() */ enum { tAGO = 260, tDAY, tDAYZONE, tAMPM, tMONTH, tMONTH_UNIT, tSEC_UNIT, tUNUMBER, tZONE, tDST }; struct token { int token; time_t value; }; /* * Parser state. */ struct gdstate { struct token *tokenp; /* Pointer to next token. */ /* HaveXxxx counts how many of this kind of phrase we've seen; * it's a fatal error to have more than one time, zone, day, * or date phrase. */ int HaveYear; int HaveMonth; int HaveDay; int HaveWeekDay; /* Day of week */ int HaveTime; /* Hour/minute/second */ int HaveZone; /* timezone and/or DST info */ int HaveRel; /* time offset; we can have more than one */ /* Absolute time values. */ time_t Timezone; /* Seconds offset from GMT */ time_t Day; time_t Hour; time_t Minutes; time_t Month; time_t Seconds; time_t Year; /* DST selection */ enum DSTMODE DSTmode; /* Day of week accounting, e.g., "3rd Tuesday" */ time_t DayOrdinal; /* "3" in "3rd Tuesday" */ time_t DayNumber; /* "Tuesday" in "3rd Tuesday" */ /* Relative time values: hour/day/week offsets are measured in * seconds, month/year are counted in months. */ time_t RelMonth; time_t RelSeconds; }; /* * A series of functions that recognize certain common time phrases. * Each function returns 1 if it managed to make sense of some of the * tokens, zero otherwise. */ /* * hour:minute or hour:minute:second with optional AM, PM, or numeric * timezone offset */ static int timephrase(struct gdstate *gds) { if (gds->tokenp[0].token == tUNUMBER && gds->tokenp[1].token == ':' && gds->tokenp[2].token == tUNUMBER && gds->tokenp[3].token == ':' && gds->tokenp[4].token == tUNUMBER) { /* "12:14:18" or "22:08:07" */ ++gds->HaveTime; gds->Hour = gds->tokenp[0].value; gds->Minutes = gds->tokenp[2].value; gds->Seconds = gds->tokenp[4].value; gds->tokenp += 5; } else if (gds->tokenp[0].token == tUNUMBER && gds->tokenp[1].token == ':' && gds->tokenp[2].token == tUNUMBER) { /* "12:14" or "22:08" */ ++gds->HaveTime; gds->Hour = gds->tokenp[0].value; gds->Minutes = gds->tokenp[2].value; gds->Seconds = 0; gds->tokenp += 3; } else if (gds->tokenp[0].token == tUNUMBER && gds->tokenp[1].token == tAMPM) { /* "7" is a time if it's followed by "am" or "pm" */ ++gds->HaveTime; gds->Hour = gds->tokenp[0].value; gds->Minutes = gds->Seconds = 0; /* We'll handle the AM/PM below. */ gds->tokenp += 1; } else { /* We can't handle this. */ return 0; } if (gds->tokenp[0].token == tAMPM) { /* "7:12pm", "12:20:13am" */ if (gds->Hour == 12) gds->Hour = 0; if (gds->tokenp[0].value == tPM) gds->Hour += 12; gds->tokenp += 1; } if (gds->tokenp[0].token == '+' && gds->tokenp[1].token == tUNUMBER) { /* "7:14+0700" */ gds->HaveZone++; gds->DSTmode = DSToff; gds->Timezone = - ((gds->tokenp[1].value / 100) * HOUR + (gds->tokenp[1].value % 100) * MINUTE); gds->tokenp += 2; } if (gds->tokenp[0].token == '-' && gds->tokenp[1].token == tUNUMBER) { /* "19:14:12-0530" */ gds->HaveZone++; gds->DSTmode = DSToff; gds->Timezone = + ((gds->tokenp[1].value / 100) * HOUR + (gds->tokenp[1].value % 100) * MINUTE); gds->tokenp += 2; } return 1; } /* * Timezone name, possibly including DST. */ static int zonephrase(struct gdstate *gds) { if (gds->tokenp[0].token == tZONE && gds->tokenp[1].token == tDST) { gds->HaveZone++; gds->Timezone = gds->tokenp[0].value; gds->DSTmode = DSTon; gds->tokenp += 1; return 1; } if (gds->tokenp[0].token == tZONE) { gds->HaveZone++; gds->Timezone = gds->tokenp[0].value; gds->DSTmode = DSToff; gds->tokenp += 1; return 1; } if (gds->tokenp[0].token == tDAYZONE) { gds->HaveZone++; gds->Timezone = gds->tokenp[0].value; gds->DSTmode = DSTon; gds->tokenp += 1; return 1; } return 0; } /* * Year/month/day in various combinations. */ static int datephrase(struct gdstate *gds) { if (gds->tokenp[0].token == tUNUMBER && gds->tokenp[1].token == '/' && gds->tokenp[2].token == tUNUMBER && gds->tokenp[3].token == '/' && gds->tokenp[4].token == tUNUMBER) { gds->HaveYear++; gds->HaveMonth++; gds->HaveDay++; if (gds->tokenp[0].value >= 13) { /* First number is big: 2004/01/29, 99/02/17 */ gds->Year = gds->tokenp[0].value; gds->Month = gds->tokenp[2].value; gds->Day = gds->tokenp[4].value; } else if ((gds->tokenp[4].value >= 13) || (gds->tokenp[2].value >= 13)) { /* Last number is big: 01/07/98 */ /* Middle number is big: 01/29/04 */ gds->Month = gds->tokenp[0].value; gds->Day = gds->tokenp[2].value; gds->Year = gds->tokenp[4].value; } else { /* No significant clues: 02/03/04 */ gds->Month = gds->tokenp[0].value; gds->Day = gds->tokenp[2].value; gds->Year = gds->tokenp[4].value; } gds->tokenp += 5; return 1; } if (gds->tokenp[0].token == tUNUMBER && gds->tokenp[1].token == '/' && gds->tokenp[2].token == tUNUMBER) { /* "1/15" */ gds->HaveMonth++; gds->HaveDay++; gds->Month = gds->tokenp[0].value; gds->Day = gds->tokenp[2].value; gds->tokenp += 3; return 1; } if (gds->tokenp[0].token == tUNUMBER && gds->tokenp[1].token == '-' && gds->tokenp[2].token == tUNUMBER && gds->tokenp[3].token == '-' && gds->tokenp[4].token == tUNUMBER) { /* ISO 8601 format. yyyy-mm-dd. */ gds->HaveYear++; gds->HaveMonth++; gds->HaveDay++; gds->Year = gds->tokenp[0].value; gds->Month = gds->tokenp[2].value; gds->Day = gds->tokenp[4].value; gds->tokenp += 5; return 1; } if (gds->tokenp[0].token == tUNUMBER && gds->tokenp[1].token == '-' && gds->tokenp[2].token == tMONTH && gds->tokenp[3].token == '-' && gds->tokenp[4].token == tUNUMBER) { gds->HaveYear++; gds->HaveMonth++; gds->HaveDay++; if (gds->tokenp[0].value > 31) { /* e.g. 1992-Jun-17 */ gds->Year = gds->tokenp[0].value; gds->Month = gds->tokenp[2].value; gds->Day = gds->tokenp[4].value; } else { /* e.g. 17-JUN-1992. */ gds->Day = gds->tokenp[0].value; gds->Month = gds->tokenp[2].value; gds->Year = gds->tokenp[4].value; } gds->tokenp += 5; return 1; } if (gds->tokenp[0].token == tMONTH && gds->tokenp[1].token == tUNUMBER && gds->tokenp[2].token == ',' && gds->tokenp[3].token == tUNUMBER) { /* "June 17, 2001" */ gds->HaveYear++; gds->HaveMonth++; gds->HaveDay++; gds->Month = gds->tokenp[0].value; gds->Day = gds->tokenp[1].value; gds->Year = gds->tokenp[3].value; gds->tokenp += 4; return 1; } if (gds->tokenp[0].token == tMONTH && gds->tokenp[1].token == tUNUMBER) { /* "May 3" */ gds->HaveMonth++; gds->HaveDay++; gds->Month = gds->tokenp[0].value; gds->Day = gds->tokenp[1].value; gds->tokenp += 2; return 1; } if (gds->tokenp[0].token == tUNUMBER && gds->tokenp[1].token == tMONTH && gds->tokenp[2].token == tUNUMBER) { /* "12 Sept 1997" */ gds->HaveYear++; gds->HaveMonth++; gds->HaveDay++; gds->Day = gds->tokenp[0].value; gds->Month = gds->tokenp[1].value; gds->Year = gds->tokenp[2].value; gds->tokenp += 3; return 1; } if (gds->tokenp[0].token == tUNUMBER && gds->tokenp[1].token == tMONTH) { /* "12 Sept" */ gds->HaveMonth++; gds->HaveDay++; gds->Day = gds->tokenp[0].value; gds->Month = gds->tokenp[1].value; gds->tokenp += 2; return 1; } return 0; } /* * Relative time phrase: "tomorrow", "yesterday", "+1 hour", etc. */ static int relunitphrase(struct gdstate *gds) { if (gds->tokenp[0].token == '-' && gds->tokenp[1].token == tUNUMBER && gds->tokenp[2].token == tSEC_UNIT) { /* "-3 hours" */ gds->HaveRel++; gds->RelSeconds -= gds->tokenp[1].value * gds->tokenp[2].value; gds->tokenp += 3; return 1; } if (gds->tokenp[0].token == '+' && gds->tokenp[1].token == tUNUMBER && gds->tokenp[2].token == tSEC_UNIT) { /* "+1 minute" */ gds->HaveRel++; gds->RelSeconds += gds->tokenp[1].value * gds->tokenp[2].value; gds->tokenp += 3; return 1; } if (gds->tokenp[0].token == tUNUMBER && gds->tokenp[1].token == tSEC_UNIT) { /* "1 day" */ gds->HaveRel++; gds->RelSeconds += gds->tokenp[0].value * gds->tokenp[1].value; gds->tokenp += 2; return 1; } if (gds->tokenp[0].token == '-' && gds->tokenp[1].token == tUNUMBER && gds->tokenp[2].token == tMONTH_UNIT) { /* "-3 months" */ gds->HaveRel++; gds->RelMonth -= gds->tokenp[1].value * gds->tokenp[2].value; gds->tokenp += 3; return 1; } if (gds->tokenp[0].token == '+' && gds->tokenp[1].token == tUNUMBER && gds->tokenp[2].token == tMONTH_UNIT) { /* "+5 years" */ gds->HaveRel++; gds->RelMonth += gds->tokenp[1].value * gds->tokenp[2].value; gds->tokenp += 3; return 1; } if (gds->tokenp[0].token == tUNUMBER && gds->tokenp[1].token == tMONTH_UNIT) { /* "2 years" */ gds->HaveRel++; gds->RelMonth += gds->tokenp[0].value * gds->tokenp[1].value; gds->tokenp += 2; return 1; } if (gds->tokenp[0].token == tSEC_UNIT) { /* "now", "tomorrow" */ gds->HaveRel++; gds->RelSeconds += gds->tokenp[0].value; gds->tokenp += 1; return 1; } if (gds->tokenp[0].token == tMONTH_UNIT) { /* "month" */ gds->HaveRel++; gds->RelMonth += gds->tokenp[0].value; gds->tokenp += 1; return 1; } return 0; } /* * Day of the week specification. */ static int dayphrase(struct gdstate *gds) { if (gds->tokenp[0].token == tDAY) { /* "tues", "wednesday," */ gds->HaveWeekDay++; gds->DayOrdinal = 1; gds->DayNumber = gds->tokenp[0].value; gds->tokenp += 1; if (gds->tokenp[0].token == ',') gds->tokenp += 1; return 1; } if (gds->tokenp[0].token == tUNUMBER && gds->tokenp[1].token == tDAY) { /* "second tues" "3 wed" */ gds->HaveWeekDay++; gds->DayOrdinal = gds->tokenp[0].value; gds->DayNumber = gds->tokenp[1].value; gds->tokenp += 2; return 1; } return 0; } /* * Try to match a phrase using one of the above functions. * This layer also deals with a couple of generic issues. */ static int phrase(struct gdstate *gds) { if (timephrase(gds)) return 1; if (zonephrase(gds)) return 1; if (datephrase(gds)) return 1; if (dayphrase(gds)) return 1; if (relunitphrase(gds)) { if (gds->tokenp[0].token == tAGO) { gds->RelSeconds = -gds->RelSeconds; gds->RelMonth = -gds->RelMonth; gds->tokenp += 1; } return 1; } /* Bare numbers sometimes have meaning. */ if (gds->tokenp[0].token == tUNUMBER) { if (gds->HaveTime && !gds->HaveYear && !gds->HaveRel) { gds->HaveYear++; gds->Year = gds->tokenp[0].value; gds->tokenp += 1; return 1; } if(gds->tokenp[0].value > 10000) { /* "20040301" */ gds->HaveYear++; gds->HaveMonth++; gds->HaveDay++; gds->Day= (gds->tokenp[0].value)%100; gds->Month= (gds->tokenp[0].value/100)%100; gds->Year = gds->tokenp[0].value/10000; gds->tokenp += 1; return 1; } if (gds->tokenp[0].value < 24) { gds->HaveTime++; gds->Hour = gds->tokenp[0].value; gds->Minutes = 0; gds->Seconds = 0; gds->tokenp += 1; return 1; } if ((gds->tokenp[0].value / 100 < 24) && (gds->tokenp[0].value % 100 < 60)) { /* "513" is same as "5:13" */ gds->Hour = gds->tokenp[0].value / 100; gds->Minutes = gds->tokenp[0].value % 100; gds->Seconds = 0; gds->tokenp += 1; return 1; } } return 0; } /* * A dictionary of time words. */ static struct LEXICON { size_t abbrev; const char *name; int type; time_t value; } const TimeWords[] = { /* am/pm */ { 0, "am", tAMPM, tAM }, { 0, "pm", tAMPM, tPM }, /* Month names. */ { 3, "january", tMONTH, 1 }, { 3, "february", tMONTH, 2 }, { 3, "march", tMONTH, 3 }, { 3, "april", tMONTH, 4 }, { 3, "may", tMONTH, 5 }, { 3, "june", tMONTH, 6 }, { 3, "july", tMONTH, 7 }, { 3, "august", tMONTH, 8 }, { 3, "september", tMONTH, 9 }, { 3, "october", tMONTH, 10 }, { 3, "november", tMONTH, 11 }, { 3, "december", tMONTH, 12 }, /* Days of the week. */ { 2, "sunday", tDAY, 0 }, { 3, "monday", tDAY, 1 }, { 2, "tuesday", tDAY, 2 }, { 3, "wednesday", tDAY, 3 }, { 2, "thursday", tDAY, 4 }, { 2, "friday", tDAY, 5 }, { 2, "saturday", tDAY, 6 }, /* Timezones: Offsets are in seconds. */ { 0, "gmt", tZONE, 0*HOUR }, /* Greenwich Mean */ { 0, "ut", tZONE, 0*HOUR }, /* Universal (Coordinated) */ { 0, "utc", tZONE, 0*HOUR }, { 0, "wet", tZONE, 0*HOUR }, /* Western European */ { 0, "bst", tDAYZONE, 0*HOUR }, /* British Summer */ { 0, "wat", tZONE, 1*HOUR }, /* West Africa */ { 0, "at", tZONE, 2*HOUR }, /* Azores */ /* { 0, "bst", tZONE, 3*HOUR }, */ /* Brazil Standard: Conflict */ /* { 0, "gst", tZONE, 3*HOUR }, */ /* Greenland Standard: Conflict*/ { 0, "nft", tZONE, 3*HOUR+30*MINUTE }, /* Newfoundland */ { 0, "nst", tZONE, 3*HOUR+30*MINUTE }, /* Newfoundland Standard */ { 0, "ndt", tDAYZONE, 3*HOUR+30*MINUTE }, /* Newfoundland Daylight */ { 0, "ast", tZONE, 4*HOUR }, /* Atlantic Standard */ { 0, "adt", tDAYZONE, 4*HOUR }, /* Atlantic Daylight */ { 0, "est", tZONE, 5*HOUR }, /* Eastern Standard */ { 0, "edt", tDAYZONE, 5*HOUR }, /* Eastern Daylight */ { 0, "cst", tZONE, 6*HOUR }, /* Central Standard */ { 0, "cdt", tDAYZONE, 6*HOUR }, /* Central Daylight */ { 0, "mst", tZONE, 7*HOUR }, /* Mountain Standard */ { 0, "mdt", tDAYZONE, 7*HOUR }, /* Mountain Daylight */ { 0, "pst", tZONE, 8*HOUR }, /* Pacific Standard */ { 0, "pdt", tDAYZONE, 8*HOUR }, /* Pacific Daylight */ { 0, "yst", tZONE, 9*HOUR }, /* Yukon Standard */ { 0, "ydt", tDAYZONE, 9*HOUR }, /* Yukon Daylight */ { 0, "hst", tZONE, 10*HOUR }, /* Hawaii Standard */ { 0, "hdt", tDAYZONE, 10*HOUR }, /* Hawaii Daylight */ { 0, "cat", tZONE, 10*HOUR }, /* Central Alaska */ { 0, "ahst", tZONE, 10*HOUR }, /* Alaska-Hawaii Standard */ { 0, "nt", tZONE, 11*HOUR }, /* Nome */ { 0, "idlw", tZONE, 12*HOUR }, /* Intl Date Line West */ { 0, "cet", tZONE, -1*HOUR }, /* Central European */ { 0, "met", tZONE, -1*HOUR }, /* Middle European */ { 0, "mewt", tZONE, -1*HOUR }, /* Middle European Winter */ { 0, "mest", tDAYZONE, -1*HOUR }, /* Middle European Summer */ { 0, "swt", tZONE, -1*HOUR }, /* Swedish Winter */ { 0, "sst", tDAYZONE, -1*HOUR }, /* Swedish Summer */ { 0, "fwt", tZONE, -1*HOUR }, /* French Winter */ { 0, "fst", tDAYZONE, -1*HOUR }, /* French Summer */ { 0, "eet", tZONE, -2*HOUR }, /* Eastern Eur, USSR Zone 1 */ { 0, "bt", tZONE, -3*HOUR }, /* Baghdad, USSR Zone 2 */ { 0, "it", tZONE, -3*HOUR-30*MINUTE },/* Iran */ { 0, "zp4", tZONE, -4*HOUR }, /* USSR Zone 3 */ { 0, "zp5", tZONE, -5*HOUR }, /* USSR Zone 4 */ { 0, "ist", tZONE, -5*HOUR-30*MINUTE },/* Indian Standard */ { 0, "zp6", tZONE, -6*HOUR }, /* USSR Zone 5 */ /* { 0, "nst", tZONE, -6.5*HOUR }, */ /* North Sumatra: Conflict */ /* { 0, "sst", tZONE, -7*HOUR }, */ /* So Sumatra, USSR 6: Conflict */ { 0, "wast", tZONE, -7*HOUR }, /* West Australian Standard */ { 0, "wadt", tDAYZONE, -7*HOUR }, /* West Australian Daylight */ { 0, "jt", tZONE, -7*HOUR-30*MINUTE },/* Java (3pm in Cronusland!)*/ { 0, "cct", tZONE, -8*HOUR }, /* China Coast, USSR Zone 7 */ { 0, "jst", tZONE, -9*HOUR }, /* Japan Std, USSR Zone 8 */ { 0, "cast", tZONE, -9*HOUR-30*MINUTE },/* Ctrl Australian Std */ { 0, "cadt", tDAYZONE, -9*HOUR-30*MINUTE },/* Ctrl Australian Daylt */ { 0, "east", tZONE, -10*HOUR }, /* Eastern Australian Std */ { 0, "eadt", tDAYZONE, -10*HOUR }, /* Eastern Australian Daylt */ { 0, "gst", tZONE, -10*HOUR }, /* Guam Std, USSR Zone 9 */ { 0, "nzt", tZONE, -12*HOUR }, /* New Zealand */ { 0, "nzst", tZONE, -12*HOUR }, /* New Zealand Standard */ { 0, "nzdt", tDAYZONE, -12*HOUR }, /* New Zealand Daylight */ { 0, "idle", tZONE, -12*HOUR }, /* Intl Date Line East */ { 0, "dst", tDST, 0 }, /* Time units. */ { 4, "years", tMONTH_UNIT, 12 }, { 5, "months", tMONTH_UNIT, 1 }, { 9, "fortnights", tSEC_UNIT, 14 * DAY }, { 4, "weeks", tSEC_UNIT, 7 * DAY }, { 3, "days", tSEC_UNIT, DAY }, { 4, "hours", tSEC_UNIT, HOUR }, { 3, "minutes", tSEC_UNIT, MINUTE }, { 3, "seconds", tSEC_UNIT, 1 }, /* Relative-time words. */ { 0, "tomorrow", tSEC_UNIT, DAY }, { 0, "yesterday", tSEC_UNIT, -DAY }, { 0, "today", tSEC_UNIT, 0 }, { 0, "now", tSEC_UNIT, 0 }, { 0, "last", tUNUMBER, -1 }, { 0, "this", tSEC_UNIT, 0 }, { 0, "next", tUNUMBER, 2 }, { 0, "first", tUNUMBER, 1 }, { 0, "1st", tUNUMBER, 1 }, /* { 0, "second", tUNUMBER, 2 }, */ { 0, "2nd", tUNUMBER, 2 }, { 0, "third", tUNUMBER, 3 }, { 0, "3rd", tUNUMBER, 3 }, { 0, "fourth", tUNUMBER, 4 }, { 0, "4th", tUNUMBER, 4 }, { 0, "fifth", tUNUMBER, 5 }, { 0, "5th", tUNUMBER, 5 }, { 0, "sixth", tUNUMBER, 6 }, { 0, "seventh", tUNUMBER, 7 }, { 0, "eighth", tUNUMBER, 8 }, { 0, "ninth", tUNUMBER, 9 }, { 0, "tenth", tUNUMBER, 10 }, { 0, "eleventh", tUNUMBER, 11 }, { 0, "twelfth", tUNUMBER, 12 }, { 0, "ago", tAGO, 1 }, /* Military timezones. */ { 0, "a", tZONE, 1*HOUR }, { 0, "b", tZONE, 2*HOUR }, { 0, "c", tZONE, 3*HOUR }, { 0, "d", tZONE, 4*HOUR }, { 0, "e", tZONE, 5*HOUR }, { 0, "f", tZONE, 6*HOUR }, { 0, "g", tZONE, 7*HOUR }, { 0, "h", tZONE, 8*HOUR }, { 0, "i", tZONE, 9*HOUR }, { 0, "k", tZONE, 10*HOUR }, { 0, "l", tZONE, 11*HOUR }, { 0, "m", tZONE, 12*HOUR }, { 0, "n", tZONE, -1*HOUR }, { 0, "o", tZONE, -2*HOUR }, { 0, "p", tZONE, -3*HOUR }, { 0, "q", tZONE, -4*HOUR }, { 0, "r", tZONE, -5*HOUR }, { 0, "s", tZONE, -6*HOUR }, { 0, "t", tZONE, -7*HOUR }, { 0, "u", tZONE, -8*HOUR }, { 0, "v", tZONE, -9*HOUR }, { 0, "w", tZONE, -10*HOUR }, { 0, "x", tZONE, -11*HOUR }, { 0, "y", tZONE, -12*HOUR }, { 0, "z", tZONE, 0*HOUR }, /* End of table. */ { 0, NULL, 0, 0 } }; /* * Year is either: * = A number from 0 to 99, which means a year from 1970 to 2069, or * = The actual year (>=100). */ static time_t Convert(time_t Month, time_t Day, time_t Year, time_t Hours, time_t Minutes, time_t Seconds, time_t Timezone, enum DSTMODE DSTmode) { signed char DaysInMonth[12] = { 31, 0, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31 }; time_t Julian; int i; struct tm *ltime; #if defined(HAVE_LOCALTIME_R) || defined(HAVE_LOCALTIME_S) struct tm tmbuf; #endif if (Year < 69) Year += 2000; else if (Year < 100) Year += 1900; DaysInMonth[1] = Year % 4 == 0 && (Year % 100 != 0 || Year % 400 == 0) ? 29 : 28; - /* Checking for 2038 bogusly assumes that time_t is 32 bits. But - I'm too lazy to try to check for time_t overflow in another way. */ - if (Year < EPOCH || Year >= 2038 + if (Year < EPOCH || (sizeof(time_t) <= 4 && Year >= 2038) || Month < 1 || Month > 12 /* Lint fluff: "conversion from long may lose accuracy" */ || Day < 1 || Day > DaysInMonth[(int)--Month] || Hours < 0 || Hours > 23 || Minutes < 0 || Minutes > 59 || Seconds < 0 || Seconds > 59) return -1; Julian = Day - 1; for (i = 0; i < Month; i++) Julian += DaysInMonth[i]; for (i = EPOCH; i < Year; i++) Julian += 365 + (i % 4 == 0); Julian *= DAY; Julian += Timezone; Julian += Hours * HOUR + Minutes * MINUTE + Seconds; #if defined(HAVE_LOCALTIME_S) ltime = localtime_s(&tmbuf, &Julian) ? NULL : &tmbuf; #elif defined(HAVE_LOCALTIME_R) ltime = localtime_r(&Julian, &tmbuf); #else ltime = localtime(&Julian); #endif if (DSTmode == DSTon || (DSTmode == DSTmaybe && ltime->tm_isdst)) Julian -= HOUR; return Julian; } static time_t DSTcorrect(time_t Start, time_t Future) { time_t StartDay; time_t FutureDay; struct tm *ltime; #if defined(HAVE_LOCALTIME_R) || defined(HAVE_LOCALTIME_S) struct tm tmbuf; #endif #if defined(HAVE_LOCALTIME_S) ltime = localtime_s(&tmbuf, &Start) ? NULL : &tmbuf; #elif defined(HAVE_LOCALTIME_R) ltime = localtime_r(&Start, &tmbuf); #else ltime = localtime(&Start); #endif StartDay = (ltime->tm_hour + 1) % 24; #if defined(HAVE_LOCALTIME_S) ltime = localtime_s(&tmbuf, &Future) ? NULL : &tmbuf; #elif defined(HAVE_LOCALTIME_R) ltime = localtime_r(&Future, &tmbuf); #else ltime = localtime(&Future); #endif FutureDay = (ltime->tm_hour + 1) % 24; return (Future - Start) + (StartDay - FutureDay) * HOUR; } static time_t RelativeDate(time_t Start, time_t zone, int dstmode, time_t DayOrdinal, time_t DayNumber) { struct tm *tm; time_t t, now; #if defined(HAVE_GMTIME_R) || defined(HAVE_GMTIME_S) struct tm tmbuf; #endif t = Start - zone; #if defined(HAVE_GMTIME_S) tm = gmtime_s(&tmbuf, &t) ? NULL : &tmbuf; #elif defined(HAVE_GMTIME_R) tm = gmtime_r(&t, &tmbuf); #else tm = gmtime(&t); #endif now = Start; now += DAY * ((DayNumber - tm->tm_wday + 7) % 7); now += 7 * DAY * (DayOrdinal <= 0 ? DayOrdinal : DayOrdinal - 1); if (dstmode == DSTmaybe) return DSTcorrect(Start, now); return now - Start; } static time_t RelativeMonth(time_t Start, time_t Timezone, time_t RelMonth) { struct tm *tm; time_t Month; time_t Year; #if defined(HAVE_LOCALTIME_R) || defined(HAVE_LOCALTIME_S) struct tm tmbuf; #endif if (RelMonth == 0) return 0; #if defined(HAVE_LOCALTIME_S) tm = localtime_s(&tmbuf, &Start) ? NULL : &tmbuf; #elif defined(HAVE_LOCALTIME_R) tm = localtime_r(&Start, &tmbuf); #else tm = localtime(&Start); #endif Month = 12 * (tm->tm_year + 1900) + tm->tm_mon + RelMonth; Year = Month / 12; Month = Month % 12 + 1; return DSTcorrect(Start, Convert(Month, (time_t)tm->tm_mday, Year, (time_t)tm->tm_hour, (time_t)tm->tm_min, (time_t)tm->tm_sec, Timezone, DSTmaybe)); } /* * Parses and consumes an unsigned number. * Returns 1 if any number is parsed. Otherwise, *value is unchanged. */ static char consume_unsigned_number(const char **in, time_t *value) { char c; if (isdigit((unsigned char)(c = **in))) { for (*value = 0; isdigit((unsigned char)(c = *(*in)++)); ) *value = 10 * *value + c - '0'; (*in)--; return 1; } return 0; } /* * Tokenizer. */ static int nexttoken(const char **in, time_t *value) { char c; char buff[64]; for ( ; ; ) { while (isspace((unsigned char)**in)) ++*in; /* Skip parenthesized comments. */ if (**in == '(') { int Count = 0; do { c = *(*in)++; if (c == '\0') return c; if (c == '(') Count++; else if (c == ')') Count--; } while (Count > 0); continue; } /* Try the next token in the word table first. */ /* This allows us to match "2nd", for example. */ { const char *src = *in; const struct LEXICON *tp; unsigned i = 0; /* Force to lowercase and strip '.' characters. */ while (*src != '\0' && (isalnum((unsigned char)*src) || *src == '.') && i < sizeof(buff)-1) { if (*src != '.') { if (isupper((unsigned char)*src)) buff[i++] = (char)tolower( (unsigned char)*src); else buff[i++] = *src; } src++; } buff[i] = '\0'; /* * Find the first match. If the word can be * abbreviated, make sure we match at least * the minimum abbreviation. */ for (tp = TimeWords; tp->name; tp++) { size_t abbrev = tp->abbrev; if (abbrev == 0) abbrev = strlen(tp->name); if (strlen(buff) >= abbrev && strncmp(tp->name, buff, strlen(buff)) == 0) { /* Skip over token. */ *in = src; /* Return the match. */ *value = tp->value; return tp->type; } } } /* * Not in the word table, maybe it's a number. Note: * Because '-' and '+' have other special meanings, I * don't deal with signed numbers here. */ if (consume_unsigned_number(in, value)) { return (tUNUMBER); } return *(*in)++; } } #define TM_YEAR_ORIGIN 1900 /* Yield A - B, measured in seconds. */ static long difftm (struct tm *a, struct tm *b) { int ay = a->tm_year + (TM_YEAR_ORIGIN - 1); int by = b->tm_year + (TM_YEAR_ORIGIN - 1); long days = ( /* difference in day of year */ a->tm_yday - b->tm_yday /* + intervening leap days */ + ((ay >> 2) - (by >> 2)) - (ay/100 - by/100) + ((ay/100 >> 2) - (by/100 >> 2)) /* + difference in years * 365 */ + (long)(ay-by) * 365 ); return (days * DAY + (a->tm_hour - b->tm_hour) * HOUR + (a->tm_min - b->tm_min) * MINUTE + (a->tm_sec - b->tm_sec)); } /* * Parses a Unix epoch timestamp (seconds). * This supports a subset of what GNU tar accepts from black box testing, * but covers common use cases. */ static time_t parse_unix_epoch(const char *p) { time_t epoch; /* may begin with + */ if (*p == '+') { p++; } /* followed by some number */ if (!consume_unsigned_number(&p, &epoch)) return (time_t)-1; /* ...and nothing else */ if (*p != '\0') return (time_t)-1; return epoch; } /* * * The public function. * * TODO: tokens[] array should be dynamically sized. */ time_t archive_parse_date(time_t now, const char *p) { struct token tokens[256]; struct gdstate _gds; struct token *lasttoken; struct gdstate *gds; struct tm local, *tm; struct tm gmt, *gmt_ptr; time_t Start; time_t tod; long tzone; /* * @-prefixed Unix epoch timestamps (seconds) * Skip the complex tokenizer - We do not want to accept strings like "@tenth" */ if (*p == '@') return parse_unix_epoch(p + 1); /* Clear out the parsed token array. */ memset(tokens, 0, sizeof(tokens)); /* Initialize the parser state. */ memset(&_gds, 0, sizeof(_gds)); gds = &_gds; /* Look up the current time. */ #if defined(HAVE_LOCALTIME_S) tm = localtime_s(&local, &now) ? NULL : &local; #elif defined(HAVE_LOCALTIME_R) tm = localtime_r(&now, &local); #else memset(&local, 0, sizeof(local)); tm = localtime(&now); #endif if (tm == NULL) return -1; #if !defined(HAVE_LOCALTIME_R) && !defined(HAVE_LOCALTIME_S) local = *tm; #endif /* Look up UTC if we can and use that to determine the current * timezone offset. */ #if defined(HAVE_GMTIME_S) gmt_ptr = gmtime_s(&gmt, &now) ? NULL : &gmt; #elif defined(HAVE_GMTIME_R) gmt_ptr = gmtime_r(&now, &gmt); #else memset(&gmt, 0, sizeof(gmt)); gmt_ptr = gmtime(&now); if (gmt_ptr != NULL) { /* Copy, in case localtime and gmtime use the same buffer. */ gmt = *gmt_ptr; } #endif if (gmt_ptr != NULL) tzone = difftm (&gmt, &local); else /* This system doesn't understand timezones; fake it. */ tzone = 0; if(local.tm_isdst) tzone += HOUR; /* Tokenize the input string. */ lasttoken = tokens; while ((lasttoken->token = nexttoken(&p, &lasttoken->value)) != 0) { ++lasttoken; if (lasttoken > tokens + 255) return -1; } gds->tokenp = tokens; /* Match phrases until we run out of input tokens. */ while (gds->tokenp < lasttoken) { if (!phrase(gds)) return -1; } /* Use current local timezone if none was specified. */ if (!gds->HaveZone) { gds->Timezone = tzone; gds->DSTmode = DSTmaybe; } /* If a timezone was specified, use that for generating the default * time components instead of the local timezone. */ if (gds->HaveZone && gmt_ptr != NULL) { now -= gds->Timezone; #if defined(HAVE_GMTIME_S) gmt_ptr = gmtime_s(&gmt, &now) ? NULL : &gmt; #elif defined(HAVE_GMTIME_R) gmt_ptr = gmtime_r(&now, &gmt); #else gmt_ptr = gmtime(&now); #endif if (gmt_ptr != NULL) local = *gmt_ptr; now += gds->Timezone; } if (!gds->HaveYear) gds->Year = local.tm_year + 1900; if (!gds->HaveMonth) gds->Month = local.tm_mon + 1; if (!gds->HaveDay) gds->Day = local.tm_mday; /* Note: No default for hour/min/sec; a specifier that just * gives date always refers to 00:00 on that date. */ /* If we saw more than one time, timezone, weekday, year, month, * or day, then give up. */ if (gds->HaveTime > 1 || gds->HaveZone > 1 || gds->HaveWeekDay > 1 || gds->HaveYear > 1 || gds->HaveMonth > 1 || gds->HaveDay > 1) return -1; /* Compute an absolute time based on whatever absolute information * we collected. */ if (gds->HaveYear || gds->HaveMonth || gds->HaveDay || gds->HaveTime || gds->HaveWeekDay) { Start = Convert(gds->Month, gds->Day, gds->Year, gds->Hour, gds->Minutes, gds->Seconds, gds->Timezone, gds->DSTmode); if (Start < 0) return -1; } else { Start = now; if (!gds->HaveRel) Start -= local.tm_hour * HOUR + local.tm_min * MINUTE + local.tm_sec; } /* Add the relative offset. */ Start += gds->RelSeconds; Start += RelativeMonth(Start, gds->Timezone, gds->RelMonth); /* Adjust for day-of-week offsets. */ if (gds->HaveWeekDay && !(gds->HaveYear || gds->HaveMonth || gds->HaveDay)) { tod = RelativeDate(Start, gds->Timezone, gds->DSTmode, gds->DayOrdinal, gds->DayNumber); Start += tod; } /* -1 is an error indicator, so return 0 instead of -1 if * that's the actual time. */ return Start == -1 ? 0 : Start; } #if defined(TEST) /* ARGSUSED */ int main(int argc, char **argv) { time_t d; time_t now = time(NULL); while (*++argv != NULL) { (void)printf("Input: %s\n", *argv); d = get_date(now, *argv); if (d == -1) (void)printf("Bad format - couldn't convert.\n"); else (void)printf("Output: %s\n", ctime(&d)); } exit(0); /* NOTREACHED */ } #endif /* defined(TEST) */ diff --git a/contrib/libarchive/libarchive/archive_platform.h b/contrib/libarchive/libarchive/archive_platform.h index f30df1104c83..33dc5582b7ed 100644 --- a/contrib/libarchive/libarchive/archive_platform.h +++ b/contrib/libarchive/libarchive/archive_platform.h @@ -1,223 +1,213 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /* !!ONLY FOR USE INTERNALLY TO LIBARCHIVE!! */ /* * This header is the first thing included in any of the libarchive * source files. As far as possible, platform-specific issues should * be dealt with here and not within individual source files. I'm * actively trying to minimize #if blocks within the main source, * since they obfuscate the code. */ #ifndef ARCHIVE_PLATFORM_H_INCLUDED #define ARCHIVE_PLATFORM_H_INCLUDED /* archive.h and archive_entry.h require this. */ #define __LIBARCHIVE_BUILD 1 #if defined(PLATFORM_CONFIG_H) /* Use hand-built config.h in environments that need it. */ #include PLATFORM_CONFIG_H #elif defined(HAVE_CONFIG_H) /* Most POSIX platforms use the 'configure' script to build config.h */ #include "config.h" #else /* Warn if the library hasn't been (automatically or manually) configured. */ #error Oops: No config.h and no pre-built configuration in archive_platform.h. #endif /* On macOS check for some symbols based on the deployment target version. */ #if defined(__APPLE__) # undef HAVE_FUTIMENS # undef HAVE_UTIMENSAT # include # if MAC_OS_X_VERSION_MIN_REQUIRED >= 101300 # define HAVE_FUTIMENS 1 # define HAVE_UTIMENSAT 1 # endif #endif /* For cygwin, to avoid missing LONG, ULONG, PUCHAR, ... definitions */ #ifdef __CYGWIN__ #include #endif /* It should be possible to get rid of this by extending the feature-test * macros to cover Windows API functions, probably along with non-trivial * refactoring of code to find structures that sit more cleanly on top of * either Windows or Posix APIs. */ #if (defined(__WIN32__) || defined(_WIN32) || defined(__WIN32)) && !defined(__CYGWIN__) #include "archive_windows.h" /* The C library on Windows specifies a calling convention for callback * functions and exports; when we interact with them (capture pointers, * call and pass function pointers) we need to match their calling * convention. * This only matters when libarchive is built with /Gr, /Gz or /Gv * (which change the default calling convention.) */ #define __LA_LIBC_CC __cdecl #else #define la_stat(path,stref) stat(path,stref) #define __LA_LIBC_CC #endif /* * The config files define a lot of feature macros. The following * uses those macros to select/define replacements and include key * headers as required. */ /* Try to get standard C99-style integer type definitions. */ #if HAVE_INTTYPES_H #include #endif #if HAVE_STDINT_H #include #endif /* Borland warns about its own constants! */ #if defined(__BORLANDC__) # if HAVE_DECL_UINT64_MAX # undef UINT64_MAX # undef HAVE_DECL_UINT64_MAX # endif # if HAVE_DECL_UINT64_MIN # undef UINT64_MIN # undef HAVE_DECL_UINT64_MIN # endif # if HAVE_DECL_INT64_MAX # undef INT64_MAX # undef HAVE_DECL_INT64_MAX # endif # if HAVE_DECL_INT64_MIN # undef INT64_MIN # undef HAVE_DECL_INT64_MIN # endif #endif /* Some platforms lack the standard *_MAX definitions. */ #if !HAVE_DECL_SIZE_MAX #define SIZE_MAX (~(size_t)0) #endif #if !HAVE_DECL_SSIZE_MAX #define SSIZE_MAX ((ssize_t)(SIZE_MAX >> 1)) #endif #if !HAVE_DECL_UINT32_MAX #define UINT32_MAX (~(uint32_t)0) #endif #if !HAVE_DECL_INT32_MAX #define INT32_MAX ((int32_t)(UINT32_MAX >> 1)) #endif #if !HAVE_DECL_INT32_MIN #define INT32_MIN ((int32_t)(~INT32_MAX)) #endif #if !HAVE_DECL_UINT64_MAX #define UINT64_MAX (~(uint64_t)0) #endif #if !HAVE_DECL_INT64_MAX #define INT64_MAX ((int64_t)(UINT64_MAX >> 1)) #endif #if !HAVE_DECL_INT64_MIN #define INT64_MIN ((int64_t)(~INT64_MAX)) #endif #if !HAVE_DECL_UINTMAX_MAX #define UINTMAX_MAX (~(uintmax_t)0) #endif #if !HAVE_DECL_INTMAX_MAX #define INTMAX_MAX ((intmax_t)(UINTMAX_MAX >> 1)) #endif #if !HAVE_DECL_INTMAX_MIN #define INTMAX_MIN ((intmax_t)(~INTMAX_MAX)) #endif /* Some platforms lack the standard PRIxN/PRIdN definitions. */ #if !HAVE_INTTYPES_H || !defined(PRIx32) || !defined(PRId32) #ifndef PRIx32 #if SIZEOF_INT == 4 #define PRIx32 "x" #elif SIZEOF_LONG == 4 #define PRIx32 "lx" #else #error No suitable 32-bit unsigned integer type found for this platform #endif #endif // PRIx32 #ifndef PRId32 #if SIZEOF_INT == 4 #define PRId32 "d" #elif SIZEOF_LONG == 4 #define PRId32 "ld" #else #error No suitable 32-bit signed integer type found for this platform #endif #endif // PRId32 #endif // !HAVE_INTTYPES_H || !defined(PRIx32) || !defined(PRId32) /* * If we can't restore metadata using a file descriptor, then * for compatibility's sake, close files before trying to restore metadata. */ #if defined(HAVE_FCHMOD) || defined(HAVE_FUTIMES) || defined(HAVE_ACL_SET_FD) || defined(HAVE_ACL_SET_FD_NP) || defined(HAVE_FCHOWN) #define CAN_RESTORE_METADATA_FD #endif -/* - * glibc 2.24 deprecates readdir_r - * bionic c deprecates readdir_r too - */ -#if defined(HAVE_READDIR_R) && (!defined(__GLIBC__) || !defined(__GLIBC_MINOR__) || __GLIBC__ < 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ < 24)) && (!defined(__ANDROID__)) -#define USE_READDIR_R 1 -#else -#undef USE_READDIR_R -#endif - /* Set up defaults for internal error codes. */ #ifndef ARCHIVE_ERRNO_FILE_FORMAT #if HAVE_EFTYPE #define ARCHIVE_ERRNO_FILE_FORMAT EFTYPE #else #if HAVE_EILSEQ #define ARCHIVE_ERRNO_FILE_FORMAT EILSEQ #else #define ARCHIVE_ERRNO_FILE_FORMAT EINVAL #endif #endif #endif #ifndef ARCHIVE_ERRNO_PROGRAMMER #define ARCHIVE_ERRNO_PROGRAMMER EINVAL #endif #ifndef ARCHIVE_ERRNO_MISC #define ARCHIVE_ERRNO_MISC (-1) #endif #if defined(__GNUC__) && (__GNUC__ >= 7) #define __LA_FALLTHROUGH __attribute__((fallthrough)) #else #define __LA_FALLTHROUGH #endif #endif /* !ARCHIVE_PLATFORM_H_INCLUDED */ diff --git a/contrib/libarchive/libarchive/archive_platform_stat.h b/contrib/libarchive/libarchive/archive_platform_stat.h new file mode 100644 index 000000000000..5432b2f6433a --- /dev/null +++ b/contrib/libarchive/libarchive/archive_platform_stat.h @@ -0,0 +1,45 @@ +/*- + * SPDX-License-Identifier: BSD-2-Clause + * + * Copyright (c) 2025 Tobias Stoeckmann + * All rights reserved. + */ + +/* !!ONLY FOR USE INTERNALLY TO LIBARCHIVE!! */ + +#ifndef ARCHIVE_PLATFORM_STAT_H_INCLUDED +#define ARCHIVE_PLATFORM_STAT_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +#if defined(_WIN32) && !defined(__CYGWIN__) +/* We use _lseeki64() on Windows. */ +typedef int64_t la_seek_t; + +struct la_seek_stat { + int64_t st_mtime; + ino_t st_ino; + unsigned short st_mode; + uint32_t st_nlink; + gid_t st_gid; + la_seek_t st_size; + uid_t st_uid; + dev_t st_dev; + dev_t st_rdev; +}; +typedef struct la_seek_stat la_seek_stat_t; + +#define la_seek_fstat(fd, st) __la_seek_fstat((fd), (st)) +#define la_seek_stat(fd, st) __la_seek_stat((fd), (st)) + +#else +typedef off_t la_seek_t; +typedef struct stat la_seek_stat_t; + +#define la_seek_fstat(fd, st) fstat((fd), (st)) +#define la_seek_stat(fd, st) stat((fd), (st)) +#endif + +#endif /* !ARCHIVE_PLATFORM_STAT_H_INCLUDED */ diff --git a/contrib/libarchive/libarchive/archive_private.h b/contrib/libarchive/libarchive/archive_private.h index 050fc63c0b2e..3a926c6886ad 100644 --- a/contrib/libarchive/libarchive/archive_private.h +++ b/contrib/libarchive/libarchive/archive_private.h @@ -1,182 +1,183 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #ifndef ARCHIVE_PRIVATE_H_INCLUDED #define ARCHIVE_PRIVATE_H_INCLUDED #ifndef __LIBARCHIVE_BUILD #ifndef __LIBARCHIVE_TEST #error This header is only to be used internally to libarchive. #endif #endif #if HAVE_ICONV_H #include #endif #include "archive.h" #include "archive_string.h" #if defined(__GNUC__) && (__GNUC__ > 2 || \ (__GNUC__ == 2 && __GNUC_MINOR__ >= 5)) #define __LA_NORETURN __attribute__((__noreturn__)) #elif defined(_MSC_VER) #define __LA_NORETURN __declspec(noreturn) #else #define __LA_NORETURN #endif #if defined(__GNUC__) && (__GNUC__ > 2 || \ (__GNUC__ == 2 && __GNUC_MINOR__ >= 7)) #define __LA_UNUSED __attribute__((__unused__)) #else #define __LA_UNUSED #endif #define ARCHIVE_WRITE_MAGIC (0xb0c5c0deU) #define ARCHIVE_READ_MAGIC (0xdeb0c5U) #define ARCHIVE_WRITE_DISK_MAGIC (0xc001b0c5U) #define ARCHIVE_READ_DISK_MAGIC (0xbadb0c5U) #define ARCHIVE_MATCH_MAGIC (0xcad11c9U) #define ARCHIVE_STATE_NEW 1U #define ARCHIVE_STATE_HEADER 2U #define ARCHIVE_STATE_DATA 4U #define ARCHIVE_STATE_EOF 0x10U #define ARCHIVE_STATE_CLOSED 0x20U #define ARCHIVE_STATE_FATAL 0x8000U #define ARCHIVE_STATE_ANY (0xFFFFU & ~ARCHIVE_STATE_FATAL) struct archive_vtable { int (*archive_close)(struct archive *); int (*archive_free)(struct archive *); int (*archive_write_header)(struct archive *, struct archive_entry *); int (*archive_write_finish_entry)(struct archive *); ssize_t (*archive_write_data)(struct archive *, const void *, size_t); ssize_t (*archive_write_data_block)(struct archive *, const void *, size_t, int64_t); int (*archive_read_next_header)(struct archive *, struct archive_entry **); int (*archive_read_next_header2)(struct archive *, struct archive_entry *); int (*archive_read_data_block)(struct archive *, const void **, size_t *, int64_t *); int (*archive_filter_count)(struct archive *); int64_t (*archive_filter_bytes)(struct archive *, int); int (*archive_filter_code)(struct archive *, int); const char * (*archive_filter_name)(struct archive *, int); }; struct archive_string_conv; struct archive { /* * The magic/state values are used to sanity-check the * client's usage. If an API function is called at a * ridiculous time, or the client passes us an invalid * pointer, these values allow me to catch that. */ unsigned int magic; unsigned int state; /* * Some public API functions depend on the "real" type of the * archive object. */ const struct archive_vtable *vtable; int archive_format; const char *archive_format_name; /* Number of file entries processed. */ int file_count; int archive_error_number; const char *error; struct archive_string error_string; char *current_code; unsigned current_codepage; /* Current ACP(ANSI CodePage). */ unsigned current_oemcp; /* Current OEMCP(OEM CodePage). */ struct archive_string_conv *sconv; /* * Used by archive_read_data() to track blocks and copy * data to client buffers, filling gaps with zero bytes. */ const char *read_data_block; int64_t read_data_offset; int64_t read_data_output_offset; size_t read_data_remaining; /* * Used by formats/filters to determine the amount of data * requested from a call to archive_read_data(). This is only * useful when the format/filter has seek support. */ char read_data_is_posix_read; size_t read_data_requested; }; /* Check magic value and state; return(ARCHIVE_FATAL) if it isn't valid. */ int __archive_check_magic(struct archive *, unsigned int magic, unsigned int state, const char *func); #define archive_check_magic(a, expected_magic, allowed_states, function_name) \ do { \ int magic_test = __archive_check_magic((a), (expected_magic), \ (allowed_states), (function_name)); \ if (magic_test == ARCHIVE_FATAL) \ return ARCHIVE_FATAL; \ } while (0) __LA_NORETURN void __archive_errx(int retvalue, const char *msg); void __archive_ensure_cloexec_flag(int fd); +int __archive_get_tempdir(struct archive_string *); int __archive_mktemp(const char *tmpdir); #if defined(_WIN32) && !defined(__CYGWIN__) int __archive_mkstemp(wchar_t *templates); #else int __archive_mkstemp(char *templates); #endif int __archive_clean(struct archive *); void __archive_reset_read_data(struct archive *); #define err_combine(a,b) ((a) < (b) ? (a) : (b)) #if defined(__BORLANDC__) || (defined(_MSC_VER) && _MSC_VER <= 1300) # define ARCHIVE_LITERAL_LL(x) x##i64 # define ARCHIVE_LITERAL_ULL(x) x##ui64 #else # define ARCHIVE_LITERAL_LL(x) x##ll # define ARCHIVE_LITERAL_ULL(x) x##ull #endif #endif diff --git a/contrib/libarchive/libarchive/archive_read.c b/contrib/libarchive/libarchive/archive_read.c index 50db87017706..c9b9d5981516 100644 --- a/contrib/libarchive/libarchive/archive_read.c +++ b/contrib/libarchive/libarchive/archive_read.c @@ -1,1749 +1,1750 @@ /*- * Copyright (c) 2003-2011 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /* * This file contains the "essential" portions of the read API, that * is, stuff that will probably always be used by any client that * actually needs to read an archive. Optional pieces have been, as * far as possible, separated out into separate files to avoid * needlessly bloating statically-linked clients. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_read_private.h" #define minimum(a, b) (a < b ? a : b) static int choose_filters(struct archive_read *); static int choose_format(struct archive_read *); static int close_filters(struct archive_read *); static int64_t _archive_filter_bytes(struct archive *, int); static int _archive_filter_code(struct archive *, int); static const char *_archive_filter_name(struct archive *, int); static int _archive_filter_count(struct archive *); static int _archive_read_close(struct archive *); static int _archive_read_data_block(struct archive *, const void **, size_t *, int64_t *); static int _archive_read_free(struct archive *); static int _archive_read_next_header(struct archive *, struct archive_entry **); static int _archive_read_next_header2(struct archive *, struct archive_entry *); static int64_t advance_file_pointer(struct archive_read_filter *, int64_t); static const struct archive_vtable archive_read_vtable = { .archive_filter_bytes = _archive_filter_bytes, .archive_filter_code = _archive_filter_code, .archive_filter_name = _archive_filter_name, .archive_filter_count = _archive_filter_count, .archive_read_data_block = _archive_read_data_block, .archive_read_next_header = _archive_read_next_header, .archive_read_next_header2 = _archive_read_next_header2, .archive_free = _archive_read_free, .archive_close = _archive_read_close, }; /* * Allocate, initialize and return a struct archive object. */ struct archive * archive_read_new(void) { struct archive_read *a; a = calloc(1, sizeof(*a)); if (a == NULL) return (NULL); a->archive.magic = ARCHIVE_READ_MAGIC; a->archive.state = ARCHIVE_STATE_NEW; a->entry = archive_entry_new2(&a->archive); a->archive.vtable = &archive_read_vtable; a->passphrases.last = &a->passphrases.first; return (&a->archive); } /* * Record the do-not-extract-to file. This belongs in archive_read_extract.c. */ void archive_read_extract_set_skip_file(struct archive *_a, la_int64_t d, la_int64_t i) { struct archive_read *a = (struct archive_read *)_a; if (ARCHIVE_OK != __archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_ANY, "archive_read_extract_set_skip_file")) return; a->skip_file_set = 1; a->skip_file_dev = d; a->skip_file_ino = i; } /* * Open the archive */ int archive_read_open(struct archive *a, void *client_data, archive_open_callback *client_opener, archive_read_callback *client_reader, archive_close_callback *client_closer) { /* Old archive_read_open() is just a thin shell around * archive_read_open1. */ archive_read_set_open_callback(a, client_opener); archive_read_set_read_callback(a, client_reader); archive_read_set_close_callback(a, client_closer); archive_read_set_callback_data(a, client_data); return archive_read_open1(a); } int archive_read_open2(struct archive *a, void *client_data, archive_open_callback *client_opener, archive_read_callback *client_reader, archive_skip_callback *client_skipper, archive_close_callback *client_closer) { /* Old archive_read_open2() is just a thin shell around * archive_read_open1. */ archive_read_set_callback_data(a, client_data); archive_read_set_open_callback(a, client_opener); archive_read_set_read_callback(a, client_reader); archive_read_set_skip_callback(a, client_skipper); archive_read_set_close_callback(a, client_closer); return archive_read_open1(a); } static ssize_t client_read_proxy(struct archive_read_filter *self, const void **buff) { ssize_t r; r = (self->archive->client.reader)(&self->archive->archive, self->data, buff); return (r); } static int64_t client_skip_proxy(struct archive_read_filter *self, int64_t request) { if (request < 0) __archive_errx(1, "Negative skip requested."); if (request == 0) return 0; if (self->archive->client.skipper != NULL) { int64_t total = 0; for (;;) { int64_t get, ask = request; get = (self->archive->client.skipper) (&self->archive->archive, self->data, ask); total += get; if (get == 0 || get == request) return (total); if (get > request) return ARCHIVE_FATAL; request -= get; } } else if (self->archive->client.seeker != NULL && request > 64 * 1024) { /* If the client provided a seeker but not a skipper, * we can use the seeker to skip forward. * * Note: This isn't always a good idea. The client * skipper is allowed to skip by less than requested * if it needs to maintain block alignment. The * seeker is not allowed to play such games, so using * the seeker here may be a performance loss compared * to just reading and discarding. That's why we * only do this for skips of over 64k. */ int64_t before = self->position; int64_t after = (self->archive->client.seeker) (&self->archive->archive, self->data, request, SEEK_CUR); if (after != before + request) return ARCHIVE_FATAL; return after - before; } return 0; } static int64_t client_seek_proxy(struct archive_read_filter *self, int64_t offset, int whence) { /* DO NOT use the skipper here! If we transparently handled * forward seek here by using the skipper, that will break * other libarchive code that assumes a successful forward * seek means it can also seek backwards. */ if (self->archive->client.seeker == NULL) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Current client reader does not support seeking a device"); return (ARCHIVE_FAILED); } return (self->archive->client.seeker)(&self->archive->archive, self->data, offset, whence); } static int read_client_close_proxy(struct archive_read *a) { int r = ARCHIVE_OK, r2; unsigned int i; if (a->client.closer == NULL) return (r); for (i = 0; i < a->client.nodes; i++) { r2 = (a->client.closer) ((struct archive *)a, a->client.dataset[i].data); if (r > r2) r = r2; } return (r); } static int client_close_proxy(struct archive_read_filter *self) { return read_client_close_proxy(self->archive); } static int client_open_proxy(struct archive_read_filter *self) { int r = ARCHIVE_OK; if (self->archive->client.opener != NULL) r = (self->archive->client.opener)( (struct archive *)self->archive, self->data); return (r); } static int client_switch_proxy(struct archive_read_filter *self, unsigned int iindex) { int r1 = ARCHIVE_OK, r2 = ARCHIVE_OK; void *data2 = NULL; /* Don't do anything if already in the specified data node */ if (self->archive->client.cursor == iindex) return (ARCHIVE_OK); self->archive->client.cursor = iindex; data2 = self->archive->client.dataset[self->archive->client.cursor].data; if (self->archive->client.switcher != NULL) { r1 = r2 = (self->archive->client.switcher) ((struct archive *)self->archive, self->data, data2); self->data = data2; } else { /* Attempt to call close and open instead */ if (self->archive->client.closer != NULL) r1 = (self->archive->client.closer) ((struct archive *)self->archive, self->data); self->data = data2; r2 = client_open_proxy(self); } return (r1 < r2) ? r1 : r2; } int archive_read_set_open_callback(struct archive *_a, archive_open_callback *client_opener) { struct archive_read *a = (struct archive_read *)_a; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_set_open_callback"); a->client.opener = client_opener; return ARCHIVE_OK; } int archive_read_set_read_callback(struct archive *_a, archive_read_callback *client_reader) { struct archive_read *a = (struct archive_read *)_a; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_set_read_callback"); a->client.reader = client_reader; return ARCHIVE_OK; } int archive_read_set_skip_callback(struct archive *_a, archive_skip_callback *client_skipper) { struct archive_read *a = (struct archive_read *)_a; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_set_skip_callback"); a->client.skipper = client_skipper; return ARCHIVE_OK; } int archive_read_set_seek_callback(struct archive *_a, archive_seek_callback *client_seeker) { struct archive_read *a = (struct archive_read *)_a; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_set_seek_callback"); a->client.seeker = client_seeker; return ARCHIVE_OK; } int archive_read_set_close_callback(struct archive *_a, archive_close_callback *client_closer) { struct archive_read *a = (struct archive_read *)_a; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_set_close_callback"); a->client.closer = client_closer; return ARCHIVE_OK; } int archive_read_set_switch_callback(struct archive *_a, archive_switch_callback *client_switcher) { struct archive_read *a = (struct archive_read *)_a; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_set_switch_callback"); a->client.switcher = client_switcher; return ARCHIVE_OK; } int archive_read_set_callback_data(struct archive *_a, void *client_data) { return archive_read_set_callback_data2(_a, client_data, 0); } int archive_read_set_callback_data2(struct archive *_a, void *client_data, unsigned int iindex) { struct archive_read *a = (struct archive_read *)_a; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_set_callback_data2"); if (a->client.nodes == 0) { a->client.dataset = (struct archive_read_data_node *) calloc(1, sizeof(*a->client.dataset)); if (a->client.dataset == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory."); return ARCHIVE_FATAL; } a->client.nodes = 1; } if (iindex > a->client.nodes - 1) { archive_set_error(&a->archive, EINVAL, "Invalid index specified."); return ARCHIVE_FATAL; } a->client.dataset[iindex].data = client_data; a->client.dataset[iindex].begin_position = -1; a->client.dataset[iindex].total_size = -1; return ARCHIVE_OK; } int archive_read_add_callback_data(struct archive *_a, void *client_data, unsigned int iindex) { struct archive_read *a = (struct archive_read *)_a; void *p; unsigned int i; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_add_callback_data"); if (iindex > a->client.nodes) { archive_set_error(&a->archive, EINVAL, "Invalid index specified."); return ARCHIVE_FATAL; } p = realloc(a->client.dataset, sizeof(*a->client.dataset) * (++(a->client.nodes))); if (p == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory."); return ARCHIVE_FATAL; } a->client.dataset = (struct archive_read_data_node *)p; for (i = a->client.nodes - 1; i > iindex; i--) { a->client.dataset[i].data = a->client.dataset[i-1].data; a->client.dataset[i].begin_position = -1; a->client.dataset[i].total_size = -1; } a->client.dataset[iindex].data = client_data; a->client.dataset[iindex].begin_position = -1; a->client.dataset[iindex].total_size = -1; return ARCHIVE_OK; } int archive_read_append_callback_data(struct archive *_a, void *client_data) { struct archive_read *a = (struct archive_read *)_a; return archive_read_add_callback_data(_a, client_data, a->client.nodes); } int archive_read_prepend_callback_data(struct archive *_a, void *client_data) { return archive_read_add_callback_data(_a, client_data, 0); } static const struct archive_read_filter_vtable none_reader_vtable = { .read = client_read_proxy, .close = client_close_proxy, }; int archive_read_open1(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct archive_read_filter *filter, *tmp; int slot, e = ARCHIVE_OK; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_open"); archive_clear_error(&a->archive); if (a->client.reader == NULL) { archive_set_error(&a->archive, EINVAL, "No reader function provided to archive_read_open"); a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } /* Open data source. */ if (a->client.opener != NULL) { e = (a->client.opener)(&a->archive, a->client.dataset[0].data); if (e != 0) { /* If the open failed, call the closer to clean up. */ read_client_close_proxy(a); return (e); } } filter = calloc(1, sizeof(*filter)); if (filter == NULL) return (ARCHIVE_FATAL); filter->bidder = NULL; filter->upstream = NULL; filter->archive = a; filter->data = a->client.dataset[0].data; filter->vtable = &none_reader_vtable; filter->name = "none"; filter->code = ARCHIVE_FILTER_NONE; filter->can_skip = 1; filter->can_seek = 1; a->client.dataset[0].begin_position = 0; if (!a->filter || !a->bypass_filter_bidding) { a->filter = filter; /* Build out the input pipeline. */ e = choose_filters(a); if (e < ARCHIVE_WARN) { a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } } else { /* Need to add "NONE" type filter at the end of the filter chain */ tmp = a->filter; while (tmp->upstream) tmp = tmp->upstream; tmp->upstream = filter; } if (!a->format) { slot = choose_format(a); if (slot < 0) { close_filters(a); a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } a->format = &(a->formats[slot]); } a->archive.state = ARCHIVE_STATE_HEADER; /* Ensure libarchive starts from the first node in a multivolume set */ client_switch_proxy(a->filter, 0); return (e); } /* * Allow each registered stream transform to bid on whether * it wants to handle this stream. Repeat until we've finished * building the pipeline. */ /* We won't build a filter pipeline with more stages than this. */ #define MAX_NUMBER_FILTERS 25 static int choose_filters(struct archive_read *a) { int number_bidders, i, bid, best_bid, number_filters; struct archive_read_filter_bidder *bidder, *best_bidder; struct archive_read_filter *filter; ssize_t avail; int r; for (number_filters = 0; number_filters < MAX_NUMBER_FILTERS; ++number_filters) { number_bidders = sizeof(a->bidders) / sizeof(a->bidders[0]); best_bid = 0; best_bidder = NULL; bidder = a->bidders; for (i = 0; i < number_bidders; i++, bidder++) { if (bidder->vtable == NULL) continue; bid = (bidder->vtable->bid)(bidder, a->filter); if (bid > best_bid) { best_bid = bid; best_bidder = bidder; } } /* If no bidder, we're done. */ if (best_bidder == NULL) { /* Verify the filter by asking it for some data. */ __archive_read_filter_ahead(a->filter, 1, &avail); if (avail < 0) { __archive_read_free_filters(a); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } - filter - = calloc(1, sizeof(*filter)); + filter = calloc(1, sizeof(*filter)); if (filter == NULL) return (ARCHIVE_FATAL); filter->bidder = best_bidder; filter->archive = a; filter->upstream = a->filter; a->filter = filter; r = (best_bidder->vtable->init)(a->filter); if (r != ARCHIVE_OK) { __archive_read_free_filters(a); return (ARCHIVE_FATAL); } } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Input requires too many filters for decoding"); return (ARCHIVE_FATAL); } int __archive_read_header(struct archive_read *a, struct archive_entry *entry) { if (!a->filter->vtable->read_header) return (ARCHIVE_OK); return a->filter->vtable->read_header(a->filter, entry); } /* * Read header of next entry. */ static int _archive_read_next_header2(struct archive *_a, struct archive_entry *entry) { struct archive_read *a = (struct archive_read *)_a; int r1 = ARCHIVE_OK, r2; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_read_next_header"); archive_entry_clear(entry); archive_clear_error(&a->archive); /* * If client didn't consume entire data, skip any remainder * (This is especially important for GNU incremental directories.) */ if (a->archive.state == ARCHIVE_STATE_DATA) { r1 = archive_read_data_skip(&a->archive); if (r1 == ARCHIVE_EOF) archive_set_error(&a->archive, EIO, "Premature end-of-file."); if (r1 == ARCHIVE_EOF || r1 == ARCHIVE_FATAL) { a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } } /* Record start-of-header offset in uncompressed stream. */ a->header_position = a->filter->position; ++_a->file_count; r2 = (a->format->read_header)(a, entry); /* * EOF and FATAL are persistent at this layer. By * modifying the state, we guarantee that future calls to * read a header or read data will fail. */ switch (r2) { case ARCHIVE_EOF: a->archive.state = ARCHIVE_STATE_EOF; --_a->file_count;/* Revert a file counter. */ break; case ARCHIVE_OK: a->archive.state = ARCHIVE_STATE_DATA; break; case ARCHIVE_WARN: a->archive.state = ARCHIVE_STATE_DATA; break; case ARCHIVE_RETRY: break; case ARCHIVE_FATAL: a->archive.state = ARCHIVE_STATE_FATAL; break; } __archive_reset_read_data(&a->archive); a->data_start_node = a->client.cursor; /* EOF always wins; otherwise return the worst error. */ return (r2 < r1 || r2 == ARCHIVE_EOF) ? r2 : r1; } static int _archive_read_next_header(struct archive *_a, struct archive_entry **entryp) { int ret; struct archive_read *a = (struct archive_read *)_a; *entryp = NULL; ret = _archive_read_next_header2(_a, a->entry); *entryp = a->entry; return ret; } /* * Allow each registered format to bid on whether it wants to handle * the next entry. Return index of winning bidder. */ static int choose_format(struct archive_read *a) { int slots; int i; int bid, best_bid; int best_bid_slot; slots = sizeof(a->formats) / sizeof(a->formats[0]); best_bid = -1; best_bid_slot = -1; /* Set up a->format for convenience of bidders. */ a->format = &(a->formats[0]); for (i = 0; i < slots; i++, a->format++) { if (a->format->bid) { bid = (a->format->bid)(a, best_bid); if (bid == ARCHIVE_FATAL) return (ARCHIVE_FATAL); if (a->filter->position != 0) __archive_read_seek(a, 0, SEEK_SET); if ((bid > best_bid) || (best_bid_slot < 0)) { best_bid = bid; best_bid_slot = i; } } } /* * There were no bidders; this is a serious programmer error * and demands a quick and definitive abort. */ if (best_bid_slot < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "No formats registered"); return (ARCHIVE_FATAL); } /* * There were bidders, but no non-zero bids; this means we * can't support this stream. */ if (best_bid < 1) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unrecognized archive format"); return (ARCHIVE_FATAL); } return (best_bid_slot); } /* * Return the file offset (within the uncompressed data stream) where * the last header started. */ la_int64_t archive_read_header_position(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_ANY, "archive_read_header_position"); return (a->header_position); } /* * Returns 1 if the archive contains at least one encrypted entry. * If the archive format not support encryption at all * ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED is returned. * If for any other reason (e.g. not enough data read so far) * we cannot say whether there are encrypted entries, then * ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW is returned. * In general, this function will return values below zero when the * reader is uncertain or totally incapable of encryption support. * When this function returns 0 you can be sure that the reader * supports encryption detection but no encrypted entries have * been found yet. * * NOTE: If the metadata/header of an archive is also encrypted, you * cannot rely on the number of encrypted entries. That is why this * function does not return the number of encrypted entries but# * just shows that there are some. */ int archive_read_has_encrypted_entries(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; int format_supports_encryption = archive_read_format_capabilities(_a) & (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); if (!_a || !format_supports_encryption) { /* Format in general doesn't support encryption */ return ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED; } /* A reader potentially has read enough data now. */ if (a->format && a->format->has_encrypted_entries) { return (a->format->has_encrypted_entries)(a); } /* For any other reason we cannot say how many entries are there. */ return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; } /* * Returns a bitmask of capabilities that are supported by the archive format reader. * If the reader has no special capabilities, ARCHIVE_READ_FORMAT_CAPS_NONE is returned. */ int archive_read_format_capabilities(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; if (a && a->format && a->format->format_capabilties) { return (a->format->format_capabilties)(a); } return ARCHIVE_READ_FORMAT_CAPS_NONE; } /* * Read data from an archive entry, using a read(2)-style interface. * This is a convenience routine that just calls * archive_read_data_block and copies the results into the client * buffer, filling any gaps with zero bytes. Clients using this * API can be completely ignorant of sparse-file issues; sparse files * will simply be padded with nulls. * * DO NOT intermingle calls to this function and archive_read_data_block * to read a single entry body. */ la_ssize_t archive_read_data(struct archive *_a, void *buff, size_t s) { struct archive *a = (struct archive *)_a; char *dest; const void *read_buf; size_t bytes_read; size_t len; int r; bytes_read = 0; dest = (char *)buff; while (s > 0) { if (a->read_data_offset == a->read_data_output_offset && a->read_data_remaining == 0) { read_buf = a->read_data_block; a->read_data_is_posix_read = 1; a->read_data_requested = s; r = archive_read_data_block(a, &read_buf, &a->read_data_remaining, &a->read_data_offset); a->read_data_block = read_buf; - if (r == ARCHIVE_EOF) + if (r == ARCHIVE_EOF && + a->read_data_offset == a->read_data_output_offset && + a->read_data_remaining == 0) return (bytes_read); /* * Error codes are all negative, so the status * return here cannot be confused with a valid * byte count. (ARCHIVE_OK is zero.) */ if (r < ARCHIVE_OK) return (r); } if (a->read_data_offset < a->read_data_output_offset) { archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, "Encountered out-of-order sparse blocks"); return (ARCHIVE_RETRY); } /* Compute the amount of zero padding needed. */ if (a->read_data_output_offset + (int64_t)s < a->read_data_offset) { len = s; } else if (a->read_data_output_offset < a->read_data_offset) { len = (size_t)(a->read_data_offset - a->read_data_output_offset); } else len = 0; /* Add zeroes. */ memset(dest, 0, len); s -= len; a->read_data_output_offset += len; dest += len; bytes_read += len; /* Copy data if there is any space left. */ if (s > 0) { len = a->read_data_remaining; if (len > s) len = s; if (len) { memcpy(dest, a->read_data_block, len); s -= len; a->read_data_block += len; a->read_data_remaining -= len; a->read_data_output_offset += len; a->read_data_offset += len; dest += len; bytes_read += len; } } } a->read_data_is_posix_read = 0; a->read_data_requested = 0; return (bytes_read); } /* * Reset the read_data_* variables, used for starting a new entry. */ void __archive_reset_read_data(struct archive * a) { a->read_data_output_offset = 0; a->read_data_remaining = 0; a->read_data_is_posix_read = 0; a->read_data_requested = 0; /* extra resets, from rar.c */ a->read_data_block = NULL; a->read_data_offset = 0; } /* * Skip over all remaining data in this entry. */ int archive_read_data_skip(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; int r; const void *buff; size_t size; int64_t offset; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_DATA, "archive_read_data_skip"); if (a->format->read_data_skip != NULL) r = (a->format->read_data_skip)(a); else { while ((r = archive_read_data_block(&a->archive, &buff, &size, &offset)) == ARCHIVE_OK) ; } if (r == ARCHIVE_EOF) r = ARCHIVE_OK; a->archive.state = ARCHIVE_STATE_HEADER; return (r); } la_int64_t archive_seek_data(struct archive *_a, int64_t offset, int whence) { struct archive_read *a = (struct archive_read *)_a; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_DATA, "archive_seek_data_block"); if (a->format->seek_data == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Internal error: " "No format_seek_data_block function registered"); return (ARCHIVE_FATAL); } return (a->format->seek_data)(a, offset, whence); } /* * Read the next block of entry data from the archive. * This is a zero-copy interface; the client receives a pointer, * size, and file offset of the next available block of data. * * Returns ARCHIVE_OK if the operation is successful, ARCHIVE_EOF if * the end of entry is encountered. */ static int _archive_read_data_block(struct archive *_a, const void **buff, size_t *size, int64_t *offset) { struct archive_read *a = (struct archive_read *)_a; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_DATA, "archive_read_data_block"); if (a->format->read_data == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Internal error: " "No format->read_data function registered"); return (ARCHIVE_FATAL); } return (a->format->read_data)(a, buff, size, offset); } static int close_filters(struct archive_read *a) { struct archive_read_filter *f = a->filter; int r = ARCHIVE_OK; /* Close each filter in the pipeline. */ while (f != NULL) { struct archive_read_filter *t = f->upstream; if (!f->closed && f->vtable != NULL) { int r1 = (f->vtable->close)(f); f->closed = 1; if (r1 < r) r = r1; } free(f->buffer); f->buffer = NULL; f = t; } return r; } void __archive_read_free_filters(struct archive_read *a) { /* Make sure filters are closed and their buffers are freed */ close_filters(a); while (a->filter != NULL) { struct archive_read_filter *t = a->filter->upstream; free(a->filter); a->filter = t; } } /* * return the count of # of filters in use */ static int _archive_filter_count(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct archive_read_filter *p = a->filter; int count = 0; while(p) { count++; p = p->upstream; } return count; } /* * Close the file and all I/O. */ static int _archive_read_close(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; int r = ARCHIVE_OK, r1 = ARCHIVE_OK; archive_check_magic(&a->archive, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_read_close"); if (a->archive.state == ARCHIVE_STATE_CLOSED) return (ARCHIVE_OK); archive_clear_error(&a->archive); a->archive.state = ARCHIVE_STATE_CLOSED; /* TODO: Clean up the formatters. */ /* Release the filter objects. */ r1 = close_filters(a); if (r1 < r) r = r1; return (r); } /* * Release memory and other resources. */ static int _archive_read_free(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct archive_read_passphrase *p; int i, n; int slots; int r = ARCHIVE_OK; if (_a == NULL) return (ARCHIVE_OK); archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_read_free"); if (a->archive.state != ARCHIVE_STATE_CLOSED && a->archive.state != ARCHIVE_STATE_FATAL) r = archive_read_close(&a->archive); /* Call cleanup functions registered by optional components. */ if (a->cleanup_archive_extract != NULL) r = (a->cleanup_archive_extract)(a); /* Cleanup format-specific data. */ slots = sizeof(a->formats) / sizeof(a->formats[0]); for (i = 0; i < slots; i++) { a->format = &(a->formats[i]); if (a->formats[i].cleanup) (a->formats[i].cleanup)(a); } /* Free the filters */ __archive_read_free_filters(a); /* Release the bidder objects. */ n = sizeof(a->bidders)/sizeof(a->bidders[0]); for (i = 0; i < n; i++) { if (a->bidders[i].vtable == NULL || a->bidders[i].vtable->free == NULL) continue; (a->bidders[i].vtable->free)(&a->bidders[i]); } /* Release passphrase list. */ p = a->passphrases.first; while (p != NULL) { struct archive_read_passphrase *np = p->next; /* A passphrase should be cleaned. */ memset(p->passphrase, 0, strlen(p->passphrase)); free(p->passphrase); free(p); p = np; } archive_string_free(&a->archive.error_string); archive_entry_free(a->entry); a->archive.magic = 0; __archive_clean(&a->archive); free(a->client.dataset); free(a); return (r); } static struct archive_read_filter * get_filter(struct archive *_a, int n) { struct archive_read *a = (struct archive_read *)_a; struct archive_read_filter *f = a->filter; /* We use n == -1 for 'the last filter', which is always the * client proxy. */ if (n == -1 && f != NULL) { struct archive_read_filter *last = f; f = f->upstream; while (f != NULL) { last = f; f = f->upstream; } return (last); } if (n < 0) return NULL; while (n > 0 && f != NULL) { f = f->upstream; --n; } return (f); } static int _archive_filter_code(struct archive *_a, int n) { struct archive_read_filter *f = get_filter(_a, n); return f == NULL ? -1 : f->code; } static const char * _archive_filter_name(struct archive *_a, int n) { struct archive_read_filter *f = get_filter(_a, n); return f != NULL ? f->name : NULL; } static int64_t _archive_filter_bytes(struct archive *_a, int n) { struct archive_read_filter *f = get_filter(_a, n); return f == NULL ? -1 : f->position; } /* * Used internally by read format handlers to register their bid and * initialization functions. */ int __archive_read_register_format(struct archive_read *a, void *format_data, const char *name, int (*bid)(struct archive_read *, int), int (*options)(struct archive_read *, const char *, const char *), int (*read_header)(struct archive_read *, struct archive_entry *), int (*read_data)(struct archive_read *, const void **, size_t *, int64_t *), int (*read_data_skip)(struct archive_read *), int64_t (*seek_data)(struct archive_read *, int64_t, int), int (*cleanup)(struct archive_read *), int (*format_capabilities)(struct archive_read *), int (*has_encrypted_entries)(struct archive_read *)) { int i, number_slots; archive_check_magic(&a->archive, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "__archive_read_register_format"); number_slots = sizeof(a->formats) / sizeof(a->formats[0]); for (i = 0; i < number_slots; i++) { if (a->formats[i].bid == bid) return (ARCHIVE_WARN); /* We've already installed */ if (a->formats[i].bid == NULL) { a->formats[i].bid = bid; a->formats[i].options = options; a->formats[i].read_header = read_header; a->formats[i].read_data = read_data; a->formats[i].read_data_skip = read_data_skip; a->formats[i].seek_data = seek_data; a->formats[i].cleanup = cleanup; a->formats[i].data = format_data; a->formats[i].name = name; a->formats[i].format_capabilties = format_capabilities; a->formats[i].has_encrypted_entries = has_encrypted_entries; return (ARCHIVE_OK); } } archive_set_error(&a->archive, ENOMEM, "Not enough slots for format registration"); return (ARCHIVE_FATAL); } /* * Used internally by decompression routines to register their bid and * initialization functions. */ int __archive_read_register_bidder(struct archive_read *a, void *bidder_data, const char *name, const struct archive_read_filter_bidder_vtable *vtable) { struct archive_read_filter_bidder *bidder; int i, number_slots; archive_check_magic(&a->archive, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "__archive_read_register_bidder"); number_slots = sizeof(a->bidders) / sizeof(a->bidders[0]); for (i = 0; i < number_slots; i++) { if (a->bidders[i].vtable != NULL) continue; memset(a->bidders + i, 0, sizeof(a->bidders[0])); bidder = (a->bidders + i); bidder->data = bidder_data; bidder->name = name; bidder->vtable = vtable; if (bidder->vtable->bid == NULL || bidder->vtable->init == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Internal error: " "no bid/init for filter bidder"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } archive_set_error(&a->archive, ENOMEM, "Not enough slots for filter registration"); return (ARCHIVE_FATAL); } /* * The next section implements the peek/consume internal I/O * system used by archive readers. This system allows simple * read-ahead for consumers while preserving zero-copy operation * most of the time. * * The two key operations: * * The read-ahead function returns a pointer to a block of data * that satisfies a minimum request. * * The consume function advances the file pointer. * * In the ideal case, filters generate blocks of data * and __archive_read_ahead() just returns pointers directly into * those blocks. Then __archive_read_consume() just bumps those * pointers. Only if your request would span blocks does the I/O * layer use a copy buffer to provide you with a contiguous block of * data. * * A couple of useful idioms: * * "I just want some data." Ask for 1 byte and pay attention to * the "number of bytes available" from __archive_read_ahead(). * Consume whatever you actually use. * * "I want to output a large block of data." As above, ask for 1 byte, * emit all that's available (up to whatever limit you have), consume * it all, then repeat until you're done. This effectively means that * you're passing along the blocks that came from your provider. * * "I want to peek ahead by a large amount." Ask for 4k or so, then * double and repeat until you get an error or have enough. Note * that the I/O layer will likely end up expanding its copy buffer * to fit your request, so use this technique cautiously. This * technique is used, for example, by some of the format tasting * code that has uncertain look-ahead needs. */ /* * Looks ahead in the input stream: * * If 'avail' pointer is provided, that returns number of bytes available * in the current buffer, which may be much larger than requested. * * If end-of-file, *avail gets set to zero. * * If error, *avail gets error code. * * If request can be met, returns pointer to data. * * If minimum request cannot be met, returns NULL. * * Note: If you just want "some data", ask for 1 byte and pay attention * to *avail, which will have the actual amount available. If you * know exactly how many bytes you need, just ask for that and treat * a NULL return as an error. * * Important: This does NOT move the file pointer. See * __archive_read_consume() below. */ const void * __archive_read_ahead(struct archive_read *a, size_t min, ssize_t *avail) { return (__archive_read_filter_ahead(a->filter, min, avail)); } const void * __archive_read_filter_ahead(struct archive_read_filter *filter, size_t min, ssize_t *avail) { ssize_t bytes_read; size_t tocopy; if (filter->fatal) { if (avail) *avail = ARCHIVE_FATAL; return (NULL); } /* * Keep pulling more data until we can satisfy the request. */ for (;;) { /* * If we can satisfy from the copy buffer (and the * copy buffer isn't empty), we're done. In particular, * note that min == 0 is a perfectly well-defined * request. */ if (filter->avail >= min && filter->avail > 0) { if (avail != NULL) *avail = filter->avail; return (filter->next); } /* * We can satisfy directly from client buffer if everything * currently in the copy buffer is still in the client buffer. */ if (filter->client_total >= filter->client_avail + filter->avail && filter->client_avail + filter->avail >= min) { /* "Roll back" to client buffer. */ filter->client_avail += filter->avail; filter->client_next -= filter->avail; /* Copy buffer is now empty. */ filter->avail = 0; filter->next = filter->buffer; /* Return data from client buffer. */ if (avail != NULL) *avail = filter->client_avail; return (filter->client_next); } /* Move data forward in copy buffer if necessary. */ if (filter->next > filter->buffer && filter->next + min > filter->buffer + filter->buffer_size) { if (filter->avail > 0) memmove(filter->buffer, filter->next, filter->avail); filter->next = filter->buffer; } /* If we've used up the client data, get more. */ if (filter->client_avail <= 0) { if (filter->end_of_file) { if (avail != NULL) *avail = filter->avail; return (NULL); } bytes_read = (filter->vtable->read)(filter, &filter->client_buff); if (bytes_read < 0) { /* Read error. */ filter->client_total = filter->client_avail = 0; filter->client_next = filter->client_buff = NULL; filter->fatal = 1; if (avail != NULL) *avail = ARCHIVE_FATAL; return (NULL); } if (bytes_read == 0) { /* Check for another client object first */ if (filter->archive->client.cursor != filter->archive->client.nodes - 1) { if (client_switch_proxy(filter, filter->archive->client.cursor + 1) == ARCHIVE_OK) continue; } /* Premature end-of-file. */ filter->client_total = filter->client_avail = 0; filter->client_next = filter->client_buff = NULL; filter->end_of_file = 1; /* Return whatever we do have. */ if (avail != NULL) *avail = filter->avail; return (NULL); } filter->client_total = bytes_read; filter->client_avail = filter->client_total; filter->client_next = filter->client_buff; } else { /* * We can't satisfy the request from the copy * buffer or the existing client data, so we * need to copy more client data over to the * copy buffer. */ /* Ensure the buffer is big enough. */ if (min > filter->buffer_size) { size_t s, t; char *p; /* Double the buffer; watch for overflow. */ s = t = filter->buffer_size; if (s == 0) s = min; while (s < min) { t *= 2; if (t <= s) { /* Integer overflow! */ archive_set_error( &filter->archive->archive, ENOMEM, "Unable to allocate copy" " buffer"); filter->fatal = 1; if (avail != NULL) *avail = ARCHIVE_FATAL; return (NULL); } s = t; } /* Now s >= min, so allocate a new buffer. */ p = malloc(s); if (p == NULL) { archive_set_error( &filter->archive->archive, ENOMEM, "Unable to allocate copy buffer"); filter->fatal = 1; if (avail != NULL) *avail = ARCHIVE_FATAL; return (NULL); } /* Move data into newly-enlarged buffer. */ if (filter->avail > 0) memmove(p, filter->next, filter->avail); free(filter->buffer); filter->next = filter->buffer = p; filter->buffer_size = s; } /* We can add client data to copy buffer. */ /* First estimate: copy to fill rest of buffer. */ tocopy = (filter->buffer + filter->buffer_size) - (filter->next + filter->avail); /* Don't waste time buffering more than we need to. */ if (tocopy + filter->avail > min) tocopy = min - filter->avail; /* Don't copy more than is available. */ if (tocopy > filter->client_avail) tocopy = filter->client_avail; memcpy(filter->next + filter->avail, filter->client_next, tocopy); /* Remove this data from client buffer. */ filter->client_next += tocopy; filter->client_avail -= tocopy; /* add it to copy buffer. */ filter->avail += tocopy; } } } /* * Move the file pointer forward. */ int64_t __archive_read_consume(struct archive_read *a, int64_t request) { return (__archive_read_filter_consume(a->filter, request)); } int64_t __archive_read_filter_consume(struct archive_read_filter * filter, int64_t request) { int64_t skipped; if (request < 0) return ARCHIVE_FATAL; if (request == 0) return 0; skipped = advance_file_pointer(filter, request); if (skipped == request) return (skipped); /* We hit EOF before we satisfied the skip request. */ if (skipped < 0) /* Map error code to 0 for error message below. */ skipped = 0; archive_set_error(&filter->archive->archive, ARCHIVE_ERRNO_MISC, "Truncated input file (needed %jd bytes, only %jd available)", (intmax_t)request, (intmax_t)skipped); return (ARCHIVE_FATAL); } /* * Advance the file pointer by the amount requested. * Returns the amount actually advanced, which may be less than the * request if EOF is encountered first. * Returns a negative value if there's an I/O error. */ static int64_t advance_file_pointer(struct archive_read_filter *filter, int64_t request) { int64_t bytes_skipped, total_bytes_skipped = 0; ssize_t bytes_read; size_t min; if (filter->fatal) return (-1); /* Use up the copy buffer first. */ if (filter->avail > 0) { min = (size_t)minimum(request, (int64_t)filter->avail); filter->next += min; filter->avail -= min; request -= min; filter->position += min; total_bytes_skipped += min; } /* Then use up the client buffer. */ if (filter->client_avail > 0) { min = (size_t)minimum(request, (int64_t)filter->client_avail); filter->client_next += min; filter->client_avail -= min; request -= min; filter->position += min; total_bytes_skipped += min; } if (request == 0) return (total_bytes_skipped); /* If there's an optimized skip function, use it. */ if (filter->can_skip != 0) { bytes_skipped = client_skip_proxy(filter, request); if (bytes_skipped < 0) { /* error */ filter->fatal = 1; return (bytes_skipped); } filter->position += bytes_skipped; total_bytes_skipped += bytes_skipped; request -= bytes_skipped; if (request == 0) return (total_bytes_skipped); } /* Use ordinary reads as necessary to complete the request. */ for (;;) { bytes_read = (filter->vtable->read)(filter, &filter->client_buff); if (bytes_read < 0) { filter->client_buff = NULL; filter->fatal = 1; return (bytes_read); } if (bytes_read == 0) { if (filter->archive->client.cursor != filter->archive->client.nodes - 1) { if (client_switch_proxy(filter, filter->archive->client.cursor + 1) == ARCHIVE_OK) continue; } filter->client_buff = NULL; filter->end_of_file = 1; return (total_bytes_skipped); } if (bytes_read >= request) { filter->client_next = ((const char *)filter->client_buff) + request; filter->client_avail = (size_t)(bytes_read - request); filter->client_total = bytes_read; total_bytes_skipped += request; filter->position += request; return (total_bytes_skipped); } filter->position += bytes_read; total_bytes_skipped += bytes_read; request -= bytes_read; } } /** * Returns ARCHIVE_FAILED if seeking isn't supported. */ int64_t __archive_read_seek(struct archive_read *a, int64_t offset, int whence) { return __archive_read_filter_seek(a->filter, offset, whence); } int64_t __archive_read_filter_seek(struct archive_read_filter *filter, int64_t offset, int whence) { struct archive_read_client *client; int64_t r; unsigned int cursor; if (filter->closed || filter->fatal) return (ARCHIVE_FATAL); if (filter->can_seek == 0) return (ARCHIVE_FAILED); client = &(filter->archive->client); switch (whence) { case SEEK_CUR: /* Adjust the offset and use SEEK_SET instead */ offset += filter->position; __LA_FALLTHROUGH; case SEEK_SET: cursor = 0; while (1) { if (client->dataset[cursor].begin_position < 0 || client->dataset[cursor].total_size < 0 || client->dataset[cursor].begin_position + client->dataset[cursor].total_size - 1 > offset || cursor + 1 >= client->nodes) break; r = client->dataset[cursor].begin_position + client->dataset[cursor].total_size; client->dataset[++cursor].begin_position = r; } while (1) { r = client_switch_proxy(filter, cursor); if (r != ARCHIVE_OK) return r; if ((r = client_seek_proxy(filter, 0, SEEK_END)) < 0) return r; client->dataset[cursor].total_size = r; if (client->dataset[cursor].begin_position + client->dataset[cursor].total_size - 1 > offset || cursor + 1 >= client->nodes) break; r = client->dataset[cursor].begin_position + client->dataset[cursor].total_size; client->dataset[++cursor].begin_position = r; } offset -= client->dataset[cursor].begin_position; if (offset < 0 || offset > client->dataset[cursor].total_size) return ARCHIVE_FATAL; if ((r = client_seek_proxy(filter, offset, SEEK_SET)) < 0) return r; break; case SEEK_END: cursor = 0; while (1) { if (client->dataset[cursor].begin_position < 0 || client->dataset[cursor].total_size < 0 || cursor + 1 >= client->nodes) break; r = client->dataset[cursor].begin_position + client->dataset[cursor].total_size; client->dataset[++cursor].begin_position = r; } while (1) { r = client_switch_proxy(filter, cursor); if (r != ARCHIVE_OK) return r; if ((r = client_seek_proxy(filter, 0, SEEK_END)) < 0) return r; client->dataset[cursor].total_size = r; r = client->dataset[cursor].begin_position + client->dataset[cursor].total_size; if (cursor + 1 >= client->nodes) break; client->dataset[++cursor].begin_position = r; } while (1) { if (r + offset >= client->dataset[cursor].begin_position) break; offset += client->dataset[cursor].total_size; if (cursor == 0) break; cursor--; r = client->dataset[cursor].begin_position + client->dataset[cursor].total_size; } offset = (r + offset) - client->dataset[cursor].begin_position; if ((r = client_switch_proxy(filter, cursor)) != ARCHIVE_OK) return r; r = client_seek_proxy(filter, offset, SEEK_SET); if (r < ARCHIVE_OK) return r; break; default: return (ARCHIVE_FATAL); } r += client->dataset[cursor].begin_position; if (r >= 0) { /* * Ouch. Clearing the buffer like this hurts, especially * at bid time. A lot of our efficiency at bid time comes * from having bidders reuse the data we've already read. * * TODO: If the seek request is in data we already * have, then don't call the seek callback. * * TODO: Zip seeks to end-of-file at bid time. If * other formats also start doing this, we may need to * find a way for clients to fudge the seek offset to * a block boundary. * * Hmmm... If whence was SEEK_END, we know the file * size is (r - offset). Can we use that to simplify * the TODO items above? */ filter->avail = filter->client_avail = 0; filter->next = filter->buffer; filter->position = r; filter->end_of_file = 0; } return r; } diff --git a/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c b/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c index 19d049770b78..42af4034b07e 100644 --- a/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c +++ b/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c @@ -1,1086 +1,1084 @@ /*- * Copyright (c) 2003-2009 Tim Kientzle * Copyright (c) 2010-2012 Michihiro NAKAJIMA * Copyright (c) 2016 Martin Matuska * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" /* This is the tree-walking code for POSIX systems. */ #if !defined(_WIN32) || defined(__CYGWIN__) #ifdef HAVE_SYS_TYPES_H #include #endif #ifdef HAVE_SYS_EXTATTR_H #include #endif #ifdef HAVE_SYS_IOCTL_H #include #endif #ifdef HAVE_SYS_PARAM_H #include #endif #ifdef HAVE_SYS_STAT_H #include #endif #if defined(HAVE_SYS_XATTR_H) #include #elif defined(HAVE_ATTR_XATTR_H) #include #endif #ifdef HAVE_SYS_EA_H #include #endif #ifdef HAVE_COPYFILE_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_LINUX_TYPES_H #include #endif #ifdef HAVE_LINUX_FIEMAP_H #include #endif #ifdef HAVE_LINUX_FS_H #include #endif /* * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. * As the include guards don't agree, the order of include is important. */ #ifdef HAVE_LINUX_EXT2_FS_H #include /* for Linux file flags */ #endif #if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) #include /* Linux file flags, broken on Cygwin */ #endif #ifdef HAVE_PATHS_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_read_disk_private.h" #ifndef O_CLOEXEC #define O_CLOEXEC 0 #endif static int setup_mac_metadata(struct archive_read_disk *, struct archive_entry *, int *fd); #ifdef ARCHIVE_XATTR_FREEBSD static int setup_xattrs_namespace(struct archive_read_disk *, struct archive_entry *, int *, int); #endif static int setup_xattrs(struct archive_read_disk *, struct archive_entry *, int *fd); static int setup_sparse(struct archive_read_disk *, struct archive_entry *, int *fd); #if defined(HAVE_LINUX_FIEMAP_H) static int setup_sparse_fiemap(struct archive_read_disk *, struct archive_entry *, int *fd); #endif #if !ARCHIVE_ACL_SUPPORT int archive_read_disk_entry_setup_acls(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { (void)a; /* UNUSED */ (void)entry; /* UNUSED */ (void)fd; /* UNUSED */ return (ARCHIVE_OK); } #endif /* * Enter working directory and return working pathname of archive_entry. * If a pointer to an integer is provided and its value is below zero * open a file descriptor on this pathname. */ const char * archive_read_disk_entry_setup_path(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { const char *path; path = archive_entry_sourcepath(entry); if (path == NULL || (a->tree != NULL && a->tree_enter_working_dir(a->tree) != 0)) path = archive_entry_pathname(entry); if (path == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Couldn't determine path"); } else if (fd != NULL && *fd < 0 && a->tree != NULL && (a->follow_symlinks || archive_entry_filetype(entry) != AE_IFLNK)) { *fd = a->open_on_current_dir(a->tree, path, O_RDONLY | O_NONBLOCK); } return (path); } int archive_read_disk_entry_from_file(struct archive *_a, struct archive_entry *entry, int fd, const struct stat *st) { struct archive_read_disk *a = (struct archive_read_disk *)_a; const char *path, *name; struct stat s; int initial_fd = fd; int r, r1; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_entry_from_file"); archive_clear_error(_a); path = archive_entry_sourcepath(entry); if (path == NULL) path = archive_entry_pathname(entry); if (a->tree == NULL) { if (st == NULL) { #if HAVE_FSTAT if (fd >= 0) { if (fstat(fd, &s) != 0) { archive_set_error(&a->archive, errno, "Can't fstat"); return (ARCHIVE_FAILED); } } else #endif #if HAVE_LSTAT if (!a->follow_symlinks) { if (lstat(path, &s) != 0) { archive_set_error(&a->archive, errno, "Can't lstat %s", path); return (ARCHIVE_FAILED); } } else #endif if (la_stat(path, &s) != 0) { archive_set_error(&a->archive, errno, "Can't stat %s", path); return (ARCHIVE_FAILED); } st = &s; } archive_entry_copy_stat(entry, st); } /* Lookup uname/gname */ name = archive_read_disk_uname(_a, archive_entry_uid(entry)); if (name != NULL) archive_entry_copy_uname(entry, name); name = archive_read_disk_gname(_a, archive_entry_gid(entry)); if (name != NULL) archive_entry_copy_gname(entry, name); #ifdef HAVE_STRUCT_STAT_ST_FLAGS /* On FreeBSD, we get flags for free with the stat. */ /* TODO: Does this belong in copy_stat()? */ if ((a->flags & ARCHIVE_READDISK_NO_FFLAGS) == 0 && st->st_flags != 0) archive_entry_set_fflags(entry, st->st_flags, 0); #endif #if (defined(FS_IOC_GETFLAGS) && defined(HAVE_WORKING_FS_IOC_GETFLAGS)) || \ (defined(EXT2_IOC_GETFLAGS) && defined(HAVE_WORKING_EXT2_IOC_GETFLAGS)) /* Linux requires an extra ioctl to pull the flags. Although * this is an extra step, it has a nice side-effect: We get an * open file descriptor which we can use in the subsequent lookups. */ if ((a->flags & ARCHIVE_READDISK_NO_FFLAGS) == 0 && (S_ISREG(st->st_mode) || S_ISDIR(st->st_mode))) { if (fd < 0) { if (a->tree != NULL) fd = a->open_on_current_dir(a->tree, path, O_RDONLY | O_NONBLOCK | O_CLOEXEC); else fd = open(path, O_RDONLY | O_NONBLOCK | O_CLOEXEC); __archive_ensure_cloexec_flag(fd); } if (fd >= 0) { int stflags; r = ioctl(fd, #if defined(FS_IOC_GETFLAGS) FS_IOC_GETFLAGS, #else EXT2_IOC_GETFLAGS, #endif &stflags); if (r == 0 && stflags != 0) archive_entry_set_fflags(entry, stflags, 0); } } #endif #if defined(HAVE_READLINK) || defined(HAVE_READLINKAT) if (S_ISLNK(st->st_mode)) { size_t linkbuffer_len = st->st_size; char *linkbuffer; int lnklen; linkbuffer = malloc(linkbuffer_len + 1); if (linkbuffer == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't read link data"); return (ARCHIVE_FAILED); } if (a->tree != NULL) { #ifdef HAVE_READLINKAT lnklen = readlinkat(a->tree_current_dir_fd(a->tree), path, linkbuffer, linkbuffer_len); #else if (a->tree_enter_working_dir(a->tree) != 0) { archive_set_error(&a->archive, errno, "Couldn't read link data"); free(linkbuffer); return (ARCHIVE_FAILED); } lnklen = readlink(path, linkbuffer, linkbuffer_len); #endif /* HAVE_READLINKAT */ } else lnklen = readlink(path, linkbuffer, linkbuffer_len); if (lnklen < 0) { archive_set_error(&a->archive, errno, "Couldn't read link data"); free(linkbuffer); return (ARCHIVE_FAILED); } linkbuffer[lnklen] = '\0'; archive_entry_set_symlink(entry, linkbuffer); free(linkbuffer); } #endif /* HAVE_READLINK || HAVE_READLINKAT */ r = 0; if ((a->flags & ARCHIVE_READDISK_NO_ACL) == 0) r = archive_read_disk_entry_setup_acls(a, entry, &fd); if ((a->flags & ARCHIVE_READDISK_NO_XATTR) == 0) { r1 = setup_xattrs(a, entry, &fd); if (r1 < r) r = r1; } if (a->flags & ARCHIVE_READDISK_MAC_COPYFILE) { r1 = setup_mac_metadata(a, entry, &fd); if (r1 < r) r = r1; } if ((a->flags & ARCHIVE_READDISK_NO_SPARSE) == 0) { r1 = setup_sparse(a, entry, &fd); if (r1 < r) r = r1; } /* If we opened the file earlier in this function, close it. */ if (initial_fd != fd) close(fd); return (r); } #if defined(__APPLE__) && defined(HAVE_COPYFILE_H) /* * The Mac OS "copyfile()" API copies the extended metadata for a * file into a separate file in AppleDouble format (see RFC 1740). * * Mac OS tar and cpio implementations store this extended * metadata as a separate entry just before the regular entry * with a "._" prefix added to the filename. * * Note that this is currently done unconditionally; the tar program has * an option to discard this information before the archive is written. * * TODO: If there's a failure, report it and return ARCHIVE_WARN. */ static int setup_mac_metadata(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { int tempfd = -1; int copyfile_flags = COPYFILE_NOFOLLOW | COPYFILE_ACL | COPYFILE_XATTR; struct stat copyfile_stat; int ret = ARCHIVE_OK; void *buff = NULL; int have_attrs; - const char *name, *tempdir; + const char *name; struct archive_string tempfile; (void)fd; /* UNUSED */ name = archive_read_disk_entry_setup_path(a, entry, NULL); if (name == NULL) return (ARCHIVE_WARN); /* Short-circuit if there's nothing to do. */ have_attrs = copyfile(name, NULL, 0, copyfile_flags | COPYFILE_CHECK); if (have_attrs == -1) { archive_set_error(&a->archive, errno, "Could not check extended attributes"); return (ARCHIVE_WARN); } if (have_attrs == 0) return (ARCHIVE_OK); - tempdir = NULL; - if (issetugid() == 0) - tempdir = getenv("TMPDIR"); - if (tempdir == NULL) - tempdir = _PATH_TMP; archive_string_init(&tempfile); - archive_strcpy(&tempfile, tempdir); + if (__archive_get_tempdir(&tempfile) != ARCHIVE_OK) { + ret = ARCHIVE_WARN; + goto cleanup; + } archive_strcat(&tempfile, "tar.md.XXXXXX"); tempfd = mkstemp(tempfile.s); if (tempfd < 0) { archive_set_error(&a->archive, errno, "Could not open extended attribute file"); ret = ARCHIVE_WARN; goto cleanup; } __archive_ensure_cloexec_flag(tempfd); /* XXX I wish copyfile() could pack directly to a memory * buffer; that would avoid the temp file here. For that * matter, it would be nice if fcopyfile() actually worked, * that would reduce the many open/close races here. */ if (copyfile(name, tempfile.s, 0, copyfile_flags | COPYFILE_PACK)) { archive_set_error(&a->archive, errno, "Could not pack extended attributes"); ret = ARCHIVE_WARN; goto cleanup; } if (fstat(tempfd, ©file_stat)) { archive_set_error(&a->archive, errno, "Could not check size of extended attributes"); ret = ARCHIVE_WARN; goto cleanup; } buff = malloc(copyfile_stat.st_size); if (buff == NULL) { archive_set_error(&a->archive, errno, "Could not allocate memory for extended attributes"); ret = ARCHIVE_WARN; goto cleanup; } if (copyfile_stat.st_size != read(tempfd, buff, copyfile_stat.st_size)) { archive_set_error(&a->archive, errno, "Could not read extended attributes into memory"); ret = ARCHIVE_WARN; goto cleanup; } archive_entry_copy_mac_metadata(entry, buff, copyfile_stat.st_size); cleanup: if (tempfd >= 0) { close(tempfd); unlink(tempfile.s); } archive_string_free(&tempfile); free(buff); return (ret); } #else /* * Stub implementation for non-Mac systems. */ static int setup_mac_metadata(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { (void)a; /* UNUSED */ (void)entry; /* UNUSED */ (void)fd; /* UNUSED */ return (ARCHIVE_OK); } #endif #if ARCHIVE_XATTR_LINUX || ARCHIVE_XATTR_DARWIN || ARCHIVE_XATTR_AIX /* * Linux, Darwin and AIX extended attribute support. * * TODO: By using a stack-allocated buffer for the first * call to getxattr(), we might be able to avoid the second * call entirely. We only need the second call if the * stack-allocated buffer is too small. But a modest buffer * of 1024 bytes or so will often be big enough. Same applies * to listxattr(). */ static int setup_xattr(struct archive_read_disk *a, struct archive_entry *entry, const char *name, int fd, const char *accpath) { ssize_t size; void *value = NULL; if (fd >= 0) { #if ARCHIVE_XATTR_LINUX size = fgetxattr(fd, name, NULL, 0); #elif ARCHIVE_XATTR_DARWIN size = fgetxattr(fd, name, NULL, 0, 0, 0); #elif ARCHIVE_XATTR_AIX size = fgetea(fd, name, NULL, 0); #endif } else if (!a->follow_symlinks) { #if ARCHIVE_XATTR_LINUX size = lgetxattr(accpath, name, NULL, 0); #elif ARCHIVE_XATTR_DARWIN size = getxattr(accpath, name, NULL, 0, 0, XATTR_NOFOLLOW); #elif ARCHIVE_XATTR_AIX size = lgetea(accpath, name, NULL, 0); #endif } else { #if ARCHIVE_XATTR_LINUX size = getxattr(accpath, name, NULL, 0); #elif ARCHIVE_XATTR_DARWIN size = getxattr(accpath, name, NULL, 0, 0, 0); #elif ARCHIVE_XATTR_AIX size = getea(accpath, name, NULL, 0); #endif } if (size == -1) { archive_set_error(&a->archive, errno, "Couldn't query extended attribute"); return (ARCHIVE_WARN); } if (size > 0 && (value = malloc(size)) == NULL) { archive_set_error(&a->archive, errno, "Out of memory"); return (ARCHIVE_FATAL); } if (fd >= 0) { #if ARCHIVE_XATTR_LINUX size = fgetxattr(fd, name, value, size); #elif ARCHIVE_XATTR_DARWIN size = fgetxattr(fd, name, value, size, 0, 0); #elif ARCHIVE_XATTR_AIX size = fgetea(fd, name, value, size); #endif } else if (!a->follow_symlinks) { #if ARCHIVE_XATTR_LINUX size = lgetxattr(accpath, name, value, size); #elif ARCHIVE_XATTR_DARWIN size = getxattr(accpath, name, value, size, 0, XATTR_NOFOLLOW); #elif ARCHIVE_XATTR_AIX size = lgetea(accpath, name, value, size); #endif } else { #if ARCHIVE_XATTR_LINUX size = getxattr(accpath, name, value, size); #elif ARCHIVE_XATTR_DARWIN size = getxattr(accpath, name, value, size, 0, 0); #elif ARCHIVE_XATTR_AIX size = getea(accpath, name, value, size); #endif } if (size == -1) { archive_set_error(&a->archive, errno, "Couldn't read extended attribute"); free(value); return (ARCHIVE_WARN); } archive_entry_xattr_add_entry(entry, name, value, size); free(value); return (ARCHIVE_OK); } static int setup_xattrs(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { char *list, *p; const char *path; ssize_t list_size; path = NULL; if (*fd < 0) { path = archive_read_disk_entry_setup_path(a, entry, fd); if (path == NULL) return (ARCHIVE_WARN); } if (*fd >= 0) { #if ARCHIVE_XATTR_LINUX list_size = flistxattr(*fd, NULL, 0); #elif ARCHIVE_XATTR_DARWIN list_size = flistxattr(*fd, NULL, 0, 0); #elif ARCHIVE_XATTR_AIX list_size = flistea(*fd, NULL, 0); #endif } else if (!a->follow_symlinks) { #if ARCHIVE_XATTR_LINUX list_size = llistxattr(path, NULL, 0); #elif ARCHIVE_XATTR_DARWIN list_size = listxattr(path, NULL, 0, XATTR_NOFOLLOW); #elif ARCHIVE_XATTR_AIX list_size = llistea(path, NULL, 0); #endif } else { #if ARCHIVE_XATTR_LINUX list_size = listxattr(path, NULL, 0); #elif ARCHIVE_XATTR_DARWIN list_size = listxattr(path, NULL, 0, 0); #elif ARCHIVE_XATTR_AIX list_size = listea(path, NULL, 0); #endif } if (list_size == -1) { if (errno == ENOTSUP || errno == ENOSYS) return (ARCHIVE_OK); archive_set_error(&a->archive, errno, "Couldn't list extended attributes"); return (ARCHIVE_WARN); } if (list_size == 0) return (ARCHIVE_OK); if ((list = malloc(list_size)) == NULL) { archive_set_error(&a->archive, errno, "Out of memory"); return (ARCHIVE_FATAL); } if (*fd >= 0) { #if ARCHIVE_XATTR_LINUX list_size = flistxattr(*fd, list, list_size); #elif ARCHIVE_XATTR_DARWIN list_size = flistxattr(*fd, list, list_size, 0); #elif ARCHIVE_XATTR_AIX list_size = flistea(*fd, list, list_size); #endif } else if (!a->follow_symlinks) { #if ARCHIVE_XATTR_LINUX list_size = llistxattr(path, list, list_size); #elif ARCHIVE_XATTR_DARWIN list_size = listxattr(path, list, list_size, XATTR_NOFOLLOW); #elif ARCHIVE_XATTR_AIX list_size = llistea(path, list, list_size); #endif } else { #if ARCHIVE_XATTR_LINUX list_size = listxattr(path, list, list_size); #elif ARCHIVE_XATTR_DARWIN list_size = listxattr(path, list, list_size, 0); #elif ARCHIVE_XATTR_AIX list_size = listea(path, list, list_size); #endif } if (list_size == -1) { archive_set_error(&a->archive, errno, "Couldn't retrieve extended attributes"); free(list); return (ARCHIVE_WARN); } for (p = list; (p - list) < list_size; p += strlen(p) + 1) { #if ARCHIVE_XATTR_LINUX /* Linux: skip POSIX.1e ACL extended attributes */ if (strncmp(p, "system.", 7) == 0 && (strcmp(p + 7, "posix_acl_access") == 0 || strcmp(p + 7, "posix_acl_default") == 0)) continue; if (strncmp(p, "trusted.SGI_", 12) == 0 && (strcmp(p + 12, "ACL_DEFAULT") == 0 || strcmp(p + 12, "ACL_FILE") == 0)) continue; /* Linux: xfsroot namespace is obsolete and unsupported */ if (strncmp(p, "xfsroot.", 8) == 0) continue; #endif setup_xattr(a, entry, p, *fd, path); } free(list); return (ARCHIVE_OK); } #elif ARCHIVE_XATTR_FREEBSD /* * FreeBSD extattr interface. */ /* TODO: Implement this. Follow the Linux model above, but * with FreeBSD-specific system calls, of course. Be careful * to not include the system extattrs that hold ACLs; we handle * those separately. */ static int setup_xattr(struct archive_read_disk *a, struct archive_entry *entry, int namespace, const char *name, const char *fullname, int fd, const char *path); static int setup_xattr(struct archive_read_disk *a, struct archive_entry *entry, int namespace, const char *name, const char *fullname, int fd, const char *accpath) { ssize_t size; void *value = NULL; if (fd >= 0) size = extattr_get_fd(fd, namespace, name, NULL, 0); else if (!a->follow_symlinks) size = extattr_get_link(accpath, namespace, name, NULL, 0); else size = extattr_get_file(accpath, namespace, name, NULL, 0); if (size == -1) { archive_set_error(&a->archive, errno, "Couldn't query extended attribute"); return (ARCHIVE_WARN); } if (size > 0 && (value = malloc(size)) == NULL) { archive_set_error(&a->archive, errno, "Out of memory"); return (ARCHIVE_FATAL); } if (fd >= 0) size = extattr_get_fd(fd, namespace, name, value, size); else if (!a->follow_symlinks) size = extattr_get_link(accpath, namespace, name, value, size); else size = extattr_get_file(accpath, namespace, name, value, size); if (size == -1) { free(value); archive_set_error(&a->archive, errno, "Couldn't read extended attribute"); return (ARCHIVE_WARN); } archive_entry_xattr_add_entry(entry, fullname, value, size); free(value); return (ARCHIVE_OK); } static int setup_xattrs_namespace(struct archive_read_disk *a, struct archive_entry *entry, int *fd, int namespace) { char buff[512]; char *list, *p; ssize_t list_size; const char *path; path = NULL; if (*fd < 0) { path = archive_read_disk_entry_setup_path(a, entry, fd); if (path == NULL) return (ARCHIVE_WARN); } if (*fd >= 0) list_size = extattr_list_fd(*fd, namespace, NULL, 0); else if (!a->follow_symlinks) list_size = extattr_list_link(path, namespace, NULL, 0); else list_size = extattr_list_file(path, namespace, NULL, 0); if (list_size == -1 && errno == EOPNOTSUPP) return (ARCHIVE_OK); if (list_size == -1 && errno == EPERM) return (ARCHIVE_OK); if (list_size == -1) { archive_set_error(&a->archive, errno, "Couldn't list extended attributes"); return (ARCHIVE_WARN); } if (list_size == 0) return (ARCHIVE_OK); if ((list = malloc(list_size)) == NULL) { archive_set_error(&a->archive, errno, "Out of memory"); return (ARCHIVE_FATAL); } if (*fd >= 0) list_size = extattr_list_fd(*fd, namespace, list, list_size); else if (!a->follow_symlinks) list_size = extattr_list_link(path, namespace, list, list_size); else list_size = extattr_list_file(path, namespace, list, list_size); if (list_size == -1) { archive_set_error(&a->archive, errno, "Couldn't retrieve extended attributes"); free(list); return (ARCHIVE_WARN); } p = list; while ((p - list) < list_size) { size_t len = 255 & (int)*p; char *name; if (namespace == EXTATTR_NAMESPACE_SYSTEM) { if (!strcmp(p + 1, "nfs4.acl") || !strcmp(p + 1, "posix1e.acl_access") || !strcmp(p + 1, "posix1e.acl_default")) { p += 1 + len; continue; } strcpy(buff, "system."); } else { strcpy(buff, "user."); } name = buff + strlen(buff); memcpy(name, p + 1, len); name[len] = '\0'; setup_xattr(a, entry, namespace, name, buff, *fd, path); p += 1 + len; } free(list); return (ARCHIVE_OK); } static int setup_xattrs(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { int namespaces[2]; int i, res; namespaces[0] = EXTATTR_NAMESPACE_USER; namespaces[1] = EXTATTR_NAMESPACE_SYSTEM; for (i = 0; i < 2; i++) { res = setup_xattrs_namespace(a, entry, fd, namespaces[i]); switch (res) { case (ARCHIVE_OK): case (ARCHIVE_WARN): break; default: return (res); } } return (ARCHIVE_OK); } #else /* * Generic (stub) extended attribute support. */ static int setup_xattrs(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { (void)a; /* UNUSED */ (void)entry; /* UNUSED */ (void)fd; /* UNUSED */ return (ARCHIVE_OK); } #endif #if defined(HAVE_LINUX_FIEMAP_H) /* * Linux FIEMAP sparse interface. * * The FIEMAP ioctl returns an "extent" for each physical allocation * on disk. We need to process those to generate a more compact list * of logical file blocks. We also need to be very careful to use * FIEMAP_FLAG_SYNC here, since there are reports that Linux sometimes * does not report allocations for newly-written data that hasn't * been synced to disk. * * It's important to return a minimal sparse file list because we want * to not trigger sparse file extensions if we don't have to, since * not all readers support them. */ static int setup_sparse_fiemap(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { char buff[4096]; struct fiemap *fm; struct fiemap_extent *fe; int64_t size; int count, do_fiemap, iters; int exit_sts = ARCHIVE_OK; const char *path; if (archive_entry_filetype(entry) != AE_IFREG || archive_entry_size(entry) <= 0 || archive_entry_hardlink(entry) != NULL) return (ARCHIVE_OK); if (*fd < 0) { path = archive_read_disk_entry_setup_path(a, entry, NULL); if (path == NULL) return (ARCHIVE_FAILED); if (a->tree != NULL) *fd = a->open_on_current_dir(a->tree, path, O_RDONLY | O_NONBLOCK | O_CLOEXEC); else *fd = open(path, O_RDONLY | O_NONBLOCK | O_CLOEXEC); if (*fd < 0) { archive_set_error(&a->archive, errno, "Can't open `%s'", path); return (ARCHIVE_FAILED); } __archive_ensure_cloexec_flag(*fd); } /* Initialize buffer to avoid the error valgrind complains about. */ memset(buff, 0, sizeof(buff)); count = (sizeof(buff) - sizeof(*fm))/sizeof(*fe); fm = (struct fiemap *)buff; fm->fm_start = 0; fm->fm_length = ~0ULL; fm->fm_flags = FIEMAP_FLAG_SYNC; fm->fm_extent_count = count; do_fiemap = 1; size = archive_entry_size(entry); for (iters = 0; ; ++iters) { int i, r; r = ioctl(*fd, FS_IOC_FIEMAP, fm); if (r < 0) { /* When something error happens, it is better we * should return ARCHIVE_OK because an earlier * version(<2.6.28) cannot perform FS_IOC_FIEMAP. */ goto exit_setup_sparse_fiemap; } if (fm->fm_mapped_extents == 0) { if (iters == 0) { /* Fully sparse file; insert a zero-length "data" entry */ archive_entry_sparse_add_entry(entry, 0, 0); } break; } fe = fm->fm_extents; for (i = 0; i < (int)fm->fm_mapped_extents; i++, fe++) { if (!(fe->fe_flags & FIEMAP_EXTENT_UNWRITTEN)) { /* The fe_length of the last block does not * adjust itself to its size files. */ int64_t length = fe->fe_length; if (fe->fe_logical + length > (uint64_t)size) length -= fe->fe_logical + length - size; if (fe->fe_logical == 0 && length == size) { /* This is not sparse. */ do_fiemap = 0; break; } if (length > 0) archive_entry_sparse_add_entry(entry, fe->fe_logical, length); } if (fe->fe_flags & FIEMAP_EXTENT_LAST) do_fiemap = 0; } if (do_fiemap) { fe = fm->fm_extents + fm->fm_mapped_extents -1; fm->fm_start = fe->fe_logical + fe->fe_length; } else break; } exit_setup_sparse_fiemap: return (exit_sts); } #if !defined(SEEK_HOLE) || !defined(SEEK_DATA) static int setup_sparse(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { return setup_sparse_fiemap(a, entry, fd); } #endif #endif /* defined(HAVE_LINUX_FIEMAP_H) */ #if defined(SEEK_HOLE) && defined(SEEK_DATA) /* * SEEK_HOLE sparse interface (FreeBSD, Linux, Solaris) */ static int setup_sparse(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { int64_t size; off_t initial_off; off_t off_s, off_e; int exit_sts = ARCHIVE_OK; int check_fully_sparse = 0; const char *path; if (archive_entry_filetype(entry) != AE_IFREG || archive_entry_size(entry) <= 0 || archive_entry_hardlink(entry) != NULL) return (ARCHIVE_OK); /* Does filesystem support the reporting of hole ? */ if (*fd < 0) path = archive_read_disk_entry_setup_path(a, entry, fd); else path = NULL; if (*fd >= 0) { #ifdef _PC_MIN_HOLE_SIZE if (fpathconf(*fd, _PC_MIN_HOLE_SIZE) <= 0) return (ARCHIVE_OK); #endif initial_off = lseek(*fd, 0, SEEK_CUR); if (initial_off != 0) lseek(*fd, 0, SEEK_SET); } else { if (path == NULL) return (ARCHIVE_FAILED); #ifdef _PC_MIN_HOLE_SIZE if (pathconf(path, _PC_MIN_HOLE_SIZE) <= 0) return (ARCHIVE_OK); #endif *fd = open(path, O_RDONLY | O_NONBLOCK | O_CLOEXEC); if (*fd < 0) { archive_set_error(&a->archive, errno, "Can't open `%s'", path); return (ARCHIVE_FAILED); } __archive_ensure_cloexec_flag(*fd); initial_off = 0; } #ifndef _PC_MIN_HOLE_SIZE /* Check if the underlying filesystem supports seek hole */ off_s = lseek(*fd, 0, SEEK_HOLE); if (off_s < 0) #if defined(HAVE_LINUX_FIEMAP_H) return setup_sparse_fiemap(a, entry, fd); #else goto exit_setup_sparse; #endif else if (off_s > 0) lseek(*fd, 0, SEEK_SET); #endif off_s = 0; size = archive_entry_size(entry); while (off_s < size) { off_s = lseek(*fd, off_s, SEEK_DATA); if (off_s == (off_t)-1) { if (errno == ENXIO) { /* no more hole */ if (archive_entry_sparse_count(entry) == 0) { /* Potentially a fully-sparse file. */ check_fully_sparse = 1; } break; } archive_set_error(&a->archive, errno, "lseek(SEEK_HOLE) failed"); exit_sts = ARCHIVE_FAILED; goto exit_setup_sparse; } off_e = lseek(*fd, off_s, SEEK_HOLE); if (off_e == (off_t)-1) { if (errno == ENXIO) { off_e = lseek(*fd, 0, SEEK_END); if (off_e != (off_t)-1) break;/* no more data */ } archive_set_error(&a->archive, errno, "lseek(SEEK_DATA) failed"); exit_sts = ARCHIVE_FAILED; goto exit_setup_sparse; } if (off_s == 0 && off_e == size) break;/* This is not sparse. */ archive_entry_sparse_add_entry(entry, off_s, off_e - off_s); off_s = off_e; } if (check_fully_sparse) { if (lseek(*fd, 0, SEEK_HOLE) == 0 && lseek(*fd, 0, SEEK_END) == size) { /* Fully sparse file; insert a zero-length "data" entry */ archive_entry_sparse_add_entry(entry, 0, 0); } } exit_setup_sparse: lseek(*fd, initial_off, SEEK_SET); return (exit_sts); } #elif !defined(HAVE_LINUX_FIEMAP_H) /* * Generic (stub) sparse support. */ static int setup_sparse(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { (void)a; /* UNUSED */ (void)entry; /* UNUSED */ (void)fd; /* UNUSED */ return (ARCHIVE_OK); } #endif #endif /* !defined(_WIN32) || defined(__CYGWIN__) */ diff --git a/contrib/libarchive/libarchive/archive_read_disk_posix.c b/contrib/libarchive/libarchive/archive_read_disk_posix.c index a7a98e9cb1cd..54a8e66188f8 100644 --- a/contrib/libarchive/libarchive/archive_read_disk_posix.c +++ b/contrib/libarchive/libarchive/archive_read_disk_posix.c @@ -1,2760 +1,2633 @@ /*- * Copyright (c) 2003-2009 Tim Kientzle * Copyright (c) 2010-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /* This is the tree-walking code for POSIX systems. */ #if !defined(_WIN32) || defined(__CYGWIN__) #include "archive_platform.h" #ifdef HAVE_SYS_PARAM_H #include #endif #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_SYS_STATFS_H #include #endif #ifdef HAVE_SYS_STATVFS_H #include #endif #ifdef HAVE_SYS_TIME_H #include #endif #ifdef HAVE_LINUX_MAGIC_H #include #endif #ifdef HAVE_LINUX_FS_H #include #elif HAVE_SYS_MOUNT_H #include #endif /* * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. * As the include guards don't agree, the order of include is important. */ #ifdef HAVE_LINUX_EXT2_FS_H #include /* for Linux file flags */ #endif #if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) #include /* Linux file flags, broken on Cygwin */ #endif #ifdef HAVE_DIRECT_H #include #endif #ifdef HAVE_DIRENT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_SYS_IOCTL_H #include #endif #include "archive.h" #include "archive_string.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_read_disk_private.h" #ifndef HAVE_FCHDIR #error fchdir function required. #endif #ifndef O_BINARY #define O_BINARY 0 #endif #ifndef O_CLOEXEC #define O_CLOEXEC 0 #endif #if defined(__hpux) && !defined(HAVE_DIRFD) #define dirfd(x) ((x)->__dd_fd) #define HAVE_DIRFD #endif /*- * This is a new directory-walking system that addresses a number * of problems I've had with fts(3). In particular, it has no * pathname-length limits (other than the size of 'int'), handles * deep logical traversals, uses considerably less memory, and has * an opaque interface (easier to modify in the future). * * Internally, it keeps a single list of "tree_entry" items that * represent filesystem objects that require further attention. * Non-directories are not kept in memory: they are pulled from * readdir(), returned to the client, then freed as soon as possible. * Any directory entry to be traversed gets pushed onto the stack. * * There is surprisingly little information that needs to be kept for * each item on the stack. Just the name, depth (represented here as the * string length of the parent directory's pathname), and some markers * indicating how to get back to the parent (via chdir("..") for a * regular dir or via fchdir(2) for a symlink). */ /* * TODO: * 1) Loop checking. * 3) Arbitrary logical traversals by closing/reopening intermediate fds. */ struct restore_time { const char *name; time_t mtime; long mtime_nsec; time_t atime; long atime_nsec; mode_t filetype; int noatime; }; struct tree_entry { int depth; struct tree_entry *next; struct tree_entry *parent; struct archive_string name; size_t dirname_length; int64_t dev; int64_t ino; int flags; int filesystem_id; /* How to return back to the parent of a symlink. */ int symlink_parent_fd; /* How to restore time of a directory. */ struct restore_time restore_time; }; struct filesystem { int64_t dev; int synthetic; int remote; int noatime; -#if defined(USE_READDIR_R) - size_t name_max; -#endif long incr_xfer_size; long max_xfer_size; long min_xfer_size; long xfer_align; /* * Buffer used for reading file contents. */ /* Exactly allocated memory pointer. */ unsigned char *allocation_ptr; /* Pointer adjusted to the filesystem alignment . */ unsigned char *buff; size_t buff_size; }; /* Definitions for tree_entry.flags bitmap. */ #define isDir 1 /* This entry is a regular directory. */ #define isDirLink 2 /* This entry is a symbolic link to a directory. */ #define needsFirstVisit 4 /* This is an initial entry. */ #define needsDescent 8 /* This entry needs to be previsited. */ #define needsOpen 16 /* This is a directory that needs to be opened. */ #define needsAscent 32 /* This entry needs to be postvisited. */ /* * Local data for this package. */ struct tree { struct tree_entry *stack; struct tree_entry *current; DIR *d; #define INVALID_DIR_HANDLE NULL struct dirent *de; -#if defined(USE_READDIR_R) - struct dirent *dirent; - size_t dirent_allocated; -#endif int flags; int visit_type; /* Error code from last failed operation. */ int tree_errno; /* Dynamically-sized buffer for holding path */ struct archive_string path; /* Last path element */ const char *basename; /* Leading dir length */ size_t dirname_length; int depth; int openCount; int maxOpenCount; int initial_dir_fd; int working_dir_fd; struct stat lst; struct stat st; int descend; int nlink; /* How to restore time of a file. */ struct restore_time restore_time; struct entry_sparse { int64_t length; int64_t offset; } *sparse_list, *current_sparse; int sparse_count; int sparse_list_size; char initial_symlink_mode; char symlink_mode; struct filesystem *current_filesystem; struct filesystem *filesystem_table; int initial_filesystem_id; int current_filesystem_id; int max_filesystem_id; int allocated_filesystem; int entry_fd; int entry_eof; int64_t entry_remaining_bytes; int64_t entry_total; unsigned char *entry_buff; size_t entry_buff_size; }; /* Definitions for tree.flags bitmap. */ #define hasStat 16 /* The st entry is valid. */ #define hasLstat 32 /* The lst entry is valid. */ #define onWorkingDir 64 /* We are on the working dir where we are * reading directory entry at this time. */ #define needsRestoreTimes 128 #define onInitialDir 256 /* We are on the initial dir. */ static int tree_dir_next_posix(struct tree *t); #ifdef HAVE_DIRENT_D_NAMLEN /* BSD extension; avoids need for a strlen() call. */ #define D_NAMELEN(dp) (dp)->d_namlen #else #define D_NAMELEN(dp) (strlen((dp)->d_name)) #endif /* Initiate/terminate a tree traversal. */ static struct tree *tree_open(const char *, char, int); static struct tree *tree_reopen(struct tree *, const char *, int); static void tree_close(struct tree *); static void tree_free(struct tree *); static void tree_push(struct tree *, const char *, int, int64_t, int64_t, struct restore_time *); static int tree_enter_initial_dir(struct tree *); static int tree_enter_working_dir(struct tree *); static int tree_current_dir_fd(struct tree *); /* * tree_next() returns Zero if there is no next entry, non-zero if * there is. Note that directories are visited three times. * Directories are always visited first as part of enumerating their * parent; that is a "regular" visit. If tree_descend() is invoked at * that time, the directory is added to a work list and will * subsequently be visited two more times: once just after descending * into the directory ("postdescent") and again just after ascending * back to the parent ("postascent"). * * TREE_ERROR_DIR is returned if the descent failed (because the * directory couldn't be opened, for instance). This is returned * instead of TREE_POSTDESCENT/TREE_POSTASCENT. TREE_ERROR_DIR is not a * fatal error, but it does imply that the relevant subtree won't be * visited. TREE_ERROR_FATAL is returned for an error that left the * traversal completely hosed. Right now, this is only returned for * chdir() failures during ascent. */ #define TREE_REGULAR 1 #define TREE_POSTDESCENT 2 #define TREE_POSTASCENT 3 #define TREE_ERROR_DIR -1 #define TREE_ERROR_FATAL -2 static int tree_next(struct tree *); /* * Return information about the current entry. */ /* * The current full pathname, length of the full pathname, and a name * that can be used to access the file. Because tree does use chdir * extensively, the access path is almost never the same as the full * current path. * * TODO: On platforms that support it, use openat()-style operations * to eliminate the chdir() operations entirely while still supporting * arbitrarily deep traversals. This makes access_path troublesome to * support, of course, which means we'll need a rich enough interface * that clients can function without it. (In particular, we'll need * tree_current_open() that returns an open file descriptor.) * */ static const char *tree_current_path(struct tree *); static const char *tree_current_access_path(struct tree *); /* * Request the lstat() or stat() data for the current path. Since the * tree package needs to do some of this anyway, and caches the * results, you should take advantage of it here if you need it rather * than make a redundant stat() or lstat() call of your own. */ static const struct stat *tree_current_stat(struct tree *); static const struct stat *tree_current_lstat(struct tree *); static int tree_current_is_symblic_link_target(struct tree *); /* The following functions use tricks to avoid a certain number of * stat()/lstat() calls. */ /* "is_physical_dir" is equivalent to S_ISDIR(tree_current_lstat()->st_mode) */ static int tree_current_is_physical_dir(struct tree *); /* "is_dir" is equivalent to S_ISDIR(tree_current_stat()->st_mode) */ static int tree_current_is_dir(struct tree *); static int update_current_filesystem(struct archive_read_disk *a, int64_t dev); static int setup_current_filesystem(struct archive_read_disk *); static int tree_target_is_same_as_parent(struct tree *, const struct stat *); static int _archive_read_disk_open(struct archive *, const char *); static int _archive_read_free(struct archive *); static int _archive_read_close(struct archive *); static int _archive_read_data_block(struct archive *, const void **, size_t *, int64_t *); static int _archive_read_next_header(struct archive *, struct archive_entry **); static int _archive_read_next_header2(struct archive *, struct archive_entry *); static const char *trivial_lookup_gname(void *, int64_t gid); static const char *trivial_lookup_uname(void *, int64_t uid); static int setup_sparse(struct archive_read_disk *, struct archive_entry *); static int close_and_restore_time(int fd, struct tree *, struct restore_time *); static int open_on_current_dir(struct tree *, const char *, int); static int tree_dup(int); static const struct archive_vtable archive_read_disk_vtable = { .archive_free = _archive_read_free, .archive_close = _archive_read_close, .archive_read_data_block = _archive_read_data_block, .archive_read_next_header = _archive_read_next_header, .archive_read_next_header2 = _archive_read_next_header2, }; const char * archive_read_disk_gname(struct archive *_a, la_int64_t gid) { struct archive_read_disk *a = (struct archive_read_disk *)_a; if (ARCHIVE_OK != __archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_gname")) return (NULL); if (a->lookup_gname == NULL) return (NULL); return ((*a->lookup_gname)(a->lookup_gname_data, gid)); } const char * archive_read_disk_uname(struct archive *_a, la_int64_t uid) { struct archive_read_disk *a = (struct archive_read_disk *)_a; if (ARCHIVE_OK != __archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_uname")) return (NULL); if (a->lookup_uname == NULL) return (NULL); return ((*a->lookup_uname)(a->lookup_uname_data, uid)); } int archive_read_disk_set_gname_lookup(struct archive *_a, void *private_data, const char * (*lookup_gname)(void *private, la_int64_t gid), void (*cleanup_gname)(void *private)) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(&a->archive, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_gname_lookup"); if (a->cleanup_gname != NULL && a->lookup_gname_data != NULL) (a->cleanup_gname)(a->lookup_gname_data); a->lookup_gname = lookup_gname; a->cleanup_gname = cleanup_gname; a->lookup_gname_data = private_data; return (ARCHIVE_OK); } int archive_read_disk_set_uname_lookup(struct archive *_a, void *private_data, const char * (*lookup_uname)(void *private, la_int64_t uid), void (*cleanup_uname)(void *private)) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(&a->archive, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_uname_lookup"); if (a->cleanup_uname != NULL && a->lookup_uname_data != NULL) (a->cleanup_uname)(a->lookup_uname_data); a->lookup_uname = lookup_uname; a->cleanup_uname = cleanup_uname; a->lookup_uname_data = private_data; return (ARCHIVE_OK); } /* * Create a new archive_read_disk object and initialize it with global state. */ struct archive * archive_read_disk_new(void) { struct archive_read_disk *a; a = calloc(1, sizeof(*a)); if (a == NULL) return (NULL); a->archive.magic = ARCHIVE_READ_DISK_MAGIC; a->archive.state = ARCHIVE_STATE_NEW; a->archive.vtable = &archive_read_disk_vtable; a->entry = archive_entry_new2(&a->archive); a->lookup_uname = trivial_lookup_uname; a->lookup_gname = trivial_lookup_gname; a->flags = ARCHIVE_READDISK_MAC_COPYFILE; a->open_on_current_dir = open_on_current_dir; a->tree_current_dir_fd = tree_current_dir_fd; a->tree_enter_working_dir = tree_enter_working_dir; return (&a->archive); } static int _archive_read_free(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; int r; if (_a == NULL) return (ARCHIVE_OK); archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_read_free"); if (a->archive.state != ARCHIVE_STATE_CLOSED) r = _archive_read_close(&a->archive); else r = ARCHIVE_OK; tree_free(a->tree); if (a->cleanup_gname != NULL && a->lookup_gname_data != NULL) (a->cleanup_gname)(a->lookup_gname_data); if (a->cleanup_uname != NULL && a->lookup_uname_data != NULL) (a->cleanup_uname)(a->lookup_uname_data); archive_string_free(&a->archive.error_string); archive_entry_free(a->entry); a->archive.magic = 0; __archive_clean(&a->archive); free(a); return (r); } static int _archive_read_close(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_read_close"); if (a->archive.state != ARCHIVE_STATE_FATAL) a->archive.state = ARCHIVE_STATE_CLOSED; tree_close(a->tree); return (ARCHIVE_OK); } static void setup_symlink_mode(struct archive_read_disk *a, char symlink_mode, char follow_symlinks) { a->symlink_mode = symlink_mode; a->follow_symlinks = follow_symlinks; if (a->tree != NULL) { a->tree->initial_symlink_mode = a->symlink_mode; a->tree->symlink_mode = a->symlink_mode; } } int archive_read_disk_set_symlink_logical(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_symlink_logical"); setup_symlink_mode(a, 'L', 1); return (ARCHIVE_OK); } int archive_read_disk_set_symlink_physical(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_symlink_physical"); setup_symlink_mode(a, 'P', 0); return (ARCHIVE_OK); } int archive_read_disk_set_symlink_hybrid(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_symlink_hybrid"); setup_symlink_mode(a, 'H', 1);/* Follow symlinks initially. */ return (ARCHIVE_OK); } int archive_read_disk_set_atime_restored(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_restore_atime"); #ifdef HAVE_UTIMES a->flags |= ARCHIVE_READDISK_RESTORE_ATIME; if (a->tree != NULL) a->tree->flags |= needsRestoreTimes; return (ARCHIVE_OK); #else /* Display warning and unset flag */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Cannot restore access time on this system"); a->flags &= ~ARCHIVE_READDISK_RESTORE_ATIME; return (ARCHIVE_WARN); #endif } int archive_read_disk_set_behavior(struct archive *_a, int flags) { struct archive_read_disk *a = (struct archive_read_disk *)_a; int r = ARCHIVE_OK; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_honor_nodump"); a->flags = flags; if (flags & ARCHIVE_READDISK_RESTORE_ATIME) r = archive_read_disk_set_atime_restored(_a); else { if (a->tree != NULL) a->tree->flags &= ~needsRestoreTimes; } return (r); } /* * Trivial implementations of gname/uname lookup functions. * These are normally overridden by the client, but these stub * versions ensure that we always have something that works. */ static const char * trivial_lookup_gname(void *private_data, int64_t gid) { (void)private_data; /* UNUSED */ (void)gid; /* UNUSED */ return (NULL); } static const char * trivial_lookup_uname(void *private_data, int64_t uid) { (void)private_data; /* UNUSED */ (void)uid; /* UNUSED */ return (NULL); } /* * Allocate memory for the reading buffer adjusted to the filesystem * alignment. */ static int setup_suitable_read_buffer(struct archive_read_disk *a) { struct tree *t = a->tree; struct filesystem *cf = t->current_filesystem; size_t asize; size_t s; if (cf->allocation_ptr == NULL) { /* If we couldn't get a filesystem alignment, * we use 4096 as default value but we won't use * O_DIRECT to open() and openat() operations. */ long xfer_align = (cf->xfer_align == -1)?4096:cf->xfer_align; if (cf->max_xfer_size != -1) asize = cf->max_xfer_size + xfer_align; else { long incr = cf->incr_xfer_size; /* Some platform does not set a proper value to * incr_xfer_size.*/ if (incr < 0) incr = cf->min_xfer_size; if (cf->min_xfer_size < 0) { incr = xfer_align; asize = xfer_align; } else asize = cf->min_xfer_size; /* Increase a buffer size up to 64K bytes in * a proper increment size. */ while (asize < 1024*64) asize += incr; /* Take a margin to adjust to the filesystem * alignment. */ asize += xfer_align; } cf->allocation_ptr = malloc(asize); if (cf->allocation_ptr == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory"); a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } /* * Calculate proper address for the filesystem. */ s = (uintptr_t)cf->allocation_ptr; s %= xfer_align; if (s > 0) s = xfer_align - s; /* * Set a read buffer pointer in the proper alignment of * the current filesystem. */ cf->buff = cf->allocation_ptr + s; cf->buff_size = asize - xfer_align; } return (ARCHIVE_OK); } static int _archive_read_data_block(struct archive *_a, const void **buff, size_t *size, int64_t *offset) { struct archive_read_disk *a = (struct archive_read_disk *)_a; struct tree *t = a->tree; int r; ssize_t bytes; int64_t sparse_bytes; size_t buffbytes; int empty_sparse_region = 0; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, "archive_read_data_block"); if (t->entry_eof || t->entry_remaining_bytes <= 0) { r = ARCHIVE_EOF; goto abort_read_data; } /* * Open the current file. */ if (t->entry_fd < 0) { int flags = O_RDONLY | O_BINARY | O_CLOEXEC; /* * Eliminate or reduce cache effects if we can. * * Carefully consider this to be enabled. */ #if defined(O_DIRECT) && 0/* Disabled for now */ if (t->current_filesystem->xfer_align != -1 && t->nlink == 1) flags |= O_DIRECT; #endif #if defined(O_NOATIME) /* * Linux has O_NOATIME flag; use it if we need. */ if ((t->flags & needsRestoreTimes) != 0 && t->restore_time.noatime == 0) flags |= O_NOATIME; #endif t->entry_fd = open_on_current_dir(t, tree_current_access_path(t), flags); __archive_ensure_cloexec_flag(t->entry_fd); #if defined(O_NOATIME) /* * When we did open the file with O_NOATIME flag, * if successful, set 1 to t->restore_time.noatime * not to restore an atime of the file later. * if failed by EPERM, retry it without O_NOATIME flag. */ if (flags & O_NOATIME) { if (t->entry_fd >= 0) t->restore_time.noatime = 1; else if (errno == EPERM) flags &= ~O_NOATIME; } #endif if (t->entry_fd < 0) { archive_set_error(&a->archive, errno, "Couldn't open %s", tree_current_path(t)); r = ARCHIVE_FAILED; tree_enter_initial_dir(t); goto abort_read_data; } tree_enter_initial_dir(t); } /* * Allocate read buffer if not allocated. */ if (t->current_filesystem->allocation_ptr == NULL) { r = setup_suitable_read_buffer(a); if (r != ARCHIVE_OK) { a->archive.state = ARCHIVE_STATE_FATAL; goto abort_read_data; } } t->entry_buff = t->current_filesystem->buff; t->entry_buff_size = t->current_filesystem->buff_size; buffbytes = t->entry_buff_size; if ((int64_t)buffbytes > t->current_sparse->length) buffbytes = t->current_sparse->length; if (t->current_sparse->length == 0) empty_sparse_region = 1; /* * Skip hole. * TODO: Should we consider t->current_filesystem->xfer_align? */ if (t->current_sparse->offset > t->entry_total) { if (lseek(t->entry_fd, (off_t)t->current_sparse->offset, SEEK_SET) != t->current_sparse->offset) { archive_set_error(&a->archive, errno, "Seek error"); r = ARCHIVE_FATAL; a->archive.state = ARCHIVE_STATE_FATAL; goto abort_read_data; } sparse_bytes = t->current_sparse->offset - t->entry_total; t->entry_remaining_bytes -= sparse_bytes; t->entry_total += sparse_bytes; } /* * Read file contents. */ if (buffbytes > 0) { bytes = read(t->entry_fd, t->entry_buff, buffbytes); if (bytes < 0) { archive_set_error(&a->archive, errno, "Read error"); r = ARCHIVE_FATAL; a->archive.state = ARCHIVE_STATE_FATAL; goto abort_read_data; } } else bytes = 0; /* * Return an EOF unless we've read a leading empty sparse region, which * is used to represent fully-sparse files. */ if (bytes == 0 && !empty_sparse_region) { /* Get EOF */ t->entry_eof = 1; r = ARCHIVE_EOF; goto abort_read_data; } *buff = t->entry_buff; *size = bytes; *offset = t->entry_total; t->entry_total += bytes; t->entry_remaining_bytes -= bytes; if (t->entry_remaining_bytes == 0) { /* Close the current file descriptor */ close_and_restore_time(t->entry_fd, t, &t->restore_time); t->entry_fd = -1; t->entry_eof = 1; } t->current_sparse->offset += bytes; t->current_sparse->length -= bytes; if (t->current_sparse->length == 0 && !t->entry_eof) t->current_sparse++; return (ARCHIVE_OK); abort_read_data: *buff = NULL; *size = 0; *offset = t->entry_total; if (t->entry_fd >= 0) { /* Close the current file descriptor */ close_and_restore_time(t->entry_fd, t, &t->restore_time); t->entry_fd = -1; } return (r); } static int next_entry(struct archive_read_disk *a, struct tree *t, struct archive_entry *entry) { const struct stat *st; /* info to use for this entry */ const struct stat *lst;/* lstat() information */ const char *name; int delayed, delayed_errno, descend, r; struct archive_string delayed_str; delayed = ARCHIVE_OK; delayed_errno = 0; archive_string_init(&delayed_str); st = NULL; lst = NULL; t->descend = 0; do { switch (tree_next(t)) { case TREE_ERROR_FATAL: archive_set_error(&a->archive, t->tree_errno, "%s: Unable to continue traversing directory tree", tree_current_path(t)); a->archive.state = ARCHIVE_STATE_FATAL; tree_enter_initial_dir(t); return (ARCHIVE_FATAL); case TREE_ERROR_DIR: - archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + archive_set_error(&a->archive, t->tree_errno, "%s: Couldn't visit directory", tree_current_path(t)); tree_enter_initial_dir(t); return (ARCHIVE_FAILED); case 0: tree_enter_initial_dir(t); return (ARCHIVE_EOF); case TREE_POSTDESCENT: case TREE_POSTASCENT: break; case TREE_REGULAR: lst = tree_current_lstat(t); if (lst == NULL) { if (errno == ENOENT && t->depth > 0) { delayed = ARCHIVE_WARN; delayed_errno = errno; if (delayed_str.length == 0) { archive_string_sprintf(&delayed_str, "%s", tree_current_path(t)); } else { archive_string_sprintf(&delayed_str, " %s", tree_current_path(t)); } } else { archive_set_error(&a->archive, errno, "%s: Cannot stat", tree_current_path(t)); tree_enter_initial_dir(t); return (ARCHIVE_FAILED); } } break; } } while (lst == NULL); #ifdef __APPLE__ if (a->flags & ARCHIVE_READDISK_MAC_COPYFILE) { /* If we're using copyfile(), ignore "._XXX" files. */ const char *bname = strrchr(tree_current_path(t), '/'); if (bname == NULL) bname = tree_current_path(t); else ++bname; if (bname[0] == '.' && bname[1] == '_') return (ARCHIVE_RETRY); } #endif archive_entry_copy_pathname(entry, tree_current_path(t)); /* * Perform path matching. */ if (a->matching) { r = archive_match_path_excluded(a->matching, entry); if (r < 0) { archive_set_error(&(a->archive), errno, "Failed : %s", archive_error_string(a->matching)); return (r); } if (r) { if (a->excluded_cb_func) a->excluded_cb_func(&(a->archive), a->excluded_cb_data, entry); return (ARCHIVE_RETRY); } } /* * Distinguish 'L'/'P'/'H' symlink following. */ switch(t->symlink_mode) { case 'H': /* 'H': After the first item, rest like 'P'. */ t->symlink_mode = 'P'; /* 'H': First item (from command line) like 'L'. */ /* FALLTHROUGH */ case 'L': /* 'L': Do descend through a symlink to dir. */ descend = tree_current_is_dir(t); /* 'L': Follow symlinks to files. */ a->symlink_mode = 'L'; a->follow_symlinks = 1; /* 'L': Archive symlinks as targets, if we can. */ st = tree_current_stat(t); if (st != NULL && !tree_target_is_same_as_parent(t, st)) break; /* If stat fails, we have a broken symlink; * in that case, don't follow the link. */ /* FALLTHROUGH */ default: /* 'P': Don't descend through a symlink to dir. */ descend = tree_current_is_physical_dir(t); /* 'P': Don't follow symlinks to files. */ a->symlink_mode = 'P'; a->follow_symlinks = 0; /* 'P': Archive symlinks as symlinks. */ st = lst; break; } if (update_current_filesystem(a, st->st_dev) != ARCHIVE_OK) { a->archive.state = ARCHIVE_STATE_FATAL; tree_enter_initial_dir(t); return (ARCHIVE_FATAL); } if (t->initial_filesystem_id == -1) t->initial_filesystem_id = t->current_filesystem_id; if (a->flags & ARCHIVE_READDISK_NO_TRAVERSE_MOUNTS) { if (t->initial_filesystem_id != t->current_filesystem_id) descend = 0; } t->descend = descend; /* * Honor nodump flag. * If the file is marked with nodump flag, do not return this entry. */ if (a->flags & ARCHIVE_READDISK_HONOR_NODUMP) { #if defined(HAVE_STRUCT_STAT_ST_FLAGS) && defined(UF_NODUMP) if (st->st_flags & UF_NODUMP) return (ARCHIVE_RETRY); #elif (defined(FS_IOC_GETFLAGS) && defined(FS_NODUMP_FL) && \ defined(HAVE_WORKING_FS_IOC_GETFLAGS)) || \ (defined(EXT2_IOC_GETFLAGS) && defined(EXT2_NODUMP_FL) && \ defined(HAVE_WORKING_EXT2_IOC_GETFLAGS)) if (S_ISREG(st->st_mode) || S_ISDIR(st->st_mode)) { int stflags; t->entry_fd = open_on_current_dir(t, tree_current_access_path(t), O_RDONLY | O_NONBLOCK | O_CLOEXEC); __archive_ensure_cloexec_flag(t->entry_fd); if (t->entry_fd >= 0) { r = ioctl(t->entry_fd, #ifdef FS_IOC_GETFLAGS FS_IOC_GETFLAGS, #else EXT2_IOC_GETFLAGS, #endif &stflags); #ifdef FS_NODUMP_FL if (r == 0 && (stflags & FS_NODUMP_FL) != 0) #else if (r == 0 && (stflags & EXT2_NODUMP_FL) != 0) #endif return (ARCHIVE_RETRY); } } #endif } archive_entry_copy_stat(entry, st); /* Save the times to be restored. This must be in before * calling archive_read_disk_descend() or any chance of it, * especially, invoking a callback. */ t->restore_time.mtime = archive_entry_mtime(entry); t->restore_time.mtime_nsec = archive_entry_mtime_nsec(entry); t->restore_time.atime = archive_entry_atime(entry); t->restore_time.atime_nsec = archive_entry_atime_nsec(entry); t->restore_time.filetype = archive_entry_filetype(entry); t->restore_time.noatime = t->current_filesystem->noatime; /* * Perform time matching. */ if (a->matching) { r = archive_match_time_excluded(a->matching, entry); if (r < 0) { archive_set_error(&(a->archive), errno, "Failed : %s", archive_error_string(a->matching)); return (r); } if (r) { if (a->excluded_cb_func) a->excluded_cb_func(&(a->archive), a->excluded_cb_data, entry); return (ARCHIVE_RETRY); } } /* Lookup uname/gname */ name = archive_read_disk_uname(&(a->archive), archive_entry_uid(entry)); if (name != NULL) archive_entry_copy_uname(entry, name); name = archive_read_disk_gname(&(a->archive), archive_entry_gid(entry)); if (name != NULL) archive_entry_copy_gname(entry, name); /* * Perform owner matching. */ if (a->matching) { r = archive_match_owner_excluded(a->matching, entry); if (r < 0) { archive_set_error(&(a->archive), errno, "Failed : %s", archive_error_string(a->matching)); return (r); } if (r) { if (a->excluded_cb_func) a->excluded_cb_func(&(a->archive), a->excluded_cb_data, entry); return (ARCHIVE_RETRY); } } /* * Invoke a meta data filter callback. */ if (a->metadata_filter_func) { if (!a->metadata_filter_func(&(a->archive), a->metadata_filter_data, entry)) return (ARCHIVE_RETRY); } /* * Populate the archive_entry with metadata from the disk. */ archive_entry_copy_sourcepath(entry, tree_current_access_path(t)); r = archive_read_disk_entry_from_file(&(a->archive), entry, t->entry_fd, st); if (r == ARCHIVE_OK) { r = delayed; if (r != ARCHIVE_OK) { archive_string_sprintf(&delayed_str, ": %s", "File removed before we read it"); archive_set_error(&(a->archive), delayed_errno, "%s", delayed_str.s); } } archive_string_free(&delayed_str); return (r); } static int _archive_read_next_header(struct archive *_a, struct archive_entry **entryp) { int ret; struct archive_read_disk *a = (struct archive_read_disk *)_a; *entryp = NULL; ret = _archive_read_next_header2(_a, a->entry); *entryp = a->entry; return ret; } static int _archive_read_next_header2(struct archive *_a, struct archive_entry *entry) { struct archive_read_disk *a = (struct archive_read_disk *)_a; struct tree *t; int r; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_read_next_header2"); t = a->tree; if (t->entry_fd >= 0) { close_and_restore_time(t->entry_fd, t, &t->restore_time); t->entry_fd = -1; } archive_entry_clear(entry); for (;;) { r = next_entry(a, t, entry); if (t->entry_fd >= 0) { close(t->entry_fd); t->entry_fd = -1; } if (r == ARCHIVE_RETRY) { archive_entry_clear(entry); continue; } break; } /* Return to the initial directory. */ tree_enter_initial_dir(t); /* * EOF and FATAL are persistent at this layer. By * modifying the state, we guarantee that future calls to * read a header or read data will fail. */ switch (r) { case ARCHIVE_EOF: a->archive.state = ARCHIVE_STATE_EOF; break; case ARCHIVE_OK: case ARCHIVE_WARN: /* Overwrite the sourcepath based on the initial directory. */ archive_entry_copy_sourcepath(entry, tree_current_path(t)); t->entry_total = 0; if (archive_entry_filetype(entry) == AE_IFREG) { t->nlink = archive_entry_nlink(entry); t->entry_remaining_bytes = archive_entry_size(entry); t->entry_eof = (t->entry_remaining_bytes == 0)? 1: 0; if (!t->entry_eof && setup_sparse(a, entry) != ARCHIVE_OK) return (ARCHIVE_FATAL); } else { t->entry_remaining_bytes = 0; t->entry_eof = 1; } a->archive.state = ARCHIVE_STATE_DATA; break; case ARCHIVE_RETRY: break; case ARCHIVE_FATAL: a->archive.state = ARCHIVE_STATE_FATAL; break; } __archive_reset_read_data(&a->archive); return (r); } static int setup_sparse(struct archive_read_disk *a, struct archive_entry *entry) { struct tree *t = a->tree; int64_t length, offset; int i; t->sparse_count = archive_entry_sparse_reset(entry); if (t->sparse_count+1 > t->sparse_list_size) { free(t->sparse_list); t->sparse_list_size = t->sparse_count + 1; t->sparse_list = malloc(sizeof(t->sparse_list[0]) * t->sparse_list_size); if (t->sparse_list == NULL) { t->sparse_list_size = 0; archive_set_error(&a->archive, ENOMEM, "Can't allocate data"); a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } } for (i = 0; i < t->sparse_count; i++) { archive_entry_sparse_next(entry, &offset, &length); t->sparse_list[i].offset = offset; t->sparse_list[i].length = length; } if (i == 0) { t->sparse_list[i].offset = 0; t->sparse_list[i].length = archive_entry_size(entry); } else { t->sparse_list[i].offset = archive_entry_size(entry); t->sparse_list[i].length = 0; } t->current_sparse = t->sparse_list; return (ARCHIVE_OK); } int archive_read_disk_set_matching(struct archive *_a, struct archive *_ma, void (*_excluded_func)(struct archive *, void *, struct archive_entry *), void *_client_data) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_matching"); a->matching = _ma; a->excluded_cb_func = _excluded_func; a->excluded_cb_data = _client_data; return (ARCHIVE_OK); } int archive_read_disk_set_metadata_filter_callback(struct archive *_a, int (*_metadata_filter_func)(struct archive *, void *, struct archive_entry *), void *_client_data) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_metadata_filter_callback"); a->metadata_filter_func = _metadata_filter_func; a->metadata_filter_data = _client_data; return (ARCHIVE_OK); } int archive_read_disk_can_descend(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; struct tree *t = a->tree; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_read_disk_can_descend"); return (t->visit_type == TREE_REGULAR && t->descend); } /* * Called by the client to mark the directory just returned from * tree_next() as needing to be visited. */ int archive_read_disk_descend(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; struct tree *t = a->tree; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_read_disk_descend"); if (!archive_read_disk_can_descend(_a)) return (ARCHIVE_OK); /* * We must not treat the initial specified path as a physical dir, * because if we do then we will try and ascend out of it by opening * ".." which is (a) wrong and (b) causes spurious permissions errors * if ".." is not readable by us. Instead, treat it as if it were a * symlink. (This uses an extra fd, but it can only happen once at the * top level of a traverse.) But we can't necessarily assume t->st is * valid here (though t->lst is), which complicates the logic a * little. */ if (tree_current_is_physical_dir(t)) { tree_push(t, t->basename, t->current_filesystem_id, t->lst.st_dev, t->lst.st_ino, &t->restore_time); if (t->stack->parent->parent != NULL) t->stack->flags |= isDir; else t->stack->flags |= isDirLink; } else if (tree_current_is_dir(t)) { tree_push(t, t->basename, t->current_filesystem_id, t->st.st_dev, t->st.st_ino, &t->restore_time); t->stack->flags |= isDirLink; } t->descend = 0; return (ARCHIVE_OK); } int archive_read_disk_open(struct archive *_a, const char *pathname) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_CLOSED, "archive_read_disk_open"); archive_clear_error(&a->archive); return (_archive_read_disk_open(_a, pathname)); } int archive_read_disk_open_w(struct archive *_a, const wchar_t *pathname) { struct archive_read_disk *a = (struct archive_read_disk *)_a; struct archive_string path; int ret; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_CLOSED, "archive_read_disk_open_w"); archive_clear_error(&a->archive); /* Make a char string from a wchar_t string. */ archive_string_init(&path); if (archive_string_append_from_wcs(&path, pathname, wcslen(pathname)) != 0) { if (errno == ENOMEM) archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't convert a path to a char string"); a->archive.state = ARCHIVE_STATE_FATAL; ret = ARCHIVE_FATAL; } else ret = _archive_read_disk_open(_a, path.s); archive_string_free(&path); return (ret); } static int _archive_read_disk_open(struct archive *_a, const char *pathname) { struct archive_read_disk *a = (struct archive_read_disk *)_a; if (a->tree != NULL) a->tree = tree_reopen(a->tree, pathname, a->flags & ARCHIVE_READDISK_RESTORE_ATIME); else a->tree = tree_open(pathname, a->symlink_mode, a->flags & ARCHIVE_READDISK_RESTORE_ATIME); if (a->tree == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate tar data"); a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } a->archive.state = ARCHIVE_STATE_HEADER; return (ARCHIVE_OK); } /* * Return a current filesystem ID which is index of the filesystem entry * you've visited through archive_read_disk. */ int archive_read_disk_current_filesystem(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, "archive_read_disk_current_filesystem"); return (a->tree->current_filesystem_id); } static int update_current_filesystem(struct archive_read_disk *a, int64_t dev) { struct tree *t = a->tree; int i, fid; if (t->current_filesystem != NULL && t->current_filesystem->dev == dev) return (ARCHIVE_OK); for (i = 0; i < t->max_filesystem_id; i++) { if (t->filesystem_table[i].dev == dev) { /* There is the filesystem ID we've already generated. */ t->current_filesystem_id = i; t->current_filesystem = &(t->filesystem_table[i]); return (ARCHIVE_OK); } } /* * This is the new filesystem which we have to generate a new ID for. */ fid = t->max_filesystem_id++; if (t->max_filesystem_id > t->allocated_filesystem) { size_t s; void *p; s = t->max_filesystem_id * 2; p = realloc(t->filesystem_table, s * sizeof(*t->filesystem_table)); if (p == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate tar data"); return (ARCHIVE_FATAL); } t->filesystem_table = (struct filesystem *)p; t->allocated_filesystem = s; } t->current_filesystem_id = fid; t->current_filesystem = &(t->filesystem_table[fid]); t->current_filesystem->dev = dev; t->current_filesystem->allocation_ptr = NULL; t->current_filesystem->buff = NULL; /* Setup the current filesystem properties which depend on * platform specific. */ return (setup_current_filesystem(a)); } /* * Returns 1 if current filesystem is generated filesystem, 0 if it is not * or -1 if it is unknown. */ int archive_read_disk_current_filesystem_is_synthetic(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, "archive_read_disk_current_filesystem"); return (a->tree->current_filesystem->synthetic); } /* * Returns 1 if current filesystem is remote filesystem, 0 if it is not * or -1 if it is unknown. */ int archive_read_disk_current_filesystem_is_remote(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, "archive_read_disk_current_filesystem"); return (a->tree->current_filesystem->remote); } #if defined(_PC_REC_INCR_XFER_SIZE) && defined(_PC_REC_MAX_XFER_SIZE) &&\ defined(_PC_REC_MIN_XFER_SIZE) && defined(_PC_REC_XFER_ALIGN) static int get_xfer_size(struct tree *t, int fd, const char *path) { t->current_filesystem->xfer_align = -1; errno = 0; if (fd >= 0) { t->current_filesystem->incr_xfer_size = fpathconf(fd, _PC_REC_INCR_XFER_SIZE); t->current_filesystem->max_xfer_size = fpathconf(fd, _PC_REC_MAX_XFER_SIZE); t->current_filesystem->min_xfer_size = fpathconf(fd, _PC_REC_MIN_XFER_SIZE); t->current_filesystem->xfer_align = fpathconf(fd, _PC_REC_XFER_ALIGN); } else if (path != NULL) { t->current_filesystem->incr_xfer_size = pathconf(path, _PC_REC_INCR_XFER_SIZE); t->current_filesystem->max_xfer_size = pathconf(path, _PC_REC_MAX_XFER_SIZE); t->current_filesystem->min_xfer_size = pathconf(path, _PC_REC_MIN_XFER_SIZE); t->current_filesystem->xfer_align = pathconf(path, _PC_REC_XFER_ALIGN); } /* At least we need an alignment size. */ if (t->current_filesystem->xfer_align == -1) return ((errno == EINVAL)?1:-1); else return (0); } #else static int get_xfer_size(struct tree *t, int fd, const char *path) { (void)t; /* UNUSED */ (void)fd; /* UNUSED */ (void)path; /* UNUSED */ return (1);/* Not supported */ } #endif #if defined(HAVE_STATVFS) static inline __LA_UNUSED void set_statvfs_transfer_size(struct filesystem *fs, const struct statvfs *sfs) { fs->xfer_align = sfs->f_frsize > 0 ? (long)sfs->f_frsize : -1; fs->max_xfer_size = -1; #if defined(HAVE_STRUCT_STATVFS_F_IOSIZE) fs->min_xfer_size = sfs->f_iosize > 0 ? (long)sfs->f_iosize : -1; fs->incr_xfer_size = sfs->f_iosize > 0 ? (long)sfs->f_iosize : -1; #else fs->min_xfer_size = sfs->f_bsize > 0 ? (long)sfs->f_bsize : -1; fs->incr_xfer_size = sfs->f_bsize > 0 ? (long)sfs->f_bsize : -1; #endif } #endif #if defined(HAVE_STRUCT_STATFS) static inline __LA_UNUSED void set_statfs_transfer_size(struct filesystem *fs, const struct statfs *sfs) { fs->xfer_align = sfs->f_bsize > 0 ? (long)sfs->f_bsize : -1; fs->max_xfer_size = -1; #if defined(HAVE_STRUCT_STATFS_F_IOSIZE) fs->min_xfer_size = sfs->f_iosize > 0 ? (long)sfs->f_iosize : -1; fs->incr_xfer_size = sfs->f_iosize > 0 ? (long)sfs->f_iosize : -1; #else fs->min_xfer_size = sfs->f_bsize > 0 ? (long)sfs->f_bsize : -1; fs->incr_xfer_size = sfs->f_bsize > 0 ? (long)sfs->f_bsize : -1; #endif } #endif #if defined(HAVE_STRUCT_STATFS) && defined(HAVE_STATFS) && \ defined(HAVE_FSTATFS) && defined(MNT_LOCAL) && !defined(ST_LOCAL) /* * Gather current filesystem properties on FreeBSD, OpenBSD and Mac OS X. */ static int setup_current_filesystem(struct archive_read_disk *a) { struct tree *t = a->tree; struct statfs sfs; #if defined(HAVE_GETVFSBYNAME) && defined(VFCF_SYNTHETIC) /* TODO: configure should set GETVFSBYNAME_ARG_TYPE to make * this accurate; some platforms have both and we need the one that's * used by getvfsbyname() * * Then the following would become: * #if defined(GETVFSBYNAME_ARG_TYPE) * GETVFSBYNAME_ARG_TYPE vfc; * #endif */ # if defined(HAVE_STRUCT_XVFSCONF) struct xvfsconf vfc; # else struct vfsconf vfc; # endif #endif int r, xr = 0; -#if !defined(HAVE_STRUCT_STATFS_F_NAMEMAX) - long nm; -#endif t->current_filesystem->synthetic = -1; t->current_filesystem->remote = -1; if (tree_current_is_symblic_link_target(t)) { #if defined(HAVE_OPENAT) /* * Get file system statistics on any directory * where current is. */ int fd = openat(tree_current_dir_fd(t), tree_current_access_path(t), O_RDONLY | O_CLOEXEC); __archive_ensure_cloexec_flag(fd); if (fd < 0) { archive_set_error(&a->archive, errno, "openat failed"); return (ARCHIVE_FAILED); } r = fstatfs(fd, &sfs); if (r == 0) xr = get_xfer_size(t, fd, NULL); close(fd); #else if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } r = statfs(tree_current_access_path(t), &sfs); if (r == 0) xr = get_xfer_size(t, -1, tree_current_access_path(t)); #endif } else { r = fstatfs(tree_current_dir_fd(t), &sfs); if (r == 0) xr = get_xfer_size(t, tree_current_dir_fd(t), NULL); } if (r == -1 || xr == -1) { archive_set_error(&a->archive, errno, "statfs failed"); return (ARCHIVE_FAILED); } else if (xr == 1) { /* pathconf(_PC_REX_*) operations are not supported. */ set_statfs_transfer_size(t->current_filesystem, &sfs); } if (sfs.f_flags & MNT_LOCAL) t->current_filesystem->remote = 0; else t->current_filesystem->remote = 1; #if defined(HAVE_GETVFSBYNAME) && defined(VFCF_SYNTHETIC) r = getvfsbyname(sfs.f_fstypename, &vfc); if (r == -1) { archive_set_error(&a->archive, errno, "getvfsbyname failed"); return (ARCHIVE_FAILED); } if (vfc.vfc_flags & VFCF_SYNTHETIC) t->current_filesystem->synthetic = 1; else t->current_filesystem->synthetic = 0; #endif #if defined(MNT_NOATIME) if (sfs.f_flags & MNT_NOATIME) t->current_filesystem->noatime = 1; else #endif t->current_filesystem->noatime = 0; -#if defined(USE_READDIR_R) - /* Set maximum filename length. */ -#if defined(HAVE_STRUCT_STATFS_F_NAMEMAX) - t->current_filesystem->name_max = sfs.f_namemax; -#else -# if defined(_PC_NAME_MAX) - /* Mac OS X does not have f_namemax in struct statfs. */ - if (tree_current_is_symblic_link_target(t)) { - if (tree_enter_working_dir(t) != 0) { - archive_set_error(&a->archive, errno, "fchdir failed"); - return (ARCHIVE_FAILED); - } - nm = pathconf(tree_current_access_path(t), _PC_NAME_MAX); - } else - nm = fpathconf(tree_current_dir_fd(t), _PC_NAME_MAX); -# else - nm = -1; -# endif - if (nm == -1) - t->current_filesystem->name_max = NAME_MAX; - else - t->current_filesystem->name_max = nm; -#endif - if (t->current_filesystem->name_max == 0) { - archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, - "Cannot determine name_max"); - return (ARCHIVE_FAILED); - } -#endif /* USE_READDIR_R */ return (ARCHIVE_OK); } #elif (defined(HAVE_STATVFS) || defined(HAVE_FSTATVFS)) && defined(ST_LOCAL) /* * Gather current filesystem properties on NetBSD */ static int setup_current_filesystem(struct archive_read_disk *a) { struct tree *t = a->tree; struct statvfs svfs; int r, xr = 0; t->current_filesystem->synthetic = -1; if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } if (tree_current_is_symblic_link_target(t)) { r = statvfs(tree_current_access_path(t), &svfs); if (r == 0) xr = get_xfer_size(t, -1, tree_current_access_path(t)); } else { #ifdef HAVE_FSTATVFS r = fstatvfs(tree_current_dir_fd(t), &svfs); if (r == 0) xr = get_xfer_size(t, tree_current_dir_fd(t), NULL); #else r = statvfs(".", &svfs); if (r == 0) xr = get_xfer_size(t, -1, "."); #endif } if (r == -1 || xr == -1) { t->current_filesystem->remote = -1; archive_set_error(&a->archive, errno, "statvfs failed"); return (ARCHIVE_FAILED); } else if (xr == 1) { /* Usually come here unless NetBSD supports _PC_REC_XFER_ALIGN * for pathconf() function. */ set_statvfs_transfer_size(t->current_filesystem, &svfs); } if (svfs.f_flag & ST_LOCAL) t->current_filesystem->remote = 0; else t->current_filesystem->remote = 1; #if defined(ST_NOATIME) if (svfs.f_flag & ST_NOATIME) t->current_filesystem->noatime = 1; else #endif t->current_filesystem->noatime = 0; /* Set maximum filename length. */ t->current_filesystem->name_max = svfs.f_namemax; return (ARCHIVE_OK); } #elif defined(HAVE_SYS_STATFS_H) && defined(HAVE_LINUX_MAGIC_H) &&\ defined(HAVE_STATFS) && defined(HAVE_FSTATFS) /* * Note: statfs is deprecated since LSB 3.2 */ #ifndef CIFS_SUPER_MAGIC #define CIFS_SUPER_MAGIC 0xFF534D42 #endif #ifndef DEVFS_SUPER_MAGIC #define DEVFS_SUPER_MAGIC 0x1373 #endif /* * Gather current filesystem properties on Linux */ static int setup_current_filesystem(struct archive_read_disk *a) { struct tree *t = a->tree; struct statfs sfs; #if defined(HAVE_STATVFS) struct statvfs svfs; #endif int r, vr = 0, xr = 0; if (tree_current_is_symblic_link_target(t)) { #if defined(HAVE_OPENAT) /* * Get file system statistics on any directory * where current is. */ int fd = openat(tree_current_dir_fd(t), tree_current_access_path(t), O_RDONLY | O_CLOEXEC); __archive_ensure_cloexec_flag(fd); if (fd < 0) { archive_set_error(&a->archive, errno, "openat failed"); return (ARCHIVE_FAILED); } #if defined(HAVE_FSTATVFS) vr = fstatvfs(fd, &svfs);/* for f_flag, mount flags */ #endif r = fstatfs(fd, &sfs); if (r == 0) xr = get_xfer_size(t, fd, NULL); close(fd); #else if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } #if defined(HAVE_STATVFS) vr = statvfs(tree_current_access_path(t), &svfs); #endif r = statfs(tree_current_access_path(t), &sfs); if (r == 0) xr = get_xfer_size(t, -1, tree_current_access_path(t)); #endif } else { #ifdef HAVE_FSTATFS #if defined(HAVE_FSTATVFS) vr = fstatvfs(tree_current_dir_fd(t), &svfs); #endif r = fstatfs(tree_current_dir_fd(t), &sfs); if (r == 0) xr = get_xfer_size(t, tree_current_dir_fd(t), NULL); #else if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } #if defined(HAVE_STATVFS) vr = statvfs(".", &svfs); #endif r = statfs(".", &sfs); if (r == 0) xr = get_xfer_size(t, -1, "."); #endif } if (r == -1 || xr == -1 || vr == -1) { t->current_filesystem->synthetic = -1; t->current_filesystem->remote = -1; archive_set_error(&a->archive, errno, "statfs failed"); return (ARCHIVE_FAILED); } else if (xr == 1) { /* pathconf(_PC_REX_*) operations are not supported. */ #if defined(HAVE_STATVFS) set_statvfs_transfer_size(t->current_filesystem, &svfs); #else set_statfs_transfer_size(t->current_filesystem, &sfs); #endif } switch (sfs.f_type) { case AFS_SUPER_MAGIC: case CIFS_SUPER_MAGIC: case CODA_SUPER_MAGIC: case NCP_SUPER_MAGIC:/* NetWare */ case NFS_SUPER_MAGIC: case SMB_SUPER_MAGIC: t->current_filesystem->remote = 1; t->current_filesystem->synthetic = 0; break; case DEVFS_SUPER_MAGIC: case PROC_SUPER_MAGIC: case USBDEVICE_SUPER_MAGIC: t->current_filesystem->remote = 0; t->current_filesystem->synthetic = 1; break; default: t->current_filesystem->remote = 0; t->current_filesystem->synthetic = 0; break; } #if defined(ST_NOATIME) #if defined(HAVE_STATVFS) if (svfs.f_flag & ST_NOATIME) #else if (sfs.f_flags & ST_NOATIME) #endif t->current_filesystem->noatime = 1; else #endif t->current_filesystem->noatime = 0; -#if defined(USE_READDIR_R) - /* Set maximum filename length. */ -#if defined(HAVE_STATVFS) - t->current_filesystem->name_max = svfs.f_namemax; -#else - t->current_filesystem->name_max = sfs.f_namelen; -#endif - if (t->current_filesystem->name_max == 0) { - archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, - "Cannot determine name_max"); - return (ARCHIVE_FAILED); - } -#endif return (ARCHIVE_OK); } #elif defined(HAVE_SYS_STATVFS_H) &&\ (defined(HAVE_STATVFS) || defined(HAVE_FSTATVFS)) /* * Gather current filesystem properties on other posix platform. */ static int setup_current_filesystem(struct archive_read_disk *a) { struct tree *t = a->tree; struct statvfs svfs; int r, xr = 0; t->current_filesystem->synthetic = -1;/* Not supported */ t->current_filesystem->remote = -1;/* Not supported */ if (tree_current_is_symblic_link_target(t)) { #if defined(HAVE_OPENAT) /* * Get file system statistics on any directory * where current is. */ int fd = openat(tree_current_dir_fd(t), tree_current_access_path(t), O_RDONLY | O_CLOEXEC); __archive_ensure_cloexec_flag(fd); if (fd < 0) { archive_set_error(&a->archive, errno, "openat failed"); return (ARCHIVE_FAILED); } r = fstatvfs(fd, &svfs); if (r == 0) xr = get_xfer_size(t, fd, NULL); close(fd); #else if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } r = statvfs(tree_current_access_path(t), &svfs); if (r == 0) xr = get_xfer_size(t, -1, tree_current_access_path(t)); #endif } else { #ifdef HAVE_FSTATVFS r = fstatvfs(tree_current_dir_fd(t), &svfs); if (r == 0) xr = get_xfer_size(t, tree_current_dir_fd(t), NULL); #else if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } r = statvfs(".", &svfs); if (r == 0) xr = get_xfer_size(t, -1, "."); #endif } if (r == -1 || xr == -1) { t->current_filesystem->synthetic = -1; t->current_filesystem->remote = -1; archive_set_error(&a->archive, errno, "statvfs failed"); return (ARCHIVE_FAILED); } else if (xr == 1) { /* pathconf(_PC_REX_*) operations are not supported. */ set_statvfs_transfer_size(t->current_filesystem, &svfs); } #if defined(ST_NOATIME) if (svfs.f_flag & ST_NOATIME) t->current_filesystem->noatime = 1; else #endif t->current_filesystem->noatime = 0; -#if defined(USE_READDIR_R) - /* Set maximum filename length. */ - t->current_filesystem->name_max = svfs.f_namemax; - if (t->current_filesystem->name_max == 0) { - archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, - "Cannot determine name_max"); - return (ARCHIVE_FAILED); - } -#endif return (ARCHIVE_OK); } #else /* * Generic: Gather current filesystem properties. * TODO: Is this generic function really needed? */ static int setup_current_filesystem(struct archive_read_disk *a) { struct tree *t = a->tree; -#if defined(_PC_NAME_MAX) && defined(USE_READDIR_R) - long nm; -#endif t->current_filesystem->synthetic = -1;/* Not supported */ t->current_filesystem->remote = -1;/* Not supported */ t->current_filesystem->noatime = 0; (void)get_xfer_size(t, -1, ".");/* Dummy call to avoid build error. */ t->current_filesystem->xfer_align = -1;/* Unknown */ t->current_filesystem->max_xfer_size = -1; t->current_filesystem->min_xfer_size = -1; t->current_filesystem->incr_xfer_size = -1; -#if defined(USE_READDIR_R) - /* Set maximum filename length. */ -# if defined(_PC_NAME_MAX) - if (tree_current_is_symblic_link_target(t)) { - if (tree_enter_working_dir(t) != 0) { - archive_set_error(&a->archive, errno, "fchdir failed"); - return (ARCHIVE_FAILED); - } - nm = pathconf(tree_current_access_path(t), _PC_NAME_MAX); - } else - nm = fpathconf(tree_current_dir_fd(t), _PC_NAME_MAX); - if (nm == -1) -# endif /* _PC_NAME_MAX */ - /* - * Some systems (HP-UX or others?) incorrectly defined - * NAME_MAX macro to be a smaller value. - */ -# if defined(NAME_MAX) && NAME_MAX >= 255 - t->current_filesystem->name_max = NAME_MAX; -# else - /* No way to get a trusted value of maximum filename - * length. */ - t->current_filesystem->name_max = PATH_MAX; -# endif /* NAME_MAX */ -# if defined(_PC_NAME_MAX) - else - t->current_filesystem->name_max = nm; -# endif /* _PC_NAME_MAX */ - if (t->current_filesystem->name_max == 0) { - archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, - "Cannot determine name_max"); - return (ARCHIVE_FAILED); - } -#endif /* USE_READDIR_R */ return (ARCHIVE_OK); } #endif static int close_and_restore_time(int fd, struct tree *t, struct restore_time *rt) { #ifndef HAVE_UTIMES (void)t; /* UNUSED */ (void)rt; /* UNUSED */ return (close(fd)); #else #if defined(HAVE_FUTIMENS) && !defined(__CYGWIN__) struct timespec timespecs[2]; #endif struct timeval times[2]; if ((t->flags & needsRestoreTimes) == 0 || rt->noatime) { if (fd >= 0) return (close(fd)); else return (0); } #if defined(HAVE_FUTIMENS) && !defined(__CYGWIN__) timespecs[1].tv_sec = rt->mtime; timespecs[1].tv_nsec = rt->mtime_nsec; timespecs[0].tv_sec = rt->atime; timespecs[0].tv_nsec = rt->atime_nsec; /* futimens() is defined in POSIX.1-2008. */ if (futimens(fd, timespecs) == 0) return (close(fd)); #endif times[1].tv_sec = rt->mtime; times[1].tv_usec = rt->mtime_nsec / 1000; times[0].tv_sec = rt->atime; times[0].tv_usec = rt->atime_nsec / 1000; #if !defined(HAVE_FUTIMENS) && defined(HAVE_FUTIMES) && !defined(__CYGWIN__) if (futimes(fd, times) == 0) return (close(fd)); #endif close(fd); #if defined(HAVE_FUTIMESAT) if (futimesat(tree_current_dir_fd(t), rt->name, times) == 0) return (0); #endif #ifdef HAVE_LUTIMES if (lutimes(rt->name, times) != 0) #else if (AE_IFLNK != rt->filetype && utimes(rt->name, times) != 0) #endif return (-1); #endif return (0); } static int open_on_current_dir(struct tree *t, const char *path, int flags) { #ifdef HAVE_OPENAT return (openat(tree_current_dir_fd(t), path, flags)); #else if (tree_enter_working_dir(t) != 0) return (-1); return (open(path, flags)); #endif } static int tree_dup(int fd) { int new_fd; #ifdef F_DUPFD_CLOEXEC static volatile int can_dupfd_cloexec = 1; if (can_dupfd_cloexec) { new_fd = fcntl(fd, F_DUPFD_CLOEXEC, 0); if (new_fd != -1) return (new_fd); /* Linux 2.6.18 - 2.6.23 declare F_DUPFD_CLOEXEC, * but it cannot be used. So we have to try dup(). */ /* We won't try F_DUPFD_CLOEXEC. */ can_dupfd_cloexec = 0; } #endif /* F_DUPFD_CLOEXEC */ new_fd = dup(fd); - __archive_ensure_cloexec_flag(new_fd); - return (new_fd); + if (new_fd != -1) { + __archive_ensure_cloexec_flag(new_fd); + return (new_fd); + } + return (-1); } /* * Add a directory path to the current stack. */ static void tree_push(struct tree *t, const char *path, int filesystem_id, int64_t dev, int64_t ino, struct restore_time *rt) { struct tree_entry *te; te = calloc(1, sizeof(*te)); if (te == NULL) __archive_errx(1, "Out of memory"); te->next = t->stack; te->parent = t->current; if (te->parent) te->depth = te->parent->depth + 1; t->stack = te; archive_string_init(&te->name); te->symlink_parent_fd = -1; archive_strcpy(&te->name, path); te->flags = needsDescent | needsOpen | needsAscent; te->filesystem_id = filesystem_id; te->dev = dev; te->ino = ino; te->dirname_length = t->dirname_length; te->restore_time.name = te->name.s; if (rt != NULL) { te->restore_time.mtime = rt->mtime; te->restore_time.mtime_nsec = rt->mtime_nsec; te->restore_time.atime = rt->atime; te->restore_time.atime_nsec = rt->atime_nsec; te->restore_time.filetype = rt->filetype; te->restore_time.noatime = rt->noatime; } } /* * Append a name to the current dir path. */ static void tree_append(struct tree *t, const char *name, size_t name_length) { size_t size_needed; t->path.s[t->dirname_length] = '\0'; t->path.length = t->dirname_length; /* Strip trailing '/' from name, unless entire name is "/". */ while (name_length > 1 && name[name_length - 1] == '/') name_length--; /* Resize pathname buffer as needed. */ size_needed = name_length + t->dirname_length + 2; archive_string_ensure(&t->path, size_needed); /* Add a separating '/' if it's needed. */ if (t->dirname_length > 0 && t->path.s[archive_strlen(&t->path)-1] != '/') archive_strappend_char(&t->path, '/'); t->basename = t->path.s + archive_strlen(&t->path); archive_strncat(&t->path, name, name_length); t->restore_time.name = t->basename; } /* * Open a directory tree for traversal. */ static struct tree * tree_open(const char *path, char symlink_mode, int restore_time) { struct tree *t; if ((t = calloc(1, sizeof(*t))) == NULL) return (NULL); archive_string_init(&t->path); archive_string_ensure(&t->path, 31); t->initial_symlink_mode = symlink_mode; return (tree_reopen(t, path, restore_time)); } static struct tree * tree_reopen(struct tree *t, const char *path, int restore_time) { #if defined(O_PATH) /* Linux */ const int o_flag = O_PATH; #elif defined(O_SEARCH) /* SunOS */ const int o_flag = O_SEARCH; #elif defined(__FreeBSD__) && defined(O_EXEC) /* FreeBSD */ const int o_flag = O_EXEC; #endif t->flags = (restore_time != 0)?needsRestoreTimes:0; t->flags |= onInitialDir; t->visit_type = 0; t->tree_errno = 0; t->dirname_length = 0; t->depth = 0; t->descend = 0; t->current = NULL; t->d = INVALID_DIR_HANDLE; t->symlink_mode = t->initial_symlink_mode; archive_string_empty(&t->path); t->entry_fd = -1; t->entry_eof = 0; t->entry_remaining_bytes = 0; t->initial_filesystem_id = -1; /* First item is set up a lot like a symlink traversal. */ tree_push(t, path, 0, 0, 0, NULL); t->stack->flags = needsFirstVisit; t->maxOpenCount = t->openCount = 1; t->initial_dir_fd = open(".", O_RDONLY | O_CLOEXEC); #if defined(O_PATH) || defined(O_SEARCH) || \ (defined(__FreeBSD__) && defined(O_EXEC)) /* * Most likely reason to fail opening "." is that it's not readable, * so try again for execute. The consequences of not opening this are * unhelpful and unnecessary errors later. */ - if (t->initial_dir_fd < 0) + if (t->initial_dir_fd < 0) { t->initial_dir_fd = open(".", o_flag | O_CLOEXEC); + if (t->initial_dir_fd < 0) + return NULL; + } #endif __archive_ensure_cloexec_flag(t->initial_dir_fd); t->working_dir_fd = tree_dup(t->initial_dir_fd); + if (t->working_dir_fd < 0) + return NULL; return (t); } static int tree_descent(struct tree *t) { int flag, new_fd, r = 0; t->dirname_length = archive_strlen(&t->path); flag = O_RDONLY | O_CLOEXEC; #if defined(O_DIRECTORY) flag |= O_DIRECTORY; #endif new_fd = open_on_current_dir(t, t->stack->name.s, flag); __archive_ensure_cloexec_flag(new_fd); if (new_fd < 0) { t->tree_errno = errno; r = TREE_ERROR_DIR; } else { t->depth++; /* If it is a link, set up fd for the ascent. */ if (t->stack->flags & isDirLink) { t->stack->symlink_parent_fd = t->working_dir_fd; t->openCount++; if (t->openCount > t->maxOpenCount) t->maxOpenCount = t->openCount; } else close(t->working_dir_fd); /* Renew the current working directory. */ t->working_dir_fd = new_fd; t->flags &= ~onWorkingDir; } return (r); } /* * We've finished a directory; ascend back to the parent. */ static int tree_ascend(struct tree *t) { struct tree_entry *te; int new_fd, r = 0, prev_dir_fd; te = t->stack; prev_dir_fd = t->working_dir_fd; if (te->flags & isDirLink) new_fd = te->symlink_parent_fd; else { new_fd = open_on_current_dir(t, "..", O_RDONLY | O_CLOEXEC); __archive_ensure_cloexec_flag(new_fd); } if (new_fd < 0) { t->tree_errno = errno; r = TREE_ERROR_FATAL; } else { /* Renew the current working directory. */ t->working_dir_fd = new_fd; t->flags &= ~onWorkingDir; /* Current directory has been changed, we should * close an fd of previous working directory. */ close_and_restore_time(prev_dir_fd, t, &te->restore_time); if (te->flags & isDirLink) { t->openCount--; te->symlink_parent_fd = -1; } t->depth--; } return (r); } /* * Return to the initial directory where tree_open() was performed. */ static int tree_enter_initial_dir(struct tree *t) { int r = 0; if ((t->flags & onInitialDir) == 0) { r = fchdir(t->initial_dir_fd); if (r == 0) { t->flags &= ~onWorkingDir; t->flags |= onInitialDir; } } return (r); } /* * Restore working directory of directory traversals. */ static int tree_enter_working_dir(struct tree *t) { int r = 0; /* * Change the current directory if really needed. * Sometimes this is unneeded when we did not do * descent. */ if (t->depth > 0 && (t->flags & onWorkingDir) == 0) { r = fchdir(t->working_dir_fd); if (r == 0) { t->flags &= ~onInitialDir; t->flags |= onWorkingDir; } } return (r); } static int tree_current_dir_fd(struct tree *t) { return (t->working_dir_fd); } /* * Pop the working stack. */ static void tree_pop(struct tree *t) { struct tree_entry *te; t->path.s[t->dirname_length] = '\0'; t->path.length = t->dirname_length; if (t->stack == t->current && t->current != NULL) t->current = t->current->parent; te = t->stack; t->stack = te->next; t->dirname_length = te->dirname_length; t->basename = t->path.s + t->dirname_length; while (t->basename[0] == '/') t->basename++; archive_string_free(&te->name); free(te); } /* * Get the next item in the tree traversal. */ static int tree_next(struct tree *t) { int r; while (t->stack != NULL) { /* If there's an open dir, get the next entry from there. */ if (t->d != INVALID_DIR_HANDLE) { r = tree_dir_next_posix(t); if (r == 0) continue; return (r); } if (t->stack->flags & needsFirstVisit) { /* Top stack item needs a regular visit. */ t->current = t->stack; tree_append(t, t->stack->name.s, archive_strlen(&(t->stack->name))); /* t->dirname_length = t->path_length; */ /* tree_pop(t); */ t->stack->flags &= ~needsFirstVisit; return (t->visit_type = TREE_REGULAR); } else if (t->stack->flags & needsDescent) { /* Top stack item is dir to descend into. */ t->current = t->stack; tree_append(t, t->stack->name.s, archive_strlen(&(t->stack->name))); t->stack->flags &= ~needsDescent; r = tree_descent(t); if (r != 0) { tree_pop(t); t->visit_type = r; } else t->visit_type = TREE_POSTDESCENT; return (t->visit_type); } else if (t->stack->flags & needsOpen) { t->stack->flags &= ~needsOpen; r = tree_dir_next_posix(t); if (r == 0) continue; return (r); } else if (t->stack->flags & needsAscent) { /* Top stack item is dir and we're done with it. */ r = tree_ascend(t); tree_pop(t); t->visit_type = r != 0 ? r : TREE_POSTASCENT; return (t->visit_type); } else { /* Top item on stack is dead. */ tree_pop(t); t->flags &= ~hasLstat; t->flags &= ~hasStat; } } return (t->visit_type = 0); } static int tree_dir_next_posix(struct tree *t) { int r; const char *name; size_t namelen; if (t->d == NULL) { -#if defined(USE_READDIR_R) - size_t dirent_size; -#endif #if defined(HAVE_FDOPENDIR) - t->d = fdopendir(tree_dup(t->working_dir_fd)); + int fd = tree_dup(t->working_dir_fd); + if (fd != -1) + t->d = fdopendir(fd); #else /* HAVE_FDOPENDIR */ if (tree_enter_working_dir(t) == 0) { t->d = opendir("."); #ifdef HAVE_DIRFD __archive_ensure_cloexec_flag(dirfd(t->d)); #endif } #endif /* HAVE_FDOPENDIR */ if (t->d == NULL) { r = tree_ascend(t); /* Undo "chdir" */ tree_pop(t); t->tree_errno = errno; t->visit_type = r != 0 ? r : TREE_ERROR_DIR; return (t->visit_type); } -#if defined(USE_READDIR_R) - dirent_size = offsetof(struct dirent, d_name) + - t->filesystem_table[t->current->filesystem_id].name_max + 1; - if (t->dirent == NULL || t->dirent_allocated < dirent_size) { - free(t->dirent); - t->dirent = malloc(dirent_size); - if (t->dirent == NULL) { - closedir(t->d); - t->d = INVALID_DIR_HANDLE; - (void)tree_ascend(t); - tree_pop(t); - t->tree_errno = ENOMEM; - t->visit_type = TREE_ERROR_DIR; - return (t->visit_type); - } - t->dirent_allocated = dirent_size; - } -#endif /* USE_READDIR_R */ } for (;;) { errno = 0; -#if defined(USE_READDIR_R) - r = readdir_r(t->d, t->dirent, &t->de); -#ifdef _AIX - /* Note: According to the man page, return value 9 indicates - * that the readdir_r was not successful and the error code - * is set to the global errno variable. And then if the end - * of directory entries was reached, the return value is 9 - * and the third parameter is set to NULL and errno is - * unchanged. */ - if (r == 9) - r = errno; -#endif /* _AIX */ - if (r != 0 || t->de == NULL) { -#else t->de = readdir(t->d); if (t->de == NULL) { r = errno; -#endif closedir(t->d); t->d = INVALID_DIR_HANDLE; if (r != 0) { t->tree_errno = r; t->visit_type = TREE_ERROR_DIR; return (t->visit_type); } else return (0); } name = t->de->d_name; namelen = D_NAMELEN(t->de); t->flags &= ~hasLstat; t->flags &= ~hasStat; if (name[0] == '.' && name[1] == '\0') continue; if (name[0] == '.' && name[1] == '.' && name[2] == '\0') continue; tree_append(t, name, namelen); return (t->visit_type = TREE_REGULAR); } } /* * Get the stat() data for the entry just returned from tree_next(). */ static const struct stat * tree_current_stat(struct tree *t) { if (!(t->flags & hasStat)) { #ifdef HAVE_FSTATAT if (fstatat(tree_current_dir_fd(t), tree_current_access_path(t), &t->st, 0) != 0) #else if (tree_enter_working_dir(t) != 0) return NULL; if (la_stat(tree_current_access_path(t), &t->st) != 0) #endif return NULL; t->flags |= hasStat; } return (&t->st); } /* * Get the lstat() data for the entry just returned from tree_next(). */ static const struct stat * tree_current_lstat(struct tree *t) { if (!(t->flags & hasLstat)) { #ifdef HAVE_FSTATAT if (fstatat(tree_current_dir_fd(t), tree_current_access_path(t), &t->lst, AT_SYMLINK_NOFOLLOW) != 0) #else if (tree_enter_working_dir(t) != 0) return NULL; #ifdef HAVE_LSTAT if (lstat(tree_current_access_path(t), &t->lst) != 0) #else if (la_stat(tree_current_access_path(t), &t->lst) != 0) #endif #endif return NULL; t->flags |= hasLstat; } return (&t->lst); } /* * Test whether current entry is a dir or link to a dir. */ static int tree_current_is_dir(struct tree *t) { const struct stat *st; /* * If we already have lstat() info, then try some * cheap tests to determine if this is a dir. */ if (t->flags & hasLstat) { /* If lstat() says it's a dir, it must be a dir. */ st = tree_current_lstat(t); if (st == NULL) return 0; if (S_ISDIR(st->st_mode)) return 1; /* Not a dir; might be a link to a dir. */ /* If it's not a link, then it's not a link to a dir. */ if (!S_ISLNK(st->st_mode)) return 0; /* * It's a link, but we don't know what it's a link to, * so we'll have to use stat(). */ } st = tree_current_stat(t); /* If we can't stat it, it's not a dir. */ if (st == NULL) return 0; /* Use the definitive test. Hopefully this is cached. */ return (S_ISDIR(st->st_mode)); } /* * Test whether current entry is a physical directory. Usually, we * already have at least one of stat() or lstat() in memory, so we * use tricks to try to avoid an extra trip to the disk. */ static int tree_current_is_physical_dir(struct tree *t) { const struct stat *st; /* * If stat() says it isn't a dir, then it's not a dir. * If stat() data is cached, this check is free, so do it first. */ if (t->flags & hasStat) { st = tree_current_stat(t); if (st == NULL) return (0); if (!S_ISDIR(st->st_mode)) return (0); } /* * Either stat() said it was a dir (in which case, we have * to determine whether it's really a link to a dir) or * stat() info wasn't available. So we use lstat(), which * hopefully is already cached. */ st = tree_current_lstat(t); /* If we can't stat it, it's not a dir. */ if (st == NULL) return 0; /* Use the definitive test. Hopefully this is cached. */ return (S_ISDIR(st->st_mode)); } /* * Test whether the same file has been in the tree as its parent. */ static int tree_target_is_same_as_parent(struct tree *t, const struct stat *st) { struct tree_entry *te; for (te = t->current->parent; te != NULL; te = te->parent) { if (te->dev == (int64_t)st->st_dev && te->ino == (int64_t)st->st_ino) return (1); } return (0); } /* * Test whether the current file is symbolic link target and * on the other filesystem. */ static int tree_current_is_symblic_link_target(struct tree *t) { static const struct stat *lst, *st; lst = tree_current_lstat(t); st = tree_current_stat(t); return (st != NULL && lst != NULL && (int64_t)st->st_dev == t->current_filesystem->dev && st->st_dev != lst->st_dev); } /* * Return the access path for the entry just returned from tree_next(). */ static const char * tree_current_access_path(struct tree *t) { return (t->basename); } /* * Return the full path for the entry just returned from tree_next(). */ static const char * tree_current_path(struct tree *t) { return (t->path.s); } /* * Terminate the traversal. */ static void tree_close(struct tree *t) { if (t == NULL) return; if (t->entry_fd >= 0) { close_and_restore_time(t->entry_fd, t, &t->restore_time); t->entry_fd = -1; } /* Close the handle of readdir(). */ if (t->d != INVALID_DIR_HANDLE) { closedir(t->d); t->d = INVALID_DIR_HANDLE; } /* Release anything remaining in the stack. */ while (t->stack != NULL) { if (t->stack->flags & isDirLink) close(t->stack->symlink_parent_fd); tree_pop(t); } if (t->working_dir_fd >= 0) { close(t->working_dir_fd); t->working_dir_fd = -1; } if (t->initial_dir_fd >= 0) { close(t->initial_dir_fd); t->initial_dir_fd = -1; } } /* * Release any resources. */ static void tree_free(struct tree *t) { int i; if (t == NULL) return; archive_string_free(&t->path); -#if defined(USE_READDIR_R) - free(t->dirent); -#endif free(t->sparse_list); for (i = 0; i < t->max_filesystem_id; i++) free(t->filesystem_table[i].allocation_ptr); free(t->filesystem_table); free(t); } #endif diff --git a/contrib/libarchive/libarchive/archive_read_open_fd.c b/contrib/libarchive/libarchive/archive_read_open_fd.c index dc7c9e52c6f6..c85a62a3e2d7 100644 --- a/contrib/libarchive/libarchive/archive_read_open_fd.c +++ b/contrib/libarchive/libarchive/archive_read_open_fd.c @@ -1,230 +1,233 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_IO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #include "archive.h" +#include "archive_platform_stat.h" struct read_fd_data { int fd; size_t block_size; int64_t size; char use_lseek; void *buffer; }; static int file_close(struct archive *, void *); static ssize_t file_read(struct archive *, void *, const void **buff); static int64_t file_seek(struct archive *, void *, int64_t request, int); static int64_t file_skip(struct archive *, void *, int64_t request); int archive_read_open_fd(struct archive *a, int fd, size_t block_size) { - struct stat st; + la_seek_stat_t st; struct read_fd_data *mine; void *b; archive_clear_error(a); - if (fstat(fd, &st) != 0) { + if (la_seek_fstat(fd, &st) != 0) { archive_set_error(a, errno, "Can't stat fd %d", fd); return (ARCHIVE_FATAL); } mine = calloc(1, sizeof(*mine)); b = malloc(block_size); if (mine == NULL || b == NULL) { archive_set_error(a, ENOMEM, "No memory"); free(mine); free(b); return (ARCHIVE_FATAL); } mine->block_size = block_size; mine->buffer = b; mine->fd = fd; /* * Skip support is a performance optimization for anything * that supports lseek(). On FreeBSD, only regular files and * raw disk devices support lseek() and there's no portable * way to determine if a device is a raw disk device, so we * only enable this optimization for regular files. */ if (S_ISREG(st.st_mode)) { archive_read_extract_set_skip_file(a, st.st_dev, st.st_ino); mine->use_lseek = 1; mine->size = st.st_size; } #if defined(__CYGWIN__) || defined(_WIN32) setmode(mine->fd, O_BINARY); #endif archive_read_set_read_callback(a, file_read); archive_read_set_skip_callback(a, file_skip); archive_read_set_seek_callback(a, file_seek); archive_read_set_close_callback(a, file_close); archive_read_set_callback_data(a, mine); return (archive_read_open1(a)); } static ssize_t file_read(struct archive *a, void *client_data, const void **buff) { struct read_fd_data *mine = (struct read_fd_data *)client_data; ssize_t bytes_read; *buff = mine->buffer; for (;;) { bytes_read = read(mine->fd, mine->buffer, mine->block_size); if (bytes_read < 0) { if (errno == EINTR) continue; archive_set_error(a, errno, "Error reading fd %d", mine->fd); } return (bytes_read); } } static int64_t file_skip(struct archive *a, void *client_data, int64_t request) { struct read_fd_data *mine = (struct read_fd_data *)client_data; - off_t skip = (off_t)request; + la_seek_t skip = (la_seek_t)request; int64_t old_offset, new_offset; int skip_bits = sizeof(skip) * 8 - 1; /* off_t is a signed type. */ if (!mine->use_lseek) return (0); /* Reduce a request that would overflow the 'skip' variable. */ if (sizeof(request) > sizeof(skip)) { const int64_t max_skip = (((int64_t)1 << (skip_bits - 1)) - 1) * 2 + 1; if (request > max_skip) skip = (off_t)max_skip; } /* Reduce 'skip' to the next smallest multiple of block_size */ - skip = (off_t)(((int64_t)skip / mine->block_size) * mine->block_size); + skip = (la_seek_t)(((int64_t)skip / mine->block_size) * mine->block_size); + if (skip == 0) return (0); if ((old_offset = lseek(mine->fd, 0, SEEK_CUR)) >= 0) { if (old_offset >= mine->size || skip > mine->size - old_offset) { /* Do not seek past end of file. */ errno = ESPIPE; } else if ((new_offset = lseek(mine->fd, skip, SEEK_CUR)) >= 0) return (new_offset - old_offset); } /* If seek failed once, it will probably fail again. */ mine->use_lseek = 0; /* Let libarchive recover with read+discard. */ if (errno == ESPIPE) return (0); /* * There's been an error other than ESPIPE. This is most * likely caused by a programmer error (too large request) * or a corrupted archive file. */ archive_set_error(a, errno, "Error seeking"); return (-1); } /* * TODO: Store the offset and use it in the read callback. */ static int64_t file_seek(struct archive *a, void *client_data, int64_t request, int whence) { struct read_fd_data *mine = (struct read_fd_data *)client_data; - off_t seek = (off_t)request; + la_seek_t seek = (la_seek_t)request; int64_t r; int seek_bits = sizeof(seek) * 8 - 1; /* off_t is a signed type. */ /* We use off_t here because lseek() is declared that way. */ - /* Reduce a request that would overflow the 'seek' variable. */ + /* Do not perform a seek which cannot be fulfilled. */ if (sizeof(request) > sizeof(seek)) { const int64_t max_seek = (((int64_t)1 << (seek_bits - 1)) - 1) * 2 + 1; const int64_t min_seek = ~max_seek; - if (request > max_seek) - seek = (off_t)max_seek; - else if (request < min_seek) - seek = (off_t)min_seek; + if (request < min_seek || request > max_seek) { + errno = EOVERFLOW; + goto err; + } } r = lseek(mine->fd, seek, whence); if (r >= 0) return r; +err: if (errno == ESPIPE) { archive_set_error(a, errno, "A file descriptor(%d) is not seekable(PIPE)", mine->fd); return (ARCHIVE_FAILED); } else { /* If the input is corrupted or truncated, fail. */ archive_set_error(a, errno, "Error seeking in a file descriptor(%d)", mine->fd); return (ARCHIVE_FATAL); } } static int file_close(struct archive *a, void *client_data) { struct read_fd_data *mine = (struct read_fd_data *)client_data; (void)a; /* UNUSED */ free(mine->buffer); free(mine); return (ARCHIVE_OK); } diff --git a/contrib/libarchive/libarchive/archive_read_open_file.c b/contrib/libarchive/libarchive/archive_read_open_file.c index 742923abbee9..6ca2ff191aa8 100644 --- a/contrib/libarchive/libarchive/archive_read_open_file.c +++ b/contrib/libarchive/libarchive/archive_read_open_file.c @@ -1,252 +1,254 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_IO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #include "archive.h" +#include "archive_platform_stat.h" struct read_FILE_data { FILE *f; size_t block_size; int64_t size; void *buffer; char can_skip; }; static int FILE_close(struct archive *, void *); static ssize_t FILE_read(struct archive *, void *, const void **buff); static int64_t FILE_seek(struct archive *, void *, int64_t, int); static int64_t FILE_skip(struct archive *, void *, int64_t); int archive_read_open_FILE(struct archive *a, FILE *f) { - struct stat st; + la_seek_stat_t st; struct read_FILE_data *mine; size_t block_size = 128 * 1024; void *b; archive_clear_error(a); mine = calloc(1, sizeof(*mine)); b = malloc(block_size); if (mine == NULL || b == NULL) { archive_set_error(a, ENOMEM, "No memory"); free(mine); free(b); return (ARCHIVE_FATAL); } mine->block_size = block_size; mine->buffer = b; mine->f = f; /* * If we can't fstat() the file, it may just be that it's not * a file. (On some platforms, FILE * objects can wrap I/O * streams that don't support fileno()). As a result, fileno() * should be used cautiously.) */ - if (fstat(fileno(mine->f), &st) == 0 && S_ISREG(st.st_mode)) { + if (la_seek_fstat(fileno(mine->f), &st) == 0 && S_ISREG(st.st_mode)) { archive_read_extract_set_skip_file(a, st.st_dev, st.st_ino); /* Enable the seek optimization only for regular files. */ mine->can_skip = 1; mine->size = st.st_size; } #if defined(__CYGWIN__) || defined(_WIN32) setmode(fileno(mine->f), O_BINARY); #endif archive_read_set_read_callback(a, FILE_read); archive_read_set_skip_callback(a, FILE_skip); archive_read_set_seek_callback(a, FILE_seek); archive_read_set_close_callback(a, FILE_close); archive_read_set_callback_data(a, mine); return (archive_read_open1(a)); } static ssize_t FILE_read(struct archive *a, void *client_data, const void **buff) { struct read_FILE_data *mine = (struct read_FILE_data *)client_data; size_t bytes_read; *buff = mine->buffer; bytes_read = fread(mine->buffer, 1, mine->block_size, mine->f); if (bytes_read < mine->block_size && ferror(mine->f)) { archive_set_error(a, errno, "Error reading file"); } return (bytes_read); } static int64_t FILE_skip(struct archive *a, void *client_data, int64_t request) { struct read_FILE_data *mine = (struct read_FILE_data *)client_data; #if HAVE__FSEEKI64 int64_t skip = request; #elif HAVE_FSEEKO off_t skip = (off_t)request; #else long skip = (long)request; #endif int64_t old_offset, new_offset = -1; int skip_bits = sizeof(skip) * 8 - 1; (void)a; /* UNUSED */ /* * If we can't skip, return 0 as the amount we did step and * the caller will work around by reading and discarding. */ if (!mine->can_skip) return (0); if (request == 0) return (0); /* If request is too big for a long or an off_t, reduce it. */ if (sizeof(request) > sizeof(skip)) { const int64_t max_skip = (((int64_t)1 << (skip_bits - 1)) - 1) * 2 + 1; if (request > max_skip) skip = max_skip; } #ifdef __ANDROID__ /* fileno() isn't safe on all platforms ... see above. */ old_offset = lseek(fileno(mine->f), 0, SEEK_CUR); #elif HAVE__FSEEKI64 old_offset = _ftelli64(mine->f); #elif HAVE_FSEEKO old_offset = ftello(mine->f); #else old_offset = ftell(mine->f); #endif if (old_offset >= 0) { if (old_offset < mine->size && skip <= mine->size - old_offset) { #ifdef __ANDROID__ new_offset = lseek(fileno(mine->f), skip, SEEK_CUR); #elif HAVE__FSEEKI64 if (_fseeki64(mine->f, skip, SEEK_CUR) == 0) new_offset = _ftelli64(mine->f); #elif HAVE_FSEEKO if (fseeko(mine->f, skip, SEEK_CUR) == 0) new_offset = ftello(mine->f); #else if (fseek(mine->f, skip, SEEK_CUR) == 0) new_offset = ftell(mine->f); #endif if (new_offset >= 0) return (new_offset - old_offset); } } mine->can_skip = 0; return (0); } /* * TODO: Store the offset and use it in the read callback. */ static int64_t FILE_seek(struct archive *a, void *client_data, int64_t request, int whence) { struct read_FILE_data *mine = (struct read_FILE_data *)client_data; #if HAVE__FSEEKI64 int64_t seek = request; #elif HAVE_FSEEKO off_t seek = (off_t)request; #else long seek = (long)request; #endif int seek_bits = sizeof(seek) * 8 - 1; (void)a; /* UNUSED */ - /* Reduce a request that would overflow the 'seek' variable. */ + /* Do not perform a seek which cannot be fulfilled. */ if (sizeof(request) > sizeof(seek)) { const int64_t max_seek = (((int64_t)1 << (seek_bits - 1)) - 1) * 2 + 1; const int64_t min_seek = ~max_seek; - if (request > max_seek) - seek = max_seek; - else if (request < min_seek) - seek = min_seek; + if (request < min_seek || request > max_seek) { + errno = EOVERFLOW; + goto err; + } } #ifdef __ANDROID__ /* Newer Android versions have fseeko...to meditate. */ int64_t ret = lseek(fileno(mine->f), seek, whence); if (ret >= 0) { return ret; } #elif HAVE__FSEEKI64 if (_fseeki64(mine->f, seek, whence) == 0) { return _ftelli64(mine->f); } #elif HAVE_FSEEKO if (fseeko(mine->f, seek, whence) == 0) { return ftello(mine->f); } #else if (fseek(mine->f, seek, whence) == 0) { return ftell(mine->f); } #endif /* If we arrive here, the input is corrupted or truncated so fail. */ +err: archive_set_error(a, errno, "Error seeking in FILE* pointer"); return (ARCHIVE_FATAL); } static int FILE_close(struct archive *a, void *client_data) { struct read_FILE_data *mine = (struct read_FILE_data *)client_data; (void)a; /* UNUSED */ free(mine->buffer); free(mine); return (ARCHIVE_OK); } diff --git a/contrib/libarchive/libarchive/archive_read_open_filename.c b/contrib/libarchive/libarchive/archive_read_open_filename.c index 5f5b3f1f7259..a910eefcbfd2 100644 --- a/contrib/libarchive/libarchive/archive_read_open_filename.c +++ b/contrib/libarchive/libarchive/archive_read_open_filename.c @@ -1,641 +1,648 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_IOCTL_H #include #endif #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_IO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__) #include #elif defined(__NetBSD__) || defined(__OpenBSD__) #include #include #elif defined(__DragonFly__) #include #endif #include "archive.h" +#include "archive_platform_stat.h" #include "archive_private.h" #include "archive_string.h" #ifndef O_BINARY #define O_BINARY 0 #endif #ifndef O_CLOEXEC #define O_CLOEXEC 0 #endif struct read_file_data { int fd; size_t block_size; void *buffer; mode_t st_mode; /* Mode bits for opened file. */ int64_t size; char use_lseek; enum fnt_e { FNT_STDIN, FNT_MBS, FNT_WCS } filename_type; union { char m[1];/* MBS filename. */ wchar_t w[1];/* WCS filename. */ } filename; /* Must be last! */ }; static int file_open(struct archive *, void *); static int file_close(struct archive *, void *); static int file_close2(struct archive *, void *); static int file_switch(struct archive *, void *, void *); static ssize_t file_read(struct archive *, void *, const void **buff); static int64_t file_seek(struct archive *, void *, int64_t request, int); static int64_t file_skip(struct archive *, void *, int64_t request); static int64_t file_skip_lseek(struct archive *, void *, int64_t request); int archive_read_open_file(struct archive *a, const char *filename, size_t block_size) { return (archive_read_open_filename(a, filename, block_size)); } int archive_read_open_filename(struct archive *a, const char *filename, size_t block_size) { const char *filenames[2]; filenames[0] = filename; filenames[1] = NULL; return archive_read_open_filenames(a, filenames, block_size); } int archive_read_open_filenames(struct archive *a, const char **filenames, size_t block_size) { struct read_file_data *mine; const char *filename = NULL; if (filenames) filename = *(filenames++); archive_clear_error(a); do { if (filename == NULL) filename = ""; mine = calloc(1, sizeof(*mine) + strlen(filename)); if (mine == NULL) goto no_memory; strcpy(mine->filename.m, filename); mine->block_size = block_size; mine->fd = -1; mine->buffer = NULL; mine->st_mode = mine->use_lseek = 0; if (filename == NULL || filename[0] == '\0') { mine->filename_type = FNT_STDIN; } else mine->filename_type = FNT_MBS; - if (archive_read_append_callback_data(a, mine) != (ARCHIVE_OK)) + if (archive_read_append_callback_data(a, mine) != (ARCHIVE_OK)) { + free(mine); return (ARCHIVE_FATAL); + } if (filenames == NULL) break; filename = *(filenames++); } while (filename != NULL && filename[0] != '\0'); archive_read_set_open_callback(a, file_open); archive_read_set_read_callback(a, file_read); archive_read_set_skip_callback(a, file_skip); archive_read_set_close_callback(a, file_close); archive_read_set_switch_callback(a, file_switch); archive_read_set_seek_callback(a, file_seek); return (archive_read_open1(a)); no_memory: archive_set_error(a, ENOMEM, "No memory"); return (ARCHIVE_FATAL); } /* * This function is an implementation detail of archive_read_open_filename_w, * which is exposed as a separate API on Windows. */ #if !defined(_WIN32) || defined(__CYGWIN__) static #endif int archive_read_open_filenames_w(struct archive *a, const wchar_t **wfilenames, size_t block_size) { struct read_file_data *mine; const wchar_t *wfilename = NULL; if (wfilenames) wfilename = *(wfilenames++); archive_clear_error(a); do { if (wfilename == NULL) wfilename = L""; mine = calloc(1, sizeof(*mine) + wcslen(wfilename) * sizeof(wchar_t)); if (mine == NULL) goto no_memory; mine->block_size = block_size; mine->fd = -1; if (wfilename == NULL || wfilename[0] == L'\0') { mine->filename_type = FNT_STDIN; } else { #if defined(_WIN32) && !defined(__CYGWIN__) mine->filename_type = FNT_WCS; wcscpy(mine->filename.w, wfilename); #else /* * POSIX system does not support a wchar_t interface for * open() system call, so we have to translate a wchar_t * filename to multi-byte one and use it. */ struct archive_string fn; archive_string_init(&fn); if (archive_string_append_from_wcs(&fn, wfilename, wcslen(wfilename)) != 0) { if (errno == ENOMEM) archive_set_error(a, errno, "Can't allocate memory"); else archive_set_error(a, EINVAL, "Failed to convert a wide-character" " filename to a multi-byte filename"); archive_string_free(&fn); free(mine); return (ARCHIVE_FATAL); } mine->filename_type = FNT_MBS; strcpy(mine->filename.m, fn.s); archive_string_free(&fn); #endif } - if (archive_read_append_callback_data(a, mine) != (ARCHIVE_OK)) + if (archive_read_append_callback_data(a, mine) != (ARCHIVE_OK)) { + free(mine); return (ARCHIVE_FATAL); + } if (wfilenames == NULL) break; wfilename = *(wfilenames++); } while (wfilename != NULL && wfilename[0] != '\0'); archive_read_set_open_callback(a, file_open); archive_read_set_read_callback(a, file_read); archive_read_set_skip_callback(a, file_skip); archive_read_set_close_callback(a, file_close); archive_read_set_switch_callback(a, file_switch); archive_read_set_seek_callback(a, file_seek); return (archive_read_open1(a)); no_memory: archive_set_error(a, ENOMEM, "No memory"); return (ARCHIVE_FATAL); } int archive_read_open_filename_w(struct archive *a, const wchar_t *wfilename, size_t block_size) { const wchar_t *wfilenames[2]; wfilenames[0] = wfilename; wfilenames[1] = NULL; return archive_read_open_filenames_w(a, wfilenames, block_size); } static int file_open(struct archive *a, void *client_data) { - struct stat st; + la_seek_stat_t st; struct read_file_data *mine = (struct read_file_data *)client_data; void *buffer; const char *filename = NULL; #if defined(_WIN32) && !defined(__CYGWIN__) const wchar_t *wfilename = NULL; #endif int fd = -1; int is_disk_like = 0; #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__) off_t mediasize = 0; /* FreeBSD-specific, so off_t okay here. */ #elif defined(__NetBSD__) || defined(__OpenBSD__) struct disklabel dl; #elif defined(__DragonFly__) struct partinfo pi; #endif archive_clear_error(a); if (mine->filename_type == FNT_STDIN) { /* We used to delegate stdin support by * directly calling archive_read_open_fd(a,0,block_size) * here, but that doesn't (and shouldn't) handle the * end-of-file flush when reading stdout from a pipe. * Basically, read_open_fd() is intended for folks who * are willing to handle such details themselves. This * API is intended to be a little smarter for folks who * want easy handling of the common case. */ fd = 0; #if defined(__CYGWIN__) || defined(_WIN32) setmode(0, O_BINARY); #endif filename = ""; } else if (mine->filename_type == FNT_MBS) { filename = mine->filename.m; fd = open(filename, O_RDONLY | O_BINARY | O_CLOEXEC); __archive_ensure_cloexec_flag(fd); if (fd < 0) { archive_set_error(a, errno, "Failed to open '%s'", filename); return (ARCHIVE_FATAL); } } else { #if defined(_WIN32) && !defined(__CYGWIN__) wfilename = mine->filename.w; fd = _wopen(wfilename, O_RDONLY | O_BINARY); if (fd < 0 && errno == ENOENT) { wchar_t *fullpath; fullpath = __la_win_permissive_name_w(wfilename); if (fullpath != NULL) { fd = _wopen(fullpath, O_RDONLY | O_BINARY); free(fullpath); } } if (fd < 0) { archive_set_error(a, errno, "Failed to open '%ls'", wfilename); return (ARCHIVE_FATAL); } #else archive_set_error(a, ARCHIVE_ERRNO_MISC, "Unexpedted operation in archive_read_open_filename"); goto fail; #endif } - if (fstat(fd, &st) != 0) { + if (la_seek_fstat(fd, &st) != 0) { #if defined(_WIN32) && !defined(__CYGWIN__) if (mine->filename_type == FNT_WCS) archive_set_error(a, errno, "Can't stat '%ls'", wfilename); else #endif archive_set_error(a, errno, "Can't stat '%s'", filename); goto fail; } /* * Determine whether the input looks like a disk device or a * tape device. The results are used below to select an I/O * strategy: * = "disk-like" devices support arbitrary lseek() and will * support I/O requests of any size. So we get easy skipping * and can cheat on block sizes to get better performance. * = "tape-like" devices require strict blocking and use * specialized ioctls for seeking. * = "socket-like" devices cannot seek at all but can improve * performance by using nonblocking I/O to read "whatever is * available right now". * * Right now, we only specially recognize disk-like devices, * but it should be straightforward to add probes and strategy * here for tape-like and socket-like devices. */ if (S_ISREG(st.st_mode)) { /* Safety: Tell the extractor not to overwrite the input. */ archive_read_extract_set_skip_file(a, st.st_dev, st.st_ino); /* Regular files act like disks. */ is_disk_like = 1; } #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__) /* FreeBSD: if it supports DIOCGMEDIASIZE ioctl, it's disk-like. */ else if (S_ISCHR(st.st_mode) && ioctl(fd, DIOCGMEDIASIZE, &mediasize) == 0 && mediasize > 0) { is_disk_like = 1; } #elif defined(__NetBSD__) || defined(__OpenBSD__) /* Net/OpenBSD: if it supports DIOCGDINFO ioctl, it's disk-like. */ else if ((S_ISCHR(st.st_mode) || S_ISBLK(st.st_mode)) && ioctl(fd, DIOCGDINFO, &dl) == 0 && dl.d_partitions[DISKPART(st.st_rdev)].p_size > 0) { is_disk_like = 1; } #elif defined(__DragonFly__) /* DragonFly BSD: if it supports DIOCGPART ioctl, it's disk-like. */ else if (S_ISCHR(st.st_mode) && ioctl(fd, DIOCGPART, &pi) == 0 && pi.media_size > 0) { is_disk_like = 1; } #elif defined(__linux__) /* Linux: All block devices are disk-like. */ else if (S_ISBLK(st.st_mode) && lseek(fd, 0, SEEK_CUR) == 0 && lseek(fd, 0, SEEK_SET) == 0 && lseek(fd, 0, SEEK_END) > 0 && lseek(fd, 0, SEEK_SET) == 0) { is_disk_like = 1; } #endif /* TODO: Add an "is_tape_like" variable and appropriate tests. */ /* Disk-like devices prefer power-of-two block sizes. */ /* Use provided block_size as a guide so users have some control. */ if (is_disk_like) { size_t new_block_size = 64 * 1024; while (new_block_size < mine->block_size && new_block_size < 64 * 1024 * 1024) new_block_size *= 2; mine->block_size = new_block_size; } buffer = malloc(mine->block_size); if (buffer == NULL) { archive_set_error(a, ENOMEM, "No memory"); goto fail; } mine->buffer = buffer; mine->fd = fd; /* Remember mode so close can decide whether to flush. */ mine->st_mode = st.st_mode; /* Disk-like inputs can use lseek(). */ if (is_disk_like) { mine->use_lseek = 1; mine->size = st.st_size; } return (ARCHIVE_OK); fail: /* * Don't close file descriptors not opened or ones pointing referring * to `FNT_STDIN`. */ if (fd != -1 && fd != 0) close(fd); return (ARCHIVE_FATAL); } static ssize_t file_read(struct archive *a, void *client_data, const void **buff) { struct read_file_data *mine = (struct read_file_data *)client_data; ssize_t bytes_read; /* TODO: If a recent lseek() operation has left us * mis-aligned, read and return a short block to try to get * us back in alignment. */ /* TODO: Someday, try mmap() here; if that succeeds, give * the entire file to libarchive as a single block. That * could be a lot faster than block-by-block manual I/O. */ /* TODO: We might be able to improve performance on pipes and * sockets by setting non-blocking I/O and just accepting * whatever we get here instead of waiting for a full block * worth of data. */ *buff = mine->buffer; for (;;) { bytes_read = read(mine->fd, mine->buffer, mine->block_size); if (bytes_read < 0) { if (errno == EINTR) continue; else if (mine->filename_type == FNT_STDIN) archive_set_error(a, errno, "Error reading stdin"); else if (mine->filename_type == FNT_MBS) archive_set_error(a, errno, "Error reading '%s'", mine->filename.m); else archive_set_error(a, errno, "Error reading '%ls'", mine->filename.w); } return (bytes_read); } } /* * Regular files and disk-like block devices can use simple lseek * without needing to round the request to the block size. * * TODO: This can leave future reads mis-aligned. Since we know the * offset here, we should store it and use it in file_read() above * to determine whether we should perform a short read to get back * into alignment. Long series of mis-aligned reads can negatively * impact disk throughput. (Of course, the performance impact should * be carefully tested; extra code complexity is only worthwhile if * it does provide measurable improvement.) * * TODO: Be lazy about the actual seek. There are a few pathological * cases where libarchive makes a bunch of seek requests in a row * without any intervening reads. This isn't a huge performance * problem, since the kernel handles seeks lazily already, but * it would be very slightly faster if we simply remembered the * seek request here and then actually performed the seek at the * top of the read callback above. */ static int64_t file_skip_lseek(struct archive *a, void *client_data, int64_t request) { struct read_file_data *mine = (struct read_file_data *)client_data; #if defined(_WIN32) && !defined(__CYGWIN__) /* We use _lseeki64() on Windows. */ - int64_t old_offset, new_offset, skip = request; + int64_t old_offset, new_offset; #else - off_t old_offset, new_offset, skip = (off_t)request; + off_t old_offset, new_offset; #endif + la_seek_t skip = (la_seek_t)request; int skip_bits = sizeof(skip) * 8 - 1; /* We use off_t here because lseek() is declared that way. */ /* Reduce a request that would overflow the 'skip' variable. */ if (sizeof(request) > sizeof(skip)) { const int64_t max_skip = (((int64_t)1 << (skip_bits - 1)) - 1) * 2 + 1; if (request > max_skip) skip = max_skip; } if ((old_offset = lseek(mine->fd, 0, SEEK_CUR)) >= 0) { if (old_offset >= mine->size || skip > mine->size - old_offset) { /* Do not seek past end of file. */ errno = ESPIPE; } else if ((new_offset = lseek(mine->fd, skip, SEEK_CUR)) >= 0) return (new_offset - old_offset); } /* If lseek() fails, don't bother trying again. */ mine->use_lseek = 0; /* Let libarchive recover with read+discard */ if (errno == ESPIPE) return (0); /* If the input is corrupted or truncated, fail. */ if (mine->filename_type == FNT_STDIN) archive_set_error(a, errno, "Error seeking in stdin"); else if (mine->filename_type == FNT_MBS) archive_set_error(a, errno, "Error seeking in '%s'", mine->filename.m); else archive_set_error(a, errno, "Error seeking in '%ls'", mine->filename.w); return (-1); } /* * TODO: Implement another file_skip_XXXX that uses MTIO ioctls to * accelerate operation on tape drives. */ static int64_t file_skip(struct archive *a, void *client_data, int64_t request) { struct read_file_data *mine = (struct read_file_data *)client_data; /* Delegate skip requests. */ if (mine->use_lseek) return (file_skip_lseek(a, client_data, request)); /* If we can't skip, return 0; libarchive will read+discard instead. */ return (0); } /* * TODO: Store the offset and use it in the read callback. */ static int64_t file_seek(struct archive *a, void *client_data, int64_t request, int whence) { struct read_file_data *mine = (struct read_file_data *)client_data; - off_t seek = (off_t)request; + la_seek_t seek = (la_seek_t)request; int64_t r; int seek_bits = sizeof(seek) * 8 - 1; /* We use off_t here because lseek() is declared that way. */ - /* Reduce a request that would overflow the 'seek' variable. */ + /* Do not perform a seek which cannot be fulfilled. */ if (sizeof(request) > sizeof(seek)) { const int64_t max_seek = (((int64_t)1 << (seek_bits - 1)) - 1) * 2 + 1; const int64_t min_seek = ~max_seek; - if (request > max_seek) - seek = (off_t)max_seek; - else if (request < min_seek) - seek = (off_t)min_seek; + if (request < min_seek || request > max_seek) { + errno = EOVERFLOW; + goto err; + } } r = lseek(mine->fd, seek, whence); if (r >= 0) return r; /* If the input is corrupted or truncated, fail. */ +err: if (mine->filename_type == FNT_STDIN) archive_set_error(a, errno, "Error seeking in stdin"); else if (mine->filename_type == FNT_MBS) archive_set_error(a, errno, "Error seeking in '%s'", mine->filename.m); else archive_set_error(a, errno, "Error seeking in '%ls'", mine->filename.w); return (ARCHIVE_FATAL); } static int file_close2(struct archive *a, void *client_data) { struct read_file_data *mine = (struct read_file_data *)client_data; (void)a; /* UNUSED */ /* Only flush and close if open succeeded. */ if (mine->fd >= 0) { /* * Sometimes, we should flush the input before closing. * Regular files: faster to just close without flush. * Disk-like devices: Ditto. * Tapes: must not flush (user might need to * read the "next" item on a non-rewind device). * Pipes and sockets: must flush (otherwise, the * program feeding the pipe or socket may complain). * Here, I flush everything except for regular files and * device nodes. */ if (!S_ISREG(mine->st_mode) && !S_ISCHR(mine->st_mode) && !S_ISBLK(mine->st_mode)) { ssize_t bytesRead; do { bytesRead = read(mine->fd, mine->buffer, mine->block_size); } while (bytesRead > 0); } /* If a named file was opened, then it needs to be closed. */ if (mine->filename_type != FNT_STDIN) close(mine->fd); } free(mine->buffer); mine->buffer = NULL; mine->fd = -1; return (ARCHIVE_OK); } static int file_close(struct archive *a, void *client_data) { struct read_file_data *mine = (struct read_file_data *)client_data; file_close2(a, client_data); free(mine); return (ARCHIVE_OK); } static int file_switch(struct archive *a, void *client_data1, void *client_data2) { file_close2(a, client_data1); return file_open(a, client_data2); } diff --git a/contrib/libarchive/libarchive/archive_read_set_format.c b/contrib/libarchive/libarchive/archive_read_set_format.c index c74361b20c13..552ab12d2f92 100644 --- a/contrib/libarchive/libarchive/archive_read_set_format.c +++ b/contrib/libarchive/libarchive/archive_read_set_format.c @@ -1,116 +1,116 @@ /*- * Copyright (c) 2003-2012 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #include "archive.h" #include "archive_private.h" #include "archive_read_private.h" int archive_read_set_format(struct archive *_a, int code) { int r1, r2, slots, i; - char str[10]; + const char *str; struct archive_read *a = (struct archive_read *)_a; if ((r1 = archive_read_support_format_by_code(_a, code)) < (ARCHIVE_OK)) return r1; r1 = r2 = (ARCHIVE_OK); if (a->format) r2 = (ARCHIVE_WARN); switch (code & ARCHIVE_FORMAT_BASE_MASK) { case ARCHIVE_FORMAT_7ZIP: - strcpy(str, "7zip"); + str = "7zip"; break; case ARCHIVE_FORMAT_AR: - strcpy(str, "ar"); + str = "ar"; break; case ARCHIVE_FORMAT_CAB: - strcpy(str, "cab"); + str = "cab"; break; case ARCHIVE_FORMAT_CPIO: - strcpy(str, "cpio"); + str = "cpio"; break; case ARCHIVE_FORMAT_EMPTY: - strcpy(str, "empty"); + str = "empty"; break; case ARCHIVE_FORMAT_ISO9660: - strcpy(str, "iso9660"); + str = "iso9660"; break; case ARCHIVE_FORMAT_LHA: - strcpy(str, "lha"); + str = "lha"; break; case ARCHIVE_FORMAT_MTREE: - strcpy(str, "mtree"); + str = "mtree"; break; case ARCHIVE_FORMAT_RAR: - strcpy(str, "rar"); + str = "rar"; break; case ARCHIVE_FORMAT_RAR_V5: - strcpy(str, "rar5"); + str = "rar5"; break; case ARCHIVE_FORMAT_RAW: - strcpy(str, "raw"); + str = "raw"; break; case ARCHIVE_FORMAT_TAR: - strcpy(str, "tar"); + str = "tar"; break; case ARCHIVE_FORMAT_WARC: - strcpy(str, "warc"); + str = "warc"; break; case ARCHIVE_FORMAT_XAR: - strcpy(str, "xar"); + str = "xar"; break; case ARCHIVE_FORMAT_ZIP: - strcpy(str, "zip"); + str = "zip"; break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Invalid format code specified"); return (ARCHIVE_FATAL); } slots = sizeof(a->formats) / sizeof(a->formats[0]); a->format = &(a->formats[0]); for (i = 0; i < slots; i++, a->format++) { if (!a->format->name || !strcmp(a->format->name, str)) break; } if (!a->format->name || strcmp(a->format->name, str)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Internal error: Unable to set format"); r1 = (ARCHIVE_FATAL); } return (r1 < r2) ? r1 : r2; } diff --git a/contrib/libarchive/libarchive/archive_read_support_filter_program.c b/contrib/libarchive/libarchive/archive_read_support_filter_program.c index 9e825223b26c..2c8e45302d8e 100644 --- a/contrib/libarchive/libarchive/archive_read_support_filter_program.c +++ b/contrib/libarchive/libarchive/archive_read_support_filter_program.c @@ -1,495 +1,492 @@ /*- * Copyright (c) 2007 Joerg Sonnenberger * Copyright (c) 2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_WAIT_H # include #endif #ifdef HAVE_ERRNO_H # include #endif #ifdef HAVE_FCNTL_H # include #endif #ifdef HAVE_LIMITS_H # include #endif #ifdef HAVE_SIGNAL_H # include #endif #ifdef HAVE_STDLIB_H # include #endif #ifdef HAVE_STRING_H # include #endif #ifdef HAVE_UNISTD_H # include #endif #include "archive.h" #include "archive_private.h" #include "archive_string.h" #include "archive_read_private.h" #include "filter_fork.h" #if ARCHIVE_VERSION_NUMBER < 4000000 /* Deprecated; remove in libarchive 4.0 */ int archive_read_support_compression_program(struct archive *a, const char *cmd) { return archive_read_support_filter_program(a, cmd); } int archive_read_support_compression_program_signature(struct archive *a, const char *cmd, const void *signature, size_t signature_len) { return archive_read_support_filter_program_signature(a, cmd, signature, signature_len); } #endif int archive_read_support_filter_program(struct archive *a, const char *cmd) { return (archive_read_support_filter_program_signature(a, cmd, NULL, 0)); } /* * The bidder object stores the command and the signature to watch for. * The 'inhibit' entry here is used to ensure that unchecked filters never * bid twice in the same pipeline. */ struct program_bidder { char *description; char *cmd; void *signature; size_t signature_len; int inhibit; }; static int program_bidder_bid(struct archive_read_filter_bidder *, struct archive_read_filter *upstream); static int program_bidder_init(struct archive_read_filter *); static void program_bidder_free(struct archive_read_filter_bidder *); /* * The actual filter needs to track input and output data. */ struct program_filter { struct archive_string description; #if defined(_WIN32) && !defined(__CYGWIN__) HANDLE child; #else pid_t child; #endif int exit_status; - int waitpid_return; + pid_t waitpid_return; int child_stdin, child_stdout; char *out_buf; size_t out_buf_len; }; static ssize_t program_filter_read(struct archive_read_filter *, const void **); static int program_filter_close(struct archive_read_filter *); static void free_state(struct program_bidder *); static const struct archive_read_filter_bidder_vtable program_bidder_vtable = { .bid = program_bidder_bid, .init = program_bidder_init, .free = program_bidder_free, }; int archive_read_support_filter_program_signature(struct archive *_a, const char *cmd, const void *signature, size_t signature_len) { struct archive_read *a = (struct archive_read *)_a; struct program_bidder *state; /* * Allocate our private state. */ state = calloc(1, sizeof (*state)); if (state == NULL) goto memerr; state->cmd = strdup(cmd); if (state->cmd == NULL) goto memerr; if (signature != NULL && signature_len > 0) { state->signature_len = signature_len; state->signature = malloc(signature_len); memcpy(state->signature, signature, signature_len); } if (__archive_read_register_bidder(a, state, NULL, &program_bidder_vtable) != ARCHIVE_OK) { free_state(state); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); memerr: free_state(state); archive_set_error(_a, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } static void program_bidder_free(struct archive_read_filter_bidder *self) { struct program_bidder *state = (struct program_bidder *)self->data; free_state(state); } static void free_state(struct program_bidder *state) { if (state) { free(state->cmd); free(state->signature); free(state); } } /* * If we do have a signature, bid only if that matches. * * If there's no signature, we bid INT_MAX the first time * we're called, then never bid again. */ static int program_bidder_bid(struct archive_read_filter_bidder *self, struct archive_read_filter *upstream) { struct program_bidder *state = self->data; const char *p; /* If we have a signature, use that to match. */ if (state->signature_len > 0) { p = __archive_read_filter_ahead(upstream, state->signature_len, NULL); if (p == NULL) return (0); /* No match, so don't bid. */ if (memcmp(p, state->signature, state->signature_len) != 0) return (0); return ((int)state->signature_len * 8); } /* Otherwise, bid once and then never bid again. */ if (state->inhibit) return (0); state->inhibit = 1; return (INT_MAX); } /* * Shut down the child, return ARCHIVE_OK if it exited normally. * * Note that the return value is sticky; if we're called again, * we won't reap the child again, but we will return the same status * (including error message if the child came to a bad end). */ static int child_stop(struct archive_read_filter *self, struct program_filter *state) { /* Close our side of the I/O with the child. */ if (state->child_stdin != -1) { close(state->child_stdin); state->child_stdin = -1; } if (state->child_stdout != -1) { close(state->child_stdout); state->child_stdout = -1; } if (state->child != 0) { /* Reap the child. */ do { state->waitpid_return = waitpid(state->child, &state->exit_status, 0); } while (state->waitpid_return == -1 && errno == EINTR); -#if defined(_WIN32) && !defined(__CYGWIN__) - CloseHandle(state->child); -#endif state->child = 0; } if (state->waitpid_return < 0) { /* waitpid() failed? This is ugly. */ archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, - "Child process exited badly"); + "Error closing child process"); return (ARCHIVE_WARN); } #if !defined(_WIN32) || defined(__CYGWIN__) if (WIFSIGNALED(state->exit_status)) { #ifdef SIGPIPE /* If the child died because we stopped reading before * it was done, that's okay. Some archive formats * have padding at the end that we routinely ignore. */ /* The alternative to this would be to add a step * before close(child_stdout) above to read from the * child until the child has no more to write. */ if (WTERMSIG(state->exit_status) == SIGPIPE) return (ARCHIVE_OK); #endif archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Child process exited with signal %d", WTERMSIG(state->exit_status)); return (ARCHIVE_WARN); } #endif /* !_WIN32 || __CYGWIN__ */ if (WIFEXITED(state->exit_status)) { if (WEXITSTATUS(state->exit_status) == 0) return (ARCHIVE_OK); archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Child process exited with status %d", WEXITSTATUS(state->exit_status)); return (ARCHIVE_WARN); } return (ARCHIVE_WARN); } /* * Use select() to decide whether the child is ready for read or write. */ static ssize_t child_read(struct archive_read_filter *self, char *buf, size_t buf_len) { struct program_filter *state = self->data; ssize_t ret, requested, avail; const char *p; #if defined(_WIN32) && !defined(__CYGWIN__) HANDLE handle = (HANDLE)_get_osfhandle(state->child_stdout); #endif requested = buf_len > SSIZE_MAX ? SSIZE_MAX : buf_len; for (;;) { do { #if defined(_WIN32) && !defined(__CYGWIN__) /* Avoid infinity wait. * Note: If there is no data in the pipe, ReadFile() * called in read() never returns and so we won't * write remaining encoded data to the pipe. * Note: This way may cause performance problem. * we are looking forward to great code to resolve * this. */ DWORD pipe_avail = -1; int cnt = 2; while (PeekNamedPipe(handle, NULL, 0, NULL, &pipe_avail, NULL) != 0 && pipe_avail == 0 && cnt--) Sleep(5); if (pipe_avail == 0) { ret = -1; errno = EAGAIN; break; } #endif ret = read(state->child_stdout, buf, requested); } while (ret == -1 && errno == EINTR); if (ret > 0) return (ret); if (ret == 0 || (ret == -1 && errno == EPIPE)) /* Child has closed its output; reap the child * and return the status. */ return (child_stop(self, state)); if (ret == -1 && errno != EAGAIN) return (-1); if (state->child_stdin == -1) { /* Block until child has some I/O ready. */ __archive_check_child(state->child_stdin, state->child_stdout); continue; } /* Get some more data from upstream. */ p = __archive_read_filter_ahead(self->upstream, 1, &avail); if (p == NULL) { close(state->child_stdin); state->child_stdin = -1; fcntl(state->child_stdout, F_SETFL, 0); if (avail < 0) return (avail); continue; } do { ret = write(state->child_stdin, p, avail); } while (ret == -1 && errno == EINTR); if (ret > 0) { /* Consume whatever we managed to write. */ __archive_read_filter_consume(self->upstream, ret); } else if (ret == -1 && errno == EAGAIN) { /* Block until child has some I/O ready. */ __archive_check_child(state->child_stdin, state->child_stdout); } else { /* Write failed. */ close(state->child_stdin); state->child_stdin = -1; fcntl(state->child_stdout, F_SETFL, 0); /* If it was a bad error, we're done; otherwise * it was EPIPE or EOF, and we can still read * from the child. */ if (ret == -1 && errno != EPIPE) return (-1); } } } static const struct archive_read_filter_vtable program_reader_vtable = { .read = program_filter_read, .close = program_filter_close, }; int __archive_read_program(struct archive_read_filter *self, const char *cmd) { struct program_filter *state; static const size_t out_buf_len = 65536; char *out_buf; const char *prefix = "Program: "; int ret; size_t l; l = strlen(prefix) + strlen(cmd) + 1; state = calloc(1, sizeof(*state)); out_buf = malloc(out_buf_len); if (state == NULL || out_buf == NULL || archive_string_ensure(&state->description, l) == NULL) { archive_set_error(&self->archive->archive, ENOMEM, "Can't allocate input data"); if (state != NULL) { archive_string_free(&state->description); free(state); } free(out_buf); return (ARCHIVE_FATAL); } archive_strcpy(&state->description, prefix); archive_strcat(&state->description, cmd); self->code = ARCHIVE_FILTER_PROGRAM; self->name = state->description.s; state->out_buf = out_buf; state->out_buf_len = out_buf_len; ret = __archive_create_child(cmd, &state->child_stdin, &state->child_stdout, &state->child); if (ret != ARCHIVE_OK) { free(state->out_buf); archive_string_free(&state->description); free(state); archive_set_error(&self->archive->archive, EINVAL, "Can't initialize filter; unable to run program \"%s\"", cmd); return (ARCHIVE_FATAL); } self->data = state; self->vtable = &program_reader_vtable; /* XXX Check that we can read at least one byte? */ return (ARCHIVE_OK); } static int program_bidder_init(struct archive_read_filter *self) { struct program_bidder *bidder_state; bidder_state = (struct program_bidder *)self->bidder->data; return (__archive_read_program(self, bidder_state->cmd)); } static ssize_t program_filter_read(struct archive_read_filter *self, const void **buff) { struct program_filter *state; ssize_t bytes; size_t total; char *p; state = (struct program_filter *)self->data; total = 0; p = state->out_buf; while (state->child_stdout != -1 && total < state->out_buf_len) { bytes = child_read(self, p, state->out_buf_len - total); if (bytes < 0) /* No recovery is possible if we can no longer * read from the child. */ return (ARCHIVE_FATAL); if (bytes == 0) /* We got EOF from the child. */ break; total += bytes; p += bytes; } *buff = state->out_buf; return (total); } static int program_filter_close(struct archive_read_filter *self) { struct program_filter *state; int e; state = (struct program_filter *)self->data; e = child_stop(self, state); /* Release our private data. */ free(state->out_buf); archive_string_free(&state->description); free(state); return (e); } diff --git a/contrib/libarchive/libarchive/archive_read_support_format_7zip.c b/contrib/libarchive/libarchive/archive_read_support_format_7zip.c index f273f84be521..595462733104 100644 --- a/contrib/libarchive/libarchive/archive_read_support_format_7zip.c +++ b/contrib/libarchive/libarchive/archive_read_support_format_7zip.c @@ -1,4523 +1,4526 @@ /*- * Copyright (c) 2011 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #if HAVE_STDINT_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_BZLIB_H #include #endif #ifdef HAVE_LZMA_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #ifdef HAVE_ZSTD_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_ppmd7_private.h" #include "archive_private.h" #include "archive_read_private.h" #include "archive_time_private.h" #include "archive_endian.h" #ifndef HAVE_ZLIB_H #include "archive_crc32.h" #endif #define _7ZIP_SIGNATURE "7z\xBC\xAF\x27\x1C" #define SFX_MIN_ADDR 0x27000 #define SFX_MAX_ADDR 0x60000 #define SFX_MAX_OFFSET (SFX_MAX_ADDR - SFX_MIN_ADDR) /* * PE format */ #define PE_DOS_HDR_LEN 0x40 #define PE_DOS_HDR_ELFANEW_OFFSET 0x3c #define PE_COFF_HDR_LEN 0x18 #define PE_COFF_HDR_SEC_CNT_OFFSET 0x6 #define PE_COFF_HDR_OPT_SZ_OFFSET 0x14 #define PE_SEC_HDR_LEN 0x28 #define PE_SEC_HDR_RAW_ADDR_OFFSET 0x14 #define PE_SEC_HDR_RAW_SZ_OFFSET 0x10 /* * ELF format */ -#define ELF_HDR_MIN_LEN 0x34 +#define ELF_HDR_MIN_LEN 0x3f #define ELF_HDR_EI_CLASS_OFFSET 0x04 #define ELF_HDR_EI_DATA_OFFSET 0x05 /* * Codec ID */ #define _7Z_COPY 0 #define _7Z_LZMA 0x030101 #define _7Z_LZMA2 0x21 #define _7Z_DEFLATE 0x040108 #define _7Z_BZ2 0x040202 #define _7Z_PPMD 0x030401 #define _7Z_DELTA 0x03 #define _7Z_CRYPTO_MAIN_ZIP 0x06F10101 /* Main Zip crypto algo */ #define _7Z_CRYPTO_RAR_29 0x06F10303 /* Rar29 AES-128 + (modified SHA-1) */ #define _7Z_CRYPTO_AES_256_SHA_256 0x06F10701 /* AES-256 + SHA-256 */ #define _7Z_X86 0x03030103 #define _7Z_X86_BCJ2 0x0303011B #define _7Z_POWERPC 0x03030205 #define _7Z_IA64 0x03030401 #define _7Z_ARM 0x03030501 #define _7Z_ARMTHUMB 0x03030701 #define _7Z_ARM64 0xa #define _7Z_RISCV 0xb #define _7Z_SPARC 0x03030805 #define _7Z_ZSTD 0x4F71101 /* Copied from https://github.com/mcmilk/7-Zip-zstd.git */ /* * 7-Zip header property IDs. */ #define kEnd 0x00 #define kHeader 0x01 #define kArchiveProperties 0x02 #define kAdditionalStreamsInfo 0x03 #define kMainStreamsInfo 0x04 #define kFilesInfo 0x05 #define kPackInfo 0x06 #define kUnPackInfo 0x07 #define kSubStreamsInfo 0x08 #define kSize 0x09 #define kCRC 0x0A #define kFolder 0x0B #define kCodersUnPackSize 0x0C #define kNumUnPackStream 0x0D #define kEmptyStream 0x0E #define kEmptyFile 0x0F #define kAnti 0x10 #define kName 0x11 #define kCTime 0x12 #define kATime 0x13 #define kMTime 0x14 #define kAttributes 0x15 #define kEncodedHeader 0x17 #define kDummy 0x19 // Check that some windows file attribute constants are defined. // Reference: https://learn.microsoft.com/en-us/windows/win32/fileio/file-attribute-constants #ifndef FILE_ATTRIBUTE_READONLY #define FILE_ATTRIBUTE_READONLY 0x00000001 #endif #ifndef FILE_ATTRIBUTE_HIDDEN #define FILE_ATTRIBUTE_HIDDEN 0x00000002 #endif #ifndef FILE_ATTRIBUTE_SYSTEM #define FILE_ATTRIBUTE_SYSTEM 0x00000004 #endif #ifndef FILE_ATTRIBUTE_DIRECTORY #define FILE_ATTRIBUTE_DIRECTORY 0x00000010 #endif // This value is defined in 7zip with the comment "trick for Unix". // // 7z archives created on unix have this bit set in the high 16 bits of // the attr field along with the unix permissions. #define FILE_ATTRIBUTE_UNIX_EXTENSION 0x8000 struct _7z_digests { unsigned char *defineds; uint32_t *digests; }; struct _7z_folder { uint64_t numCoders; struct _7z_coder { unsigned long codec; uint64_t numInStreams; uint64_t numOutStreams; uint64_t propertiesSize; unsigned char *properties; } *coders; uint64_t numBindPairs; struct { uint64_t inIndex; uint64_t outIndex; } *bindPairs; uint64_t numPackedStreams; uint64_t *packedStreams; uint64_t numInStreams; uint64_t numOutStreams; uint64_t *unPackSize; unsigned char digest_defined; uint32_t digest; uint64_t numUnpackStreams; uint32_t packIndex; /* Unoperated bytes. */ uint64_t skipped_bytes; }; struct _7z_coders_info { uint64_t numFolders; struct _7z_folder *folders; uint64_t dataStreamIndex; }; struct _7z_pack_info { uint64_t pos; uint64_t numPackStreams; uint64_t *sizes; struct _7z_digests digest; /* Calculated from pos and numPackStreams. */ uint64_t *positions; }; struct _7z_substream_info { size_t unpack_streams; uint64_t *unpackSizes; unsigned char *digestsDefined; uint32_t *digests; }; struct _7z_stream_info { struct _7z_pack_info pi; struct _7z_coders_info ci; struct _7z_substream_info ss; }; struct _7z_header_info { uint64_t dataIndex; unsigned char *emptyStreamBools; unsigned char *emptyFileBools; unsigned char *antiBools; unsigned char *attrBools; }; struct _7zip_entry { size_t name_len; unsigned char *utf16name; #if defined(_WIN32) && !defined(__CYGWIN__) && defined(_DEBUG) const wchar_t *wname; #endif uint32_t folderIndex; uint32_t ssIndex; unsigned flg; #define MTIME_IS_SET (1<<0) #define ATIME_IS_SET (1<<1) #define CTIME_IS_SET (1<<2) #define CRC32_IS_SET (1<<3) #define HAS_STREAM (1<<4) int64_t mtime; int64_t atime; int64_t ctime; uint32_t mtime_ns; uint32_t atime_ns; uint32_t ctime_ns; __LA_MODE_T mode; uint32_t attr; }; struct _7zip { /* Structural information about the archive. */ struct _7z_stream_info si; int header_is_being_read; int header_is_encoded; uint64_t header_bytes_remaining; unsigned long header_crc32; /* Header offset to check that reading points of the file contents * will not exceed the header. */ uint64_t header_offset; /* Base offset of the archive file for a seek in case reading SFX. */ uint64_t seek_base; /* List of entries */ size_t entries_remaining; uint64_t numFiles; struct _7zip_entry *entries; struct _7zip_entry *entry; unsigned char *entry_names; /* entry_bytes_remaining is the number of bytes we expect. */ int64_t entry_offset; uint64_t entry_bytes_remaining; /* Running CRC32 of the decompressed data */ unsigned long entry_crc32; /* Flags to mark progress of decompression. */ char end_of_entry; /* Uncompressed buffer control. */ #define UBUFF_SIZE (64 * 1024) unsigned char *uncompressed_buffer; unsigned char *uncompressed_buffer_pointer; size_t uncompressed_buffer_size; size_t uncompressed_buffer_bytes_remaining; /* Offset of the compressed data. */ int64_t stream_offset; /* * Decompressing control data. */ unsigned folder_index; uint64_t folder_outbytes_remaining; unsigned pack_stream_index; unsigned pack_stream_remaining; uint64_t pack_stream_inbytes_remaining; size_t pack_stream_bytes_unconsumed; /* The codec information of a folder. */ unsigned long codec; unsigned long codec2; /* * Decompressor controllers. */ /* Decoding LZMA1 and LZMA2 data. */ #ifdef HAVE_LZMA_H lzma_stream lzstream; int lzstream_valid; #endif /* Decoding bzip2 data. */ #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) bz_stream bzstream; int bzstream_valid; #endif /* Decoding deflate data. */ #ifdef HAVE_ZLIB_H z_stream stream; int stream_valid; #endif /* Decoding Zstandard data. */ #if HAVE_ZSTD_H ZSTD_DStream *zstd_dstream; int zstdstream_valid; #endif /* Decoding PPMd data. */ int ppmd7_stat; CPpmd7 ppmd7_context; CPpmd7z_RangeDec range_dec; IByteIn bytein; struct { const unsigned char *next_in; int64_t avail_in; int64_t total_in; int64_t stream_in; unsigned char *next_out; int64_t avail_out; int64_t total_out; int overconsumed; } ppstream; int ppmd7_valid; /* Decoding BCJ and BCJ2 data. */ uint32_t bcj_state; size_t odd_bcj_size; unsigned char odd_bcj[4]; /* Decoding BCJ data. */ size_t bcj_prevPosT; uint32_t bcj_prevMask; uint32_t bcj_ip; /* Decoding BCJ2 data. */ size_t main_stream_bytes_remaining; unsigned char *sub_stream_buff[3]; size_t sub_stream_size[3]; size_t sub_stream_bytes_remaining[3]; unsigned char *tmp_stream_buff; size_t tmp_stream_buff_size; size_t tmp_stream_bytes_avail; size_t tmp_stream_bytes_remaining; #ifdef _LZMA_PROB32 #define CProb uint32_t #else #define CProb uint16_t #endif CProb bcj2_p[256 + 2]; uint8_t bcj2_prevByte; uint32_t bcj2_range; uint32_t bcj2_code; uint64_t bcj2_outPos; /* Filename character-set conversion data. */ struct archive_string_conv *sconv; char format_name[64]; /* Custom value that is non-zero if this archive contains encrypted entries. */ int has_encrypted_entries; }; /* Maximum entry size. This limitation prevents reading intentional * corrupted 7-zip files on assuming there are not so many entries in * the files. */ #define UMAX_ENTRY ARCHIVE_LITERAL_ULL(100000000) static int archive_read_format_7zip_has_encrypted_entries(struct archive_read *); static int archive_read_support_format_7zip_capabilities(struct archive_read *a); static int archive_read_format_7zip_bid(struct archive_read *, int); static int archive_read_format_7zip_cleanup(struct archive_read *); static int archive_read_format_7zip_read_data(struct archive_read *, const void **, size_t *, int64_t *); static int archive_read_format_7zip_read_data_skip(struct archive_read *); static int archive_read_format_7zip_read_header(struct archive_read *, struct archive_entry *); static int check_7zip_header_in_sfx(const char *); static unsigned long decode_codec_id(const unsigned char *, size_t); static int decode_encoded_header_info(struct archive_read *, struct _7z_stream_info *); static int decompress(struct archive_read *, struct _7zip *, void *, size_t *, const void *, size_t *); static ssize_t extract_pack_stream(struct archive_read *, size_t); static uint64_t folder_uncompressed_size(struct _7z_folder *); static void free_CodersInfo(struct _7z_coders_info *); static void free_Digest(struct _7z_digests *); static void free_Folder(struct _7z_folder *); static void free_Header(struct _7z_header_info *); static void free_PackInfo(struct _7z_pack_info *); static void free_StreamsInfo(struct _7z_stream_info *); static void free_SubStreamsInfo(struct _7z_substream_info *); static int free_decompression(struct archive_read *, struct _7zip *); static ssize_t get_uncompressed_data(struct archive_read *, const void **, size_t, size_t); static const unsigned char * header_bytes(struct archive_read *, size_t); static int init_decompression(struct archive_read *, struct _7zip *, const struct _7z_coder *, const struct _7z_coder *); static int parse_7zip_uint64(struct archive_read *, uint64_t *); static int read_Bools(struct archive_read *, unsigned char *, size_t); static int read_CodersInfo(struct archive_read *, struct _7z_coders_info *); static int read_Digests(struct archive_read *, struct _7z_digests *, size_t); static int read_Folder(struct archive_read *, struct _7z_folder *); static int read_Header(struct archive_read *, struct _7z_header_info *, int); static int read_PackInfo(struct archive_read *, struct _7z_pack_info *); static int read_StreamsInfo(struct archive_read *, struct _7z_stream_info *); static int read_SubStreamsInfo(struct archive_read *, struct _7z_substream_info *, struct _7z_folder *, size_t); static int read_Times(struct archive_read *, struct _7z_header_info *, int); static void read_consume(struct archive_read *); static ssize_t read_stream(struct archive_read *, const void **, size_t, size_t); static int seek_pack(struct archive_read *); static int64_t skip_stream(struct archive_read *, size_t); static int skip_sfx(struct archive_read *, const ssize_t); static ssize_t find_pe_overlay(struct archive_read *); static ssize_t find_elf_data_sec(struct archive_read *); static int slurp_central_directory(struct archive_read *, struct _7zip *, struct _7z_header_info *); static int setup_decode_folder(struct archive_read *, struct _7z_folder *, int); static void x86_Init(struct _7zip *); static size_t x86_Convert(struct _7zip *, uint8_t *, size_t); static void arm_Init(struct _7zip *); static size_t arm_Convert(struct _7zip *, uint8_t *, size_t); static size_t arm64_Convert(struct _7zip *, uint8_t *, size_t); static ssize_t Bcj2_Decode(struct _7zip *, uint8_t *, size_t); static size_t sparc_Convert(struct _7zip *, uint8_t *, size_t); static size_t powerpc_Convert(struct _7zip *, uint8_t *, size_t); int archive_read_support_format_7zip(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct _7zip *zip; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_7zip"); zip = calloc(1, sizeof(*zip)); if (zip == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate 7zip data"); return (ARCHIVE_FATAL); } /* * Until enough data has been read, we cannot tell about * any encrypted entries yet. */ zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; r = __archive_read_register_format(a, zip, "7zip", archive_read_format_7zip_bid, NULL, archive_read_format_7zip_read_header, archive_read_format_7zip_read_data, archive_read_format_7zip_read_data_skip, NULL, archive_read_format_7zip_cleanup, archive_read_support_format_7zip_capabilities, archive_read_format_7zip_has_encrypted_entries); if (r != ARCHIVE_OK) free(zip); return (ARCHIVE_OK); } static int archive_read_support_format_7zip_capabilities(struct archive_read * a) { (void)a; /* UNUSED */ return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); } static int archive_read_format_7zip_has_encrypted_entries(struct archive_read *_a) { if (_a && _a->format) { struct _7zip * zip = (struct _7zip *)_a->format->data; if (zip) { return zip->has_encrypted_entries; } } return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; } static int archive_read_format_7zip_bid(struct archive_read *a, int best_bid) { const char *p; /* If someone has already bid more than 32, then avoid trashing the look-ahead buffers with a seek. */ if (best_bid > 32) return (-1); if ((p = __archive_read_ahead(a, 6, NULL)) == NULL) return (0); /* If first six bytes are the 7-Zip signature, * return the bid right now. */ if (memcmp(p, _7ZIP_SIGNATURE, 6) == 0) return (48); /* * It may a 7-Zip SFX archive file. If first two bytes are * 'M' and 'Z' available on Windows or first four bytes are * "\x7F\x45LF" available on posix like system, seek the 7-Zip * signature. While find_pe_overlay can be performed without * performing a seek, find_elf_data_sec requires one, * thus a performance difference between the two is expected. */ if ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0) { const ssize_t min_addr = p[0] == 'M' ? find_pe_overlay(a) : find_elf_data_sec(a); ssize_t offset = min_addr; ssize_t window = 4096; ssize_t bytes_avail; while (offset + window <= (min_addr + SFX_MAX_OFFSET)) { const char *buff = __archive_read_ahead(a, offset + window, &bytes_avail); if (buff == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < 0x40) return (0); continue; } p = buff + offset; while (p + 32 < buff + bytes_avail) { int step = check_7zip_header_in_sfx(p); if (step == 0) return (48); p += step; } offset = p - buff; } } return (0); } static int check_7zip_header_in_sfx(const char *p) { switch ((unsigned char)p[5]) { case 0x1C: if (memcmp(p, _7ZIP_SIGNATURE, 6) != 0) return (6); /* * Test the CRC because its extraction code has 7-Zip * Magic Code, so we should do this in order not to * make a mis-detection. */ if (crc32(0, (const unsigned char *)p + 12, 20) != archive_le32dec(p + 8)) return (6); /* Hit the header! */ return (0); case 0x37: return (5); case 0x7A: return (4); case 0xBC: return (3); case 0xAF: return (2); case 0x27: return (1); default: return (6); } } static int skip_sfx(struct archive_read *a, const ssize_t min_addr) { const void *h; const char *p, *q; size_t skip, offset; ssize_t bytes, window; if (__archive_read_seek(a, min_addr, SEEK_SET) < 0) return (ARCHIVE_FATAL); offset = 0; window = 1; while (offset + window <= SFX_MAX_ADDR - SFX_MIN_ADDR) { h = __archive_read_ahead(a, window, &bytes); if (h == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < 0x40) goto fatal; continue; } if (bytes < 6) { /* This case might happen when window == 1. */ window = 4096; continue; } p = (const char *)h; q = p + bytes; /* * Scan ahead until we find something that looks * like the 7-Zip header. */ while (p + 32 < q) { int step = check_7zip_header_in_sfx(p); if (step == 0) { struct _7zip *zip = (struct _7zip *)a->format->data; skip = p - (const char *)h; __archive_read_consume(a, skip); zip->seek_base = min_addr + offset + skip; return (ARCHIVE_OK); } p += step; } skip = p - (const char *)h; __archive_read_consume(a, skip); offset += skip; if (window == 1) window = 4096; } fatal: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Couldn't find out 7-Zip header"); return (ARCHIVE_FATAL); } static ssize_t find_pe_overlay(struct archive_read *a) { const char *h; ssize_t bytes, max_offset, offset, sec_end; ssize_t opt_hdr_sz, sec_cnt; for (;;) { /* * Read Dos header to find e_lfanew */ h = __archive_read_ahead(a, PE_DOS_HDR_LEN, &bytes); if (h == NULL || h[0] != 'M' || h[1] != 'Z') { break; } offset = archive_le32dec(h + PE_DOS_HDR_ELFANEW_OFFSET); /* * Read COFF header to find opt header size and sec cnt */ if (bytes < offset + PE_COFF_HDR_LEN) { h = __archive_read_ahead(a, offset + PE_COFF_HDR_LEN, &bytes); if (h == NULL || h[offset] != 'P' || h[offset + 1] != 'E') { break; } } sec_cnt = archive_le16dec( h + offset + PE_COFF_HDR_SEC_CNT_OFFSET); opt_hdr_sz = archive_le16dec( h + offset + PE_COFF_HDR_OPT_SZ_OFFSET); /* * Skip optional header */ if (opt_hdr_sz != 0) { offset += PE_COFF_HDR_LEN + opt_hdr_sz; } else { break; } /* * Traverse sec table to find max raw offset (i.e., overlay) */ if (bytes < offset + sec_cnt * PE_SEC_HDR_LEN) { h = __archive_read_ahead(a, offset + sec_cnt * PE_SEC_HDR_LEN, NULL); if (h == NULL) { break; } } max_offset = offset; while (sec_cnt > 0) { sec_end = archive_le32dec( h + offset + PE_SEC_HDR_RAW_SZ_OFFSET) + archive_le32dec( h + offset + PE_SEC_HDR_RAW_ADDR_OFFSET); if (sec_end > max_offset) { max_offset = sec_end; } offset += PE_SEC_HDR_LEN; sec_cnt--; } return (max_offset); } /* * If encounter any weirdness, revert to old brute-force style search */ return (SFX_MIN_ADDR); } static ssize_t find_elf_data_sec(struct archive_read *a) { const char *h; char big_endian, format_64; ssize_t bytes, min_addr = SFX_MIN_ADDR; uint64_t e_shoff, strtab_offset, strtab_size; uint16_t e_shentsize, e_shnum, e_shstrndx; uint16_t (*dec16)(const void *); uint32_t (*dec32)(const void *); uint64_t (*dec64)(const void *); for (;;) { /* * Read Elf header to find bitness & endianness */ h = __archive_read_ahead(a, ELF_HDR_MIN_LEN, &bytes); if (h == NULL || memcmp(h, "\x7F\x45LF", 4) != 0) { break; } format_64 = h[ELF_HDR_EI_CLASS_OFFSET] == 0x2; big_endian = h[ELF_HDR_EI_DATA_OFFSET] == 0x2; if (big_endian) { dec16 = &archive_be16dec; dec32 = &archive_be32dec; dec64 = &archive_be64dec; } else { dec16 = &archive_le16dec; dec32 = &archive_le32dec; dec64 = &archive_le64dec; } /* * Read section header table info */ if (format_64) { e_shoff = (*dec64)(h + 0x28); e_shentsize = (*dec16)(h + 0x3A); e_shnum = (*dec16)(h + 0x3C); e_shstrndx = (*dec16)(h + 0x3E); if (e_shnum < e_shstrndx || e_shentsize < 0x28) break; } else { e_shoff = (*dec32)(h + 0x20); e_shentsize = (*dec16)(h + 0x2E); e_shnum = (*dec16)(h + 0x30); e_shstrndx = (*dec16)(h + 0x32); if (e_shnum < e_shstrndx || e_shentsize < 0x18) break; } /* * Reading the section table to find strtab section */ if (__archive_read_seek(a, e_shoff, SEEK_SET) < 0) { break; } h = __archive_read_ahead(a, (size_t)e_shnum * (size_t)e_shentsize, NULL); if (h == NULL) { break; } if (format_64) { strtab_offset = (*dec64)( h + e_shstrndx * e_shentsize + 0x18); strtab_size = (*dec64)( h + e_shstrndx * e_shentsize + 0x20); } else { strtab_offset = (*dec32)( h + e_shstrndx * e_shentsize + 0x10); strtab_size = (*dec32)( h + e_shstrndx * e_shentsize + 0x14); } + if (strtab_size < 6 || strtab_size > SIZE_MAX) + break; /* * Read the STRTAB section to find the .data offset */ if (__archive_read_seek(a, strtab_offset, SEEK_SET) < 0) { break; } h = __archive_read_ahead(a, strtab_size, NULL); if (h == NULL) { break; } ssize_t data_sym_offset = -1; for (size_t offset = 0; offset < strtab_size - 6; offset++) { if (memcmp(h + offset, ".data\00", 6) == 0) { data_sym_offset = offset; break; } } if (data_sym_offset == -1) { break; } /* * Find the section with the .data name */ if (__archive_read_seek(a, e_shoff, SEEK_SET) < 0) { break; } h = __archive_read_ahead(a, (size_t)e_shnum * (size_t)e_shentsize, NULL); if (h == NULL) { break; } ssize_t sec_tbl_offset = 0, name_offset; while (e_shnum > 0) { name_offset = (*dec32)(h + sec_tbl_offset); if (name_offset == data_sym_offset) { if (format_64) { min_addr = (*dec64)( h + sec_tbl_offset + 0x18); } else { min_addr = (*dec32)( h + sec_tbl_offset + 0x10); } break; } sec_tbl_offset += e_shentsize; e_shnum--; } break; } __archive_read_seek(a, 0, SEEK_SET); return (min_addr); } static int archive_read_format_7zip_read_header(struct archive_read *a, struct archive_entry *entry) { struct _7zip *zip = (struct _7zip *)a->format->data; struct _7zip_entry *zip_entry; int r, ret = ARCHIVE_OK; struct _7z_folder *folder = 0; uint64_t fidx = 0; /* * It should be sufficient to call archive_read_next_header() for * a reader to determine if an entry is encrypted or not. If the * encryption of an entry is only detectable when calling * archive_read_data(), so be it. We'll do the same check there * as well. */ if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { zip->has_encrypted_entries = 0; } a->archive.archive_format = ARCHIVE_FORMAT_7ZIP; if (a->archive.archive_format_name == NULL) a->archive.archive_format_name = "7-Zip"; if (zip->entries == NULL) { struct _7z_header_info header; memset(&header, 0, sizeof(header)); r = slurp_central_directory(a, zip, &header); free_Header(&header); if (r != ARCHIVE_OK) return (r); zip->entries_remaining = (size_t)zip->numFiles; zip->entry = zip->entries; } else { ++zip->entry; } zip_entry = zip->entry; if (zip->entries_remaining <= 0 || zip_entry == NULL) return ARCHIVE_EOF; --zip->entries_remaining; zip->entry_offset = 0; zip->end_of_entry = 0; zip->entry_crc32 = crc32(0, NULL, 0); /* Setup a string conversion for a filename. */ if (zip->sconv == NULL) { zip->sconv = archive_string_conversion_from_charset( &a->archive, "UTF-16LE", 1); if (zip->sconv == NULL) return (ARCHIVE_FATAL); } /* Figure out if the entry is encrypted by looking at the folder that is associated to the current 7zip entry. If the folder has a coder with a _7Z_CRYPTO codec then the folder is encrypted. Hence the entry must also be encrypted. */ if (zip_entry && zip_entry->folderIndex < zip->si.ci.numFolders) { folder = &(zip->si.ci.folders[zip_entry->folderIndex]); for (fidx=0; folder && fidxnumCoders; fidx++) { switch(folder->coders[fidx].codec) { case _7Z_CRYPTO_MAIN_ZIP: case _7Z_CRYPTO_RAR_29: case _7Z_CRYPTO_AES_256_SHA_256: { archive_entry_set_is_data_encrypted(entry, 1); zip->has_encrypted_entries = 1; break; } } } } /* Now that we've checked for encryption, if there were still no * encrypted entries found we can say for sure that there are none. */ if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { zip->has_encrypted_entries = 0; } if (archive_entry_copy_pathname_l(entry, (const char *)zip_entry->utf16name, zip_entry->name_len, zip->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Pathname cannot be converted " "from %s to current locale.", archive_string_conversion_charset_name(zip->sconv)); ret = ARCHIVE_WARN; } /* Populate some additional entry fields: */ archive_entry_set_mode(entry, zip_entry->mode); if (zip_entry->flg & MTIME_IS_SET) archive_entry_set_mtime(entry, zip_entry->mtime, zip_entry->mtime_ns); if (zip_entry->flg & CTIME_IS_SET) archive_entry_set_ctime(entry, zip_entry->ctime, zip_entry->ctime_ns); if (zip_entry->flg & ATIME_IS_SET) archive_entry_set_atime(entry, zip_entry->atime, zip_entry->atime_ns); if (zip_entry->ssIndex != (uint32_t)-1) { zip->entry_bytes_remaining = zip->si.ss.unpackSizes[zip_entry->ssIndex]; archive_entry_set_size(entry, zip->entry_bytes_remaining); } else { zip->entry_bytes_remaining = 0; archive_entry_set_size(entry, 0); } // These attributes are supported by the windows implementation of archive_write_disk. const int supported_attrs = FILE_ATTRIBUTE_READONLY | FILE_ATTRIBUTE_HIDDEN | FILE_ATTRIBUTE_SYSTEM; if (zip_entry->attr & supported_attrs) { char *fflags_text, *ptr; /* allocate for ",rdonly,hidden,system" */ fflags_text = malloc(22 * sizeof(*fflags_text)); if (fflags_text != NULL) { ptr = fflags_text; if (zip_entry->attr & FILE_ATTRIBUTE_READONLY) { strcpy(ptr, ",rdonly"); ptr = ptr + 7; } if (zip_entry->attr & FILE_ATTRIBUTE_HIDDEN) { strcpy(ptr, ",hidden"); ptr = ptr + 7; } if (zip_entry->attr & FILE_ATTRIBUTE_SYSTEM) { strcpy(ptr, ",system"); ptr = ptr + 7; } if (ptr > fflags_text) { archive_entry_copy_fflags_text(entry, fflags_text + 1); } free(fflags_text); } } /* If there's no body, force read_data() to return EOF immediately. */ if (zip->entry_bytes_remaining < 1) zip->end_of_entry = 1; if ((zip_entry->mode & AE_IFMT) == AE_IFLNK) { unsigned char *symname = NULL; size_t symsize = 0; /* * Symbolic-name is recorded as its contents. We have to * read the contents at this time. */ while (zip->entry_bytes_remaining > 0) { const void *buff; unsigned char *mem; size_t size; int64_t offset; r = archive_read_format_7zip_read_data(a, &buff, &size, &offset); if (r < ARCHIVE_WARN) { free(symname); return (r); } mem = realloc(symname, symsize + size + 1); if (mem == NULL) { free(symname); archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Symname"); return (ARCHIVE_FATAL); } symname = mem; memcpy(symname+symsize, buff, size); symsize += size; } if (symsize == 0) { /* If there is no symname, handle it as a regular * file. */ zip_entry->mode &= ~AE_IFMT; zip_entry->mode |= AE_IFREG; archive_entry_set_mode(entry, zip_entry->mode); } else { struct archive_string_conv* utf8_conv; symname[symsize] = '\0'; /* Symbolic links are embedded as UTF-8 strings */ utf8_conv = archive_string_conversion_from_charset(&a->archive, "UTF-8", 1); if (utf8_conv == NULL) { free(symname); return ARCHIVE_FATAL; } archive_entry_copy_symlink_l(entry, (const char*)symname, symsize, utf8_conv); } free(symname); archive_entry_set_size(entry, 0); } /* Set up a more descriptive format name. */ snprintf(zip->format_name, sizeof(zip->format_name), "7-Zip"); a->archive.archive_format_name = zip->format_name; return (ret); } static int archive_read_format_7zip_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct _7zip *zip; ssize_t bytes; int ret = ARCHIVE_OK; zip = (struct _7zip *)(a->format->data); if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { zip->has_encrypted_entries = 0; } if (zip->pack_stream_bytes_unconsumed) read_consume(a); *offset = zip->entry_offset; *size = 0; *buff = NULL; /* * If we hit end-of-entry last time, clean up and return * ARCHIVE_EOF this time. */ if (zip->end_of_entry) return (ARCHIVE_EOF); size_t bytes_to_read = 16 * 1024 * 1024; // Don't try to read more than 16 MB at a time if ((uint64_t)bytes_to_read > zip->entry_bytes_remaining) { bytes_to_read = (size_t)zip->entry_bytes_remaining; } bytes = read_stream(a, buff, bytes_to_read, 0); if (bytes < 0) return ((int)bytes); if (bytes == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file body"); return (ARCHIVE_FATAL); } zip->entry_bytes_remaining -= bytes; if (zip->entry_bytes_remaining == 0) zip->end_of_entry = 1; /* Update checksum */ if ((zip->entry->flg & CRC32_IS_SET) && bytes) zip->entry_crc32 = crc32(zip->entry_crc32, *buff, (unsigned)bytes); /* If we hit the end, swallow any end-of-data marker. */ if (zip->end_of_entry) { /* Check computed CRC against file contents. */ if ((zip->entry->flg & CRC32_IS_SET) && zip->si.ss.digests[zip->entry->ssIndex] != zip->entry_crc32) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "7-Zip bad CRC: 0x%lx should be 0x%lx", (unsigned long)zip->entry_crc32, (unsigned long)zip->si.ss.digests[ zip->entry->ssIndex]); ret = ARCHIVE_WARN; } } *size = bytes; *offset = zip->entry_offset; zip->entry_offset += bytes; return (ret); } static int archive_read_format_7zip_read_data_skip(struct archive_read *a) { struct _7zip *zip; int64_t bytes_skipped; zip = (struct _7zip *)(a->format->data); if (zip->pack_stream_bytes_unconsumed) read_consume(a); /* If we've already read to end of data, we're done. */ if (zip->end_of_entry) return (ARCHIVE_OK); /* * If the length is at the beginning, we can skip the * compressed data much more quickly. */ bytes_skipped = skip_stream(a, (size_t)zip->entry_bytes_remaining); if (bytes_skipped < 0) return (ARCHIVE_FATAL); zip->entry_bytes_remaining = 0; /* This entry is finished and done. */ zip->end_of_entry = 1; return (ARCHIVE_OK); } static int archive_read_format_7zip_cleanup(struct archive_read *a) { struct _7zip *zip; zip = (struct _7zip *)(a->format->data); free_StreamsInfo(&(zip->si)); free(zip->entries); free(zip->entry_names); free_decompression(a, zip); free(zip->uncompressed_buffer); free(zip->sub_stream_buff[0]); free(zip->sub_stream_buff[1]); free(zip->sub_stream_buff[2]); free(zip->tmp_stream_buff); free(zip); (a->format->data) = NULL; return (ARCHIVE_OK); } static void read_consume(struct archive_read *a) { struct _7zip *zip = (struct _7zip *)a->format->data; if (zip->pack_stream_bytes_unconsumed) { __archive_read_consume(a, zip->pack_stream_bytes_unconsumed); zip->stream_offset += zip->pack_stream_bytes_unconsumed; zip->pack_stream_bytes_unconsumed = 0; } } #ifdef HAVE_LZMA_H /* * Set an error code and choose an error message for liblzma. */ static void set_error(struct archive_read *a, int ret) { switch (ret) { case LZMA_STREAM_END: /* Found end of stream. */ case LZMA_OK: /* Decompressor made some progress. */ break; case LZMA_MEM_ERROR: archive_set_error(&a->archive, ENOMEM, "Lzma library error: Cannot allocate memory"); break; case LZMA_MEMLIMIT_ERROR: archive_set_error(&a->archive, ENOMEM, "Lzma library error: Out of memory"); break; case LZMA_FORMAT_ERROR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Lzma library error: format not recognized"); break; case LZMA_OPTIONS_ERROR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Lzma library error: Invalid options"); break; case LZMA_DATA_ERROR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Lzma library error: Corrupted input data"); break; case LZMA_BUF_ERROR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Lzma library error: No progress is possible"); break; default: /* Return an error. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Lzma decompression failed: Unknown error"); break; } } #endif static unsigned long decode_codec_id(const unsigned char *codecId, size_t id_size) { unsigned i; unsigned long id = 0; for (i = 0; i < id_size; i++) { id <<= 8; id += codecId[i]; } return (id); } static Byte ppmd_read(void *p) { struct archive_read *a = ((IByteIn*)p)->a; struct _7zip *zip = (struct _7zip *)(a->format->data); Byte b; if (zip->ppstream.avail_in <= 0) { /* * Ppmd7_DecodeSymbol might require reading multiple bytes * and we are on boundary; * last resort to read using __archive_read_ahead. */ ssize_t bytes_avail = 0; const uint8_t* data = __archive_read_ahead(a, (size_t)zip->ppstream.stream_in+1, &bytes_avail); if(data == NULL || bytes_avail < zip->ppstream.stream_in+1) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7z file data"); zip->ppstream.overconsumed = 1; return (0); } zip->ppstream.next_in++; b = data[zip->ppstream.stream_in]; } else { b = *zip->ppstream.next_in++; } zip->ppstream.avail_in--; zip->ppstream.total_in++; zip->ppstream.stream_in++; return (b); } static int init_decompression(struct archive_read *a, struct _7zip *zip, const struct _7z_coder *coder1, const struct _7z_coder *coder2) { int r; zip->codec = coder1->codec; zip->codec2 = -1; switch (zip->codec) { case _7Z_COPY: case _7Z_BZ2: case _7Z_DEFLATE: case _7Z_ZSTD: case _7Z_PPMD: if (coder2 != NULL) { if (coder2->codec != _7Z_X86 && coder2->codec != _7Z_X86_BCJ2 && coder2->codec != _7Z_ARM && coder2->codec != _7Z_ARM64 && coder2->codec != _7Z_POWERPC && coder2->codec != _7Z_SPARC) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unsupported filter %lx for %lx", coder2->codec, coder1->codec); return (ARCHIVE_FAILED); } zip->codec2 = coder2->codec; zip->bcj_state = 0; if (coder2->codec == _7Z_X86) x86_Init(zip); else if (coder2->codec == _7Z_ARM) arm_Init(zip); } break; default: break; } switch (zip->codec) { case _7Z_COPY: break; case _7Z_LZMA: case _7Z_LZMA2: #ifdef HAVE_LZMA_H #if LZMA_VERSION_MAJOR >= 5 /* Effectively disable the limiter. */ #define LZMA_MEMLIMIT UINT64_MAX #else /* NOTE: This needs to check memory size which running system has. */ #define LZMA_MEMLIMIT (1U << 30) #endif { lzma_options_delta delta_opt; lzma_filter filters[LZMA_FILTERS_MAX], *ff; int fi = 0; if (zip->lzstream_valid) { lzma_end(&(zip->lzstream)); zip->lzstream_valid = 0; } /* * NOTE: liblzma incompletely handle the BCJ+LZMA compressed * data made by 7-Zip because 7-Zip does not add End-Of- * Payload Marker(EOPM) at the end of LZMA compressed data, * and so liblzma cannot know the end of the compressed data * without EOPM. So consequently liblzma will not return last * three or four bytes of uncompressed data because * LZMA_FILTER_X86 filter does not handle input data if its * data size is less than five bytes. If liblzma detect EOPM * or know the uncompressed data size, liblzma will flush out * the remaining that three or four bytes of uncompressed * data. That is why we have to use our converting program * for BCJ+LZMA. If we were able to tell the uncompressed * size to liblzma when using lzma_raw_decoder() liblzma * could correctly deal with BCJ+LZMA. But unfortunately * there is no way to do that. - * Discussion about this can be found at XZ Utils forum. + * + * Reference: https://web.archive.org/web/20240405171610/https://www.mail-archive.com/xz-devel@tukaani.org/msg00373.html */ if (coder2 != NULL) { zip->codec2 = coder2->codec; filters[fi].options = NULL; switch (zip->codec2) { case _7Z_X86: if (zip->codec == _7Z_LZMA2) { filters[fi].id = LZMA_FILTER_X86; fi++; } else /* Use our filter. */ x86_Init(zip); break; case _7Z_X86_BCJ2: /* Use our filter. */ zip->bcj_state = 0; break; case _7Z_DELTA: if (coder2->propertiesSize != 1) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid Delta parameter"); return (ARCHIVE_FAILED); } filters[fi].id = LZMA_FILTER_DELTA; memset(&delta_opt, 0, sizeof(delta_opt)); delta_opt.type = LZMA_DELTA_TYPE_BYTE; delta_opt.dist = (uint32_t)coder2->properties[0] + 1; filters[fi].options = &delta_opt; fi++; break; /* Following filters have not been tested yet. */ case _7Z_POWERPC: filters[fi].id = LZMA_FILTER_POWERPC; fi++; break; case _7Z_IA64: filters[fi].id = LZMA_FILTER_IA64; fi++; break; case _7Z_ARM: filters[fi].id = LZMA_FILTER_ARM; fi++; break; case _7Z_ARMTHUMB: filters[fi].id = LZMA_FILTER_ARMTHUMB; fi++; break; #ifdef LZMA_FILTER_ARM64 case _7Z_ARM64: filters[fi].id = LZMA_FILTER_ARM64; fi++; break; #endif #ifdef LZMA_FILTER_RISCV case _7Z_RISCV: filters[fi].id = LZMA_FILTER_RISCV; fi++; break; #endif case _7Z_SPARC: filters[fi].id = LZMA_FILTER_SPARC; fi++; break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unexpected codec ID: %lX", zip->codec2); return (ARCHIVE_FAILED); } } if (zip->codec == _7Z_LZMA2) filters[fi].id = LZMA_FILTER_LZMA2; else filters[fi].id = LZMA_FILTER_LZMA1; filters[fi].options = NULL; ff = &filters[fi]; r = lzma_properties_decode(&filters[fi], NULL, coder1->properties, (size_t)coder1->propertiesSize); if (r != LZMA_OK) { set_error(a, r); return (ARCHIVE_FAILED); } fi++; filters[fi].id = LZMA_VLI_UNKNOWN; filters[fi].options = NULL; r = lzma_raw_decoder(&(zip->lzstream), filters); free(ff->options); if (r != LZMA_OK) { set_error(a, r); return (ARCHIVE_FAILED); } zip->lzstream_valid = 1; zip->lzstream.total_in = 0; zip->lzstream.total_out = 0; break; } #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "LZMA codec is unsupported"); return (ARCHIVE_FAILED); #endif case _7Z_BZ2: #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) if (zip->bzstream_valid) { BZ2_bzDecompressEnd(&(zip->bzstream)); zip->bzstream_valid = 0; } r = BZ2_bzDecompressInit(&(zip->bzstream), 0, 0); if (r == BZ_MEM_ERROR) r = BZ2_bzDecompressInit(&(zip->bzstream), 0, 1); if (r != BZ_OK) { int err = ARCHIVE_ERRNO_MISC; const char *detail = NULL; switch (r) { case BZ_PARAM_ERROR: detail = "invalid setup parameter"; break; case BZ_MEM_ERROR: err = ENOMEM; detail = "out of memory"; break; case BZ_CONFIG_ERROR: detail = "mis-compiled library"; break; } archive_set_error(&a->archive, err, "Internal error initializing decompressor: %s", detail != NULL ? detail : "??"); zip->bzstream_valid = 0; return (ARCHIVE_FAILED); } zip->bzstream_valid = 1; zip->bzstream.total_in_lo32 = 0; zip->bzstream.total_in_hi32 = 0; zip->bzstream.total_out_lo32 = 0; zip->bzstream.total_out_hi32 = 0; break; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "BZ2 codec is unsupported"); return (ARCHIVE_FAILED); #endif case _7Z_ZSTD: { #if defined(HAVE_ZSTD_H) if (zip->zstdstream_valid) { ZSTD_freeDStream(zip->zstd_dstream); zip->zstdstream_valid = 0; } zip->zstd_dstream = ZSTD_createDStream(); zip->zstdstream_valid = 1; break; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZSTD codec is unsupported"); return (ARCHIVE_FAILED); #endif } case _7Z_DEFLATE: #ifdef HAVE_ZLIB_H if (zip->stream_valid) r = inflateReset(&(zip->stream)); else r = inflateInit2(&(zip->stream), -15 /* Don't check for zlib header */); if (r != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Couldn't initialize zlib stream."); return (ARCHIVE_FAILED); } zip->stream_valid = 1; zip->stream.total_in = 0; zip->stream.total_out = 0; break; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "DEFLATE codec is unsupported"); return (ARCHIVE_FAILED); #endif case _7Z_PPMD: { unsigned order; uint32_t msize; if (zip->ppmd7_valid) { __archive_ppmd7_functions.Ppmd7_Free( &zip->ppmd7_context); zip->ppmd7_valid = 0; } if (coder1->propertiesSize < 5) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed PPMd parameter"); return (ARCHIVE_FAILED); } order = coder1->properties[0]; msize = archive_le32dec(&(coder1->properties[1])); if (order < PPMD7_MIN_ORDER || order > PPMD7_MAX_ORDER || msize < PPMD7_MIN_MEM_SIZE || msize > PPMD7_MAX_MEM_SIZE) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed PPMd parameter"); return (ARCHIVE_FAILED); } __archive_ppmd7_functions.Ppmd7_Construct(&zip->ppmd7_context); r = __archive_ppmd7_functions.Ppmd7_Alloc( &zip->ppmd7_context, msize); if (r == 0) { archive_set_error(&a->archive, ENOMEM, "Coludn't allocate memory for PPMd"); return (ARCHIVE_FATAL); } __archive_ppmd7_functions.Ppmd7_Init( &zip->ppmd7_context, order); __archive_ppmd7_functions.Ppmd7z_RangeDec_CreateVTable( &zip->range_dec); zip->ppmd7_valid = 1; zip->ppmd7_stat = 0; zip->ppstream.overconsumed = 0; zip->ppstream.total_in = 0; zip->ppstream.total_out = 0; break; } case _7Z_X86: case _7Z_X86_BCJ2: case _7Z_POWERPC: case _7Z_IA64: case _7Z_ARM: case _7Z_ARMTHUMB: case _7Z_ARM64: case _7Z_RISCV: case _7Z_SPARC: case _7Z_DELTA: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unexpected codec ID: %lX", zip->codec); return (ARCHIVE_FAILED); case _7Z_CRYPTO_MAIN_ZIP: case _7Z_CRYPTO_RAR_29: case _7Z_CRYPTO_AES_256_SHA_256: if (a->entry) { archive_entry_set_is_metadata_encrypted(a->entry, 1); archive_entry_set_is_data_encrypted(a->entry, 1); zip->has_encrypted_entries = 1; } archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Crypto codec not supported yet (ID: 0x%lX)", zip->codec); return (ARCHIVE_FAILED); default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unknown codec ID: %lX", zip->codec); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } static int decompress(struct archive_read *a, struct _7zip *zip, void *buff, size_t *outbytes, const void *b, size_t *used) { const uint8_t *t_next_in; uint8_t *t_next_out; size_t o_avail_in, o_avail_out; size_t t_avail_in, t_avail_out; uint8_t *bcj2_next_out; size_t bcj2_avail_out; int r, ret = ARCHIVE_OK; t_avail_in = o_avail_in = *used; t_avail_out = o_avail_out = *outbytes; t_next_in = b; t_next_out = buff; if (zip->codec != _7Z_LZMA2 && zip->codec2 == _7Z_X86) { int i; /* Do not copy out the BCJ remaining bytes when the output * buffer size is less than five bytes. */ if (o_avail_in != 0 && t_avail_out < 5 && zip->odd_bcj_size) { *used = 0; *outbytes = 0; return (ret); } for (i = 0; zip->odd_bcj_size > 0 && t_avail_out; i++) { *t_next_out++ = zip->odd_bcj[i]; t_avail_out--; zip->odd_bcj_size--; } if (o_avail_in == 0 || t_avail_out == 0) { *used = o_avail_in - t_avail_in; *outbytes = o_avail_out - t_avail_out; if (o_avail_in == 0) ret = ARCHIVE_EOF; return (ret); } } bcj2_next_out = t_next_out; bcj2_avail_out = t_avail_out; if (zip->codec2 == _7Z_X86_BCJ2) { /* * Decord a remaining decompressed main stream for BCJ2. */ if (zip->tmp_stream_bytes_remaining) { ssize_t bytes; size_t remaining = zip->tmp_stream_bytes_remaining; bytes = Bcj2_Decode(zip, t_next_out, t_avail_out); if (bytes < 0) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "BCJ2 conversion Failed"); return (ARCHIVE_FAILED); } zip->main_stream_bytes_remaining -= remaining - zip->tmp_stream_bytes_remaining; t_avail_out -= bytes; if (o_avail_in == 0 || t_avail_out == 0) { *used = 0; *outbytes = o_avail_out - t_avail_out; if (o_avail_in == 0 && zip->tmp_stream_bytes_remaining) ret = ARCHIVE_EOF; return (ret); } t_next_out += bytes; bcj2_next_out = t_next_out; bcj2_avail_out = t_avail_out; } t_next_out = zip->tmp_stream_buff; t_avail_out = zip->tmp_stream_buff_size; } switch (zip->codec) { case _7Z_COPY: { size_t bytes = (t_avail_in > t_avail_out)?t_avail_out:t_avail_in; memcpy(t_next_out, t_next_in, bytes); t_avail_in -= bytes; t_avail_out -= bytes; if (o_avail_in == 0) ret = ARCHIVE_EOF; break; } #ifdef HAVE_LZMA_H case _7Z_LZMA: case _7Z_LZMA2: zip->lzstream.next_in = t_next_in; zip->lzstream.avail_in = t_avail_in; zip->lzstream.next_out = t_next_out; zip->lzstream.avail_out = t_avail_out; r = lzma_code(&(zip->lzstream), LZMA_RUN); switch (r) { case LZMA_STREAM_END: /* Found end of stream. */ lzma_end(&(zip->lzstream)); zip->lzstream_valid = 0; ret = ARCHIVE_EOF; break; case LZMA_OK: /* Decompressor made some progress. */ break; default: archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Decompression failed(%d)", r); return (ARCHIVE_FAILED); } t_avail_in = zip->lzstream.avail_in; t_avail_out = zip->lzstream.avail_out; break; #endif #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) case _7Z_BZ2: zip->bzstream.next_in = (char *)(uintptr_t)t_next_in; zip->bzstream.avail_in = (uint32_t)t_avail_in; zip->bzstream.next_out = (char *)(uintptr_t)t_next_out; zip->bzstream.avail_out = (uint32_t)t_avail_out; r = BZ2_bzDecompress(&(zip->bzstream)); switch (r) { case BZ_STREAM_END: /* Found end of stream. */ switch (BZ2_bzDecompressEnd(&(zip->bzstream))) { case BZ_OK: break; default: archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Failed to clean up decompressor"); return (ARCHIVE_FAILED); } zip->bzstream_valid = 0; ret = ARCHIVE_EOF; break; case BZ_OK: /* Decompressor made some progress. */ break; default: archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "bzip decompression failed"); return (ARCHIVE_FAILED); } t_avail_in = zip->bzstream.avail_in; t_avail_out = zip->bzstream.avail_out; break; #endif #ifdef HAVE_ZLIB_H case _7Z_DEFLATE: zip->stream.next_in = (Bytef *)(uintptr_t)t_next_in; zip->stream.avail_in = (uInt)t_avail_in; zip->stream.next_out = t_next_out; zip->stream.avail_out = (uInt)t_avail_out; r = inflate(&(zip->stream), 0); switch (r) { case Z_STREAM_END: /* Found end of stream. */ ret = ARCHIVE_EOF; break; case Z_OK: /* Decompressor made some progress.*/ break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "File decompression failed (%d)", r); return (ARCHIVE_FAILED); } t_avail_in = zip->stream.avail_in; t_avail_out = zip->stream.avail_out; break; #endif #ifdef HAVE_ZSTD_H case _7Z_ZSTD: { ZSTD_inBuffer input = { t_next_in, t_avail_in, 0 }; // src, size, pos ZSTD_outBuffer output = { t_next_out, t_avail_out, 0 }; // dst, size, pos size_t const zret = ZSTD_decompressStream(zip->zstd_dstream, &output, &input); if (ZSTD_isError(zret)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Zstd decompression failed: %s", ZSTD_getErrorName(zret)); return ARCHIVE_FAILED; } t_avail_in -= input.pos; t_avail_out -= output.pos; break; } #endif case _7Z_PPMD: { uint64_t flush_bytes; if (!zip->ppmd7_valid || zip->ppmd7_stat < 0 || t_avail_out <= 0) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Decompression internal error"); return (ARCHIVE_FAILED); } zip->ppstream.next_in = t_next_in; zip->ppstream.avail_in = t_avail_in; zip->ppstream.stream_in = 0; zip->ppstream.next_out = t_next_out; zip->ppstream.avail_out = t_avail_out; if (zip->ppmd7_stat == 0) { zip->bytein.a = a; zip->bytein.Read = &ppmd_read; zip->range_dec.Stream = &zip->bytein; r = __archive_ppmd7_functions.Ppmd7z_RangeDec_Init( &(zip->range_dec)); if (r == 0) { zip->ppmd7_stat = -1; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to initialize PPMd range decoder"); return (ARCHIVE_FAILED); } if (zip->ppstream.overconsumed) { zip->ppmd7_stat = -1; return (ARCHIVE_FAILED); } zip->ppmd7_stat = 1; } if (t_avail_in == 0) /* XXX Flush out remaining decoded data XXX */ flush_bytes = zip->folder_outbytes_remaining; else flush_bytes = 0; do { int sym; sym = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( &(zip->ppmd7_context), &(zip->range_dec.p)); if (sym < 0) { zip->ppmd7_stat = -1; archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Failed to decode PPMd"); return (ARCHIVE_FAILED); } if (zip->ppstream.overconsumed) { zip->ppmd7_stat = -1; return (ARCHIVE_FAILED); } *zip->ppstream.next_out++ = (unsigned char)sym; zip->ppstream.avail_out--; zip->ppstream.total_out++; if (flush_bytes) flush_bytes--; } while (zip->ppstream.avail_out && (zip->ppstream.avail_in || flush_bytes)); t_avail_in = (size_t)zip->ppstream.avail_in; t_avail_out = (size_t)zip->ppstream.avail_out; break; } default: archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Decompression internal error"); return (ARCHIVE_FAILED); } if (ret != ARCHIVE_OK && ret != ARCHIVE_EOF) return (ret); *used = o_avail_in - t_avail_in; *outbytes = o_avail_out - t_avail_out; /* * Decord BCJ. */ if (zip->codec != _7Z_LZMA2) { if (zip->codec2 == _7Z_X86) { size_t l = x86_Convert(zip, buff, *outbytes); zip->odd_bcj_size = *outbytes - l; if (zip->odd_bcj_size > 0 && zip->odd_bcj_size <= 4 && o_avail_in && ret != ARCHIVE_EOF) { memcpy(zip->odd_bcj, ((unsigned char *)buff) + l, zip->odd_bcj_size); *outbytes = l; } else zip->odd_bcj_size = 0; } else if (zip->codec2 == _7Z_ARM) { *outbytes = arm_Convert(zip, buff, *outbytes); } else if (zip->codec2 == _7Z_ARM64) { *outbytes = arm64_Convert(zip, buff, *outbytes); } else if (zip->codec2 == _7Z_SPARC) { *outbytes = sparc_Convert(zip, buff, *outbytes); } else if (zip->codec2 == _7Z_POWERPC) { *outbytes = powerpc_Convert(zip, buff, *outbytes); } } /* * Decord BCJ2 with a decompressed main stream. */ if (zip->codec2 == _7Z_X86_BCJ2) { ssize_t bytes; zip->tmp_stream_bytes_avail = zip->tmp_stream_buff_size - t_avail_out; if (zip->tmp_stream_bytes_avail > zip->main_stream_bytes_remaining) zip->tmp_stream_bytes_avail = zip->main_stream_bytes_remaining; zip->tmp_stream_bytes_remaining = zip->tmp_stream_bytes_avail; bytes = Bcj2_Decode(zip, bcj2_next_out, bcj2_avail_out); if (bytes < 0) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "BCJ2 conversion Failed"); return (ARCHIVE_FAILED); } zip->main_stream_bytes_remaining -= zip->tmp_stream_bytes_avail - zip->tmp_stream_bytes_remaining; bcj2_avail_out -= bytes; *outbytes = o_avail_out - bcj2_avail_out; } return (ret); } static int free_decompression(struct archive_read *a, struct _7zip *zip) { int r = ARCHIVE_OK; #if !defined(HAVE_ZLIB_H) &&\ !(defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR)) (void)a;/* UNUSED */ #endif #ifdef HAVE_LZMA_H if (zip->lzstream_valid) lzma_end(&(zip->lzstream)); #endif #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) if (zip->bzstream_valid) { if (BZ2_bzDecompressEnd(&(zip->bzstream)) != BZ_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to clean up bzip2 decompressor"); r = ARCHIVE_FATAL; } zip->bzstream_valid = 0; } #endif #ifdef HAVE_ZLIB_H if (zip->stream_valid) { if (inflateEnd(&(zip->stream)) != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to clean up zlib decompressor"); r = ARCHIVE_FATAL; } zip->stream_valid = 0; } #endif #ifdef HAVE_ZSTD_H if (zip->zstdstream_valid) ZSTD_freeDStream(zip->zstd_dstream); #endif if (zip->ppmd7_valid) { __archive_ppmd7_functions.Ppmd7_Free( &zip->ppmd7_context); zip->ppmd7_valid = 0; } return (r); } static int parse_7zip_uint64(struct archive_read *a, uint64_t *val) { const unsigned char *p; unsigned char avail, mask; int i; if ((p = header_bytes(a, 1)) == NULL) return (-1); avail = *p; mask = 0x80; *val = 0; for (i = 0; i < 8; i++) { if (avail & mask) { if ((p = header_bytes(a, 1)) == NULL) return (-1); *val |= ((uint64_t)*p) << (8 * i); mask >>= 1; continue; } *val += ((uint64_t)(avail & (mask -1))) << (8 * i); break; } return (0); } static int read_Bools(struct archive_read *a, unsigned char *data, size_t num) { const unsigned char *p; unsigned i, mask = 0, avail = 0; for (i = 0; i < num; i++) { if (mask == 0) { if ((p = header_bytes(a, 1)) == NULL) return (-1); avail = *p; mask = 0x80; } data[i] = (avail & mask)?1:0; mask >>= 1; } return (0); } static void free_Digest(struct _7z_digests *d) { free(d->defineds); free(d->digests); } static int read_Digests(struct archive_read *a, struct _7z_digests *d, size_t num) { const unsigned char *p; unsigned i; if (num == 0) return (-1); memset(d, 0, sizeof(*d)); d->defineds = malloc(num); if (d->defineds == NULL) return (-1); /* * Read Bools. */ if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p == 0) { if (read_Bools(a, d->defineds, num) < 0) return (-1); } else /* All are defined */ memset(d->defineds, 1, num); d->digests = calloc(num, sizeof(*d->digests)); if (d->digests == NULL) return (-1); for (i = 0; i < num; i++) { if (d->defineds[i]) { if ((p = header_bytes(a, 4)) == NULL) return (-1); d->digests[i] = archive_le32dec(p); } } return (0); } static void free_PackInfo(struct _7z_pack_info *pi) { free(pi->sizes); free(pi->positions); free_Digest(&(pi->digest)); } static int read_PackInfo(struct archive_read *a, struct _7z_pack_info *pi) { const unsigned char *p; unsigned i; memset(pi, 0, sizeof(*pi)); /* * Read PackPos. */ if (parse_7zip_uint64(a, &(pi->pos)) < 0) return (-1); /* * Read NumPackStreams. */ if (parse_7zip_uint64(a, &(pi->numPackStreams)) < 0) return (-1); if (pi->numPackStreams == 0) return (-1); if (UMAX_ENTRY < pi->numPackStreams) return (-1); /* * Read PackSizes[num] */ if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p == kEnd) /* PackSizes[num] are not present. */ return (0); if (*p != kSize) return (-1); pi->sizes = calloc((size_t)pi->numPackStreams, sizeof(uint64_t)); pi->positions = calloc((size_t)pi->numPackStreams, sizeof(uint64_t)); if (pi->sizes == NULL || pi->positions == NULL) return (-1); for (i = 0; i < pi->numPackStreams; i++) { if (parse_7zip_uint64(a, &(pi->sizes[i])) < 0) return (-1); } /* * Read PackStreamDigests[num] */ if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p == kEnd) { /* PackStreamDigests[num] are not present. */ pi->digest.defineds = calloc((size_t)pi->numPackStreams, sizeof(*pi->digest.defineds)); pi->digest.digests = calloc((size_t)pi->numPackStreams, sizeof(*pi->digest.digests)); if (pi->digest.defineds == NULL || pi->digest.digests == NULL) return (-1); return (0); } if (*p != kCRC) return (-1); if (read_Digests(a, &(pi->digest), (size_t)pi->numPackStreams) < 0) return (-1); /* * Must be marked by kEnd. */ if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p != kEnd) return (-1); return (0); } static void free_Folder(struct _7z_folder *f) { unsigned i; if (f->coders) { for (i = 0; i< f->numCoders; i++) { free(f->coders[i].properties); } free(f->coders); } free(f->bindPairs); free(f->packedStreams); free(f->unPackSize); } static int read_Folder(struct archive_read *a, struct _7z_folder *f) { struct _7zip *zip = (struct _7zip *)a->format->data; const unsigned char *p; uint64_t numInStreamsTotal = 0; uint64_t numOutStreamsTotal = 0; unsigned i; memset(f, 0, sizeof(*f)); /* * Read NumCoders. */ if (parse_7zip_uint64(a, &(f->numCoders)) < 0) return (-1); if (f->numCoders > 4) /* Too many coders. */ return (-1); f->coders = calloc((size_t)f->numCoders, sizeof(*f->coders)); if (f->coders == NULL) return (-1); for (i = 0; i< f->numCoders; i++) { size_t codec_size; int simple, attr; if ((p = header_bytes(a, 1)) == NULL) return (-1); /* * 0:3 CodecIdSize * 4: 0 - IsSimple * 1 - Is not Simple * 5: 0 - No Attributes * 1 - There are Attributes; * 7: Must be zero. */ codec_size = *p & 0xf; simple = (*p & 0x10)?0:1; attr = *p & 0x20; if (*p & 0x80) return (-1);/* Not supported. */ /* * Read Decompression Method IDs. */ if ((p = header_bytes(a, codec_size)) == NULL) return (-1); f->coders[i].codec = decode_codec_id(p, codec_size); if (simple) { f->coders[i].numInStreams = 1; f->coders[i].numOutStreams = 1; } else { if (parse_7zip_uint64( a, &(f->coders[i].numInStreams)) < 0) return (-1); if (UMAX_ENTRY < f->coders[i].numInStreams) return (-1); if (parse_7zip_uint64( a, &(f->coders[i].numOutStreams)) < 0) return (-1); if (UMAX_ENTRY < f->coders[i].numOutStreams) return (-1); } if (attr) { if (parse_7zip_uint64( a, &(f->coders[i].propertiesSize)) < 0) return (-1); if (UMAX_ENTRY < f->coders[i].propertiesSize) return (-1); if ((p = header_bytes( a, (size_t)f->coders[i].propertiesSize)) == NULL) return (-1); f->coders[i].properties = malloc((size_t)f->coders[i].propertiesSize); if (f->coders[i].properties == NULL) return (-1); memcpy(f->coders[i].properties, p, (size_t)f->coders[i].propertiesSize); } numInStreamsTotal += f->coders[i].numInStreams; numOutStreamsTotal += f->coders[i].numOutStreams; } if (numOutStreamsTotal == 0 || numInStreamsTotal < numOutStreamsTotal-1) return (-1); f->numBindPairs = numOutStreamsTotal - 1; if (zip->header_bytes_remaining < f->numBindPairs) return (-1); if (f->numBindPairs > 0) { f->bindPairs = calloc((size_t)f->numBindPairs, sizeof(*f->bindPairs)); if (f->bindPairs == NULL) return (-1); } else f->bindPairs = NULL; for (i = 0; i < f->numBindPairs; i++) { if (parse_7zip_uint64(a, &(f->bindPairs[i].inIndex)) < 0) return (-1); if (UMAX_ENTRY < f->bindPairs[i].inIndex) return (-1); if (parse_7zip_uint64(a, &(f->bindPairs[i].outIndex)) < 0) return (-1); if (UMAX_ENTRY < f->bindPairs[i].outIndex) return (-1); } f->numPackedStreams = numInStreamsTotal - f->numBindPairs; f->packedStreams = calloc((size_t)f->numPackedStreams, sizeof(*f->packedStreams)); if (f->packedStreams == NULL) return (-1); if (f->numPackedStreams == 1) { for (i = 0; i < numInStreamsTotal; i++) { unsigned j; for (j = 0; j < f->numBindPairs; j++) { if (f->bindPairs[j].inIndex == i) break; } if (j == f->numBindPairs) break; } if (i == numInStreamsTotal) return (-1); f->packedStreams[0] = i; } else { for (i = 0; i < f->numPackedStreams; i++) { if (parse_7zip_uint64(a, &(f->packedStreams[i])) < 0) return (-1); if (UMAX_ENTRY < f->packedStreams[i]) return (-1); } } f->numInStreams = numInStreamsTotal; f->numOutStreams = numOutStreamsTotal; return (0); } static void free_CodersInfo(struct _7z_coders_info *ci) { unsigned i; if (ci->folders) { for (i = 0; i < ci->numFolders; i++) free_Folder(&(ci->folders[i])); free(ci->folders); } } static int read_CodersInfo(struct archive_read *a, struct _7z_coders_info *ci) { const unsigned char *p; struct _7z_digests digest; unsigned i; memset(ci, 0, sizeof(*ci)); memset(&digest, 0, sizeof(digest)); if ((p = header_bytes(a, 1)) == NULL) goto failed; if (*p != kFolder) goto failed; /* * Read NumFolders. */ if (parse_7zip_uint64(a, &(ci->numFolders)) < 0) goto failed; if (UMAX_ENTRY < ci->numFolders) return (-1); /* * Read External. */ if ((p = header_bytes(a, 1)) == NULL) goto failed; switch (*p) { case 0: ci->folders = calloc((size_t)ci->numFolders, sizeof(*ci->folders)); if (ci->folders == NULL) return (-1); for (i = 0; i < ci->numFolders; i++) { if (read_Folder(a, &(ci->folders[i])) < 0) goto failed; } break; case 1: if (parse_7zip_uint64(a, &(ci->dataStreamIndex)) < 0) return (-1); if (UMAX_ENTRY < ci->dataStreamIndex) return (-1); if (ci->numFolders > 0) { archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); goto failed; } break; default: archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); goto failed; } if ((p = header_bytes(a, 1)) == NULL) goto failed; if (*p != kCodersUnPackSize) goto failed; for (i = 0; i < ci->numFolders; i++) { struct _7z_folder *folder = &(ci->folders[i]); unsigned j; folder->unPackSize = calloc((size_t)folder->numOutStreams, sizeof(*folder->unPackSize)); if (folder->unPackSize == NULL) goto failed; for (j = 0; j < folder->numOutStreams; j++) { if (parse_7zip_uint64(a, &(folder->unPackSize[j])) < 0) goto failed; } } /* * Read CRCs. */ if ((p = header_bytes(a, 1)) == NULL) goto failed; if (*p == kEnd) return (0); if (*p != kCRC) goto failed; if (read_Digests(a, &digest, (size_t)ci->numFolders) < 0) goto failed; for (i = 0; i < ci->numFolders; i++) { ci->folders[i].digest_defined = digest.defineds[i]; ci->folders[i].digest = digest.digests[i]; } /* * Must be kEnd. */ if ((p = header_bytes(a, 1)) == NULL) goto failed; if (*p != kEnd) goto failed; free_Digest(&digest); return (0); failed: free_Digest(&digest); return (-1); } static uint64_t folder_uncompressed_size(struct _7z_folder *f) { int n = (int)f->numOutStreams; unsigned pairs = (unsigned)f->numBindPairs; while (--n >= 0) { unsigned i; for (i = 0; i < pairs; i++) { if (f->bindPairs[i].outIndex == (uint64_t)n) break; } if (i >= pairs) return (f->unPackSize[n]); } return (0); } static void free_SubStreamsInfo(struct _7z_substream_info *ss) { free(ss->unpackSizes); free(ss->digestsDefined); free(ss->digests); } static int read_SubStreamsInfo(struct archive_read *a, struct _7z_substream_info *ss, struct _7z_folder *f, size_t numFolders) { const unsigned char *p; uint64_t *usizes; size_t unpack_streams; int type; unsigned i; uint32_t numDigests; memset(ss, 0, sizeof(*ss)); for (i = 0; i < numFolders; i++) f[i].numUnpackStreams = 1; if ((p = header_bytes(a, 1)) == NULL) return (-1); type = *p; if (type == kNumUnPackStream) { unpack_streams = 0; for (i = 0; i < numFolders; i++) { if (parse_7zip_uint64(a, &(f[i].numUnpackStreams)) < 0) return (-1); if (UMAX_ENTRY < f[i].numUnpackStreams) return (-1); if (unpack_streams > SIZE_MAX - UMAX_ENTRY) { return (-1); } unpack_streams += (size_t)f[i].numUnpackStreams; } if ((p = header_bytes(a, 1)) == NULL) return (-1); type = *p; } else unpack_streams = numFolders; ss->unpack_streams = unpack_streams; if (unpack_streams) { ss->unpackSizes = calloc(unpack_streams, sizeof(*ss->unpackSizes)); ss->digestsDefined = calloc(unpack_streams, sizeof(*ss->digestsDefined)); ss->digests = calloc(unpack_streams, sizeof(*ss->digests)); if (ss->unpackSizes == NULL || ss->digestsDefined == NULL || ss->digests == NULL) return (-1); } usizes = ss->unpackSizes; for (i = 0; i < numFolders; i++) { unsigned pack; uint64_t size, sum; if (f[i].numUnpackStreams == 0) continue; sum = 0; if (type == kSize) { for (pack = 1; pack < f[i].numUnpackStreams; pack++) { if (parse_7zip_uint64(a, usizes) < 0) return (-1); if (*usizes > UINT64_MAX - sum) return (-1); sum += *usizes++; } } size = folder_uncompressed_size(&f[i]); if (size < sum) return (-1); *usizes++ = size - sum; } if (type == kSize) { if ((p = header_bytes(a, 1)) == NULL) return (-1); type = *p; } for (i = 0; i < unpack_streams; i++) { ss->digestsDefined[i] = 0; ss->digests[i] = 0; } numDigests = 0; for (i = 0; i < numFolders; i++) { if (f[i].numUnpackStreams != 1 || !f[i].digest_defined) numDigests += (uint32_t)f[i].numUnpackStreams; } if (type == kCRC) { struct _7z_digests tmpDigests; unsigned char *digestsDefined = ss->digestsDefined; uint32_t * digests = ss->digests; int di = 0; memset(&tmpDigests, 0, sizeof(tmpDigests)); if (read_Digests(a, &(tmpDigests), numDigests) < 0) { free_Digest(&tmpDigests); return (-1); } for (i = 0; i < numFolders; i++) { if (f[i].numUnpackStreams == 1 && f[i].digest_defined) { *digestsDefined++ = 1; *digests++ = f[i].digest; } else { unsigned j; for (j = 0; j < f[i].numUnpackStreams; j++, di++) { *digestsDefined++ = tmpDigests.defineds[di]; *digests++ = tmpDigests.digests[di]; } } } free_Digest(&tmpDigests); if ((p = header_bytes(a, 1)) == NULL) return (-1); type = *p; } /* * Must be kEnd. */ if (type != kEnd) return (-1); return (0); } static void free_StreamsInfo(struct _7z_stream_info *si) { free_PackInfo(&(si->pi)); free_CodersInfo(&(si->ci)); free_SubStreamsInfo(&(si->ss)); } static int read_StreamsInfo(struct archive_read *a, struct _7z_stream_info *si) { struct _7zip *zip = (struct _7zip *)a->format->data; const unsigned char *p; unsigned i; memset(si, 0, sizeof(*si)); if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p == kPackInfo) { uint64_t packPos; if (read_PackInfo(a, &(si->pi)) < 0) return (-1); if (si->pi.positions == NULL || si->pi.sizes == NULL) return (-1); /* * Calculate packed stream positions. */ packPos = si->pi.pos; for (i = 0; i < si->pi.numPackStreams; i++) { si->pi.positions[i] = packPos; if (packPos > UINT64_MAX - si->pi.sizes[i]) return (-1); packPos += si->pi.sizes[i]; if (packPos > zip->header_offset) return (-1); } if ((p = header_bytes(a, 1)) == NULL) return (-1); } if (*p == kUnPackInfo) { uint32_t packIndex; struct _7z_folder *f; if (read_CodersInfo(a, &(si->ci)) < 0) return (-1); /* * Calculate packed stream indexes. */ packIndex = 0; f = si->ci.folders; for (i = 0; i < si->ci.numFolders; i++) { f[i].packIndex = packIndex; if (f[i].numPackedStreams > UINT32_MAX) return (-1); if (packIndex > UINT32_MAX - (uint32_t)f[i].numPackedStreams) return (-1); packIndex += (uint32_t)f[i].numPackedStreams; if (packIndex > si->pi.numPackStreams) return (-1); } if ((p = header_bytes(a, 1)) == NULL) return (-1); } if (*p == kSubStreamsInfo) { if (read_SubStreamsInfo(a, &(si->ss), si->ci.folders, (size_t)si->ci.numFolders) < 0) return (-1); if ((p = header_bytes(a, 1)) == NULL) return (-1); } /* * Must be kEnd. */ if (*p != kEnd) return (-1); return (0); } static void free_Header(struct _7z_header_info *h) { free(h->emptyStreamBools); free(h->emptyFileBools); free(h->antiBools); free(h->attrBools); } static int read_Header(struct archive_read *a, struct _7z_header_info *h, int check_header_id) { struct _7zip *zip = (struct _7zip *)a->format->data; const unsigned char *p; struct _7z_folder *folders; struct _7z_stream_info *si = &(zip->si); struct _7zip_entry *entries; uint32_t folderIndex, indexInFolder; unsigned i; int eindex, empty_streams, sindex; if (check_header_id) { /* * Read Header. */ if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p != kHeader) return (-1); } /* * Read ArchiveProperties. */ if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p == kArchiveProperties) { for (;;) { uint64_t size; if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p == 0) break; if (parse_7zip_uint64(a, &size) < 0) return (-1); } if ((p = header_bytes(a, 1)) == NULL) return (-1); } /* * Read MainStreamsInfo. */ if (*p == kMainStreamsInfo) { if (read_StreamsInfo(a, &(zip->si)) < 0) return (-1); if ((p = header_bytes(a, 1)) == NULL) return (-1); } if (*p == kEnd) return (0); /* * Read FilesInfo. */ if (*p != kFilesInfo) return (-1); if (parse_7zip_uint64(a, &(zip->numFiles)) < 0) return (-1); if (UMAX_ENTRY < zip->numFiles) return (-1); zip->entries = calloc((size_t)zip->numFiles, sizeof(*zip->entries)); if (zip->entries == NULL) return (-1); entries = zip->entries; empty_streams = 0; for (;;) { int type; uint64_t size; size_t ll; if ((p = header_bytes(a, 1)) == NULL) return (-1); type = *p; if (type == kEnd) break; if (parse_7zip_uint64(a, &size) < 0) return (-1); if (zip->header_bytes_remaining < size) return (-1); ll = (size_t)size; switch (type) { case kEmptyStream: if (h->emptyStreamBools != NULL) return (-1); h->emptyStreamBools = calloc((size_t)zip->numFiles, sizeof(*h->emptyStreamBools)); if (h->emptyStreamBools == NULL) return (-1); if (read_Bools( a, h->emptyStreamBools, (size_t)zip->numFiles) < 0) return (-1); empty_streams = 0; for (i = 0; i < zip->numFiles; i++) { if (h->emptyStreamBools[i]) empty_streams++; } break; case kEmptyFile: if (empty_streams <= 0) { /* Unexcepted sequence. Skip this. */ if (header_bytes(a, ll) == NULL) return (-1); break; } if (h->emptyFileBools != NULL) return (-1); h->emptyFileBools = calloc(empty_streams, sizeof(*h->emptyFileBools)); if (h->emptyFileBools == NULL) return (-1); if (read_Bools(a, h->emptyFileBools, empty_streams) < 0) return (-1); break; case kAnti: if (empty_streams <= 0) { /* Unexcepted sequence. Skip this. */ if (header_bytes(a, ll) == NULL) return (-1); break; } if (h->antiBools != NULL) return (-1); h->antiBools = calloc(empty_streams, sizeof(*h->antiBools)); if (h->antiBools == NULL) return (-1); if (read_Bools(a, h->antiBools, empty_streams) < 0) return (-1); break; case kCTime: case kATime: case kMTime: if (read_Times(a, h, type) < 0) return (-1); break; case kName: { unsigned char *np; size_t nl, nb; /* Skip one byte. */ if ((p = header_bytes(a, 1)) == NULL) return (-1); ll--; if ((ll & 1) || ll < zip->numFiles * 4) return (-1); if (zip->entry_names != NULL) return (-1); zip->entry_names = malloc(ll); if (zip->entry_names == NULL) return (-1); np = zip->entry_names; nb = ll; /* * Copy whole file names. * NOTE: This loop prevents from expanding * the uncompressed buffer in order not to * use extra memory resource. */ while (nb) { size_t b; if (nb > UBUFF_SIZE) b = UBUFF_SIZE; else b = nb; if ((p = header_bytes(a, b)) == NULL) return (-1); memcpy(np, p, b); np += b; nb -= b; } np = zip->entry_names; nl = ll; for (i = 0; i < zip->numFiles; i++) { entries[i].utf16name = np; #if defined(_WIN32) && !defined(__CYGWIN__) && defined(_DEBUG) entries[i].wname = (wchar_t *)np; #endif /* Find a terminator. */ while (nl >= 2 && (np[0] || np[1])) { np += 2; nl -= 2; } if (nl < 2) return (-1);/* Terminator not found */ entries[i].name_len = np - entries[i].utf16name; np += 2; nl -= 2; } break; } case kAttributes: { int allAreDefined; if ((p = header_bytes(a, 2)) == NULL) return (-1); allAreDefined = *p; if (h->attrBools != NULL) return (-1); h->attrBools = calloc((size_t)zip->numFiles, sizeof(*h->attrBools)); if (h->attrBools == NULL) return (-1); if (allAreDefined) memset(h->attrBools, 1, (size_t)zip->numFiles); else { if (read_Bools(a, h->attrBools, (size_t)zip->numFiles) < 0) return (-1); } for (i = 0; i < zip->numFiles; i++) { if (h->attrBools[i]) { if ((p = header_bytes(a, 4)) == NULL) return (-1); entries[i].attr = archive_le32dec(p); } } break; } case kDummy: if (ll == 0) break; __LA_FALLTHROUGH; default: if (header_bytes(a, ll) == NULL) return (-1); break; } } /* * Set up entry's attributes. */ folders = si->ci.folders; eindex = sindex = 0; folderIndex = indexInFolder = 0; for (i = 0; i < zip->numFiles; i++) { if (h->emptyStreamBools == NULL || h->emptyStreamBools[i] == 0) entries[i].flg |= HAS_STREAM; /* The high 16 bits of attributes is a posix file mode. */ entries[i].mode = entries[i].attr >> 16; if (!(entries[i].attr & FILE_ATTRIBUTE_UNIX_EXTENSION)) { // Only windows permissions specified for this entry. Translate to // reasonable corresponding unix permissions. if (entries[i].attr & FILE_ATTRIBUTE_DIRECTORY) { if (entries[i].attr & FILE_ATTRIBUTE_READONLY) { // Read-only directory. entries[i].mode = AE_IFDIR | 0555; } else { // Read-write directory. entries[i].mode = AE_IFDIR | 0755; } } else if (entries[i].attr & FILE_ATTRIBUTE_READONLY) { // Readonly file. entries[i].mode = AE_IFREG | 0444; } else { // Assume read-write file. entries[i].mode = AE_IFREG | 0644; } } if (entries[i].flg & HAS_STREAM) { if ((size_t)sindex >= si->ss.unpack_streams) return (-1); if (entries[i].mode == 0) entries[i].mode = AE_IFREG | 0666; if (si->ss.digestsDefined[sindex]) entries[i].flg |= CRC32_IS_SET; entries[i].ssIndex = sindex; sindex++; } else { int dir; if (h->emptyFileBools == NULL) dir = 1; else { if (h->emptyFileBools[eindex]) dir = 0; else dir = 1; eindex++; } if (entries[i].mode == 0) { if (dir) entries[i].mode = AE_IFDIR | 0777; else entries[i].mode = AE_IFREG | 0666; } else if (dir && (entries[i].mode & AE_IFMT) != AE_IFDIR) { entries[i].mode &= ~AE_IFMT; entries[i].mode |= AE_IFDIR; } if ((entries[i].mode & AE_IFMT) == AE_IFDIR && entries[i].name_len >= 2 && (entries[i].utf16name[entries[i].name_len-2] != '/' || entries[i].utf16name[entries[i].name_len-1] != 0)) { entries[i].utf16name[entries[i].name_len] = '/'; entries[i].utf16name[entries[i].name_len+1] = 0; entries[i].name_len += 2; } entries[i].ssIndex = -1; } if (entries[i].attr & FILE_ATTRIBUTE_READONLY) entries[i].mode &= ~0222;/* Read only. */ if ((entries[i].flg & HAS_STREAM) == 0 && indexInFolder == 0) { /* * The entry is an empty file or a directory file, * those both have no contents. */ entries[i].folderIndex = -1; continue; } if (indexInFolder == 0) { for (;;) { if (folderIndex >= si->ci.numFolders) return (-1); if (folders[folderIndex].numUnpackStreams) break; folderIndex++; } } entries[i].folderIndex = folderIndex; if ((entries[i].flg & HAS_STREAM) == 0) continue; indexInFolder++; if (indexInFolder >= folders[folderIndex].numUnpackStreams) { folderIndex++; indexInFolder = 0; } } return (0); } static int read_Times(struct archive_read *a, struct _7z_header_info *h, int type) { struct _7zip *zip = (struct _7zip *)a->format->data; const unsigned char *p; struct _7zip_entry *entries = zip->entries; unsigned char *timeBools; int allAreDefined; unsigned i; timeBools = calloc((size_t)zip->numFiles, sizeof(*timeBools)); if (timeBools == NULL) return (-1); /* Read allAreDefined. */ if ((p = header_bytes(a, 1)) == NULL) goto failed; allAreDefined = *p; if (allAreDefined) memset(timeBools, 1, (size_t)zip->numFiles); else { if (read_Bools(a, timeBools, (size_t)zip->numFiles) < 0) goto failed; } /* Read external. */ if ((p = header_bytes(a, 1)) == NULL) goto failed; if (*p) { if (parse_7zip_uint64(a, &(h->dataIndex)) < 0) goto failed; if (UMAX_ENTRY < h->dataIndex) goto failed; } for (i = 0; i < zip->numFiles; i++) { if (!timeBools[i]) continue; if ((p = header_bytes(a, 8)) == NULL) goto failed; switch (type) { case kCTime: ntfs_to_unix(archive_le64dec(p), &(entries[i].ctime), &(entries[i].ctime_ns)); entries[i].flg |= CTIME_IS_SET; break; case kATime: ntfs_to_unix(archive_le64dec(p), &(entries[i].atime), &(entries[i].atime_ns)); entries[i].flg |= ATIME_IS_SET; break; case kMTime: ntfs_to_unix(archive_le64dec(p), &(entries[i].mtime), &(entries[i].mtime_ns)); entries[i].flg |= MTIME_IS_SET; break; } } free(timeBools); return (0); failed: free(timeBools); return (-1); } static int decode_encoded_header_info(struct archive_read *a, struct _7z_stream_info *si) { struct _7zip *zip = (struct _7zip *)a->format->data; errno = 0; if (read_StreamsInfo(a, si) < 0) { if (errno == ENOMEM) archive_set_error(&a->archive, -1, "Couldn't allocate memory"); else archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); return (ARCHIVE_FATAL); } if (si->pi.numPackStreams == 0 || si->ci.numFolders == 0) { archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); return (ARCHIVE_FATAL); } if (zip->header_offset < si->pi.pos + si->pi.sizes[0] || (int64_t)(si->pi.pos + si->pi.sizes[0]) < 0 || si->pi.sizes[0] == 0 || (int64_t)si->pi.pos < 0) { archive_set_error(&a->archive, -1, "Malformed Header offset"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } static const unsigned char * header_bytes(struct archive_read *a, size_t rbytes) { struct _7zip *zip = (struct _7zip *)a->format->data; const unsigned char *p; if (zip->header_bytes_remaining < rbytes) return (NULL); if (zip->pack_stream_bytes_unconsumed) read_consume(a); if (zip->header_is_encoded == 0) { p = __archive_read_ahead(a, rbytes, NULL); if (p == NULL) return (NULL); zip->header_bytes_remaining -= rbytes; zip->pack_stream_bytes_unconsumed = rbytes; } else { const void *buff; ssize_t bytes; bytes = read_stream(a, &buff, rbytes, rbytes); if (bytes <= 0) return (NULL); zip->header_bytes_remaining -= bytes; p = buff; } /* Update checksum */ zip->header_crc32 = crc32(zip->header_crc32, p, (unsigned)rbytes); return (p); } static int slurp_central_directory(struct archive_read *a, struct _7zip *zip, struct _7z_header_info *header) { const unsigned char *p; uint64_t next_header_offset; uint64_t next_header_size; uint32_t next_header_crc; ssize_t bytes_avail; int check_header_crc, r; if ((p = __archive_read_ahead(a, 32, &bytes_avail)) == NULL) return (ARCHIVE_FATAL); if ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0) { /* This is an executable ? Must be self-extracting... */ const ssize_t min_addr = p[0] == 'M' ? find_pe_overlay(a) : find_elf_data_sec(a); r = skip_sfx(a, min_addr); if (r < ARCHIVE_WARN) return (r); if ((p = __archive_read_ahead(a, 32, &bytes_avail)) == NULL) return (ARCHIVE_FATAL); } zip->seek_base += 32; if (memcmp(p, _7ZIP_SIGNATURE, 6) != 0) { archive_set_error(&a->archive, -1, "Not 7-Zip archive file"); return (ARCHIVE_FATAL); } /* CRC check. */ if (crc32(0, (const unsigned char *)p + 12, 20) != archive_le32dec(p + 8)) { #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&a->archive, -1, "Header CRC error"); return (ARCHIVE_FATAL); #endif } next_header_offset = archive_le64dec(p + 12); next_header_size = archive_le64dec(p + 20); next_header_crc = archive_le32dec(p + 28); if (next_header_size == 0) /* There is no entry in an archive file. */ return (ARCHIVE_EOF); if (((int64_t)next_header_offset) < 0) { archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); return (ARCHIVE_FATAL); } __archive_read_consume(a, 32); if (next_header_offset != 0) { if (bytes_avail >= (ssize_t)next_header_offset) __archive_read_consume(a, next_header_offset); else if (__archive_read_seek(a, next_header_offset + zip->seek_base, SEEK_SET) < 0) return (ARCHIVE_FATAL); } zip->stream_offset = next_header_offset; zip->header_offset = next_header_offset; zip->header_bytes_remaining = next_header_size; zip->header_crc32 = 0; zip->header_is_encoded = 0; zip->header_is_being_read = 1; zip->has_encrypted_entries = 0; check_header_crc = 1; if ((p = header_bytes(a, 1)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file body"); return (ARCHIVE_FATAL); } /* Parse ArchiveProperties. */ switch (p[0]) { case kEncodedHeader: /* * The archive has an encoded header and we have to decode it * in order to parse the header correctly. */ r = decode_encoded_header_info(a, &(zip->si)); /* Check the EncodedHeader CRC.*/ if (r == 0 && zip->header_crc32 != next_header_crc) { #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&a->archive, -1, "Damaged 7-Zip archive"); r = -1; #endif } if (r == 0) { if (zip->si.ci.folders[0].digest_defined) next_header_crc = zip->si.ci.folders[0].digest; else check_header_crc = 0; if (zip->pack_stream_bytes_unconsumed) read_consume(a); r = setup_decode_folder(a, zip->si.ci.folders, 1); if (r == 0) { zip->header_bytes_remaining = zip->folder_outbytes_remaining; r = seek_pack(a); } } /* Clean up StreamsInfo. */ free_StreamsInfo(&(zip->si)); memset(&(zip->si), 0, sizeof(zip->si)); if (r < 0) return (ARCHIVE_FATAL); zip->header_is_encoded = 1; zip->header_crc32 = 0; /* FALL THROUGH */ case kHeader: /* * Parse the header. */ errno = 0; r = read_Header(a, header, zip->header_is_encoded); if (r < 0) { if (errno == ENOMEM) archive_set_error(&a->archive, -1, "Couldn't allocate memory"); else archive_set_error(&a->archive, -1, "Damaged 7-Zip archive"); return (ARCHIVE_FATAL); } /* * Must be kEnd. */ if ((p = header_bytes(a, 1)) == NULL ||*p != kEnd) { archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); return (ARCHIVE_FATAL); } /* Check the Header CRC.*/ if (check_header_crc && zip->header_crc32 != next_header_crc) { #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); return (ARCHIVE_FATAL); #endif } break; default: archive_set_error(&a->archive, -1, "Unexpected Property ID = %X", p[0]); return (ARCHIVE_FATAL); } /* Clean up variables be used for decoding the archive header */ zip->pack_stream_remaining = 0; zip->pack_stream_index = 0; zip->folder_outbytes_remaining = 0; zip->uncompressed_buffer_bytes_remaining = 0; zip->pack_stream_bytes_unconsumed = 0; zip->header_is_being_read = 0; return (ARCHIVE_OK); } static ssize_t get_uncompressed_data(struct archive_read *a, const void **buff, size_t size, size_t minimum) { struct _7zip *zip = (struct _7zip *)a->format->data; ssize_t bytes_avail; if (zip->codec == _7Z_COPY && zip->codec2 == (unsigned long)-1) { /* Copy mode. */ *buff = __archive_read_ahead(a, minimum, &bytes_avail); if (*buff == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file data"); return (ARCHIVE_FATAL); } if ((size_t)bytes_avail > zip->uncompressed_buffer_bytes_remaining) bytes_avail = (ssize_t) zip->uncompressed_buffer_bytes_remaining; if ((size_t)bytes_avail > size) bytes_avail = (ssize_t)size; zip->pack_stream_bytes_unconsumed = bytes_avail; } else if (zip->uncompressed_buffer_pointer == NULL) { /* Decompression has failed. */ archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Damaged 7-Zip archive"); return (ARCHIVE_FATAL); } else { /* Packed mode. */ if (minimum > zip->uncompressed_buffer_bytes_remaining) { /* * If remaining uncompressed data size is less than * the minimum size, fill the buffer up to the * minimum size. */ if (extract_pack_stream(a, minimum) < 0) return (ARCHIVE_FATAL); } if (size > zip->uncompressed_buffer_bytes_remaining) bytes_avail = (ssize_t) zip->uncompressed_buffer_bytes_remaining; else bytes_avail = (ssize_t)size; *buff = zip->uncompressed_buffer_pointer; zip->uncompressed_buffer_pointer += bytes_avail; } zip->uncompressed_buffer_bytes_remaining -= bytes_avail; return (bytes_avail); } static ssize_t extract_pack_stream(struct archive_read *a, size_t minimum) { struct _7zip *zip = (struct _7zip *)a->format->data; ssize_t bytes_avail; int r; if (zip->codec == _7Z_COPY && zip->codec2 == (unsigned long)-1) { if (minimum == 0) minimum = 1; if (__archive_read_ahead(a, minimum, &bytes_avail) == NULL || bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file body"); return (ARCHIVE_FATAL); } if ((uint64_t)bytes_avail > zip->pack_stream_inbytes_remaining) bytes_avail = (ssize_t)zip->pack_stream_inbytes_remaining; zip->pack_stream_inbytes_remaining -= bytes_avail; if ((uint64_t)bytes_avail > zip->folder_outbytes_remaining) bytes_avail = (ssize_t)zip->folder_outbytes_remaining; zip->folder_outbytes_remaining -= bytes_avail; zip->uncompressed_buffer_bytes_remaining = bytes_avail; return (ARCHIVE_OK); } /* If the buffer hasn't been allocated, allocate it now. */ if (zip->uncompressed_buffer == NULL) { zip->uncompressed_buffer_size = UBUFF_SIZE; if (zip->uncompressed_buffer_size < minimum) { zip->uncompressed_buffer_size = minimum + 1023; zip->uncompressed_buffer_size &= ~0x3ff; } zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); if (zip->uncompressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for 7-Zip decompression"); return (ARCHIVE_FATAL); } zip->uncompressed_buffer_bytes_remaining = 0; } else if (zip->uncompressed_buffer_size < minimum || zip->uncompressed_buffer_bytes_remaining < minimum) { /* * Make sure the uncompressed buffer can have bytes * at least `minimum' bytes. * NOTE: This case happen when reading the header. */ size_t used; if (zip->uncompressed_buffer_pointer != 0) used = zip->uncompressed_buffer_pointer - zip->uncompressed_buffer; else used = 0; if (zip->uncompressed_buffer_size < minimum) { /* * Expand the uncompressed buffer up to * the minimum size. */ void *p; size_t new_size; new_size = minimum + 1023; new_size &= ~0x3ff; p = realloc(zip->uncompressed_buffer, new_size); if (p == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for 7-Zip decompression"); return (ARCHIVE_FATAL); } zip->uncompressed_buffer = (unsigned char *)p; zip->uncompressed_buffer_size = new_size; } /* * Move unconsumed bytes to the head. */ if (used) { memmove(zip->uncompressed_buffer, zip->uncompressed_buffer + used, zip->uncompressed_buffer_bytes_remaining); } } else zip->uncompressed_buffer_bytes_remaining = 0; zip->uncompressed_buffer_pointer = NULL; for (;;) { size_t bytes_in, bytes_out; const void *buff_in; unsigned char *buff_out; int end_of_data; /* * Note: '1' here is a performance optimization. * Recall that the decompression layer returns a count of * available bytes; asking for more than that forces the * decompressor to combine reads by copying data. */ buff_in = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file body"); return (ARCHIVE_FATAL); } buff_out = zip->uncompressed_buffer + zip->uncompressed_buffer_bytes_remaining; bytes_out = zip->uncompressed_buffer_size - zip->uncompressed_buffer_bytes_remaining; bytes_in = bytes_avail; if (bytes_in > zip->pack_stream_inbytes_remaining) bytes_in = (size_t)zip->pack_stream_inbytes_remaining; /* Drive decompression. */ r = decompress(a, zip, buff_out, &bytes_out, buff_in, &bytes_in); switch (r) { case ARCHIVE_OK: end_of_data = 0; break; case ARCHIVE_EOF: end_of_data = 1; break; default: return (ARCHIVE_FATAL); } zip->pack_stream_inbytes_remaining -= bytes_in; if (bytes_out > zip->folder_outbytes_remaining) bytes_out = (size_t)zip->folder_outbytes_remaining; zip->folder_outbytes_remaining -= bytes_out; zip->uncompressed_buffer_bytes_remaining += bytes_out; zip->pack_stream_bytes_unconsumed = bytes_in; /* * Continue decompression until uncompressed_buffer is full. */ if (zip->uncompressed_buffer_bytes_remaining == zip->uncompressed_buffer_size) break; if (zip->codec2 == _7Z_X86 && zip->odd_bcj_size && zip->uncompressed_buffer_bytes_remaining + 5 > zip->uncompressed_buffer_size) break; if (zip->pack_stream_inbytes_remaining == 0 && zip->folder_outbytes_remaining == 0) break; if (end_of_data || (bytes_in == 0 && bytes_out == 0)) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Damaged 7-Zip archive"); return (ARCHIVE_FATAL); } read_consume(a); } if (zip->uncompressed_buffer_bytes_remaining < minimum) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Damaged 7-Zip archive"); return (ARCHIVE_FATAL); } zip->uncompressed_buffer_pointer = zip->uncompressed_buffer; return (ARCHIVE_OK); } static int seek_pack(struct archive_read *a) { struct _7zip *zip = (struct _7zip *)a->format->data; int64_t pack_offset; if (zip->pack_stream_remaining <= 0) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Damaged 7-Zip archive"); return (ARCHIVE_FATAL); } zip->pack_stream_inbytes_remaining = zip->si.pi.sizes[zip->pack_stream_index]; pack_offset = zip->si.pi.positions[zip->pack_stream_index]; if (zip->stream_offset != pack_offset) { if (0 > __archive_read_seek(a, pack_offset + zip->seek_base, SEEK_SET)) return (ARCHIVE_FATAL); zip->stream_offset = pack_offset; } zip->pack_stream_index++; zip->pack_stream_remaining--; return (ARCHIVE_OK); } static ssize_t read_stream(struct archive_read *a, const void **buff, size_t size, size_t minimum) { struct _7zip *zip = (struct _7zip *)a->format->data; uint64_t skip_bytes = 0; ssize_t r; if (zip->uncompressed_buffer_bytes_remaining == 0) { if (zip->pack_stream_inbytes_remaining > 0) { r = extract_pack_stream(a, 0); if (r < 0) return (r); return (get_uncompressed_data(a, buff, size, minimum)); } else if (zip->folder_outbytes_remaining > 0) { /* Extract a remaining pack stream. */ r = extract_pack_stream(a, 0); if (r < 0) return (r); return (get_uncompressed_data(a, buff, size, minimum)); } } else return (get_uncompressed_data(a, buff, size, minimum)); /* * Current pack stream has been consumed. */ if (zip->pack_stream_remaining == 0) { if (zip->header_is_being_read) { /* Invalid sequence. This might happen when * reading a malformed archive. */ archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Malformed 7-Zip archive"); return (ARCHIVE_FATAL); } /* * All current folder's pack streams have been * consumed. Switch to next folder. */ if (zip->folder_index == 0 && (zip->si.ci.folders[zip->entry->folderIndex].skipped_bytes || zip->folder_index != zip->entry->folderIndex)) { zip->folder_index = zip->entry->folderIndex; skip_bytes = zip->si.ci.folders[zip->folder_index].skipped_bytes; } if (zip->folder_index >= zip->si.ci.numFolders) { /* * We have consumed all folders and its pack streams. */ *buff = NULL; return (0); } r = setup_decode_folder(a, &(zip->si.ci.folders[zip->folder_index]), 0); if (r != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->folder_index++; } /* * Switch to next pack stream. */ r = seek_pack(a); if (r < 0) return (r); /* Extract a new pack stream. */ r = extract_pack_stream(a, 0); if (r < 0) return (r); /* * Skip the bytes we already has skipped in skip_stream(). */ while (1) { ssize_t skipped; if (zip->uncompressed_buffer_bytes_remaining == 0) { if (zip->pack_stream_inbytes_remaining > 0) { r = extract_pack_stream(a, 0); if (r < 0) return (r); } else if (zip->folder_outbytes_remaining > 0) { /* Extract a remaining pack stream. */ r = extract_pack_stream(a, 0); if (r < 0) return (r); } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file body"); return (ARCHIVE_FATAL); } } if (!skip_bytes) break; skipped = get_uncompressed_data( a, buff, (size_t)skip_bytes, 0); if (skipped < 0) return (skipped); skip_bytes -= skipped; if (zip->pack_stream_bytes_unconsumed) read_consume(a); } return (get_uncompressed_data(a, buff, size, minimum)); } static int setup_decode_folder(struct archive_read *a, struct _7z_folder *folder, int header) { struct _7zip *zip = (struct _7zip *)a->format->data; const struct _7z_coder *coder1, *coder2; const char *cname = (header)?"archive header":"file content"; unsigned i; int r, found_bcj2 = 0; /* * Release the memory which the previous folder used for BCJ2. */ for (i = 0; i < 3; i++) { free(zip->sub_stream_buff[i]); zip->sub_stream_buff[i] = NULL; } /* * Initialize a stream reader. */ zip->pack_stream_remaining = (unsigned)folder->numPackedStreams; zip->pack_stream_index = (unsigned)folder->packIndex; zip->folder_outbytes_remaining = folder_uncompressed_size(folder); zip->uncompressed_buffer_bytes_remaining = 0; /* * Check coder types. */ for (i = 0; i < folder->numCoders; i++) { switch(folder->coders[i].codec) { case _7Z_CRYPTO_MAIN_ZIP: case _7Z_CRYPTO_RAR_29: case _7Z_CRYPTO_AES_256_SHA_256: { /* For entry that is associated with this folder, mark it as encrypted (data+metadata). */ zip->has_encrypted_entries = 1; if (a->entry) { archive_entry_set_is_data_encrypted(a->entry, 1); archive_entry_set_is_metadata_encrypted(a->entry, 1); } archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "The %s is encrypted, " "but currently not supported", cname); return (ARCHIVE_FATAL); } case _7Z_X86_BCJ2: { found_bcj2++; break; } } } /* Now that we've checked for encryption, if there were still no * encrypted entries found we can say for sure that there are none. */ if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { zip->has_encrypted_entries = 0; } if ((folder->numCoders > 2 && !found_bcj2) || found_bcj2 > 1) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "The %s is encoded with many filters, " "but currently not supported", cname); return (ARCHIVE_FATAL); } coder1 = &(folder->coders[0]); if (folder->numCoders == 2) coder2 = &(folder->coders[1]); else coder2 = NULL; if (found_bcj2) { /* * Preparation to decode BCJ2. * Decoding BCJ2 requires four sources. Those are at least, * as far as I know, two types of the storage form. */ const struct _7z_coder *fc = folder->coders; static const struct _7z_coder coder_copy = {0, 1, 1, 0, NULL}; const struct _7z_coder *scoder[3] = {&coder_copy, &coder_copy, &coder_copy}; const void *buff; ssize_t bytes; unsigned char *b[3] = {NULL, NULL, NULL}; uint64_t sunpack[3] ={-1, -1, -1}; size_t s[3] = {0, 0, 0}; int idx[3] = {0, 1, 2}; if (folder->numCoders == 4 && fc[3].codec == _7Z_X86_BCJ2 && folder->numInStreams == 7 && folder->numOutStreams == 4 && zip->pack_stream_remaining == 4) { /* Source type 1 made by 7zr or 7z with -m options. */ if (folder->bindPairs[0].inIndex == 5) { /* The form made by 7zr */ idx[0] = 1; idx[1] = 2; idx[2] = 0; scoder[1] = &(fc[1]); scoder[2] = &(fc[0]); sunpack[1] = folder->unPackSize[1]; sunpack[2] = folder->unPackSize[0]; coder1 = &(fc[2]); } else { /* * NOTE: Some patterns do not work. * work: * 7z a -m0=BCJ2 -m1=COPY -m2=COPY * -m3=(any) * 7z a -m0=BCJ2 -m1=COPY -m2=(any) * -m3=COPY * 7z a -m0=BCJ2 -m1=(any) -m2=COPY * -m3=COPY * not work: * other patterns. * * We have to handle this like `pipe' or * our libarchive7s filter frame work, * decoding the BCJ2 main stream sequentially, * m3 -> m2 -> m1 -> BCJ2. * */ if (fc[0].codec == _7Z_COPY && fc[1].codec == _7Z_COPY) coder1 = &(folder->coders[2]); else if (fc[0].codec == _7Z_COPY && fc[2].codec == _7Z_COPY) coder1 = &(folder->coders[1]); else if (fc[1].codec == _7Z_COPY && fc[2].codec == _7Z_COPY) coder1 = &(folder->coders[0]); else { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Unsupported form of " "BCJ2 streams"); return (ARCHIVE_FATAL); } } coder2 = &(fc[3]); zip->main_stream_bytes_remaining = (size_t)folder->unPackSize[2]; } else if (coder2 != NULL && coder2->codec == _7Z_X86_BCJ2 && zip->pack_stream_remaining == 4 && folder->numInStreams == 5 && folder->numOutStreams == 2) { /* Source type 0 made by 7z */ zip->main_stream_bytes_remaining = (size_t)folder->unPackSize[0]; } else { /* We got an unexpected form. */ archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Unsupported form of BCJ2 streams"); return (ARCHIVE_FATAL); } /* Skip the main stream at this time. */ if ((r = seek_pack(a)) < 0) return (r); zip->pack_stream_bytes_unconsumed = (size_t)zip->pack_stream_inbytes_remaining; read_consume(a); /* Read following three sub streams. */ for (i = 0; i < 3; i++) { const struct _7z_coder *coder = scoder[i]; if ((r = seek_pack(a)) < 0) { free(b[0]); free(b[1]); free(b[2]); return (r); } if (sunpack[i] == (uint64_t)-1) zip->folder_outbytes_remaining = zip->pack_stream_inbytes_remaining; else zip->folder_outbytes_remaining = sunpack[i]; r = init_decompression(a, zip, coder, NULL); if (r != ARCHIVE_OK) { free(b[0]); free(b[1]); free(b[2]); return (ARCHIVE_FATAL); } /* Allocate memory for the decoded data of a sub * stream. */ b[i] = malloc((size_t)zip->folder_outbytes_remaining); if (b[i] == NULL) { free(b[0]); free(b[1]); free(b[2]); archive_set_error(&a->archive, ENOMEM, "No memory for 7-Zip decompression"); return (ARCHIVE_FATAL); } /* Extract a sub stream. */ while (zip->pack_stream_inbytes_remaining > 0) { r = (int)extract_pack_stream(a, 0); if (r < 0) { free(b[0]); free(b[1]); free(b[2]); return (r); } bytes = get_uncompressed_data(a, &buff, zip->uncompressed_buffer_bytes_remaining, 0); if (bytes < 0) { free(b[0]); free(b[1]); free(b[2]); return ((int)bytes); } memcpy(b[i]+s[i], buff, bytes); s[i] += bytes; if (zip->pack_stream_bytes_unconsumed) read_consume(a); } } /* Set the sub streams to the right place. */ for (i = 0; i < 3; i++) { zip->sub_stream_buff[i] = b[idx[i]]; zip->sub_stream_size[i] = s[idx[i]]; zip->sub_stream_bytes_remaining[i] = s[idx[i]]; } /* Allocate memory used for decoded main stream bytes. */ if (zip->tmp_stream_buff == NULL) { zip->tmp_stream_buff_size = 32 * 1024; zip->tmp_stream_buff = malloc(zip->tmp_stream_buff_size); if (zip->tmp_stream_buff == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for 7-Zip decompression"); return (ARCHIVE_FATAL); } } zip->tmp_stream_bytes_avail = 0; zip->tmp_stream_bytes_remaining = 0; zip->odd_bcj_size = 0; zip->bcj2_outPos = 0; /* * Reset a stream reader in order to read the main stream * of BCJ2. */ zip->pack_stream_remaining = 1; zip->pack_stream_index = (unsigned)folder->packIndex; zip->folder_outbytes_remaining = folder_uncompressed_size(folder); zip->uncompressed_buffer_bytes_remaining = 0; } /* * Initialize the decompressor for the new folder's pack streams. */ r = init_decompression(a, zip, coder1, coder2); if (r != ARCHIVE_OK) return (ARCHIVE_FATAL); return (ARCHIVE_OK); } static int64_t skip_stream(struct archive_read *a, size_t skip_bytes) { struct _7zip *zip = (struct _7zip *)a->format->data; const void *p; int64_t skipped_bytes; size_t bytes = skip_bytes; if (zip->folder_index == 0) { /* * Optimization for a list mode. * Avoid unnecessary decoding operations. */ zip->si.ci.folders[zip->entry->folderIndex].skipped_bytes += skip_bytes; return (skip_bytes); } while (bytes) { skipped_bytes = read_stream(a, &p, bytes, 0); if (skipped_bytes < 0) return (skipped_bytes); if (skipped_bytes == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file body"); return (ARCHIVE_FATAL); } bytes -= (size_t)skipped_bytes; if (zip->pack_stream_bytes_unconsumed) read_consume(a); } return (skip_bytes); } /* * Brought from LZMA SDK. * * Bra86.c -- Converter for x86 code (BCJ) * 2008-10-04 : Igor Pavlov : Public domain * */ #define Test86MSByte(b) ((b) == 0 || (b) == 0xFF) static void x86_Init(struct _7zip *zip) { zip->bcj_state = 0; zip->bcj_prevPosT = (size_t)0 - 1; zip->bcj_prevMask = 0; zip->bcj_ip = 5; } static size_t x86_Convert(struct _7zip *zip, uint8_t *data, size_t size) { static const uint8_t kMaskToAllowedStatus[8] = {1, 1, 1, 0, 1, 0, 0, 0}; static const uint8_t kMaskToBitNumber[8] = {0, 1, 2, 2, 3, 3, 3, 3}; size_t bufferPos, prevPosT; uint32_t ip, prevMask; if (size < 5) return 0; bufferPos = 0; prevPosT = zip->bcj_prevPosT; prevMask = zip->bcj_prevMask; ip = zip->bcj_ip; for (;;) { uint8_t *p = data + bufferPos; uint8_t *limit = data + size - 4; for (; p < limit; p++) if ((*p & 0xFE) == 0xE8) break; bufferPos = (size_t)(p - data); if (p >= limit) break; prevPosT = bufferPos - prevPosT; if (prevPosT > 3) prevMask = 0; else { prevMask = (prevMask << ((int)prevPosT - 1)) & 0x7; if (prevMask != 0) { unsigned char b = p[4 - kMaskToBitNumber[prevMask]]; if (!kMaskToAllowedStatus[prevMask] || Test86MSByte(b)) { prevPosT = bufferPos; prevMask = ((prevMask << 1) & 0x7) | 1; bufferPos++; continue; } } } prevPosT = bufferPos; if (Test86MSByte(p[4])) { uint32_t src = ((uint32_t)p[4] << 24) | ((uint32_t)p[3] << 16) | ((uint32_t)p[2] << 8) | ((uint32_t)p[1]); uint32_t dest; for (;;) { uint8_t b; int b_index; dest = src - (ip + (uint32_t)bufferPos); if (prevMask == 0) break; b_index = kMaskToBitNumber[prevMask] * 8; b = (uint8_t)(dest >> (24 - b_index)); if (!Test86MSByte(b)) break; src = dest ^ ((1 << (32 - b_index)) - 1); } p[4] = (uint8_t)(~(((dest >> 24) & 1) - 1)); p[3] = (uint8_t)(dest >> 16); p[2] = (uint8_t)(dest >> 8); p[1] = (uint8_t)dest; bufferPos += 5; } else { prevMask = ((prevMask << 1) & 0x7) | 1; bufferPos++; } } zip->bcj_prevPosT = prevPosT; zip->bcj_prevMask = prevMask; zip->bcj_ip += (uint32_t)bufferPos; return (bufferPos); } static void arm_Init(struct _7zip *zip) { zip->bcj_ip = 8; } static size_t arm_Convert(struct _7zip *zip, uint8_t *buf, size_t size) { // This function was adapted from // static size_t bcj_arm(struct xz_dec_bcj *s, uint8_t *buf, size_t size) // in https://git.tukaani.org/xz-embedded.git /* * Branch/Call/Jump (BCJ) filter decoders * * Authors: Lasse Collin * Igor Pavlov * * This file has been put into the public domain. * You can do whatever you want with this file. */ size_t i; uint32_t addr; for (i = 0; i + 4 <= size; i += 4) { if (buf[i + 3] == 0xEB) { // Calculate the transformed addr. addr = (uint32_t)buf[i] | ((uint32_t)buf[i + 1] << 8) | ((uint32_t)buf[i + 2] << 16); addr <<= 2; addr -= zip->bcj_ip + (uint32_t)i; addr >>= 2; // Store the transformed addr in buf. buf[i] = (uint8_t)addr; buf[i + 1] = (uint8_t)(addr >> 8); buf[i + 2] = (uint8_t)(addr >> 16); } } zip->bcj_ip += (uint32_t)i; return i; } static size_t arm64_Convert(struct _7zip *zip, uint8_t *buf, size_t size) { // This function was adapted from // static size_t bcj_arm64(struct xz_dec_bcj *s, uint8_t *buf, size_t size) // in https://git.tukaani.org/xz-embedded.git /* * Branch/Call/Jump (BCJ) filter decoders * * Authors: Lasse Collin * Igor Pavlov * * This file has been put into the public domain. * You can do whatever you want with this file. */ size_t i; uint32_t instr; uint32_t addr; for (i = 0; i + 4 <= size; i += 4) { instr = (uint32_t)buf[i] | ((uint32_t)buf[i+1] << 8) | ((uint32_t)buf[i+2] << 16) | ((uint32_t)buf[i+3] << 24); if ((instr >> 26) == 0x25) { /* BL instruction */ addr = instr - ((zip->bcj_ip + (uint32_t)i) >> 2); instr = 0x94000000 | (addr & 0x03FFFFFF); buf[i] = (uint8_t)instr; buf[i+1] = (uint8_t)(instr >> 8); buf[i+2] = (uint8_t)(instr >> 16); buf[i+3] = (uint8_t)(instr >> 24); } else if ((instr & 0x9F000000) == 0x90000000) { /* ADRP instruction */ addr = ((instr >> 29) & 3) | ((instr >> 3) & 0x1FFFFC); /* Only convert values in the range +/-512 MiB. */ if ((addr + 0x020000) & 0x1C0000) continue; addr -= (zip->bcj_ip + (uint32_t)i) >> 12; instr &= 0x9000001F; instr |= (addr & 3) << 29; instr |= (addr & 0x03FFFC) << 3; instr |= (0U - (addr & 0x020000)) & 0xE00000; buf[i] = (uint8_t)instr; buf[i+1] = (uint8_t)(instr >> 8); buf[i+2] = (uint8_t)(instr >> 16); buf[i+3] = (uint8_t)(instr >> 24); } } zip->bcj_ip += (uint32_t)i; return i; } static size_t sparc_Convert(struct _7zip *zip, uint8_t *buf, size_t size) { // This function was adapted from // static size_t bcj_sparc(struct xz_dec_bcj *s, uint8_t *buf, size_t size) // in https://git.tukaani.org/xz-embedded.git /* * Branch/Call/Jump (BCJ) filter decoders * * Authors: Lasse Collin * Igor Pavlov * * Copyright (C) The XZ Embedded authors and contributors * * Permission to use, copy, modify, and/or distribute this * software for any purpose with or without fee is hereby granted. * * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL * WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED * WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL * THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM * LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, * NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ size_t i; uint32_t instr; size &= ~(size_t)3; for (i = 0; i < size; i += 4) { instr = (uint32_t)(buf[i] << 24) | ((uint32_t)buf[i+1] << 16) | ((uint32_t)buf[i+2] << 8) | (uint32_t)buf[i+3]; if ((instr >> 22) == 0x100 || (instr >> 22) == 0x1FF) { instr <<= 2; instr -= zip->bcj_ip + (uint32_t)i; instr >>= 2; instr = ((uint32_t)0x40000000 - (instr & 0x400000)) | 0x40000000 | (instr & 0x3FFFFF); buf[i] = (uint8_t)(instr >> 24); buf[i+1] = (uint8_t)(instr >> 16); buf[i+2] = (uint8_t)(instr >> 8); buf[i+3] = (uint8_t)instr; } } zip->bcj_ip += (uint32_t)i; return i; } static size_t powerpc_Convert(struct _7zip *zip, uint8_t *buf, size_t size) { // This function was adapted from // static size_t powerpc_code(void *simple, uint32_t now_pos, bool is_encoder, uint8_t *buffer, size_t size) // in https://git.tukaani.org/xz.git /* * Filter for PowerPC (big endian) binaries * * Authors: Igor Pavlov * Lasse Collin * * Copyright (C) The XZ Utils authors and contributors * * Permission to use, copy, modify, and/or distribute this * software for any purpose with or without fee is hereby granted. * * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL * WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED * WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL * THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM * LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, * NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ size &= ~(size_t)3; size_t i; for (i = 0; i < size; i += 4) { // PowerPC branch 6(48) 24(Offset) 1(Abs) 1(Link) if ((buf[i] >> 2) == 0x12 && ((buf[i + 3] & 3) == 1)) { const uint32_t src = (((uint32_t)(buf[i + 0]) & 3) << 24) | ((uint32_t)(buf[i + 1]) << 16) | ((uint32_t)(buf[i + 2]) << 8) | ((uint32_t)(buf[i + 3]) & ~UINT32_C(3)); uint32_t dest = src - (zip->bcj_ip + (uint32_t)(i)); buf[i + 0] = 0x48 | ((dest >> 24) & 0x03); buf[i + 1] = (dest >> 16); buf[i + 2] = (dest >> 8); buf[i + 3] &= 0x03; buf[i + 3] |= dest; } } zip->bcj_ip += (uint32_t)i; return i; } /* * Brought from LZMA SDK. * * Bcj2.c -- Converter for x86 code (BCJ2) * 2008-10-04 : Igor Pavlov : Public domain * */ #define SZ_ERROR_DATA ARCHIVE_FAILED #define IsJcc(b0, b1) ((b0) == 0x0F && ((b1) & 0xF0) == 0x80) #define IsJ(b0, b1) ((b1 & 0xFE) == 0xE8 || IsJcc(b0, b1)) #define kNumTopBits 24 #define kTopValue ((uint32_t)1 << kNumTopBits) #define kNumBitModelTotalBits 11 #define kBitModelTotal (1 << kNumBitModelTotalBits) #define kNumMoveBits 5 #define RC_READ_BYTE (*buffer++) #define RC_TEST { if (buffer == bufferLim) return SZ_ERROR_DATA; } #define RC_INIT2 do { \ zip->bcj2_code = 0; \ zip->bcj2_range = 0xFFFFFFFF; \ { \ int ii; \ for (ii = 0; ii < 5; ii++) { \ RC_TEST; \ zip->bcj2_code = (zip->bcj2_code << 8) | RC_READ_BYTE; \ } \ } \ } while (0) #define NORMALIZE if (zip->bcj2_range < kTopValue) { RC_TEST; zip->bcj2_range <<= 8; zip->bcj2_code = (zip->bcj2_code << 8) | RC_READ_BYTE; } #define IF_BIT_0(p) ttt = *(p); bound = (zip->bcj2_range >> kNumBitModelTotalBits) * ttt; if (zip->bcj2_code < bound) #define UPDATE_0(p) zip->bcj2_range = bound; *(p) = (CProb)(ttt + ((kBitModelTotal - ttt) >> kNumMoveBits)); NORMALIZE; #define UPDATE_1(p) zip->bcj2_range -= bound; zip->bcj2_code -= bound; *(p) = (CProb)(ttt - (ttt >> kNumMoveBits)); NORMALIZE; static ssize_t Bcj2_Decode(struct _7zip *zip, uint8_t *outBuf, size_t outSize) { size_t inPos = 0, outPos = 0; const uint8_t *buf0, *buf1, *buf2, *buf3; size_t size0, size1, size2, size3; const uint8_t *buffer, *bufferLim; unsigned int i, j; size0 = zip->tmp_stream_bytes_remaining; buf0 = zip->tmp_stream_buff + zip->tmp_stream_bytes_avail - size0; size1 = zip->sub_stream_bytes_remaining[0]; buf1 = zip->sub_stream_buff[0] + zip->sub_stream_size[0] - size1; size2 = zip->sub_stream_bytes_remaining[1]; buf2 = zip->sub_stream_buff[1] + zip->sub_stream_size[1] - size2; size3 = zip->sub_stream_bytes_remaining[2]; buf3 = zip->sub_stream_buff[2] + zip->sub_stream_size[2] - size3; buffer = buf3; bufferLim = buffer + size3; if (zip->bcj_state == 0) { /* * Initialize. */ zip->bcj2_prevByte = 0; for (i = 0; i < sizeof(zip->bcj2_p) / sizeof(zip->bcj2_p[0]); i++) zip->bcj2_p[i] = kBitModelTotal >> 1; RC_INIT2; zip->bcj_state = 1; } /* * Gather the odd bytes of a previous call. */ for (i = 0; zip->odd_bcj_size > 0 && outPos < outSize; i++) { outBuf[outPos++] = zip->odd_bcj[i]; zip->odd_bcj_size--; } if (outSize == 0) { zip->bcj2_outPos += outPos; return (outPos); } for (;;) { uint8_t b; CProb *prob; uint32_t bound; uint32_t ttt; size_t limit = size0 - inPos; if (outSize - outPos < limit) limit = outSize - outPos; if (zip->bcj_state == 1) { while (limit != 0) { uint8_t bb = buf0[inPos]; outBuf[outPos++] = bb; if (IsJ(zip->bcj2_prevByte, bb)) { zip->bcj_state = 2; break; } inPos++; zip->bcj2_prevByte = bb; limit--; } } if (limit == 0 || outPos == outSize) break; zip->bcj_state = 1; b = buf0[inPos++]; if (b == 0xE8) prob = zip->bcj2_p + zip->bcj2_prevByte; else if (b == 0xE9) prob = zip->bcj2_p + 256; else prob = zip->bcj2_p + 257; IF_BIT_0(prob) { UPDATE_0(prob) zip->bcj2_prevByte = b; } else { uint32_t dest; const uint8_t *v; uint8_t out[4]; UPDATE_1(prob) if (b == 0xE8) { v = buf1; if (size1 < 4) return SZ_ERROR_DATA; buf1 += 4; size1 -= 4; } else { v = buf2; if (size2 < 4) return SZ_ERROR_DATA; buf2 += 4; size2 -= 4; } dest = (((uint32_t)v[0] << 24) | ((uint32_t)v[1] << 16) | ((uint32_t)v[2] << 8) | ((uint32_t)v[3])) - ((uint32_t)zip->bcj2_outPos + (uint32_t)outPos + 4); out[0] = (uint8_t)dest; out[1] = (uint8_t)(dest >> 8); out[2] = (uint8_t)(dest >> 16); out[3] = zip->bcj2_prevByte = (uint8_t)(dest >> 24); for (i = 0; i < 4 && outPos < outSize; i++) outBuf[outPos++] = out[i]; if (i < 4) { /* * Save odd bytes which we could not add into * the output buffer because of out of space. */ zip->odd_bcj_size = 4 -i; for (; i < 4; i++) { j = i - 4 + (unsigned)zip->odd_bcj_size; zip->odd_bcj[j] = out[i]; } break; } } } zip->tmp_stream_bytes_remaining -= inPos; zip->sub_stream_bytes_remaining[0] = size1; zip->sub_stream_bytes_remaining[1] = size2; zip->sub_stream_bytes_remaining[2] = bufferLim - buffer; zip->bcj2_outPos += outPos; return ((ssize_t)outPos); } diff --git a/contrib/libarchive/libarchive/archive_read_support_format_mtree.c b/contrib/libarchive/libarchive/archive_read_support_format_mtree.c index ba0e49de2408..ded13bee79a3 100644 --- a/contrib/libarchive/libarchive/archive_read_support_format_mtree.c +++ b/contrib/libarchive/libarchive/archive_read_support_format_mtree.c @@ -1,2155 +1,2170 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * Copyright (c) 2008 Joerg Sonnenberger * Copyright (c) 2011-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #include /* #include */ /* See archive_platform.h */ #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_CTYPE_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_private.h" +#include "archive_platform_stat.h" #include "archive_private.h" #include "archive_rb.h" #include "archive_read_private.h" #include "archive_string.h" #include "archive_pack_dev.h" #ifndef O_BINARY #define O_BINARY 0 #endif #ifndef O_CLOEXEC #define O_CLOEXEC 0 #endif #define MTREE_HAS_DEVICE 0x0001 #define MTREE_HAS_FFLAGS 0x0002 #define MTREE_HAS_GID 0x0004 #define MTREE_HAS_GNAME 0x0008 #define MTREE_HAS_MTIME 0x0010 #define MTREE_HAS_NLINK 0x0020 #define MTREE_HAS_PERM 0x0040 #define MTREE_HAS_SIZE 0x0080 #define MTREE_HAS_TYPE 0x0100 #define MTREE_HAS_UID 0x0200 #define MTREE_HAS_UNAME 0x0400 #define MTREE_HAS_OPTIONAL 0x0800 #define MTREE_HAS_NOCHANGE 0x1000 /* FreeBSD specific */ #define MAX_LINE_LEN (1024 * 1024) struct mtree_option { struct mtree_option *next; char *value; }; struct mtree_entry { struct archive_rb_node rbnode; struct mtree_entry *next_dup; struct mtree_entry *next; struct mtree_option *options; char *name; char full; char used; }; struct mtree { struct archive_string line; size_t buffsize; char *buff; int64_t offset; int fd; int archive_format; const char *archive_format_name; struct mtree_entry *entries; struct mtree_entry *this_entry; struct archive_rb_tree entry_rbtree; struct archive_string current_dir; struct archive_string contents_name; struct archive_entry_linkresolver *resolver; struct archive_rb_tree rbtree; int64_t cur_size; char checkfs; }; static int bid_keycmp(const char *, const char *, ssize_t); static int cleanup(struct archive_read *); static int detect_form(struct archive_read *, int *); static int mtree_bid(struct archive_read *, int); static int parse_file(struct archive_read *, struct archive_entry *, struct mtree *, struct mtree_entry *, int *); static void parse_escapes(char *, struct mtree_entry *); static int parse_line(struct archive_read *, struct archive_entry *, struct mtree *, struct mtree_entry *, int *); static int parse_keyword(struct archive_read *, struct mtree *, struct archive_entry *, struct mtree_option *, int *); static int read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset); static ssize_t readline(struct archive_read *, struct mtree *, char **, ssize_t); static int skip(struct archive_read *a); static int read_header(struct archive_read *, struct archive_entry *); static int64_t mtree_atol(char **, int base); #ifndef HAVE_STRNLEN static size_t mtree_strnlen(const char *, size_t); #endif /* * There's no standard for TIME_T_MAX/TIME_T_MIN. So we compute them * here. TODO: Move this to configure time, but be careful * about cross-compile environments. */ static int64_t get_time_t_max(void) { #if defined(TIME_T_MAX) return TIME_T_MAX; #else /* ISO C allows time_t to be a floating-point type, but POSIX requires an integer type. The following should work on any system that follows the POSIX conventions. */ if (((time_t)0) < ((time_t)-1)) { /* Time_t is unsigned */ return (~(time_t)0); } else { /* Time_t is signed. */ /* Assume it's the same as int64_t or int32_t */ if (sizeof(time_t) == sizeof(int64_t)) { return (time_t)INT64_MAX; } else { return (time_t)INT32_MAX; } } #endif } static int64_t get_time_t_min(void) { #if defined(TIME_T_MIN) return TIME_T_MIN; #else if (((time_t)0) < ((time_t)-1)) { /* Time_t is unsigned */ return (time_t)0; } else { /* Time_t is signed. */ if (sizeof(time_t) == sizeof(int64_t)) { return (time_t)INT64_MIN; } else { return (time_t)INT32_MIN; } } #endif } #ifdef HAVE_STRNLEN #define mtree_strnlen(a,b) strnlen(a,b) #else static size_t mtree_strnlen(const char *p, size_t maxlen) { size_t i; for (i = 0; i <= maxlen; i++) { if (p[i] == 0) break; } if (i > maxlen) return (-1);/* invalid */ return (i); } #endif static int archive_read_format_mtree_options(struct archive_read *a, const char *key, const char *val) { struct mtree *mtree; mtree = (struct mtree *)(a->format->data); if (strcmp(key, "checkfs") == 0) { /* Allows to read information missing from the mtree from the file system */ if (val == NULL || val[0] == 0) { mtree->checkfs = 0; } else { mtree->checkfs = 1; } return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static void free_options(struct mtree_option *head) { struct mtree_option *next; for (; head != NULL; head = next) { next = head->next; free(head->value); free(head); } } static int mtree_cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct mtree_entry *e1 = (const struct mtree_entry *)n1; const struct mtree_entry *e2 = (const struct mtree_entry *)n2; return (strcmp(e1->name, e2->name)); } static int mtree_cmp_key(const struct archive_rb_node *n, const void *key) { const struct mtree_entry *e = (const struct mtree_entry *)n; return (strcmp(e->name, key)); } int archive_read_support_format_mtree(struct archive *_a) { static const struct archive_rb_tree_ops rb_ops = { mtree_cmp_node, mtree_cmp_key, }; struct archive_read *a = (struct archive_read *)_a; struct mtree *mtree; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_mtree"); mtree = calloc(1, sizeof(*mtree)); if (mtree == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate mtree data"); return (ARCHIVE_FATAL); } mtree->checkfs = 0; mtree->fd = -1; __archive_rb_tree_init(&mtree->rbtree, &rb_ops); r = __archive_read_register_format(a, mtree, "mtree", mtree_bid, archive_read_format_mtree_options, read_header, read_data, skip, NULL, cleanup, NULL, NULL); if (r != ARCHIVE_OK) free(mtree); return (ARCHIVE_OK); } static int cleanup(struct archive_read *a) { struct mtree *mtree; struct mtree_entry *p, *q; mtree = (struct mtree *)(a->format->data); p = mtree->entries; while (p != NULL) { q = p->next; free(p->name); free_options(p->options); free(p); p = q; } archive_string_free(&mtree->line); archive_string_free(&mtree->current_dir); archive_string_free(&mtree->contents_name); archive_entry_linkresolver_free(mtree->resolver); free(mtree->buff); free(mtree); (a->format->data) = NULL; return (ARCHIVE_OK); } static ssize_t get_line_size(const char *b, ssize_t avail, ssize_t *nlsize) { ssize_t len; len = 0; while (len < avail) { switch (*b) { case '\0':/* Non-ascii character or control character. */ if (nlsize != NULL) *nlsize = 0; return (-1); case '\r': if (avail-len > 1 && b[1] == '\n') { if (nlsize != NULL) *nlsize = 2; return (len+2); } /* FALL THROUGH */ case '\n': if (nlsize != NULL) *nlsize = 1; return (len+1); default: b++; len++; break; } } if (nlsize != NULL) *nlsize = 0; return (avail); } /* * <---------------- ravail ---------------------> * <-- diff ------> <--- avail -----------------> * <---- len -----------> * | Previous lines | line being parsed nl extra | * ^ * b * */ static ssize_t next_line(struct archive_read *a, const char **b, ssize_t *avail, ssize_t *ravail, ssize_t *nl) { ssize_t len; int quit; quit = 0; if (*avail == 0) { *nl = 0; len = 0; } else len = get_line_size(*b, *avail, nl); /* * Read bytes more while it does not reach the end of line. */ while (*nl == 0 && len == *avail && !quit) { ssize_t diff = *ravail - *avail; size_t nbytes_req = (*ravail+1023) & ~1023U; ssize_t tested; /* * Place an arbitrary limit on the line length. * mtree is almost free-form input and without line length limits, * it can consume a lot of memory. */ if (len >= MAX_LINE_LEN) return (-1); /* Increase reading bytes if it is not enough to at least * new two lines. */ if (nbytes_req < (size_t)*ravail + 160) nbytes_req <<= 1; *b = __archive_read_ahead(a, nbytes_req, avail); if (*b == NULL) { if (*ravail >= *avail) return (0); /* Reading bytes reaches the end of file. */ *b = __archive_read_ahead(a, *avail, avail); quit = 1; } *ravail = *avail; *b += diff; *avail -= diff; tested = len;/* Skip some bytes we already determined. */ len = get_line_size(*b + len, *avail - len, nl); if (len >= 0) len += tested; } return (len); } /* * Compare characters with an mtree keyword. * Returns the length of an mtree keyword if matched. * Returns 0 if not matched. */ static int bid_keycmp(const char *p, const char *key, ssize_t len) { int match_len = 0; while (len > 0 && *p && *key) { if (*p == *key) { --len; ++p; ++key; ++match_len; continue; } return (0);/* Not match */ } if (*key != '\0') return (0);/* Not match */ /* A following character should be specified characters */ if (p[0] == '=' || p[0] == ' ' || p[0] == '\t' || p[0] == '\n' || p[0] == '\r' || (p[0] == '\\' && (p[1] == '\n' || p[1] == '\r'))) return (match_len); return (0);/* Not match */ } /* * Test whether the characters 'p' has is mtree keyword. * Returns the length of a detected keyword. * Returns 0 if any keywords were not found. */ static int bid_keyword(const char *p, ssize_t len) { static const char * const keys_c[] = { "content", "contents", "cksum", NULL }; static const char * const keys_df[] = { "device", "flags", NULL }; static const char * const keys_g[] = { "gid", "gname", NULL }; static const char * const keys_il[] = { "ignore", "inode", "link", NULL }; static const char * const keys_m[] = { "md5", "md5digest", "mode", NULL }; static const char * const keys_no[] = { "nlink", "nochange", "optional", NULL }; static const char * const keys_r[] = { "resdevice", "rmd160", "rmd160digest", NULL }; static const char * const keys_s[] = { "sha1", "sha1digest", "sha256", "sha256digest", "sha384", "sha384digest", "sha512", "sha512digest", "size", NULL }; static const char * const keys_t[] = { "tags", "time", "type", NULL }; static const char * const keys_u[] = { "uid", "uname", NULL }; const char * const *keys; int i; switch (*p) { case 'c': keys = keys_c; break; case 'd': case 'f': keys = keys_df; break; case 'g': keys = keys_g; break; case 'i': case 'l': keys = keys_il; break; case 'm': keys = keys_m; break; case 'n': case 'o': keys = keys_no; break; case 'r': keys = keys_r; break; case 's': keys = keys_s; break; case 't': keys = keys_t; break; case 'u': keys = keys_u; break; default: return (0);/* Unknown key */ } for (i = 0; keys[i] != NULL; i++) { int l = bid_keycmp(p, keys[i], len); if (l > 0) return (l); } return (0);/* Unknown key */ } /* * Test whether there is a set of mtree keywords. * Returns the number of keywords. * Returns -1 if we got incorrect sequence. * This function expects a set of "keyword=value". * When "unset" is specified, expects a set of "keyword". */ static int bid_keyword_list(const char *p, ssize_t len, int unset, int last_is_path) { int l; int keycnt = 0; while (len > 0 && *p) { int blank = 0; /* Test whether there are blank characters in the line. */ while (len >0 && (*p == ' ' || *p == '\t')) { ++p; --len; blank = 1; } if (*p == '\n' || *p == '\r') break; if (p[0] == '\\' && (p[1] == '\n' || p[1] == '\r')) break; if (!blank && !last_is_path) /* No blank character. */ return (-1); if (last_is_path && len == 0) return (keycnt); if (unset) { l = bid_keycmp(p, "all", len); if (l > 0) return (1); } /* Test whether there is a correct key in the line. */ l = bid_keyword(p, len); if (l == 0) return (-1);/* Unknown keyword was found. */ p += l; len -= l; keycnt++; /* Skip value */ if (*p == '=') { int value = 0; ++p; --len; while (len > 0 && *p != ' ' && *p != '\t') { ++p; --len; value = 1; } /* A keyword should have a its value unless * "/unset" operation. */ if (!unset && value == 0) return (-1); } } return (keycnt); } static int bid_entry(const char *p, ssize_t len, ssize_t nl, int *last_is_path) { int f = 0; static const unsigned char safe_char[256] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ /* !"$%&'()*+,-./ EXCLUSION:( )(#) */ 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 20 - 2F */ /* 0123456789:;<>? EXCLUSION:(=) */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, /* 30 - 3F */ /* @ABCDEFGHIJKLMNO */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 40 - 4F */ /* PQRSTUVWXYZ[\]^_ */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 50 - 5F */ /* `abcdefghijklmno */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 60 - 6F */ /* pqrstuvwxyz{|}~ */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, /* 70 - 7F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ }; ssize_t ll; const char *pp = p; const char * const pp_end = pp + len; *last_is_path = 0; /* * Skip the path-name which is quoted. */ for (;pp < pp_end; ++pp) { if (!safe_char[*(const unsigned char *)pp]) { if (*pp != ' ' && *pp != '\t' && *pp != '\r' && *pp != '\n') f = 0; break; } f = 1; } ll = pp_end - pp; /* If a path-name was not found at the first, try to check * a mtree format(a.k.a form D) ``NetBSD's mtree -D'' creates, * which places the path-name at the last. */ if (f == 0) { const char *pb = p + len - nl; int name_len = 0; int slash; /* The form D accepts only a single line for an entry. */ if (pb-2 >= p && pb[-1] == '\\' && (pb[-2] == ' ' || pb[-2] == '\t')) return (-1); if (pb-1 >= p && pb[-1] == '\\') return (-1); slash = 0; while (p <= --pb && *pb != ' ' && *pb != '\t') { if (!safe_char[*(const unsigned char *)pb]) return (-1); name_len++; /* The pathname should have a slash in this * format. */ if (*pb == '/') slash = 1; } if (name_len == 0 || slash == 0) return (-1); /* If '/' is placed at the first in this field, this is not * a valid filename. */ if (pb[1] == '/') return (-1); ll = len - nl - name_len; pp = p; *last_is_path = 1; } return (bid_keyword_list(pp, ll, 0, *last_is_path)); } #define MAX_BID_ENTRY 3 static int mtree_bid(struct archive_read *a, int best_bid) { const char *signature = "#mtree"; const char *p; (void)best_bid; /* UNUSED */ /* Now let's look at the actual header and see if it matches. */ p = __archive_read_ahead(a, strlen(signature), NULL); if (p == NULL) return (-1); if (memcmp(p, signature, strlen(signature)) == 0) return (8 * (int)strlen(signature)); /* * There is not a mtree signature. Let's try to detect mtree format. */ return (detect_form(a, NULL)); } static int detect_form(struct archive_read *a, int *is_form_d) { const char *p; ssize_t avail, ravail; ssize_t len, nl; int entry_cnt = 0, multiline = 0; int form_D = 0;/* The archive is generated by `NetBSD mtree -D' * (In this source we call it `form D') . */ if (is_form_d != NULL) *is_form_d = 0; p = __archive_read_ahead(a, 1, &avail); if (p == NULL) return (-1); ravail = avail; for (;;) { len = next_line(a, &p, &avail, &ravail, &nl); /* The terminal character of the line should be * a new line character, '\r\n' or '\n'. */ if (len <= 0 || nl == 0) break; if (!multiline) { /* Leading whitespace is never significant, * ignore it. */ while (len > 0 && (*p == ' ' || *p == '\t')) { ++p; --avail; --len; } /* Skip comment or empty line. */ if (p[0] == '#' || p[0] == '\n' || p[0] == '\r') { p += len; avail -= len; continue; } } else { /* A continuance line; the terminal * character of previous line was '\' character. */ if (bid_keyword_list(p, len, 0, 0) <= 0) break; if (p[len-nl-1] != '\\') { if (multiline == 1 && ++entry_cnt >= MAX_BID_ENTRY) break; multiline = 0; } p += len; avail -= len; continue; } if (p[0] != '/') { int last_is_path, keywords; keywords = bid_entry(p, len, nl, &last_is_path); if (keywords >= 0) { if (form_D == 0) { if (last_is_path) form_D = 1; else if (keywords > 0) /* This line is not `form D'. */ form_D = -1; } else if (form_D == 1) { if (!last_is_path && keywords > 0) /* This this is not `form D' * and We cannot accept mixed * format. */ break; } if (!last_is_path && p[len-nl-1] == '\\') /* This line continues. */ multiline = 1; else { /* We've got plenty of correct lines * to assume that this file is an mtree * format. */ if (++entry_cnt >= MAX_BID_ENTRY) break; } } else break; } else if (len > 4 && strncmp(p, "/set", 4) == 0) { if (bid_keyword_list(p+4, len-4, 0, 0) <= 0) break; /* This line continues. */ if (p[len-nl-1] == '\\') multiline = 2; } else if (len > 6 && strncmp(p, "/unset", 6) == 0) { if (bid_keyword_list(p+6, len-6, 1, 0) <= 0) break; /* This line continues. */ if (p[len-nl-1] == '\\') multiline = 2; } else break; /* Test next line. */ p += len; avail -= len; } if (entry_cnt >= MAX_BID_ENTRY || (entry_cnt > 0 && len == 0)) { if (is_form_d != NULL) { if (form_D == 1) *is_form_d = 1; } return (32); } return (0); } /* * The extended mtree format permits multiple lines specifying * attributes for each file. For those entries, only the last line * is actually used. Practically speaking, that means we have * to read the entire mtree file into memory up front. * * The parsing is done in two steps. First, it is decided if a line * changes the global defaults and if it is, processed accordingly. * Otherwise, the options of the line are merged with the current * global options. */ static int add_option(struct archive_read *a, struct mtree_option **global, const char *value, size_t len) { struct mtree_option *opt; if ((opt = malloc(sizeof(*opt))) == NULL) { archive_set_error(&a->archive, errno, "Can't allocate memory"); return (ARCHIVE_FATAL); } if ((opt->value = malloc(len + 1)) == NULL) { free(opt); archive_set_error(&a->archive, errno, "Can't allocate memory"); return (ARCHIVE_FATAL); } memcpy(opt->value, value, len); opt->value[len] = '\0'; opt->next = *global; *global = opt; return (ARCHIVE_OK); } static void remove_option(struct mtree_option **global, const char *value, size_t len) { struct mtree_option *iter, *last; last = NULL; for (iter = *global; iter != NULL; last = iter, iter = iter->next) { if (strncmp(iter->value, value, len) == 0 && (iter->value[len] == '\0' || iter->value[len] == '=')) break; } if (iter == NULL) return; if (last == NULL) *global = iter->next; else last->next = iter->next; free(iter->value); free(iter); } static int process_global_set(struct archive_read *a, struct mtree_option **global, const char *line) { const char *next, *eq; size_t len; int r; line += 4; for (;;) { next = line + strspn(line, " \t\r\n"); if (*next == '\0') return (ARCHIVE_OK); line = next; next = line + strcspn(line, " \t\r\n"); eq = strchr(line, '='); if (eq > next) len = next - line; else len = eq - line; remove_option(global, line, len); r = add_option(a, global, line, next - line); if (r != ARCHIVE_OK) return (r); line = next; } } static int process_global_unset(struct archive_read *a, struct mtree_option **global, const char *line) { const char *next; size_t len; line += 6; if (strchr(line, '=') != NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "/unset shall not contain `='"); return ARCHIVE_FATAL; } for (;;) { next = line + strspn(line, " \t\r\n"); if (*next == '\0') return (ARCHIVE_OK); line = next; len = strcspn(line, " \t\r\n"); if (len == 3 && strncmp(line, "all", 3) == 0) { free_options(*global); *global = NULL; } else { remove_option(global, line, len); } line += len; } } static int process_add_entry(struct archive_read *a, struct mtree *mtree, struct mtree_option **global, const char *line, ssize_t line_len, struct mtree_entry **last_entry, int is_form_d) { struct mtree_entry *entry; struct mtree_option *iter; const char *next, *eq, *name, *end; size_t name_len, len; int r, i; if ((entry = malloc(sizeof(*entry))) == NULL) { archive_set_error(&a->archive, errno, "Can't allocate memory"); return (ARCHIVE_FATAL); } entry->next = NULL; entry->options = NULL; entry->name = NULL; entry->used = 0; entry->full = 0; /* Add this entry to list. */ if (*last_entry == NULL) mtree->entries = entry; else (*last_entry)->next = entry; *last_entry = entry; if (is_form_d) { /* Filename is last item on line. */ /* Adjust line_len to trim trailing whitespace */ while (line_len > 0) { char last_character = line[line_len - 1]; if (last_character == '\r' || last_character == '\n' || last_character == '\t' || last_character == ' ') { line_len--; } else { break; } } /* Name starts after the last whitespace separator */ name = line; for (i = 0; i < line_len; i++) { if (line[i] == '\r' || line[i] == '\n' || line[i] == '\t' || line[i] == ' ') { name = line + i + 1; } } name_len = line + line_len - name; end = name; } else { /* Filename is first item on line */ name_len = strcspn(line, " \t\r\n"); name = line; line += name_len; end = line + line_len; } /* name/name_len is the name within the line. */ /* line..end brackets the entire line except the name */ if ((entry->name = malloc(name_len + 1)) == NULL) { archive_set_error(&a->archive, errno, "Can't allocate memory"); return (ARCHIVE_FATAL); } memcpy(entry->name, name, name_len); entry->name[name_len] = '\0'; parse_escapes(entry->name, entry); entry->next_dup = NULL; if (entry->full) { if (!__archive_rb_tree_insert_node(&mtree->rbtree, &entry->rbnode)) { struct mtree_entry *alt; alt = (struct mtree_entry *)__archive_rb_tree_find_node( &mtree->rbtree, entry->name); if (alt != NULL) { while (alt->next_dup) alt = alt->next_dup; alt->next_dup = entry; } } } for (iter = *global; iter != NULL; iter = iter->next) { r = add_option(a, &entry->options, iter->value, strlen(iter->value)); if (r != ARCHIVE_OK) return (r); } for (;;) { next = line + strspn(line, " \t\r\n"); if (*next == '\0') return (ARCHIVE_OK); if (next >= end) return (ARCHIVE_OK); line = next; next = line + strcspn(line, " \t\r\n"); eq = strchr(line, '='); if (eq == NULL || eq > next) len = next - line; else len = eq - line; remove_option(&entry->options, line, len); r = add_option(a, &entry->options, line, next - line); if (r != ARCHIVE_OK) return (r); line = next; } } static int read_mtree(struct archive_read *a, struct mtree *mtree) { ssize_t len; uintmax_t counter; char *p, *s; struct mtree_option *global; struct mtree_entry *last_entry; int r, is_form_d; mtree->archive_format = ARCHIVE_FORMAT_MTREE; mtree->archive_format_name = "mtree"; global = NULL; last_entry = NULL; (void)detect_form(a, &is_form_d); for (counter = 1; ; ++counter) { r = ARCHIVE_OK; len = readline(a, mtree, &p, 65536); if (len == 0) { mtree->this_entry = mtree->entries; free_options(global); return (ARCHIVE_OK); } if (len < 0) { free_options(global); return ((int)len); } /* Leading whitespace is never significant, ignore it. */ while (*p == ' ' || *p == '\t') { ++p; --len; } /* Skip content lines and blank lines. */ if (*p == '#') continue; if (*p == '\r' || *p == '\n' || *p == '\0') continue; /* Non-printable characters are not allowed */ for (s = p;s < p + len - 1; s++) { if (!isprint((unsigned char)*s) && *s != '\t') { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Non-printable character 0x%02X", (unsigned char)(*s)); r = ARCHIVE_FATAL; break; } } if (r != ARCHIVE_OK) break; if (*p != '/') { r = process_add_entry(a, mtree, &global, p, len, &last_entry, is_form_d); } else if (len > 4 && strncmp(p, "/set", 4) == 0) { if (p[4] != ' ' && p[4] != '\t') break; r = process_global_set(a, &global, p); } else if (len > 6 && strncmp(p, "/unset", 6) == 0) { if (p[6] != ' ' && p[6] != '\t') break; r = process_global_unset(a, &global, p); } else break; if (r != ARCHIVE_OK) { free_options(global); return r; } } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't parse line %ju", counter); free_options(global); return (ARCHIVE_FATAL); } /* * Read in the entire mtree file into memory on the first request. * Then use the next unused file to satisfy each header request. */ static int read_header(struct archive_read *a, struct archive_entry *entry) { struct mtree *mtree; char *p; int r, use_next; mtree = (struct mtree *)(a->format->data); if (mtree->fd >= 0) { close(mtree->fd); mtree->fd = -1; } if (mtree->entries == NULL) { mtree->resolver = archive_entry_linkresolver_new(); if (mtree->resolver == NULL) return ARCHIVE_FATAL; archive_entry_linkresolver_set_strategy(mtree->resolver, ARCHIVE_FORMAT_MTREE); r = read_mtree(a, mtree); if (r != ARCHIVE_OK) return (r); } a->archive.archive_format = mtree->archive_format; a->archive.archive_format_name = mtree->archive_format_name; for (;;) { if (mtree->this_entry == NULL) return (ARCHIVE_EOF); if (strcmp(mtree->this_entry->name, "..") == 0) { mtree->this_entry->used = 1; if (archive_strlen(&mtree->current_dir) > 0) { /* Roll back current path. */ p = mtree->current_dir.s + mtree->current_dir.length - 1; while (p >= mtree->current_dir.s && *p != '/') --p; if (p >= mtree->current_dir.s) --p; mtree->current_dir.length = p - mtree->current_dir.s + 1; } } if (!mtree->this_entry->used) { use_next = 0; r = parse_file(a, entry, mtree, mtree->this_entry, &use_next); if (use_next == 0) return (r); } mtree->this_entry = mtree->this_entry->next; } } /* * A single file can have multiple lines contribute specifications. * Parse as many lines as necessary, then pull additional information * from a backing file on disk as necessary. */ static int parse_file(struct archive_read *a, struct archive_entry *entry, struct mtree *mtree, struct mtree_entry *mentry, int *use_next) { const char *path; - struct stat st_storage, *st; + la_seek_stat_t st_storage, *st; struct mtree_entry *mp; struct archive_entry *sparse_entry; int r = ARCHIVE_OK, r1, parsed_kws; mentry->used = 1; /* Initialize reasonable defaults. */ archive_entry_set_filetype(entry, AE_IFREG); archive_entry_set_size(entry, 0); archive_string_empty(&mtree->contents_name); /* Parse options from this line. */ parsed_kws = 0; r = parse_line(a, entry, mtree, mentry, &parsed_kws); if (mentry->full) { archive_entry_copy_pathname(entry, mentry->name); /* * "Full" entries are allowed to have multiple lines * and those lines aren't required to be adjacent. We * don't support multiple lines for "relative" entries * nor do we make any attempt to merge data from * separate "relative" and "full" entries. (Merging * "relative" and "full" entries would require dealing * with pathname canonicalization, which is a very * tricky subject.) */ mp = (struct mtree_entry *)__archive_rb_tree_find_node( &mtree->rbtree, mentry->name); for (; mp; mp = mp->next_dup) { if (mp->full && !mp->used) { /* Later lines override earlier ones. */ mp->used = 1; r1 = parse_line(a, entry, mtree, mp, &parsed_kws); if (r1 < r) r = r1; } } } else { /* * Relative entries require us to construct * the full path and possibly update the * current directory. */ size_t n = archive_strlen(&mtree->current_dir); if (n > 0) archive_strcat(&mtree->current_dir, "/"); archive_strcat(&mtree->current_dir, mentry->name); archive_entry_copy_pathname(entry, mtree->current_dir.s); if (archive_entry_filetype(entry) != AE_IFDIR) mtree->current_dir.length = n; } if (mtree->checkfs) { /* * Try to open and stat the file to get the real size * and other file info. It would be nice to avoid * this here so that getting a listing of an mtree * wouldn't require opening every referenced contents * file. But then we wouldn't know the actual * contents size, so I don't see a really viable way * around this. (Also, we may want to someday pull * other unspecified info from the contents file on * disk.) */ mtree->fd = -1; if (archive_strlen(&mtree->contents_name) > 0) path = mtree->contents_name.s; else path = archive_entry_pathname(entry); if (archive_entry_filetype(entry) == AE_IFREG || archive_entry_filetype(entry) == AE_IFDIR) { mtree->fd = open(path, O_RDONLY | O_BINARY | O_CLOEXEC); __archive_ensure_cloexec_flag(mtree->fd); - if (mtree->fd == -1 && ( + if (mtree->fd < 0 && ( #if defined(_WIN32) && !defined(__CYGWIN__) /* * On Windows, attempting to open a file with an * invalid name result in EINVAL (Error 22) */ (errno != ENOENT && errno != EINVAL) #else errno != ENOENT #endif || archive_strlen(&mtree->contents_name) > 0)) { archive_set_error(&a->archive, errno, "Can't open %s", path); r = ARCHIVE_WARN; } } st = &st_storage; if (mtree->fd >= 0) { - if (fstat(mtree->fd, st) == -1) { + if (la_seek_fstat(mtree->fd, st) == -1) { archive_set_error(&a->archive, errno, "Could not fstat %s", path); r = ARCHIVE_WARN; /* If we can't stat it, don't keep it open. */ close(mtree->fd); mtree->fd = -1; st = NULL; } } #ifdef HAVE_LSTAT else if (lstat(path, st) == -1) #else - else if (la_stat(path, st) == -1) + else if (la_seek_stat(path, st) == -1) #endif { st = NULL; } /* * Check for a mismatch between the type in the specification * and the type of the contents object on disk. */ if (st != NULL) { if (((st->st_mode & S_IFMT) == S_IFREG && archive_entry_filetype(entry) == AE_IFREG) #ifdef S_IFLNK ||((st->st_mode & S_IFMT) == S_IFLNK && archive_entry_filetype(entry) == AE_IFLNK) #endif #ifdef S_IFSOCK ||((st->st_mode & S_IFSOCK) == S_IFSOCK && archive_entry_filetype(entry) == AE_IFSOCK) #endif #ifdef S_IFCHR ||((st->st_mode & S_IFMT) == S_IFCHR && archive_entry_filetype(entry) == AE_IFCHR) #endif #ifdef S_IFBLK ||((st->st_mode & S_IFMT) == S_IFBLK && archive_entry_filetype(entry) == AE_IFBLK) #endif ||((st->st_mode & S_IFMT) == S_IFDIR && archive_entry_filetype(entry) == AE_IFDIR) #ifdef S_IFIFO ||((st->st_mode & S_IFMT) == S_IFIFO && archive_entry_filetype(entry) == AE_IFIFO) #endif ) { /* Types match. */ } else { /* Types don't match; bail out gracefully. */ if (mtree->fd >= 0) close(mtree->fd); mtree->fd = -1; if (parsed_kws & MTREE_HAS_OPTIONAL) { /* It's not an error for an optional * entry to not match disk. */ *use_next = 1; } else if (r == ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "mtree specification has different" " type for %s", archive_entry_pathname(entry)); r = ARCHIVE_WARN; } return (r); } } /* * If there is a contents file on disk, pick some of the * metadata from that file. For most of these, we only * set it from the contents if it wasn't already parsed * from the specification. */ if (st != NULL) { if (((parsed_kws & MTREE_HAS_DEVICE) == 0 || (parsed_kws & MTREE_HAS_NOCHANGE) != 0) && (archive_entry_filetype(entry) == AE_IFCHR || archive_entry_filetype(entry) == AE_IFBLK)) archive_entry_set_rdev(entry, st->st_rdev); if ((parsed_kws & (MTREE_HAS_GID | MTREE_HAS_GNAME)) == 0 || (parsed_kws & MTREE_HAS_NOCHANGE) != 0) archive_entry_set_gid(entry, st->st_gid); if ((parsed_kws & (MTREE_HAS_UID | MTREE_HAS_UNAME)) == 0 || (parsed_kws & MTREE_HAS_NOCHANGE) != 0) archive_entry_set_uid(entry, st->st_uid); if ((parsed_kws & MTREE_HAS_MTIME) == 0 || (parsed_kws & MTREE_HAS_NOCHANGE) != 0) { #if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC archive_entry_set_mtime(entry, st->st_mtime, st->st_mtimespec.tv_nsec); #elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC archive_entry_set_mtime(entry, st->st_mtime, st->st_mtim.tv_nsec); #elif HAVE_STRUCT_STAT_ST_MTIME_N archive_entry_set_mtime(entry, st->st_mtime, st->st_mtime_n); #elif HAVE_STRUCT_STAT_ST_UMTIME archive_entry_set_mtime(entry, st->st_mtime, st->st_umtime*1000); #elif HAVE_STRUCT_STAT_ST_MTIME_USEC archive_entry_set_mtime(entry, st->st_mtime, st->st_mtime_usec*1000); #else archive_entry_set_mtime(entry, st->st_mtime, 0); #endif } if ((parsed_kws & MTREE_HAS_NLINK) == 0 || (parsed_kws & MTREE_HAS_NOCHANGE) != 0) archive_entry_set_nlink(entry, st->st_nlink); if ((parsed_kws & MTREE_HAS_PERM) == 0 || (parsed_kws & MTREE_HAS_NOCHANGE) != 0) archive_entry_set_perm(entry, st->st_mode); if ((parsed_kws & MTREE_HAS_SIZE) == 0 || (parsed_kws & MTREE_HAS_NOCHANGE) != 0) archive_entry_set_size(entry, st->st_size); archive_entry_set_ino(entry, st->st_ino); archive_entry_set_dev(entry, st->st_dev); archive_entry_linkify(mtree->resolver, &entry, &sparse_entry); } else if (parsed_kws & MTREE_HAS_OPTIONAL) { /* * Couldn't open the entry, stat it or the on-disk type * didn't match. If this entry is optional, just * ignore it and read the next header entry. */ *use_next = 1; return ARCHIVE_OK; } } mtree->cur_size = archive_entry_size(entry); mtree->offset = 0; return r; } /* * Each line contains a sequence of keywords. */ static int parse_line(struct archive_read *a, struct archive_entry *entry, struct mtree *mtree, struct mtree_entry *mp, int *parsed_kws) { struct mtree_option *iter; int r = ARCHIVE_OK, r1; for (iter = mp->options; iter != NULL; iter = iter->next) { r1 = parse_keyword(a, mtree, entry, iter, parsed_kws); if (r1 < r) r = r1; } if (r == ARCHIVE_OK && (*parsed_kws & MTREE_HAS_TYPE) == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Missing type keyword in mtree specification"); return (ARCHIVE_WARN); } return (r); } /* * Device entries have one of the following forms: * - raw dev_t * - format,major,minor[,subdevice] * When parsing succeeded, `pdev' will contain the appropriate dev_t value. */ /* strsep() is not in C90, but strcspn() is. */ /* Taken from http://unixpapa.com/incnote/string.html */ static char * la_strsep(char **sp, const char *sep) { char *p, *s; if (sp == NULL || *sp == NULL || **sp == '\0') return(NULL); s = *sp; p = s + strcspn(s, sep); if (*p != '\0') *p++ = '\0'; *sp = p; return(s); } static int parse_device(dev_t *pdev, struct archive *a, char *val) { #define MAX_PACK_ARGS 3 unsigned long numbers[MAX_PACK_ARGS]; char *p, *dev; int argc; pack_t *pack; dev_t result; const char *error = NULL; memset(pdev, 0, sizeof(*pdev)); if ((dev = strchr(val, ',')) != NULL) { /* * Device's major/minor are given in a specified format. * Decode and pack it accordingly. */ *dev++ = '\0'; if ((pack = pack_find(val)) == NULL) { archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, "Unknown format `%s'", val); return ARCHIVE_WARN; } argc = 0; while ((p = la_strsep(&dev, ",")) != NULL) { if (*p == '\0') { archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, "Missing number"); return ARCHIVE_WARN; } if (argc >= MAX_PACK_ARGS) { archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, "Too many arguments"); return ARCHIVE_WARN; } numbers[argc++] = (unsigned long)mtree_atol(&p, 0); } if (argc < 2) { archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, "Not enough arguments"); return ARCHIVE_WARN; } result = (*pack)(argc, numbers, &error); if (error != NULL) { archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, "%s", error); return ARCHIVE_WARN; } } else { /* file system raw value. */ result = (dev_t)mtree_atol(&val, 0); } *pdev = result; return ARCHIVE_OK; #undef MAX_PACK_ARGS } static int parse_hex_nibble(char c) { if (c >= '0' && c <= '9') return c - '0'; if (c >= 'a' && c <= 'f') return 10 + c - 'a'; #if 0 /* XXX: Is uppercase something we should support? */ if (c >= 'A' && c <= 'F') return 10 + c - 'A'; #endif return -1; } static int parse_digest(struct archive_read *a, struct archive_entry *entry, const char *digest, int type) { unsigned char digest_buf[64]; int high, low; size_t i, j, len; switch (type) { case ARCHIVE_ENTRY_DIGEST_MD5: len = sizeof(entry->digest.md5); break; case ARCHIVE_ENTRY_DIGEST_RMD160: len = sizeof(entry->digest.rmd160); break; case ARCHIVE_ENTRY_DIGEST_SHA1: len = sizeof(entry->digest.sha1); break; case ARCHIVE_ENTRY_DIGEST_SHA256: len = sizeof(entry->digest.sha256); break; case ARCHIVE_ENTRY_DIGEST_SHA384: len = sizeof(entry->digest.sha384); break; case ARCHIVE_ENTRY_DIGEST_SHA512: len = sizeof(entry->digest.sha512); break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Internal error: Unknown digest type"); return ARCHIVE_FATAL; } if (len > sizeof(digest_buf)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Internal error: Digest storage too large"); return ARCHIVE_FATAL; } len *= 2; if (mtree_strnlen(digest, len+1) != len) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "incorrect digest length, ignoring"); return ARCHIVE_WARN; } for (i = 0, j = 0; i < len; i += 2, j++) { high = parse_hex_nibble(digest[i]); low = parse_hex_nibble(digest[i+1]); if (high == -1 || low == -1) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "invalid digest data, ignoring"); return ARCHIVE_WARN; } digest_buf[j] = high << 4 | low; } return archive_entry_set_digest(entry, type, digest_buf); } /* * Parse a single keyword and its value. */ static int parse_keyword(struct archive_read *a, struct mtree *mtree, struct archive_entry *entry, struct mtree_option *opt, int *parsed_kws) { char *val, *key; key = opt->value; if (*key == '\0') return (ARCHIVE_OK); if (strcmp(key, "nochange") == 0) { *parsed_kws |= MTREE_HAS_NOCHANGE; return (ARCHIVE_OK); } if (strcmp(key, "optional") == 0) { *parsed_kws |= MTREE_HAS_OPTIONAL; return (ARCHIVE_OK); } if (strcmp(key, "ignore") == 0) { /* * The mtree processing is not recursive, so * recursion will only happen for explicitly listed * entries. */ return (ARCHIVE_OK); } val = strchr(key, '='); if (val == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Malformed attribute \"%s\" (%d)", key, key[0]); return (ARCHIVE_WARN); } *val = '\0'; ++val; switch (key[0]) { case 'c': if (strcmp(key, "content") == 0 || strcmp(key, "contents") == 0) { parse_escapes(val, NULL); archive_strcpy(&mtree->contents_name, val); return (ARCHIVE_OK); } if (strcmp(key, "cksum") == 0) return (ARCHIVE_OK); break; case 'd': if (strcmp(key, "device") == 0) { /* stat(2) st_rdev field, e.g. the major/minor IDs * of a char/block special file */ int r; dev_t dev; *parsed_kws |= MTREE_HAS_DEVICE; r = parse_device(&dev, &a->archive, val); if (r == ARCHIVE_OK) archive_entry_set_rdev(entry, dev); return r; } break; case 'f': if (strcmp(key, "flags") == 0) { *parsed_kws |= MTREE_HAS_FFLAGS; archive_entry_copy_fflags_text(entry, val); return (ARCHIVE_OK); } break; case 'g': if (strcmp(key, "gid") == 0) { *parsed_kws |= MTREE_HAS_GID; archive_entry_set_gid(entry, mtree_atol(&val, 10)); return (ARCHIVE_OK); } if (strcmp(key, "gname") == 0) { *parsed_kws |= MTREE_HAS_GNAME; archive_entry_copy_gname(entry, val); return (ARCHIVE_OK); } break; case 'i': if (strcmp(key, "inode") == 0) { archive_entry_set_ino(entry, mtree_atol(&val, 10)); return (ARCHIVE_OK); } break; case 'l': if (strcmp(key, "link") == 0) { parse_escapes(val, NULL); archive_entry_copy_symlink(entry, val); return (ARCHIVE_OK); } break; case 'm': if (strcmp(key, "md5") == 0 || strcmp(key, "md5digest") == 0) { return parse_digest(a, entry, val, ARCHIVE_ENTRY_DIGEST_MD5); } if (strcmp(key, "mode") == 0) { if (val[0] < '0' || val[0] > '7') { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Symbolic or non-octal mode \"%s\" unsupported", val); return (ARCHIVE_WARN); } *parsed_kws |= MTREE_HAS_PERM; archive_entry_set_perm(entry, (mode_t)mtree_atol(&val, 8)); return (ARCHIVE_OK); } break; case 'n': if (strcmp(key, "nlink") == 0) { *parsed_kws |= MTREE_HAS_NLINK; archive_entry_set_nlink(entry, (unsigned int)mtree_atol(&val, 10)); return (ARCHIVE_OK); } break; case 'r': if (strcmp(key, "resdevice") == 0) { /* stat(2) st_dev field, e.g. the device ID where the * inode resides */ int r; dev_t dev; r = parse_device(&dev, &a->archive, val); if (r == ARCHIVE_OK) archive_entry_set_dev(entry, dev); return r; } if (strcmp(key, "rmd160") == 0 || strcmp(key, "rmd160digest") == 0) { return parse_digest(a, entry, val, ARCHIVE_ENTRY_DIGEST_RMD160); } break; case 's': if (strcmp(key, "sha1") == 0 || strcmp(key, "sha1digest") == 0) { return parse_digest(a, entry, val, ARCHIVE_ENTRY_DIGEST_SHA1); } if (strcmp(key, "sha256") == 0 || strcmp(key, "sha256digest") == 0) { return parse_digest(a, entry, val, ARCHIVE_ENTRY_DIGEST_SHA256); } if (strcmp(key, "sha384") == 0 || strcmp(key, "sha384digest") == 0) { return parse_digest(a, entry, val, ARCHIVE_ENTRY_DIGEST_SHA384); } if (strcmp(key, "sha512") == 0 || strcmp(key, "sha512digest") == 0) { return parse_digest(a, entry, val, ARCHIVE_ENTRY_DIGEST_SHA512); } if (strcmp(key, "size") == 0) { archive_entry_set_size(entry, mtree_atol(&val, 10)); return (ARCHIVE_OK); } break; case 't': if (strcmp(key, "tags") == 0) { /* * Comma delimited list of tags. * Ignore the tags for now, but the interface * should be extended to allow inclusion/exclusion. */ return (ARCHIVE_OK); } if (strcmp(key, "time") == 0) { int64_t m; int64_t my_time_t_max = get_time_t_max(); int64_t my_time_t_min = get_time_t_min(); long ns = 0; *parsed_kws |= MTREE_HAS_MTIME; m = mtree_atol(&val, 10); /* Replicate an old mtree bug: * 123456789.1 represents 123456789 * seconds and 1 nanosecond. */ if (*val == '.') { ++val; ns = (long)mtree_atol(&val, 10); if (ns < 0) ns = 0; else if (ns > 999999999) ns = 999999999; } if (m > my_time_t_max) m = my_time_t_max; else if (m < my_time_t_min) m = my_time_t_min; archive_entry_set_mtime(entry, (time_t)m, ns); return (ARCHIVE_OK); } if (strcmp(key, "type") == 0) { switch (val[0]) { case 'b': if (strcmp(val, "block") == 0) { *parsed_kws |= MTREE_HAS_TYPE; archive_entry_set_filetype(entry, AE_IFBLK); return (ARCHIVE_OK); } break; case 'c': if (strcmp(val, "char") == 0) { *parsed_kws |= MTREE_HAS_TYPE; archive_entry_set_filetype(entry, AE_IFCHR); return (ARCHIVE_OK); } break; case 'd': if (strcmp(val, "dir") == 0) { *parsed_kws |= MTREE_HAS_TYPE; archive_entry_set_filetype(entry, AE_IFDIR); return (ARCHIVE_OK); } break; case 'f': if (strcmp(val, "fifo") == 0) { *parsed_kws |= MTREE_HAS_TYPE; archive_entry_set_filetype(entry, AE_IFIFO); return (ARCHIVE_OK); } if (strcmp(val, "file") == 0) { *parsed_kws |= MTREE_HAS_TYPE; archive_entry_set_filetype(entry, AE_IFREG); return (ARCHIVE_OK); } break; case 'l': if (strcmp(val, "link") == 0) { *parsed_kws |= MTREE_HAS_TYPE; archive_entry_set_filetype(entry, AE_IFLNK); return (ARCHIVE_OK); } break; default: break; } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unrecognized file type \"%s\"; " "assuming \"file\"", val); archive_entry_set_filetype(entry, AE_IFREG); return (ARCHIVE_WARN); } break; case 'u': if (strcmp(key, "uid") == 0) { *parsed_kws |= MTREE_HAS_UID; archive_entry_set_uid(entry, mtree_atol(&val, 10)); return (ARCHIVE_OK); } if (strcmp(key, "uname") == 0) { *parsed_kws |= MTREE_HAS_UNAME; archive_entry_copy_uname(entry, val); return (ARCHIVE_OK); } break; default: break; } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unrecognized key %s=%s", key, val); return (ARCHIVE_WARN); } static int read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { size_t bytes_to_read; ssize_t bytes_read; struct mtree *mtree; mtree = (struct mtree *)(a->format->data); if (mtree->fd < 0) { *buff = NULL; *offset = 0; *size = 0; return (ARCHIVE_EOF); } if (mtree->buff == NULL) { mtree->buffsize = 64 * 1024; mtree->buff = malloc(mtree->buffsize); if (mtree->buff == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } } *buff = mtree->buff; *offset = mtree->offset; if ((int64_t)mtree->buffsize > mtree->cur_size - mtree->offset) bytes_to_read = (size_t)(mtree->cur_size - mtree->offset); else bytes_to_read = mtree->buffsize; bytes_read = read(mtree->fd, mtree->buff, bytes_to_read); if (bytes_read < 0) { archive_set_error(&a->archive, errno, "Can't read"); return (ARCHIVE_WARN); } if (bytes_read == 0) { *size = 0; return (ARCHIVE_EOF); } mtree->offset += bytes_read; *size = bytes_read; return (ARCHIVE_OK); } /* Skip does nothing except possibly close the contents file. */ static int skip(struct archive_read *a) { struct mtree *mtree; mtree = (struct mtree *)(a->format->data); if (mtree->fd >= 0) { close(mtree->fd); mtree->fd = -1; } return (ARCHIVE_OK); } /* * Since parsing backslash sequences always makes strings shorter, * we can always do this conversion in-place. */ static void parse_escapes(char *src, struct mtree_entry *mentry) { char *dest = src; char c; if (mentry != NULL && strcmp(src, ".") == 0) mentry->full = 1; while (*src != '\0') { c = *src++; if (c == '/' && mentry != NULL) mentry->full = 1; if (c == '\\') { switch (src[0]) { case '0': if (src[1] < '0' || src[1] > '7') { c = 0; ++src; break; } /* FALLTHROUGH */ case '1': case '2': case '3': if (src[1] >= '0' && src[1] <= '7' && src[2] >= '0' && src[2] <= '7') { c = (src[0] - '0') << 6; c |= (src[1] - '0') << 3; c |= (src[2] - '0'); src += 3; } break; case 'a': c = '\a'; ++src; break; case 'b': c = '\b'; ++src; break; case 'f': c = '\f'; ++src; break; case 'n': c = '\n'; ++src; break; case 'r': c = '\r'; ++src; break; case 's': c = ' '; ++src; break; case 't': c = '\t'; ++src; break; case 'v': c = '\v'; ++src; break; case '\\': c = '\\'; ++src; break; } } *dest++ = c; } *dest = '\0'; } /* Parse a hex digit. */ static int parsedigit(char c) { if (c >= '0' && c <= '9') return c - '0'; else if (c >= 'a' && c <= 'f') return c - 'a'; else if (c >= 'A' && c <= 'F') return c - 'A'; else return -1; } /* * Note that this implementation does not (and should not!) obey * locale settings; you cannot simply substitute strtol here, since * it does obey locale. */ static int64_t mtree_atol(char **p, int base) { int64_t l, limit; int digit, last_digit_limit; if (base == 0) { if (**p != '0') base = 10; else if ((*p)[1] == 'x' || (*p)[1] == 'X') { *p += 2; base = 16; } else { base = 8; } } if (**p == '-') { limit = INT64_MIN / base; last_digit_limit = -(INT64_MIN % base); ++(*p); l = 0; digit = parsedigit(**p); while (digit >= 0 && digit < base) { if (l < limit || (l == limit && digit >= last_digit_limit)) return INT64_MIN; l = (l * base) - digit; digit = parsedigit(*++(*p)); } return l; } else { limit = INT64_MAX / base; last_digit_limit = INT64_MAX % base; l = 0; digit = parsedigit(**p); while (digit >= 0 && digit < base) { if (l > limit || (l == limit && digit > last_digit_limit)) return INT64_MAX; l = (l * base) + digit; digit = parsedigit(*++(*p)); } return l; } } /* * Returns length of line (including trailing newline) * or negative on error. 'start' argument is updated to * point to first character of line. */ static ssize_t readline(struct archive_read *a, struct mtree *mtree, char **start, ssize_t limit) { ssize_t bytes_read; ssize_t total_size = 0; ssize_t find_off = 0; const void *t; void *nl; char *u; /* Accumulate line in a line buffer. */ for (;;) { /* Read some more. */ t = __archive_read_ahead(a, 1, &bytes_read); if (t == NULL) return (0); if (bytes_read < 0) return (ARCHIVE_FATAL); nl = memchr(t, '\n', bytes_read); /* If we found '\n', trim the read to end exactly there. */ if (nl != NULL) { bytes_read = ((const char *)nl) - ((const char *)t) + 1; } if (total_size + bytes_read + 1 > limit) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Line too long"); return (ARCHIVE_FATAL); } if (archive_string_ensure(&mtree->line, total_size + bytes_read + 1) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate working buffer"); return (ARCHIVE_FATAL); } /* Append new bytes to string. */ memcpy(mtree->line.s + total_size, t, bytes_read); __archive_read_consume(a, bytes_read); total_size += bytes_read; mtree->line.s[total_size] = '\0'; for (u = mtree->line.s + find_off; *u; ++u) { if (u[0] == '\n') { /* Ends with unescaped newline. */ + /* Check if preceded by '\r' for CRLF handling */ + if (u > mtree->line.s && u[-1] == '\r') { + /* CRLF ending - remove the '\r' */ + u[-1] = '\n'; + u[0] = '\0'; + total_size--; + } *start = mtree->line.s; return total_size; } else if (u[0] == '#') { /* Ends with comment sequence #...\n */ if (nl == NULL) { /* But we've not found the \n yet */ break; } } else if (u[0] == '\\') { if (u[1] == '\n') { /* Trim escaped newline. */ total_size -= 2; mtree->line.s[total_size] = '\0'; break; + } else if (u[1] == '\r' && u[2] == '\n') { + /* Trim escaped CRLF. */ + total_size -= 3; + mtree->line.s[total_size] = '\0'; + break; } else if (u[1] != '\0') { /* Skip the two-char escape sequence */ ++u; } } } find_off = u - mtree->line.s; } } diff --git a/contrib/libarchive/libarchive/archive_read_support_format_rar.c b/contrib/libarchive/libarchive/archive_read_support_format_rar.c index 923ae5c65e17..9b401c00ba34 100644 --- a/contrib/libarchive/libarchive/archive_read_support_format_rar.c +++ b/contrib/libarchive/libarchive/archive_read_support_format_rar.c @@ -1,3920 +1,3918 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * Copyright (c) 2011 Andres Mejia * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #include #include #ifdef HAVE_ZLIB_H #include /* crc32 */ #endif #include "archive.h" #ifndef HAVE_ZLIB_H #include "archive_crc32.h" #endif #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_ppmd7_private.h" #include "archive_private.h" #include "archive_read_private.h" /* RAR signature, also known as the mark header */ #define RAR_SIGNATURE "\x52\x61\x72\x21\x1A\x07\x00" /* Header types */ #define MARK_HEAD 0x72 #define MAIN_HEAD 0x73 #define FILE_HEAD 0x74 #define COMM_HEAD 0x75 #define AV_HEAD 0x76 #define SUB_HEAD 0x77 #define PROTECT_HEAD 0x78 #define SIGN_HEAD 0x79 #define NEWSUB_HEAD 0x7a #define ENDARC_HEAD 0x7b /* Main Header Flags */ #define MHD_VOLUME 0x0001 #define MHD_COMMENT 0x0002 #define MHD_LOCK 0x0004 #define MHD_SOLID 0x0008 #define MHD_NEWNUMBERING 0x0010 #define MHD_AV 0x0020 #define MHD_PROTECT 0x0040 #define MHD_PASSWORD 0x0080 #define MHD_FIRSTVOLUME 0x0100 #define MHD_ENCRYPTVER 0x0200 /* Flags common to all headers */ #define HD_MARKDELETION 0x4000 #define HD_ADD_SIZE_PRESENT 0x8000 /* File Header Flags */ #define FHD_SPLIT_BEFORE 0x0001 #define FHD_SPLIT_AFTER 0x0002 #define FHD_PASSWORD 0x0004 #define FHD_COMMENT 0x0008 #define FHD_SOLID 0x0010 #define FHD_LARGE 0x0100 #define FHD_UNICODE 0x0200 #define FHD_SALT 0x0400 #define FHD_VERSION 0x0800 #define FHD_EXTTIME 0x1000 #define FHD_EXTFLAGS 0x2000 /* File dictionary sizes */ #define DICTIONARY_SIZE_64 0x00 #define DICTIONARY_SIZE_128 0x20 #define DICTIONARY_SIZE_256 0x40 #define DICTIONARY_SIZE_512 0x60 #define DICTIONARY_SIZE_1024 0x80 #define DICTIONARY_SIZE_2048 0xA0 #define DICTIONARY_SIZE_4096 0xC0 #define FILE_IS_DIRECTORY 0xE0 #define DICTIONARY_MASK FILE_IS_DIRECTORY /* OS Flags */ #define OS_MSDOS 0 #define OS_OS2 1 #define OS_WIN32 2 #define OS_UNIX 3 #define OS_MAC_OS 4 #define OS_BEOS 5 /* Compression Methods */ #define COMPRESS_METHOD_STORE 0x30 /* LZSS */ #define COMPRESS_METHOD_FASTEST 0x31 #define COMPRESS_METHOD_FAST 0x32 #define COMPRESS_METHOD_NORMAL 0x33 /* PPMd Variant H */ #define COMPRESS_METHOD_GOOD 0x34 #define COMPRESS_METHOD_BEST 0x35 #define CRC_POLYNOMIAL 0xEDB88320 #define NS_UNIT 10000000 #define DICTIONARY_MAX_SIZE 0x400000 #define MAINCODE_SIZE 299 #define OFFSETCODE_SIZE 60 #define LOWOFFSETCODE_SIZE 17 #define LENGTHCODE_SIZE 28 #define HUFFMAN_TABLE_SIZE \ MAINCODE_SIZE + OFFSETCODE_SIZE + LOWOFFSETCODE_SIZE + LENGTHCODE_SIZE #define MAX_SYMBOL_LENGTH 0xF #define MAX_SYMBOLS 20 /* Virtual Machine Properties */ #define VM_MEMORY_SIZE 0x40000 #define VM_MEMORY_MASK (VM_MEMORY_SIZE - 1) #define PROGRAM_WORK_SIZE 0x3C000 #define PROGRAM_GLOBAL_SIZE 0x2000 #define PROGRAM_SYSTEM_GLOBAL_ADDRESS PROGRAM_WORK_SIZE #define PROGRAM_SYSTEM_GLOBAL_SIZE 0x40 #define PROGRAM_USER_GLOBAL_ADDRESS (PROGRAM_SYSTEM_GLOBAL_ADDRESS + PROGRAM_SYSTEM_GLOBAL_SIZE) #define PROGRAM_USER_GLOBAL_SIZE (PROGRAM_GLOBAL_SIZE - PROGRAM_SYSTEM_GLOBAL_SIZE) /* * Considering L1,L2 cache miss and a calling of write system-call, * the best size of the output buffer(uncompressed buffer) is 128K. * If the structure of extracting process is changed, this value * might be researched again. */ #define UNP_BUFFER_SIZE (128 * 1024) /* Define this here for non-Windows platforms */ #if !((defined(__WIN32__) || defined(_WIN32) || defined(__WIN32)) && !defined(__CYGWIN__)) #define FILE_ATTRIBUTE_DIRECTORY 0x10 #endif #undef minimum #define minimum(a, b) ((a)<(b)?(a):(b)) /* Stack overflow check */ #define MAX_COMPRESS_DEPTH 1024 /* Fields common to all headers */ struct rar_header { char crc[2]; char type; char flags[2]; char size[2]; }; /* Fields common to all file headers */ struct rar_file_header { char pack_size[4]; char unp_size[4]; char host_os; char file_crc[4]; char file_time[4]; char unp_ver; char method; char name_size[2]; char file_attr[4]; }; struct huffman_tree_node { int branches[2]; }; struct huffman_table_entry { unsigned int length; int value; }; struct huffman_code { struct huffman_tree_node *tree; int numentries; int numallocatedentries; int minlength; int maxlength; int tablesize; struct huffman_table_entry *table; }; struct lzss { unsigned char *window; int mask; int64_t position; }; struct data_block_offsets { int64_t header_size; int64_t start_offset; int64_t end_offset; }; struct rar_program_code { uint8_t *staticdata; uint32_t staticdatalen; uint8_t *globalbackup; uint32_t globalbackuplen; uint64_t fingerprint; uint32_t usagecount; uint32_t oldfilterlength; struct rar_program_code *next; }; struct rar_filter { struct rar_program_code *prog; uint32_t initialregisters[8]; uint8_t *globaldata; uint32_t globaldatalen; size_t blockstartpos; uint32_t blocklength; uint32_t filteredblockaddress; uint32_t filteredblocklength; struct rar_filter *next; }; struct memory_bit_reader { const uint8_t *bytes; size_t length; size_t offset; uint64_t bits; int available; int at_eof; }; struct rar_virtual_machine { uint32_t registers[8]; uint8_t memory[VM_MEMORY_SIZE + sizeof(uint32_t)]; }; struct rar_filters { struct rar_virtual_machine *vm; struct rar_program_code *progs; struct rar_filter *stack; int64_t filterstart; uint32_t lastfilternum; int64_t lastend; uint8_t *bytes; size_t bytes_ready; }; struct audio_state { int8_t weight[5]; int16_t delta[4]; int8_t lastdelta; int error[11]; int count; uint8_t lastbyte; }; struct rar { /* Entries from main RAR header */ unsigned main_flags; unsigned long file_crc; char reserved1[2]; char reserved2[4]; char encryptver; /* File header entries */ char compression_method; unsigned file_flags; int64_t packed_size; int64_t unp_size; time_t mtime; long mnsec; mode_t mode; char *filename; char *filename_save; size_t filename_save_size; size_t filename_allocated; /* File header optional entries */ char salt[8]; time_t atime; long ansec; time_t ctime; long cnsec; time_t arctime; long arcnsec; /* Fields to help with tracking decompression of files. */ int64_t bytes_unconsumed; int64_t bytes_remaining; int64_t bytes_uncopied; int64_t offset; int64_t offset_outgoing; int64_t offset_seek; char valid; unsigned int unp_offset; unsigned int unp_buffer_size; unsigned char *unp_buffer; unsigned int dictionary_size; char start_new_block; char entry_eof; unsigned long crc_calculated; int found_first_header; char has_endarc_header; struct data_block_offsets *dbo; size_t cursor; size_t nodes; char filename_must_match; /* LZSS members */ struct huffman_code maincode; struct huffman_code offsetcode; struct huffman_code lowoffsetcode; struct huffman_code lengthcode; unsigned char lengthtable[HUFFMAN_TABLE_SIZE]; struct lzss lzss; unsigned int lastlength; unsigned int lastoffset; unsigned int oldoffset[4]; unsigned int lastlowoffset; unsigned int numlowoffsetrepeats; char start_new_table; /* Filters */ struct rar_filters filters; /* PPMd Variant H members */ char ppmd_valid; char ppmd_eod; char is_ppmd_block; int ppmd_escape; CPpmd7 ppmd7_context; CPpmd7z_RangeDec range_dec; IByteIn bytein; /* * String conversion object. */ int init_default_conversion; struct archive_string_conv *sconv_default; struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv_utf8; struct archive_string_conv *sconv_utf16be; /* * Bit stream reader. */ struct rar_br { #define CACHE_TYPE uint64_t #define CACHE_BITS (8 * sizeof(CACHE_TYPE)) /* Cache buffer. */ CACHE_TYPE cache_buffer; /* Indicates how many bits avail in cache_buffer. */ int cache_avail; ssize_t avail_in; const unsigned char *next_in; } br; /* * Custom field to denote that this archive contains encrypted entries */ int has_encrypted_entries; }; static int archive_read_support_format_rar_capabilities(struct archive_read *); static int archive_read_format_rar_has_encrypted_entries(struct archive_read *); static int archive_read_format_rar_bid(struct archive_read *, int); static int archive_read_format_rar_options(struct archive_read *, const char *, const char *); static int archive_read_format_rar_read_header(struct archive_read *, struct archive_entry *); static int archive_read_format_rar_read_data(struct archive_read *, const void **, size_t *, int64_t *); static int archive_read_format_rar_read_data_skip(struct archive_read *a); static int64_t archive_read_format_rar_seek_data(struct archive_read *, int64_t, int); static int archive_read_format_rar_cleanup(struct archive_read *); /* Support functions */ static int read_header(struct archive_read *, struct archive_entry *, char); static time_t get_time(int); static int read_exttime(const char *, struct rar *, const char *); static int read_symlink_stored(struct archive_read *, struct archive_entry *, struct archive_string_conv *); static int read_data_stored(struct archive_read *, const void **, size_t *, int64_t *); static int read_data_compressed(struct archive_read *, const void **, size_t *, int64_t *, size_t); static int rar_br_preparation(struct archive_read *, struct rar_br *); static int parse_codes(struct archive_read *); static void free_codes(struct archive_read *); static int read_next_symbol(struct archive_read *, struct huffman_code *); static int create_code(struct archive_read *, struct huffman_code *, unsigned char *, int, char); static int add_value(struct archive_read *, struct huffman_code *, int, int, int); static int new_node(struct huffman_code *); static int make_table(struct archive_read *, struct huffman_code *); static int make_table_recurse(struct archive_read *, struct huffman_code *, int, struct huffman_table_entry *, int, int); static int expand(struct archive_read *, int64_t *); static int copy_from_lzss_window_to_unp(struct archive_read *, const void **, int64_t, size_t); static const void *rar_read_ahead(struct archive_read *, size_t, ssize_t *); static int parse_filter(struct archive_read *, const uint8_t *, uint16_t, uint8_t); static int run_filters(struct archive_read *); static void clear_filters(struct rar_filters *); static struct rar_filter *create_filter(struct rar_program_code *, const uint8_t *, uint32_t, uint32_t[8], size_t, uint32_t); static void delete_filter(struct rar_filter *filter); static struct rar_program_code *compile_program(const uint8_t *, size_t); static void delete_program_code(struct rar_program_code *prog); static uint32_t membr_next_rarvm_number(struct memory_bit_reader *br); static inline uint32_t membr_bits(struct memory_bit_reader *br, int bits); static int membr_fill(struct memory_bit_reader *br, int bits); static int read_filter(struct archive_read *, int64_t *); static int rar_decode_byte(struct archive_read*, uint8_t *); static int execute_filter(struct archive_read*, struct rar_filter *, struct rar_virtual_machine *, size_t); static int copy_from_lzss_window(struct archive_read *, uint8_t *, int64_t, int); static inline void vm_write_32(struct rar_virtual_machine*, size_t, uint32_t); static inline uint32_t vm_read_32(struct rar_virtual_machine*, size_t); /* * Bit stream reader. */ /* Check that the cache buffer has enough bits. */ #define rar_br_has(br, n) ((br)->cache_avail >= n) /* Get compressed data by bit. */ #define rar_br_bits(br, n) \ (((uint32_t)((br)->cache_buffer >> \ ((br)->cache_avail - (n)))) & cache_masks[n]) #define rar_br_bits_forced(br, n) \ (((uint32_t)((br)->cache_buffer << \ ((n) - (br)->cache_avail))) & cache_masks[n]) /* Read ahead to make sure the cache buffer has enough compressed data we * will use. * True : completed, there is enough data in the cache buffer. * False : there is no data in the stream. */ #define rar_br_read_ahead(a, br, n) \ ((rar_br_has(br, (n)) || rar_br_fillup(a, br)) || rar_br_has(br, (n))) /* Notify how many bits we consumed. */ #define rar_br_consume(br, n) ((br)->cache_avail -= (n)) #define rar_br_consume_unaligned_bits(br) ((br)->cache_avail &= ~7) static const uint32_t cache_masks[] = { 0x00000000, 0x00000001, 0x00000003, 0x00000007, 0x0000000F, 0x0000001F, 0x0000003F, 0x0000007F, 0x000000FF, 0x000001FF, 0x000003FF, 0x000007FF, 0x00000FFF, 0x00001FFF, 0x00003FFF, 0x00007FFF, 0x0000FFFF, 0x0001FFFF, 0x0003FFFF, 0x0007FFFF, 0x000FFFFF, 0x001FFFFF, 0x003FFFFF, 0x007FFFFF, 0x00FFFFFF, 0x01FFFFFF, 0x03FFFFFF, 0x07FFFFFF, 0x0FFFFFFF, 0x1FFFFFFF, 0x3FFFFFFF, 0x7FFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF }; /* * Shift away used bits in the cache data and fill it up with following bits. * Call this when cache buffer does not have enough bits you need. * * Returns 1 if the cache buffer is full. * Returns 0 if the cache buffer is not full; input buffer is empty. */ static int rar_br_fillup(struct archive_read *a, struct rar_br *br) { struct rar *rar = (struct rar *)(a->format->data); int n = CACHE_BITS - br->cache_avail; for (;;) { switch (n >> 3) { case 8: if (br->avail_in >= 8) { br->cache_buffer = ((uint64_t)br->next_in[0]) << 56 | ((uint64_t)br->next_in[1]) << 48 | ((uint64_t)br->next_in[2]) << 40 | ((uint64_t)br->next_in[3]) << 32 | ((uint32_t)br->next_in[4]) << 24 | ((uint32_t)br->next_in[5]) << 16 | ((uint32_t)br->next_in[6]) << 8 | (uint32_t)br->next_in[7]; br->next_in += 8; br->avail_in -= 8; br->cache_avail += 8 * 8; rar->bytes_unconsumed += 8; rar->bytes_remaining -= 8; return (1); } break; case 7: if (br->avail_in >= 7) { br->cache_buffer = (br->cache_buffer << 56) | ((uint64_t)br->next_in[0]) << 48 | ((uint64_t)br->next_in[1]) << 40 | ((uint64_t)br->next_in[2]) << 32 | ((uint32_t)br->next_in[3]) << 24 | ((uint32_t)br->next_in[4]) << 16 | ((uint32_t)br->next_in[5]) << 8 | (uint32_t)br->next_in[6]; br->next_in += 7; br->avail_in -= 7; br->cache_avail += 7 * 8; rar->bytes_unconsumed += 7; rar->bytes_remaining -= 7; return (1); } break; case 6: if (br->avail_in >= 6) { br->cache_buffer = (br->cache_buffer << 48) | ((uint64_t)br->next_in[0]) << 40 | ((uint64_t)br->next_in[1]) << 32 | ((uint32_t)br->next_in[2]) << 24 | ((uint32_t)br->next_in[3]) << 16 | ((uint32_t)br->next_in[4]) << 8 | (uint32_t)br->next_in[5]; br->next_in += 6; br->avail_in -= 6; br->cache_avail += 6 * 8; rar->bytes_unconsumed += 6; rar->bytes_remaining -= 6; return (1); } break; case 0: /* We have enough compressed data in * the cache buffer.*/ return (1); default: break; } if (br->avail_in <= 0) { if (rar->bytes_unconsumed > 0) { /* Consume as much as the decompressor * actually used. */ __archive_read_consume(a, rar->bytes_unconsumed); rar->bytes_unconsumed = 0; } br->next_in = rar_read_ahead(a, 1, &(br->avail_in)); if (br->next_in == NULL) return (0); if (br->avail_in == 0) return (0); } br->cache_buffer = (br->cache_buffer << 8) | *br->next_in++; br->avail_in--; br->cache_avail += 8; n -= 8; rar->bytes_unconsumed++; rar->bytes_remaining--; } } static int rar_br_preparation(struct archive_read *a, struct rar_br *br) { struct rar *rar = (struct rar *)(a->format->data); if (rar->bytes_remaining > 0) { br->next_in = rar_read_ahead(a, 1, &(br->avail_in)); if (br->next_in == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); return (ARCHIVE_FATAL); } if (br->cache_avail == 0) (void)rar_br_fillup(a, br); } return (ARCHIVE_OK); } /* Find last bit set */ static inline int rar_fls(unsigned int word) { word |= (word >> 1); word |= (word >> 2); word |= (word >> 4); word |= (word >> 8); word |= (word >> 16); return word - (word >> 1); } /* LZSS functions */ static inline int64_t lzss_position(struct lzss *lzss) { return lzss->position; } static inline int lzss_mask(struct lzss *lzss) { return lzss->mask; } static inline int lzss_size(struct lzss *lzss) { return lzss->mask + 1; } static inline int lzss_offset_for_position(struct lzss *lzss, int64_t pos) { return (int)(pos & lzss->mask); } static inline unsigned char * lzss_pointer_for_position(struct lzss *lzss, int64_t pos) { return &lzss->window[lzss_offset_for_position(lzss, pos)]; } static inline int lzss_current_offset(struct lzss *lzss) { return lzss_offset_for_position(lzss, lzss->position); } static inline uint8_t * lzss_current_pointer(struct lzss *lzss) { return lzss_pointer_for_position(lzss, lzss->position); } static inline void lzss_emit_literal(struct rar *rar, uint8_t literal) { *lzss_current_pointer(&rar->lzss) = literal; rar->lzss.position++; } static inline void lzss_emit_match(struct rar *rar, int offset, int length) { int dstoffs = lzss_current_offset(&rar->lzss); int srcoffs = (dstoffs - offset) & lzss_mask(&rar->lzss); int l, li, remaining; unsigned char *d, *s; remaining = length; while (remaining > 0) { l = remaining; if (dstoffs > srcoffs) { if (l > lzss_size(&rar->lzss) - dstoffs) l = lzss_size(&rar->lzss) - dstoffs; } else { if (l > lzss_size(&rar->lzss) - srcoffs) l = lzss_size(&rar->lzss) - srcoffs; } d = &(rar->lzss.window[dstoffs]); s = &(rar->lzss.window[srcoffs]); if ((dstoffs + l < srcoffs) || (srcoffs + l < dstoffs)) memcpy(d, s, l); else { for (li = 0; li < l; li++) d[li] = s[li]; } remaining -= l; dstoffs = (dstoffs + l) & lzss_mask(&(rar->lzss)); srcoffs = (srcoffs + l) & lzss_mask(&(rar->lzss)); } rar->lzss.position += length; } static Byte ppmd_read(void *p) { struct archive_read *a = ((IByteIn*)p)->a; struct rar *rar = (struct rar *)(a->format->data); struct rar_br *br = &(rar->br); Byte b; if (!rar_br_read_ahead(a, br, 8)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); rar->valid = 0; return 0; } b = rar_br_bits(br, 8); rar_br_consume(br, 8); return b; } int archive_read_support_format_rar(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct rar *rar; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_rar"); rar = calloc(1, sizeof(*rar)); if (rar == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate rar data"); return (ARCHIVE_FATAL); } /* * Until enough data has been read, we cannot tell about * any encrypted entries yet. */ rar->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; r = __archive_read_register_format(a, rar, "rar", archive_read_format_rar_bid, archive_read_format_rar_options, archive_read_format_rar_read_header, archive_read_format_rar_read_data, archive_read_format_rar_read_data_skip, archive_read_format_rar_seek_data, archive_read_format_rar_cleanup, archive_read_support_format_rar_capabilities, archive_read_format_rar_has_encrypted_entries); if (r != ARCHIVE_OK) free(rar); return (r); } static int archive_read_support_format_rar_capabilities(struct archive_read * a) { (void)a; /* UNUSED */ return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); } static int archive_read_format_rar_has_encrypted_entries(struct archive_read *_a) { if (_a && _a->format) { struct rar * rar = (struct rar *)_a->format->data; if (rar) { return rar->has_encrypted_entries; } } return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; } static int archive_read_format_rar_bid(struct archive_read *a, int best_bid) { const char *p; /* If there's already a bid > 30, we'll never win. */ if (best_bid > 30) return (-1); if ((p = __archive_read_ahead(a, 7, NULL)) == NULL) return (-1); if (memcmp(p, RAR_SIGNATURE, 7) == 0) return (30); if ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0) { /* This is a PE file */ ssize_t offset = 0x10000; ssize_t window = 4096; ssize_t bytes_avail; while (offset + window <= (1024 * 128)) { const char *buff = __archive_read_ahead(a, offset + window, &bytes_avail); if (buff == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < 0x40) return (0); continue; } p = buff + offset; while (p + 7 < buff + bytes_avail) { if (memcmp(p, RAR_SIGNATURE, 7) == 0) return (30); p += 0x10; } offset = p - buff; } } return (0); } static int skip_sfx(struct archive_read *a) { const void *h; const char *p, *q; size_t skip, total; ssize_t bytes, window; total = 0; window = 4096; while (total + window <= (1024 * 128)) { h = __archive_read_ahead(a, window, &bytes); if (h == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < 0x40) goto fatal; continue; } if (bytes < 0x40) goto fatal; p = h; q = p + bytes; /* * Scan ahead until we find something that looks * like the RAR header. */ while (p + 7 < q) { if (memcmp(p, RAR_SIGNATURE, 7) == 0) { skip = p - (const char *)h; __archive_read_consume(a, skip); return (ARCHIVE_OK); } p += 0x10; } skip = p - (const char *)h; __archive_read_consume(a, skip); total += skip; } fatal: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Couldn't find out RAR header"); return (ARCHIVE_FATAL); } static int archive_read_format_rar_options(struct archive_read *a, const char *key, const char *val) { struct rar *rar; int ret = ARCHIVE_FAILED; rar = (struct rar *)(a->format->data); if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "rar: hdrcharset option needs a character-set name"); else { rar->opt_sconv = archive_string_conversion_from_charset( &a->archive, val, 0); if (rar->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int archive_read_format_rar_read_header(struct archive_read *a, struct archive_entry *entry) { const void *h; const char *p; struct rar *rar; int64_t skip; char head_type; int ret; unsigned flags; unsigned long crc32_expected; a->archive.archive_format = ARCHIVE_FORMAT_RAR; if (a->archive.archive_format_name == NULL) a->archive.archive_format_name = "RAR"; rar = (struct rar *)(a->format->data); /* * It should be sufficient to call archive_read_next_header() for * a reader to determine if an entry is encrypted or not. If the * encryption of an entry is only detectable when calling * archive_read_data(), so be it. We'll do the same check there * as well. */ if (rar->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { rar->has_encrypted_entries = 0; } /* RAR files can be generated without EOF headers, so return ARCHIVE_EOF if * this fails. */ if ((h = __archive_read_ahead(a, 7, NULL)) == NULL) return (ARCHIVE_EOF); p = h; if (rar->found_first_header == 0 && ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0)) { /* This is an executable ? Must be self-extracting... */ ret = skip_sfx(a); if (ret < ARCHIVE_WARN) return (ret); } rar->found_first_header = 1; while (1) { unsigned long crc32_val; if ((h = __archive_read_ahead(a, 7, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Failed to read next header."); return (ARCHIVE_FATAL); } p = h; head_type = p[2]; switch(head_type) { case MARK_HEAD: if (memcmp(p, RAR_SIGNATURE, 7) != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid marker header"); return (ARCHIVE_FATAL); } __archive_read_consume(a, 7); break; case MAIN_HEAD: rar->main_flags = archive_le16dec(p + 3); skip = archive_le16dec(p + 5); if ((size_t)skip < 7 + sizeof(rar->reserved1) + sizeof(rar->reserved2)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size"); return (ARCHIVE_FATAL); } if ((h = __archive_read_ahead(a, skip, NULL)) == NULL) return (ARCHIVE_FATAL); p = h; memcpy(rar->reserved1, p + 7, sizeof(rar->reserved1)); memcpy(rar->reserved2, p + 7 + sizeof(rar->reserved1), sizeof(rar->reserved2)); if (rar->main_flags & MHD_ENCRYPTVER) { if ((size_t)skip < 7 + sizeof(rar->reserved1) + sizeof(rar->reserved2) + 1) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size"); return (ARCHIVE_FATAL); } rar->encryptver = *(p + 7 + sizeof(rar->reserved1) + sizeof(rar->reserved2)); } /* Main header is password encrypted, so we cannot read any file names or any other info about files from the header. */ if (rar->main_flags & MHD_PASSWORD) { archive_entry_set_is_metadata_encrypted(entry, 1); archive_entry_set_is_data_encrypted(entry, 1); rar->has_encrypted_entries = 1; archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "RAR encryption support unavailable."); return (ARCHIVE_FATAL); } crc32_val = crc32(0, (const unsigned char *)p + 2, (unsigned)skip - 2); if ((crc32_val & 0xffff) != archive_le16dec(p)) { #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Header CRC error"); return (ARCHIVE_FATAL); #endif } __archive_read_consume(a, skip); break; case FILE_HEAD: return read_header(a, entry, head_type); case COMM_HEAD: case AV_HEAD: case SUB_HEAD: case PROTECT_HEAD: case SIGN_HEAD: case ENDARC_HEAD: flags = archive_le16dec(p + 3); skip = archive_le16dec(p + 5); if (skip < 7) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size too small"); return (ARCHIVE_FATAL); } if (flags & HD_ADD_SIZE_PRESENT) { if (skip < 7 + 4) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size too small"); return (ARCHIVE_FATAL); } if ((h = __archive_read_ahead(a, skip, NULL)) == NULL) return (ARCHIVE_FATAL); p = h; skip += archive_le32dec(p + 7); } /* Skip over the 2-byte CRC at the beginning of the header. */ crc32_expected = archive_le16dec(p); __archive_read_consume(a, 2); skip -= 2; /* Skim the entire header and compute the CRC. */ crc32_val = 0; while (skip > 0) { unsigned to_read; if (skip > 32 * 1024) to_read = 32 * 1024; else to_read = (unsigned)skip; if ((h = __archive_read_ahead(a, to_read, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad RAR file"); return (ARCHIVE_FATAL); } p = h; crc32_val = crc32(crc32_val, (const unsigned char *)p, to_read); __archive_read_consume(a, to_read); skip -= to_read; } if ((crc32_val & 0xffff) != crc32_expected) { #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Header CRC error"); return (ARCHIVE_FATAL); #endif } if (head_type == ENDARC_HEAD) return (ARCHIVE_EOF); break; case NEWSUB_HEAD: if ((ret = read_header(a, entry, head_type)) < ARCHIVE_WARN) return ret; break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad RAR file"); return (ARCHIVE_FATAL); } } } static int archive_read_format_rar_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct rar *rar = (struct rar *)(a->format->data); int ret; if (rar->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { rar->has_encrypted_entries = 0; } if (rar->bytes_unconsumed > 0) { /* Consume as much as the decompressor actually used. */ __archive_read_consume(a, rar->bytes_unconsumed); rar->bytes_unconsumed = 0; } *buff = NULL; if (rar->entry_eof || rar->offset_seek >= rar->unp_size) { *size = 0; *offset = rar->offset; - if (*offset < rar->unp_size) - *offset = rar->unp_size; return (ARCHIVE_EOF); } switch (rar->compression_method) { case COMPRESS_METHOD_STORE: ret = read_data_stored(a, buff, size, offset); break; case COMPRESS_METHOD_FASTEST: case COMPRESS_METHOD_FAST: case COMPRESS_METHOD_NORMAL: case COMPRESS_METHOD_GOOD: case COMPRESS_METHOD_BEST: ret = read_data_compressed(a, buff, size, offset, 0); if (ret != ARCHIVE_OK && ret != ARCHIVE_WARN) { __archive_ppmd7_functions.Ppmd7_Free(&rar->ppmd7_context); rar->start_new_table = 1; rar->ppmd_valid = 0; } break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported compression method for RAR file."); ret = ARCHIVE_FATAL; break; } return (ret); } static int archive_read_format_rar_read_data_skip(struct archive_read *a) { struct rar *rar; int64_t bytes_skipped; int ret; rar = (struct rar *)(a->format->data); if (rar->bytes_unconsumed > 0) { /* Consume as much as the decompressor actually used. */ __archive_read_consume(a, rar->bytes_unconsumed); rar->bytes_unconsumed = 0; } if (rar->bytes_remaining > 0) { bytes_skipped = __archive_read_consume(a, rar->bytes_remaining); if (bytes_skipped < 0) return (ARCHIVE_FATAL); } /* Compressed data to skip must be read from each header in a multivolume * archive. */ if (rar->main_flags & MHD_VOLUME && rar->file_flags & FHD_SPLIT_AFTER) { ret = archive_read_format_rar_read_header(a, a->entry); if (ret == (ARCHIVE_EOF)) ret = archive_read_format_rar_read_header(a, a->entry); if (ret != (ARCHIVE_OK)) return ret; return archive_read_format_rar_read_data_skip(a); } return (ARCHIVE_OK); } static int64_t archive_read_format_rar_seek_data(struct archive_read *a, int64_t offset, int whence) { int64_t client_offset, ret; size_t i; struct rar *rar = (struct rar *)(a->format->data); if (rar->compression_method == COMPRESS_METHOD_STORE) { /* Modify the offset for use with SEEK_SET */ switch (whence) { case SEEK_CUR: client_offset = rar->offset_seek; break; case SEEK_END: client_offset = rar->unp_size; break; case SEEK_SET: default: client_offset = 0; } client_offset += offset; if (client_offset < 0) { /* Can't seek past beginning of data block */ return -1; } else if (client_offset > rar->unp_size) { /* * Set the returned offset but only seek to the end of * the data block. */ rar->offset_seek = client_offset; client_offset = rar->unp_size; } client_offset += rar->dbo[0].start_offset; i = 0; while (i < rar->cursor) { i++; client_offset += rar->dbo[i].start_offset - rar->dbo[i-1].end_offset; } if (rar->main_flags & MHD_VOLUME) { /* Find the appropriate offset among the multivolume archive */ while (1) { if (client_offset < rar->dbo[rar->cursor].start_offset && rar->file_flags & FHD_SPLIT_BEFORE) { /* Search backwards for the correct data block */ if (rar->cursor == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Attempt to seek past beginning of RAR data block"); return (ARCHIVE_FAILED); } rar->cursor--; client_offset -= rar->dbo[rar->cursor+1].start_offset - rar->dbo[rar->cursor].end_offset; if (client_offset < rar->dbo[rar->cursor].start_offset) continue; ret = __archive_read_seek(a, rar->dbo[rar->cursor].start_offset - rar->dbo[rar->cursor].header_size, SEEK_SET); if (ret < (ARCHIVE_OK)) return ret; ret = archive_read_format_rar_read_header(a, a->entry); if (ret != (ARCHIVE_OK)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Error during seek of RAR file"); return (ARCHIVE_FAILED); } rar->cursor--; break; } else if (client_offset > rar->dbo[rar->cursor].end_offset && rar->file_flags & FHD_SPLIT_AFTER) { /* Search forward for the correct data block */ rar->cursor++; if (rar->cursor < rar->nodes && client_offset > rar->dbo[rar->cursor].end_offset) { client_offset += rar->dbo[rar->cursor].start_offset - rar->dbo[rar->cursor-1].end_offset; continue; } rar->cursor--; ret = __archive_read_seek(a, rar->dbo[rar->cursor].end_offset, SEEK_SET); if (ret < (ARCHIVE_OK)) return ret; ret = archive_read_format_rar_read_header(a, a->entry); if (ret == (ARCHIVE_EOF)) { rar->has_endarc_header = 1; ret = archive_read_format_rar_read_header(a, a->entry); } if (ret != (ARCHIVE_OK)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Error during seek of RAR file"); return (ARCHIVE_FAILED); } client_offset += rar->dbo[rar->cursor].start_offset - rar->dbo[rar->cursor-1].end_offset; continue; } break; } } ret = __archive_read_seek(a, client_offset, SEEK_SET); if (ret < (ARCHIVE_OK)) return ret; rar->bytes_remaining = rar->dbo[rar->cursor].end_offset - ret; i = rar->cursor; while (i > 0) { i--; ret -= rar->dbo[i+1].start_offset - rar->dbo[i].end_offset; } ret -= rar->dbo[0].start_offset; /* Always restart reading the file after a seek */ __archive_reset_read_data(&a->archive); rar->bytes_unconsumed = 0; rar->offset = 0; /* * If a seek past the end of file was requested, return the requested * offset. */ if (ret == rar->unp_size && rar->offset_seek > rar->unp_size) return rar->offset_seek; /* Return the new offset */ rar->offset_seek = ret; return rar->offset_seek; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Seeking of compressed RAR files is unsupported"); } return (ARCHIVE_FAILED); } static int archive_read_format_rar_cleanup(struct archive_read *a) { struct rar *rar; rar = (struct rar *)(a->format->data); free_codes(a); clear_filters(&rar->filters); free(rar->filename); free(rar->filename_save); free(rar->dbo); free(rar->unp_buffer); free(rar->lzss.window); __archive_ppmd7_functions.Ppmd7_Free(&rar->ppmd7_context); free(rar); (a->format->data) = NULL; return (ARCHIVE_OK); } static int read_header(struct archive_read *a, struct archive_entry *entry, char head_type) { const void *h; const char *p, *endp; struct rar *rar; struct rar_header rar_header; struct rar_file_header file_header; int64_t header_size; unsigned filename_size, end; char *filename; char *strp; char packed_size[8]; char unp_size[8]; int ttime; struct archive_string_conv *sconv, *fn_sconv; uint32_t crc32_computed, crc32_read; int ret = (ARCHIVE_OK), ret2; char *newptr; size_t newsize; rar = (struct rar *)(a->format->data); /* Setup a string conversion object for non-rar-unicode filenames. */ sconv = rar->opt_sconv; if (sconv == NULL) { if (!rar->init_default_conversion) { rar->sconv_default = archive_string_default_conversion_for_read( &(a->archive)); rar->init_default_conversion = 1; } sconv = rar->sconv_default; } if ((h = __archive_read_ahead(a, 7, NULL)) == NULL) return (ARCHIVE_FATAL); p = h; memcpy(&rar_header, p, sizeof(rar_header)); rar->file_flags = archive_le16dec(rar_header.flags); header_size = archive_le16dec(rar_header.size); if (header_size < (int64_t)sizeof(file_header) + 7) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size"); return (ARCHIVE_FATAL); } crc32_computed = crc32(0, (const unsigned char *)p + 2, 7 - 2); __archive_read_consume(a, 7); if (!(rar->file_flags & FHD_SOLID)) { rar->compression_method = 0; rar->packed_size = 0; rar->unp_size = 0; rar->mtime = 0; rar->ctime = 0; rar->atime = 0; rar->arctime = 0; rar->mode = 0; memset(&rar->salt, 0, sizeof(rar->salt)); rar->atime = 0; rar->ansec = 0; rar->ctime = 0; rar->cnsec = 0; rar->mtime = 0; rar->mnsec = 0; rar->arctime = 0; rar->arcnsec = 0; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "RAR solid archive support unavailable."); return (ARCHIVE_FATAL); } if ((h = __archive_read_ahead(a, (size_t)header_size - 7, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Failed to read full header content."); return (ARCHIVE_FATAL); } /* File Header CRC check. */ crc32_computed = crc32(crc32_computed, h, (unsigned)(header_size - 7)); crc32_read = archive_le16dec(rar_header.crc); if ((crc32_computed & 0xffff) != crc32_read) { #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Header CRC error"); return (ARCHIVE_FATAL); #endif } - /* If no CRC error, Go on parsing File Header. */ + /* If no CRC error, go on parsing File Header. */ p = h; endp = p + header_size - 7; memcpy(&file_header, p, sizeof(file_header)); p += sizeof(file_header); rar->compression_method = file_header.method; ttime = archive_le32dec(file_header.file_time); rar->mtime = get_time(ttime); rar->file_crc = archive_le32dec(file_header.file_crc); if (rar->file_flags & FHD_PASSWORD) { archive_entry_set_is_data_encrypted(entry, 1); rar->has_encrypted_entries = 1; archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "RAR encryption support unavailable."); /* Since it is only the data part itself that is encrypted we can at least extract information about the currently processed entry and don't need to return ARCHIVE_FATAL here. */ /*return (ARCHIVE_FATAL);*/ } if (rar->file_flags & FHD_LARGE) { if (p + 8 > endp) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size"); return (ARCHIVE_FATAL); } memcpy(packed_size, file_header.pack_size, 4); memcpy(packed_size + 4, p, 4); /* High pack size */ p += 4; memcpy(unp_size, file_header.unp_size, 4); memcpy(unp_size + 4, p, 4); /* High unpack size */ p += 4; rar->packed_size = archive_le64dec(&packed_size); rar->unp_size = archive_le64dec(&unp_size); } else { rar->packed_size = archive_le32dec(file_header.pack_size); rar->unp_size = archive_le32dec(file_header.unp_size); } if (rar->packed_size < 0 || rar->unp_size < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid sizes specified."); return (ARCHIVE_FATAL); } rar->bytes_remaining = rar->packed_size; /* TODO: RARv3 subblocks contain comments. For now the complete block is * consumed at the end. */ if (head_type == NEWSUB_HEAD) { size_t distance = p - (const char *)h; if (rar->packed_size > INT64_MAX - header_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Extended header size too large."); return (ARCHIVE_FATAL); } header_size += rar->packed_size; if ((uintmax_t)header_size > SIZE_MAX) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unable to read extended header data."); return (ARCHIVE_FATAL); } /* Make sure we have the extended data. */ if ((h = __archive_read_ahead(a, (size_t)header_size - 7, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Failed to read extended header data."); return (ARCHIVE_FATAL); } p = h; endp = p + header_size - 7; p += distance; } filename_size = archive_le16dec(file_header.name_size); if (p + filename_size > endp) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid filename size"); return (ARCHIVE_FATAL); } if (rar->filename_allocated < filename_size * 2 + 2) { newsize = filename_size * 2 + 2; newptr = realloc(rar->filename, newsize); if (newptr == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory."); return (ARCHIVE_FATAL); } rar->filename = newptr; rar->filename_allocated = newsize; } filename = rar->filename; memcpy(filename, p, filename_size); filename[filename_size] = '\0'; if (rar->file_flags & FHD_UNICODE) { if (filename_size != strlen(filename)) { unsigned char highbyte, flagbits, flagbyte; unsigned fn_end, offset; end = filename_size; fn_end = filename_size * 2; filename_size = 0; offset = (unsigned)strlen(filename) + 1; highbyte = offset >= end ? 0 : *(p + offset++); flagbits = 0; flagbyte = 0; while (offset < end && filename_size < fn_end) { if (!flagbits) { flagbyte = *(p + offset++); flagbits = 8; } flagbits -= 2; switch((flagbyte >> flagbits) & 3) { case 0: if (offset >= end) continue; filename[filename_size++] = '\0'; filename[filename_size++] = *(p + offset++); break; case 1: if (offset >= end) continue; filename[filename_size++] = highbyte; filename[filename_size++] = *(p + offset++); break; case 2: if (offset >= end - 1) { offset = end; continue; } filename[filename_size++] = *(p + offset + 1); filename[filename_size++] = *(p + offset); offset += 2; break; case 3: { char extra, high; uint8_t length; if (offset >= end) continue; length = *(p + offset++); if (length & 0x80) { if (offset >= end) continue; extra = *(p + offset++); high = (char)highbyte; } else extra = high = 0; length = (length & 0x7f) + 2; while (length && filename_size < fn_end) { unsigned cp = filename_size >> 1; filename[filename_size++] = high; filename[filename_size++] = p[cp] + extra; length--; } } break; } } if (filename_size > fn_end) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid filename"); return (ARCHIVE_FATAL); } filename[filename_size++] = '\0'; /* * Do not increment filename_size here as the computations below * add the space for the terminating NUL explicitly. */ filename[filename_size] = '\0'; /* Decoded unicode form is UTF-16BE, so we have to update a string * conversion object for it. */ if (rar->sconv_utf16be == NULL) { rar->sconv_utf16be = archive_string_conversion_from_charset( &a->archive, "UTF-16BE", 1); if (rar->sconv_utf16be == NULL) return (ARCHIVE_FATAL); } fn_sconv = rar->sconv_utf16be; strp = filename; while (memcmp(strp, "\x00\x00", 2)) { if (!memcmp(strp, "\x00\\", 2)) *(strp + 1) = '/'; strp += 2; } p += offset; } else { /* * If FHD_UNICODE is set but no unicode data, this file name form * is UTF-8, so we have to update a string conversion object for * it accordingly. */ if (rar->sconv_utf8 == NULL) { rar->sconv_utf8 = archive_string_conversion_from_charset( &a->archive, "UTF-8", 1); if (rar->sconv_utf8 == NULL) return (ARCHIVE_FATAL); } fn_sconv = rar->sconv_utf8; while ((strp = strchr(filename, '\\')) != NULL) *strp = '/'; p += filename_size; } } else { fn_sconv = sconv; while ((strp = strchr(filename, '\\')) != NULL) *strp = '/'; p += filename_size; } /* Split file in multivolume RAR. No more need to process header. */ if (rar->filename_save && filename_size == rar->filename_save_size && !memcmp(rar->filename, rar->filename_save, filename_size + 1)) { __archive_read_consume(a, header_size - 7); rar->br.avail_in = 0; rar->br.next_in = NULL; rar->cursor++; if (rar->cursor >= rar->nodes) { struct data_block_offsets *newdbo; newsize = sizeof(*rar->dbo) * (rar->nodes + 1); if ((newdbo = realloc(rar->dbo, newsize)) == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory."); return (ARCHIVE_FATAL); } rar->dbo = newdbo; rar->nodes++; rar->dbo[rar->cursor].header_size = header_size; rar->dbo[rar->cursor].start_offset = -1; rar->dbo[rar->cursor].end_offset = -1; } if (rar->dbo[rar->cursor].start_offset < 0) { if (rar->packed_size > INT64_MAX - a->filter->position) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unable to store offsets."); return (ARCHIVE_FATAL); } rar->dbo[rar->cursor].start_offset = a->filter->position; rar->dbo[rar->cursor].end_offset = rar->dbo[rar->cursor].start_offset + rar->packed_size; } return ret; } else if (rar->filename_must_match) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Mismatch of file parts split across multi-volume archive"); return (ARCHIVE_FATAL); } newsize = filename_size + 1; if ((newptr = realloc(rar->filename_save, newsize)) == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory."); return (ARCHIVE_FATAL); } rar->filename_save = newptr; memcpy(rar->filename_save, rar->filename, newsize); rar->filename_save_size = filename_size; /* Set info for seeking */ free(rar->dbo); if ((rar->dbo = calloc(1, sizeof(*rar->dbo))) == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory."); return (ARCHIVE_FATAL); } rar->dbo[0].header_size = header_size; rar->dbo[0].start_offset = -1; rar->dbo[0].end_offset = -1; rar->cursor = 0; rar->nodes = 1; if (rar->file_flags & FHD_SALT) { if (p + 8 > endp) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size"); return (ARCHIVE_FATAL); } memcpy(rar->salt, p, 8); p += 8; } if (rar->file_flags & FHD_EXTTIME) { if (read_exttime(p, rar, endp) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size"); return (ARCHIVE_FATAL); } } __archive_read_consume(a, header_size - 7); if (rar->packed_size > INT64_MAX - a->filter->position) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unable to store offsets."); return (ARCHIVE_FATAL); } rar->dbo[0].start_offset = a->filter->position; rar->dbo[0].end_offset = rar->dbo[0].start_offset + rar->packed_size; switch(file_header.host_os) { case OS_MSDOS: case OS_OS2: case OS_WIN32: rar->mode = (__LA_MODE_T)archive_le32dec(file_header.file_attr); if (rar->mode & FILE_ATTRIBUTE_DIRECTORY) rar->mode = AE_IFDIR | S_IXUSR | S_IXGRP | S_IXOTH; else rar->mode = AE_IFREG; rar->mode |= S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH; break; case OS_UNIX: case OS_MAC_OS: case OS_BEOS: rar->mode = (__LA_MODE_T)archive_le32dec(file_header.file_attr); break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unknown file attributes from RAR file's host OS"); return (ARCHIVE_FATAL); } rar->bytes_uncopied = rar->bytes_unconsumed = 0; rar->lzss.position = rar->offset = 0; rar->offset_seek = 0; rar->dictionary_size = 0; rar->offset_outgoing = 0; rar->br.cache_avail = 0; rar->br.avail_in = 0; rar->br.next_in = NULL; rar->crc_calculated = 0; rar->entry_eof = 0; rar->valid = 1; rar->is_ppmd_block = 0; rar->start_new_table = 1; free(rar->unp_buffer); rar->unp_buffer = NULL; rar->unp_offset = 0; rar->unp_buffer_size = UNP_BUFFER_SIZE; memset(rar->lengthtable, 0, sizeof(rar->lengthtable)); __archive_ppmd7_functions.Ppmd7_Free(&rar->ppmd7_context); rar->ppmd_valid = rar->ppmd_eod = 0; rar->filters.filterstart = INT64_MAX; /* Don't set any archive entries for non-file header types */ if (head_type == NEWSUB_HEAD) return ret; archive_entry_set_mtime(entry, rar->mtime, rar->mnsec); archive_entry_set_ctime(entry, rar->ctime, rar->cnsec); archive_entry_set_atime(entry, rar->atime, rar->ansec); archive_entry_set_size(entry, rar->unp_size); archive_entry_set_mode(entry, rar->mode); if (archive_entry_copy_pathname_l(entry, filename, filename_size, fn_sconv)) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Pathname cannot be converted from %s to current locale.", archive_string_conversion_charset_name(fn_sconv)); ret = (ARCHIVE_WARN); } if (((rar->mode) & AE_IFMT) == AE_IFLNK) { /* Make sure a symbolic-link file does not have its body. */ rar->bytes_remaining = 0; archive_entry_set_size(entry, 0); /* Read a symbolic-link name. */ if ((ret2 = read_symlink_stored(a, entry, sconv)) < (ARCHIVE_WARN)) return ret2; if (ret > ret2) ret = ret2; } if (rar->bytes_remaining == 0) rar->entry_eof = 1; return ret; } static time_t get_time(int ttime) { struct tm tm; tm.tm_sec = 2 * (ttime & 0x1f); tm.tm_min = (ttime >> 5) & 0x3f; tm.tm_hour = (ttime >> 11) & 0x1f; tm.tm_mday = (ttime >> 16) & 0x1f; tm.tm_mon = ((ttime >> 21) & 0x0f) - 1; tm.tm_year = ((ttime >> 25) & 0x7f) + 80; tm.tm_isdst = -1; return mktime(&tm); } static int read_exttime(const char *p, struct rar *rar, const char *endp) { unsigned rmode, flags, rem, j, count; int ttime, i; struct tm *tm; time_t t; long nsec; #if defined(HAVE_LOCALTIME_R) || defined(HAVE_LOCALTIME_S) struct tm tmbuf; #endif if (p + 2 > endp) return (-1); flags = archive_le16dec(p); p += 2; for (i = 3; i >= 0; i--) { t = 0; if (i == 3) t = rar->mtime; rmode = flags >> i * 4; if (rmode & 8) { if (!t) { if (p + 4 > endp) return (-1); ttime = archive_le32dec(p); t = get_time(ttime); p += 4; } rem = 0; count = rmode & 3; if (p + count > endp) return (-1); for (j = 0; j < count; j++) { rem = (((unsigned)(unsigned char)*p) << 16) | (rem >> 8); p++; } #if defined(HAVE_LOCALTIME_S) tm = localtime_s(&tmbuf, &t) ? NULL : &tmbuf; #elif defined(HAVE_LOCALTIME_R) tm = localtime_r(&t, &tmbuf); #else tm = localtime(&t); #endif nsec = tm->tm_sec + rem / NS_UNIT; if (rmode & 4) { tm->tm_sec++; t = mktime(tm); } if (i == 3) { rar->mtime = t; rar->mnsec = nsec; } else if (i == 2) { rar->ctime = t; rar->cnsec = nsec; } else if (i == 1) { rar->atime = t; rar->ansec = nsec; } else { rar->arctime = t; rar->arcnsec = nsec; } } } return (0); } static int read_symlink_stored(struct archive_read *a, struct archive_entry *entry, struct archive_string_conv *sconv) { const void *h; const char *p; struct rar *rar; int ret = (ARCHIVE_OK); rar = (struct rar *)(a->format->data); if ((uintmax_t)rar->packed_size > SIZE_MAX) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unable to read link."); return (ARCHIVE_FATAL); } if ((h = rar_read_ahead(a, (size_t)rar->packed_size, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Failed to read link."); return (ARCHIVE_FATAL); } p = h; if (archive_entry_copy_symlink_l(entry, p, (size_t)rar->packed_size, sconv)) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for link"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "link cannot be converted from %s to current locale.", archive_string_conversion_charset_name(sconv)); ret = (ARCHIVE_WARN); } __archive_read_consume(a, rar->packed_size); return ret; } static int read_data_stored(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct rar *rar; ssize_t bytes_avail; rar = (struct rar *)(a->format->data); if (rar->bytes_remaining == 0 && !(rar->main_flags & MHD_VOLUME && rar->file_flags & FHD_SPLIT_AFTER)) { *buff = NULL; *size = 0; *offset = rar->offset; if (rar->file_crc != rar->crc_calculated) { #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "File CRC error"); return (ARCHIVE_FATAL); #endif } rar->entry_eof = 1; return (ARCHIVE_EOF); } *buff = rar_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); return (ARCHIVE_FATAL); } *size = bytes_avail; *offset = rar->offset; rar->offset += bytes_avail; rar->offset_seek += bytes_avail; rar->bytes_remaining -= bytes_avail; rar->bytes_unconsumed = bytes_avail; /* Calculate File CRC. */ rar->crc_calculated = crc32(rar->crc_calculated, *buff, (unsigned)bytes_avail); return (ARCHIVE_OK); } static int read_data_compressed(struct archive_read *a, const void **buff, size_t *size, int64_t *offset, size_t looper) { if (looper++ > MAX_COMPRESS_DEPTH) return (ARCHIVE_FATAL); struct rar *rar; int64_t start, end; size_t bs; int ret = (ARCHIVE_OK), sym, code, lzss_offset, length, i; rar = (struct rar *)(a->format->data); do { if (!rar->valid) return (ARCHIVE_FATAL); if (rar->filters.bytes_ready > 0) { /* Flush unp_buffer first */ if (rar->unp_offset > 0) { *buff = rar->unp_buffer; *size = rar->unp_offset; rar->unp_offset = 0; *offset = rar->offset_outgoing; rar->offset_outgoing += *size; } else { *buff = rar->filters.bytes; *size = rar->filters.bytes_ready; rar->offset += *size; *offset = rar->offset_outgoing; rar->offset_outgoing += *size; rar->filters.bytes_ready -= *size; rar->filters.bytes += *size; } goto ending_block; } if (rar->ppmd_eod || (rar->dictionary_size && rar->offset >= rar->unp_size)) { if (rar->unp_offset > 0) { /* * We have unprocessed extracted data. write it out. */ *buff = rar->unp_buffer; *size = rar->unp_offset; *offset = rar->offset_outgoing; rar->offset_outgoing += *size; /* Calculate File CRC. */ rar->crc_calculated = crc32(rar->crc_calculated, *buff, (unsigned)*size); rar->unp_offset = 0; return (ARCHIVE_OK); } *buff = NULL; *size = 0; *offset = rar->offset; if (rar->file_crc != rar->crc_calculated) { #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "File CRC error"); return (ARCHIVE_FATAL); #endif } rar->entry_eof = 1; return (ARCHIVE_EOF); } if (!rar->is_ppmd_block && rar->dictionary_size && rar->bytes_uncopied > 0) { if (rar->bytes_uncopied > (rar->unp_buffer_size - rar->unp_offset)) bs = rar->unp_buffer_size - rar->unp_offset; else bs = (size_t)rar->bytes_uncopied; ret = copy_from_lzss_window_to_unp(a, buff, rar->offset, bs); if (ret != ARCHIVE_OK) return (ret); rar->offset += bs; rar->bytes_uncopied -= bs; if (*buff != NULL) { rar->unp_offset = 0; *size = rar->unp_buffer_size; *offset = rar->offset_outgoing; rar->offset_outgoing += *size; /* Calculate File CRC. */ rar->crc_calculated = crc32(rar->crc_calculated, *buff, (unsigned)*size); return (ret); } continue; } if (rar->filters.lastend == rar->filters.filterstart) { if (!run_filters(a)) return (ARCHIVE_FATAL); continue; } if (!rar->br.next_in && (ret = rar_br_preparation(a, &(rar->br))) < ARCHIVE_WARN) return (ret); if (rar->start_new_table && ((ret = parse_codes(a)) < (ARCHIVE_WARN))) return (ret); if (rar->is_ppmd_block) { if ((sym = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( &rar->ppmd7_context, &rar->range_dec.p)) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid symbol"); return (ARCHIVE_FATAL); } if(sym != rar->ppmd_escape) { lzss_emit_literal(rar, sym); rar->bytes_uncopied++; } else { if ((code = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( &rar->ppmd7_context, &rar->range_dec.p)) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid symbol"); return (ARCHIVE_FATAL); } switch(code) { case 0: rar->start_new_table = 1; return read_data_compressed(a, buff, size, offset, looper); case 2: rar->ppmd_eod = 1;/* End Of ppmd Data. */ continue; case 3: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Parsing filters is unsupported."); return (ARCHIVE_FAILED); case 4: lzss_offset = 0; for (i = 2; i >= 0; i--) { if ((code = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( &rar->ppmd7_context, &rar->range_dec.p)) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid symbol"); return (ARCHIVE_FATAL); } lzss_offset |= code << (i * 8); } if ((length = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( &rar->ppmd7_context, &rar->range_dec.p)) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid symbol"); return (ARCHIVE_FATAL); } lzss_emit_match(rar, lzss_offset + 2, length + 32); rar->bytes_uncopied += length + 32; break; case 5: if ((length = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( &rar->ppmd7_context, &rar->range_dec.p)) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid symbol"); return (ARCHIVE_FATAL); } lzss_emit_match(rar, 1, length + 4); rar->bytes_uncopied += length + 4; break; default: lzss_emit_literal(rar, sym); rar->bytes_uncopied++; } } } else { start = rar->offset; end = start + rar->dictionary_size; /* We don't want to overflow the window and overwrite data that we write * at 'start'. Therefore, reduce the end length by the maximum match size, * which is 260 bytes. You can compute this maximum by looking at the * definition of 'expand', in particular when 'symbol >= 271'. */ /* NOTE: It's possible for 'dictionary_size' to be less than this 260 * value, however that will only be the case when 'unp_size' is small, * which should only happen when the entry size is small and there's no * risk of overflowing the buffer */ if (rar->dictionary_size > 260) { end -= 260; } if (rar->filters.filterstart < end) { end = rar->filters.filterstart; } ret = expand(a, &end); if (ret != ARCHIVE_OK) return (ret); rar->bytes_uncopied = end - start; rar->filters.lastend = end; if (rar->filters.lastend != rar->filters.filterstart && rar->bytes_uncopied == 0) { /* Broken RAR files cause this case. * NOTE: If this case were possible on a normal RAR file * we would find out where it was actually bad and * what we would do to solve it. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Internal error extracting RAR file"); return (ARCHIVE_FATAL); } } if (rar->bytes_uncopied > (rar->unp_buffer_size - rar->unp_offset)) bs = rar->unp_buffer_size - rar->unp_offset; else bs = (size_t)rar->bytes_uncopied; ret = copy_from_lzss_window_to_unp(a, buff, rar->offset, bs); if (ret != ARCHIVE_OK) return (ret); rar->offset += bs; rar->bytes_uncopied -= bs; /* * If *buff is NULL, it means unp_buffer is not full. * So we have to continue extracting a RAR file. */ } while (*buff == NULL); rar->unp_offset = 0; *size = rar->unp_buffer_size; *offset = rar->offset_outgoing; rar->offset_outgoing += *size; ending_block: /* Calculate File CRC. */ rar->crc_calculated = crc32(rar->crc_calculated, *buff, (unsigned)*size); return ret; } static int parse_codes(struct archive_read *a) { int i, j, val, n, r; unsigned char bitlengths[MAX_SYMBOLS], zerocount, ppmd_flags; unsigned int maxorder; struct huffman_code precode; struct rar *rar = (struct rar *)(a->format->data); struct rar_br *br = &(rar->br); free_codes(a); /* Skip to the next byte */ rar_br_consume_unaligned_bits(br); /* PPMd block flag */ if (!rar_br_read_ahead(a, br, 1)) goto truncated_data; if ((rar->is_ppmd_block = rar_br_bits(br, 1)) != 0) { rar_br_consume(br, 1); if (!rar_br_read_ahead(a, br, 7)) goto truncated_data; ppmd_flags = rar_br_bits(br, 7); rar_br_consume(br, 7); /* Memory is allocated in MB */ if (ppmd_flags & 0x20) { if (!rar_br_read_ahead(a, br, 8)) goto truncated_data; rar->dictionary_size = (rar_br_bits(br, 8) + 1) << 20; rar_br_consume(br, 8); } if (ppmd_flags & 0x40) { if (!rar_br_read_ahead(a, br, 8)) goto truncated_data; rar->ppmd_escape = rar->ppmd7_context.InitEsc = rar_br_bits(br, 8); rar_br_consume(br, 8); } else rar->ppmd_escape = 2; if (ppmd_flags & 0x20) { maxorder = (ppmd_flags & 0x1F) + 1; if(maxorder > 16) maxorder = 16 + (maxorder - 16) * 3; if (maxorder == 1) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); return (ARCHIVE_FATAL); } - /* Make sure ppmd7_contest is freed before Ppmd7_Construct - * because reading a broken file cause this abnormal sequence. */ + /* Make sure ppmd7_context is freed before Ppmd7_Construct + * because reading a broken file causes this abnormal sequence. */ __archive_ppmd7_functions.Ppmd7_Free(&rar->ppmd7_context); rar->bytein.a = a; rar->bytein.Read = &ppmd_read; __archive_ppmd7_functions.PpmdRAR_RangeDec_CreateVTable(&rar->range_dec); rar->range_dec.Stream = &rar->bytein; __archive_ppmd7_functions.Ppmd7_Construct(&rar->ppmd7_context); if (rar->dictionary_size == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid zero dictionary size"); return (ARCHIVE_FATAL); } if (!__archive_ppmd7_functions.Ppmd7_Alloc(&rar->ppmd7_context, rar->dictionary_size)) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } if (!__archive_ppmd7_functions.PpmdRAR_RangeDec_Init(&rar->range_dec)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unable to initialize PPMd range decoder"); return (ARCHIVE_FATAL); } __archive_ppmd7_functions.Ppmd7_Init(&rar->ppmd7_context, maxorder); rar->ppmd_valid = 1; } else { if (!rar->ppmd_valid) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid PPMd sequence"); return (ARCHIVE_FATAL); } if (!__archive_ppmd7_functions.PpmdRAR_RangeDec_Init(&rar->range_dec)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unable to initialize PPMd range decoder"); return (ARCHIVE_FATAL); } } } else { rar_br_consume(br, 1); /* Keep existing table flag */ if (!rar_br_read_ahead(a, br, 1)) goto truncated_data; if (!rar_br_bits(br, 1)) memset(rar->lengthtable, 0, sizeof(rar->lengthtable)); rar_br_consume(br, 1); memset(&bitlengths, 0, sizeof(bitlengths)); for (i = 0; i < MAX_SYMBOLS;) { if (!rar_br_read_ahead(a, br, 4)) goto truncated_data; bitlengths[i++] = rar_br_bits(br, 4); rar_br_consume(br, 4); if (bitlengths[i-1] == 0xF) { if (!rar_br_read_ahead(a, br, 4)) goto truncated_data; zerocount = rar_br_bits(br, 4); rar_br_consume(br, 4); if (zerocount) { i--; for (j = 0; j < zerocount + 2 && i < MAX_SYMBOLS; j++) bitlengths[i++] = 0; } } } memset(&precode, 0, sizeof(precode)); r = create_code(a, &precode, bitlengths, MAX_SYMBOLS, MAX_SYMBOL_LENGTH); if (r != ARCHIVE_OK) { free(precode.tree); free(precode.table); return (r); } for (i = 0; i < HUFFMAN_TABLE_SIZE;) { if ((val = read_next_symbol(a, &precode)) < 0) { free(precode.tree); free(precode.table); return (ARCHIVE_FATAL); } if (val < 16) { rar->lengthtable[i] = (rar->lengthtable[i] + val) & 0xF; i++; } else if (val < 18) { if (i == 0) { free(precode.tree); free(precode.table); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Internal error extracting RAR file."); return (ARCHIVE_FATAL); } if(val == 16) { if (!rar_br_read_ahead(a, br, 3)) { free(precode.tree); free(precode.table); goto truncated_data; } n = rar_br_bits(br, 3) + 3; rar_br_consume(br, 3); } else { if (!rar_br_read_ahead(a, br, 7)) { free(precode.tree); free(precode.table); goto truncated_data; } n = rar_br_bits(br, 7) + 11; rar_br_consume(br, 7); } for (j = 0; j < n && i < HUFFMAN_TABLE_SIZE; j++) { rar->lengthtable[i] = rar->lengthtable[i-1]; i++; } } else { if(val == 18) { if (!rar_br_read_ahead(a, br, 3)) { free(precode.tree); free(precode.table); goto truncated_data; } n = rar_br_bits(br, 3) + 3; rar_br_consume(br, 3); } else { if (!rar_br_read_ahead(a, br, 7)) { free(precode.tree); free(precode.table); goto truncated_data; } n = rar_br_bits(br, 7) + 11; rar_br_consume(br, 7); } for(j = 0; j < n && i < HUFFMAN_TABLE_SIZE; j++) rar->lengthtable[i++] = 0; } } free(precode.tree); free(precode.table); r = create_code(a, &rar->maincode, &rar->lengthtable[0], MAINCODE_SIZE, MAX_SYMBOL_LENGTH); if (r != ARCHIVE_OK) return (r); r = create_code(a, &rar->offsetcode, &rar->lengthtable[MAINCODE_SIZE], OFFSETCODE_SIZE, MAX_SYMBOL_LENGTH); if (r != ARCHIVE_OK) return (r); r = create_code(a, &rar->lowoffsetcode, &rar->lengthtable[MAINCODE_SIZE + OFFSETCODE_SIZE], LOWOFFSETCODE_SIZE, MAX_SYMBOL_LENGTH); if (r != ARCHIVE_OK) return (r); r = create_code(a, &rar->lengthcode, &rar->lengthtable[MAINCODE_SIZE + OFFSETCODE_SIZE + LOWOFFSETCODE_SIZE], LENGTHCODE_SIZE, MAX_SYMBOL_LENGTH); if (r != ARCHIVE_OK) return (r); } if (!rar->dictionary_size || !rar->lzss.window) { /* Seems as though dictionary sizes are not used. Even so, minimize * memory usage as much as possible. */ void *new_window; unsigned int new_size; if (rar->unp_size >= DICTIONARY_MAX_SIZE) new_size = DICTIONARY_MAX_SIZE; else new_size = rar_fls((unsigned int)rar->unp_size) << 1; if (new_size == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Zero window size is invalid."); return (ARCHIVE_FATAL); } new_window = realloc(rar->lzss.window, new_size); if (new_window == NULL) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for uncompressed data."); return (ARCHIVE_FATAL); } rar->lzss.window = (unsigned char *)new_window; rar->dictionary_size = new_size; memset(rar->lzss.window, 0, rar->dictionary_size); rar->lzss.mask = rar->dictionary_size - 1; } rar->start_new_table = 0; return (ARCHIVE_OK); truncated_data: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); rar->valid = 0; return (ARCHIVE_FATAL); } static void free_codes(struct archive_read *a) { struct rar *rar = (struct rar *)(a->format->data); free(rar->maincode.tree); free(rar->offsetcode.tree); free(rar->lowoffsetcode.tree); free(rar->lengthcode.tree); free(rar->maincode.table); free(rar->offsetcode.table); free(rar->lowoffsetcode.table); free(rar->lengthcode.table); memset(&rar->maincode, 0, sizeof(rar->maincode)); memset(&rar->offsetcode, 0, sizeof(rar->offsetcode)); memset(&rar->lowoffsetcode, 0, sizeof(rar->lowoffsetcode)); memset(&rar->lengthcode, 0, sizeof(rar->lengthcode)); } static int read_next_symbol(struct archive_read *a, struct huffman_code *code) { unsigned char bit; unsigned int bits; int length, value, node; struct rar *rar; struct rar_br *br; if (!code->table) { if (make_table(a, code) != (ARCHIVE_OK)) return -1; } rar = (struct rar *)(a->format->data); br = &(rar->br); /* Look ahead (peek) at bits */ if (!rar_br_read_ahead(a, br, code->tablesize)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); rar->valid = 0; return -1; } bits = rar_br_bits(br, code->tablesize); length = code->table[bits].length; value = code->table[bits].value; if (length < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid prefix code in bitstream"); return -1; } if (length <= code->tablesize) { /* Skip length bits */ rar_br_consume(br, length); return value; } /* Skip tablesize bits */ rar_br_consume(br, code->tablesize); node = value; while (code->tree[node].branches[0] != code->tree[node].branches[1]) { if (!rar_br_read_ahead(a, br, 1)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); rar->valid = 0; return -1; } bit = rar_br_bits(br, 1); rar_br_consume(br, 1); if (code->tree[node].branches[bit] < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid prefix code in bitstream"); return -1; } node = code->tree[node].branches[bit]; } return code->tree[node].branches[0]; } static int create_code(struct archive_read *a, struct huffman_code *code, unsigned char *lengths, int numsymbols, char maxlength) { int i, j, codebits = 0, symbolsleft = numsymbols; code->numentries = 0; code->numallocatedentries = 0; if (new_node(code) < 0) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for node data."); return (ARCHIVE_FATAL); } code->numentries = 1; code->minlength = INT_MAX; code->maxlength = INT_MIN; codebits = 0; for(i = 1; i <= maxlength; i++) { for(j = 0; j < numsymbols; j++) { if (lengths[j] != i) continue; if (add_value(a, code, j, codebits, i) != ARCHIVE_OK) return (ARCHIVE_FATAL); codebits++; if (--symbolsleft <= 0) break; } if (symbolsleft <= 0) break; codebits <<= 1; } return (ARCHIVE_OK); } static int add_value(struct archive_read *a, struct huffman_code *code, int value, int codebits, int length) { int lastnode, bitpos, bit; /* int repeatpos, repeatnode, nextnode; */ free(code->table); code->table = NULL; if(length > code->maxlength) code->maxlength = length; if(length < code->minlength) code->minlength = length; /* * Dead code, repeatpos was is -1 * repeatpos = -1; if (repeatpos == 0 || (repeatpos >= 0 && (((codebits >> (repeatpos - 1)) & 3) == 0 || ((codebits >> (repeatpos - 1)) & 3) == 3))) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid repeat position"); return (ARCHIVE_FATAL); } */ lastnode = 0; for (bitpos = length - 1; bitpos >= 0; bitpos--) { bit = (codebits >> bitpos) & 1; /* Leaf node check */ if (code->tree[lastnode].branches[0] == code->tree[lastnode].branches[1]) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Prefix found"); return (ARCHIVE_FATAL); } /* * Dead code, repeatpos was -1, bitpos >=0 * if (bitpos == repeatpos) { * Open branch check * if (!(code->tree[lastnode].branches[bit] < 0)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid repeating code"); return (ARCHIVE_FATAL); } if ((repeatnode = new_node(code)) < 0) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for node data."); return (ARCHIVE_FATAL); } if ((nextnode = new_node(code)) < 0) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for node data."); return (ARCHIVE_FATAL); } * Set branches * code->tree[lastnode].branches[bit] = repeatnode; code->tree[repeatnode].branches[bit] = repeatnode; code->tree[repeatnode].branches[bit^1] = nextnode; lastnode = nextnode; bitpos++; * terminating bit already handled, skip it * } else { */ /* Open branch check */ if (code->tree[lastnode].branches[bit] < 0) { if (new_node(code) < 0) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for node data."); return (ARCHIVE_FATAL); } code->tree[lastnode].branches[bit] = code->numentries++; } /* set to branch */ lastnode = code->tree[lastnode].branches[bit]; /* } */ } if (!(code->tree[lastnode].branches[0] == -1 && code->tree[lastnode].branches[1] == -2)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Prefix found"); return (ARCHIVE_FATAL); } /* Set leaf value */ code->tree[lastnode].branches[0] = value; code->tree[lastnode].branches[1] = value; return (ARCHIVE_OK); } static int new_node(struct huffman_code *code) { void *new_tree; if (code->numallocatedentries == code->numentries) { int new_num_entries = 256; if (code->numentries > 0) { new_num_entries = code->numentries * 2; } new_tree = realloc(code->tree, new_num_entries * sizeof(*code->tree)); if (new_tree == NULL) return (-1); code->tree = (struct huffman_tree_node *)new_tree; code->numallocatedentries = new_num_entries; } code->tree[code->numentries].branches[0] = -1; code->tree[code->numentries].branches[1] = -2; return 1; } static int make_table(struct archive_read *a, struct huffman_code *code) { if (code->maxlength < code->minlength || code->maxlength > 10) code->tablesize = 10; else code->tablesize = code->maxlength; code->table = calloc(((size_t)1U) << code->tablesize, sizeof(*code->table)); return make_table_recurse(a, code, 0, code->table, 0, code->tablesize); } static int make_table_recurse(struct archive_read *a, struct huffman_code *code, int node, struct huffman_table_entry *table, int depth, int maxdepth) { int currtablesize, i, ret = (ARCHIVE_OK); if (!code->tree) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Huffman tree was not created."); return (ARCHIVE_FATAL); } if (node < 0 || node >= code->numentries) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid location to Huffman tree specified."); return (ARCHIVE_FATAL); } currtablesize = 1 << (maxdepth - depth); if (code->tree[node].branches[0] == code->tree[node].branches[1]) { for(i = 0; i < currtablesize; i++) { table[i].length = depth; table[i].value = code->tree[node].branches[0]; } } /* * Dead code, node >= 0 * else if (node < 0) { for(i = 0; i < currtablesize; i++) table[i].length = -1; } */ else { if(depth == maxdepth) { table[0].length = maxdepth + 1; table[0].value = node; } else { ret |= make_table_recurse(a, code, code->tree[node].branches[0], table, depth + 1, maxdepth); ret |= make_table_recurse(a, code, code->tree[node].branches[1], table + currtablesize / 2, depth + 1, maxdepth); } } return ret; } static int expand(struct archive_read *a, int64_t *end) { static const unsigned char lengthbases[] = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 24, 28, 32, 40, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224 }; static const unsigned char lengthbits[] = { 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5 }; static const int lengthb_min = minimum( (int)(sizeof(lengthbases)/sizeof(lengthbases[0])), (int)(sizeof(lengthbits)/sizeof(lengthbits[0])) ); static const unsigned int offsetbases[] = { 0, 1, 2, 3, 4, 6, 8, 12, 16, 24, 32, 48, 64, 96, 128, 192, 256, 384, 512, 768, 1024, 1536, 2048, 3072, 4096, 6144, 8192, 12288, 16384, 24576, 32768, 49152, 65536, 98304, 131072, 196608, 262144, 327680, 393216, 458752, 524288, 589824, 655360, 720896, 786432, 851968, 917504, 983040, 1048576, 1310720, 1572864, 1835008, 2097152, 2359296, 2621440, 2883584, 3145728, 3407872, 3670016, 3932160 }; static const unsigned char offsetbits[] = { 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9, 10, 10, 11, 11, 12, 12, 13, 13, 14, 14, 15, 15, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18 }; static const int offsetb_min = minimum( (int)(sizeof(offsetbases)/sizeof(offsetbases[0])), (int)(sizeof(offsetbits)/sizeof(offsetbits[0])) ); static const unsigned char shortbases[] = { 0, 4, 8, 16, 32, 64, 128, 192 }; static const unsigned char shortbits[] = { 2, 2, 3, 4, 5, 6, 6, 6 }; int symbol, offs, len, offsindex, lensymbol, i, offssymbol, lowoffsetsymbol; unsigned char newfile; struct rar *rar = (struct rar *)(a->format->data); struct rar_br *br = &(rar->br); if (rar->filters.filterstart < *end) *end = rar->filters.filterstart; while (1) { if(lzss_position(&rar->lzss) >= *end) { return (ARCHIVE_OK); } if(rar->is_ppmd_block) { *end = lzss_position(&rar->lzss); return (ARCHIVE_OK); } if ((symbol = read_next_symbol(a, &rar->maincode)) < 0) goto bad_data; if (symbol < 256) { lzss_emit_literal(rar, (uint8_t)symbol); continue; } else if (symbol == 256) { if (!rar_br_read_ahead(a, br, 1)) goto truncated_data; newfile = !rar_br_bits(br, 1); rar_br_consume(br, 1); if(newfile) { rar->start_new_block = 1; if (!rar_br_read_ahead(a, br, 1)) goto truncated_data; rar->start_new_table = rar_br_bits(br, 1); rar_br_consume(br, 1); *end = lzss_position(&rar->lzss); return (ARCHIVE_OK); } else { if (parse_codes(a) != ARCHIVE_OK) goto bad_data; continue; } } else if(symbol==257) { if (!read_filter(a, end)) goto bad_data; continue; } else if(symbol==258) { if(rar->lastlength == 0) continue; offs = rar->lastoffset; len = rar->lastlength; } else if (symbol <= 262) { offsindex = symbol - 259; offs = rar->oldoffset[offsindex]; if ((lensymbol = read_next_symbol(a, &rar->lengthcode)) < 0) goto bad_data; if (lensymbol >= lengthb_min) goto bad_data; len = lengthbases[lensymbol] + 2; if (lengthbits[lensymbol] > 0) { if (!rar_br_read_ahead(a, br, lengthbits[lensymbol])) goto truncated_data; len += rar_br_bits(br, lengthbits[lensymbol]); rar_br_consume(br, lengthbits[lensymbol]); } for (i = offsindex; i > 0; i--) rar->oldoffset[i] = rar->oldoffset[i-1]; rar->oldoffset[0] = offs; } else if(symbol<=270) { offs = shortbases[symbol-263] + 1; if(shortbits[symbol-263] > 0) { if (!rar_br_read_ahead(a, br, shortbits[symbol-263])) goto truncated_data; offs += rar_br_bits(br, shortbits[symbol-263]); rar_br_consume(br, shortbits[symbol-263]); } len = 2; for(i = 3; i > 0; i--) rar->oldoffset[i] = rar->oldoffset[i-1]; rar->oldoffset[0] = offs; } else { if (symbol-271 >= lengthb_min) goto bad_data; len = lengthbases[symbol-271]+3; if(lengthbits[symbol-271] > 0) { if (!rar_br_read_ahead(a, br, lengthbits[symbol-271])) goto truncated_data; len += rar_br_bits(br, lengthbits[symbol-271]); rar_br_consume(br, lengthbits[symbol-271]); } if ((offssymbol = read_next_symbol(a, &rar->offsetcode)) < 0) goto bad_data; if (offssymbol >= offsetb_min) goto bad_data; offs = offsetbases[offssymbol]+1; if(offsetbits[offssymbol] > 0) { if(offssymbol > 9) { if(offsetbits[offssymbol] > 4) { if (!rar_br_read_ahead(a, br, offsetbits[offssymbol] - 4)) goto truncated_data; offs += rar_br_bits(br, offsetbits[offssymbol] - 4) << 4; rar_br_consume(br, offsetbits[offssymbol] - 4); } if(rar->numlowoffsetrepeats > 0) { rar->numlowoffsetrepeats--; offs += rar->lastlowoffset; } else { if ((lowoffsetsymbol = read_next_symbol(a, &rar->lowoffsetcode)) < 0) goto bad_data; if(lowoffsetsymbol == 16) { rar->numlowoffsetrepeats = 15; offs += rar->lastlowoffset; } else { offs += lowoffsetsymbol; rar->lastlowoffset = lowoffsetsymbol; } } } else { if (!rar_br_read_ahead(a, br, offsetbits[offssymbol])) goto truncated_data; offs += rar_br_bits(br, offsetbits[offssymbol]); rar_br_consume(br, offsetbits[offssymbol]); } } if (offs >= 0x40000) len++; if (offs >= 0x2000) len++; for(i = 3; i > 0; i--) rar->oldoffset[i] = rar->oldoffset[i-1]; rar->oldoffset[0] = offs; } rar->lastoffset = offs; rar->lastlength = len; lzss_emit_match(rar, rar->lastoffset, rar->lastlength); } truncated_data: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); rar->valid = 0; return (ARCHIVE_FATAL); bad_data: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad RAR file data"); return (ARCHIVE_FATAL); } static int copy_from_lzss_window(struct archive_read *a, uint8_t *buffer, int64_t startpos, int length) { int windowoffs, firstpart; struct rar *rar = (struct rar *)(a->format->data); windowoffs = lzss_offset_for_position(&rar->lzss, startpos); firstpart = lzss_size(&rar->lzss) - windowoffs; if (firstpart < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad RAR file data"); return (ARCHIVE_FATAL); } if (firstpart < length) { memcpy(buffer, &rar->lzss.window[windowoffs], firstpart); memcpy(buffer + firstpart, &rar->lzss.window[0], length - firstpart); } else { memcpy(buffer, &rar->lzss.window[windowoffs], length); } return (ARCHIVE_OK); } static int copy_from_lzss_window_to_unp(struct archive_read *a, const void **buffer, int64_t startpos, size_t length) { int windowoffs, firstpart; struct rar *rar = (struct rar *)(a->format->data); if (length > rar->unp_buffer_size) { goto fatal; } if (!rar->unp_buffer) { if ((rar->unp_buffer = malloc(rar->unp_buffer_size)) == NULL) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for uncompressed data."); return (ARCHIVE_FATAL); } } windowoffs = lzss_offset_for_position(&rar->lzss, startpos); if(windowoffs + length <= (size_t)lzss_size(&rar->lzss)) { memcpy(&rar->unp_buffer[rar->unp_offset], &rar->lzss.window[windowoffs], length); } else if (length <= (size_t)lzss_size(&rar->lzss)) { firstpart = lzss_size(&rar->lzss) - windowoffs; if (firstpart < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad RAR file data"); return (ARCHIVE_FATAL); } if ((size_t)firstpart < length) { memcpy(&rar->unp_buffer[rar->unp_offset], &rar->lzss.window[windowoffs], firstpart); memcpy(&rar->unp_buffer[rar->unp_offset + firstpart], &rar->lzss.window[0], length - firstpart); } else { memcpy(&rar->unp_buffer[rar->unp_offset], &rar->lzss.window[windowoffs], length); } } else { goto fatal; } rar->unp_offset += (unsigned int) length; if (rar->unp_offset >= rar->unp_buffer_size) *buffer = rar->unp_buffer; else *buffer = NULL; return (ARCHIVE_OK); fatal: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad RAR file data"); return (ARCHIVE_FATAL); } static const void * rar_read_ahead(struct archive_read *a, size_t min, ssize_t *avail) { struct rar *rar = (struct rar *)(a->format->data); const void *h; int ret; again: h = __archive_read_ahead(a, min, avail); if (avail) { if (a->archive.read_data_is_posix_read && *avail > (ssize_t)a->archive.read_data_requested) *avail = a->archive.read_data_requested; if (*avail > rar->bytes_remaining) *avail = (ssize_t)rar->bytes_remaining; if (*avail < 0) return NULL; else if (*avail == 0 && rar->main_flags & MHD_VOLUME && rar->file_flags & FHD_SPLIT_AFTER) { rar->filename_must_match = 1; ret = archive_read_format_rar_read_header(a, a->entry); if (ret == (ARCHIVE_EOF)) { rar->has_endarc_header = 1; ret = archive_read_format_rar_read_header(a, a->entry); } rar->filename_must_match = 0; if (ret != (ARCHIVE_OK)) return NULL; goto again; } } return h; } static int parse_filter(struct archive_read *a, const uint8_t *bytes, uint16_t length, uint8_t flags) { struct rar *rar = (struct rar *)(a->format->data); struct rar_filters *filters = &rar->filters; struct memory_bit_reader br = { 0 }; struct rar_program_code *prog; struct rar_filter *filter, **nextfilter; uint32_t numprogs, num, blocklength, globaldatalen; uint8_t *globaldata; size_t blockstartpos; uint32_t registers[8] = { 0 }; uint32_t i; br.bytes = bytes; br.length = length; numprogs = 0; for (prog = filters->progs; prog; prog = prog->next) numprogs++; if ((flags & 0x80)) { num = membr_next_rarvm_number(&br); if (num == 0) { delete_filter(filters->stack); filters->stack = NULL; delete_program_code(filters->progs); filters->progs = NULL; } else num--; if (num > numprogs) { return 0; } filters->lastfilternum = num; } else num = filters->lastfilternum; prog = filters->progs; for (i = 0; i < num; i++) prog = prog->next; if (prog) prog->usagecount++; blockstartpos = membr_next_rarvm_number(&br) + (size_t)lzss_position(&rar->lzss); if ((flags & 0x40)) blockstartpos += 258; if ((flags & 0x20)) blocklength = membr_next_rarvm_number(&br); else blocklength = prog ? prog->oldfilterlength : 0; if (blocklength > rar->dictionary_size) return 0; registers[3] = PROGRAM_SYSTEM_GLOBAL_ADDRESS; registers[4] = blocklength; registers[5] = prog ? prog->usagecount : 0; registers[7] = VM_MEMORY_SIZE; if ((flags & 0x10)) { uint8_t mask = (uint8_t)membr_bits(&br, 7); for (i = 0; i < 7; i++) if ((mask & (1 << i))) registers[i] = membr_next_rarvm_number(&br); } if (!prog) { uint32_t len = membr_next_rarvm_number(&br); uint8_t *bytecode; struct rar_program_code **next; if (len == 0 || len > 0x10000) return 0; bytecode = malloc(len); if (!bytecode) return 0; for (i = 0; i < len; i++) bytecode[i] = (uint8_t)membr_bits(&br, 8); prog = compile_program(bytecode, len); if (!prog) { free(bytecode); return 0; } free(bytecode); next = &filters->progs; while (*next) next = &(*next)->next; *next = prog; } prog->oldfilterlength = blocklength; globaldata = NULL; globaldatalen = 0; if ((flags & 0x08)) { globaldatalen = membr_next_rarvm_number(&br); if (globaldatalen > PROGRAM_USER_GLOBAL_SIZE) return 0; globaldata = malloc(globaldatalen + PROGRAM_SYSTEM_GLOBAL_SIZE); if (!globaldata) return 0; for (i = 0; i < globaldatalen; i++) globaldata[i + PROGRAM_SYSTEM_GLOBAL_SIZE] = (uint8_t)membr_bits(&br, 8); } if (br.at_eof) { free(globaldata); return 0; } filter = create_filter(prog, globaldata, globaldatalen, registers, blockstartpos, blocklength); free(globaldata); if (!filter) return 0; for (i = 0; i < 7; i++) archive_le32enc(&filter->globaldata[i * 4], registers[i]); archive_le32enc(&filter->globaldata[0x1C], blocklength); archive_le32enc(&filter->globaldata[0x20], 0); archive_le32enc(&filter->globaldata[0x2C], prog->usagecount); nextfilter = &filters->stack; while (*nextfilter) nextfilter = &(*nextfilter)->next; *nextfilter = filter; if (!filters->stack->next) filters->filterstart = blockstartpos; return 1; } static struct rar_filter * create_filter(struct rar_program_code *prog, const uint8_t *globaldata, uint32_t globaldatalen, uint32_t registers[8], size_t startpos, uint32_t length) { struct rar_filter *filter; filter = calloc(1, sizeof(*filter)); if (!filter) return NULL; filter->prog = prog; filter->globaldatalen = globaldatalen > PROGRAM_SYSTEM_GLOBAL_SIZE ? globaldatalen : PROGRAM_SYSTEM_GLOBAL_SIZE; filter->globaldata = calloc(1, filter->globaldatalen); if (!filter->globaldata) { free(filter); return NULL; } if (globaldata) memcpy(filter->globaldata, globaldata, globaldatalen); if (registers) memcpy(filter->initialregisters, registers, sizeof(filter->initialregisters)); filter->blockstartpos = startpos; filter->blocklength = length; return filter; } static int run_filters(struct archive_read *a) { struct rar *rar = (struct rar *)(a->format->data); struct rar_filters *filters = &rar->filters; struct rar_filter *filter = filters->stack; struct rar_filter *f; size_t start, end; int64_t tend; uint32_t lastfilteraddress; uint32_t lastfilterlength; int ret; if (filters == NULL || filter == NULL) return (0); start = (size_t)filters->filterstart; end = start + filter->blocklength; filters->filterstart = INT64_MAX; tend = (int64_t)end; ret = expand(a, &tend); if (ret != ARCHIVE_OK) return 0; /* Check if filter stack was modified in expand() */ ret = ARCHIVE_FATAL; f = filters->stack; while (f) { if (f == filter) { ret = ARCHIVE_OK; break; } f = f->next; } if (ret != ARCHIVE_OK) return 0; if (tend < 0) return 0; end = (size_t)tend; if (end != start + filter->blocklength) return 0; if (!filters->vm) { filters->vm = calloc(1, sizeof(*filters->vm)); if (!filters->vm) return 0; } if (filter->blocklength > VM_MEMORY_SIZE) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad RAR file data"); return 0; } ret = copy_from_lzss_window(a, filters->vm->memory, start, filter->blocklength); if (ret != ARCHIVE_OK) return 0; if (!execute_filter(a, filter, filters->vm, (size_t)rar->offset)) return 0; lastfilteraddress = filter->filteredblockaddress; lastfilterlength = filter->filteredblocklength; filters->stack = filter->next; filter->next = NULL; delete_filter(filter); while ((filter = filters->stack) != NULL && (int64_t)filter->blockstartpos == filters->filterstart && filter->blocklength == lastfilterlength) { memmove(&filters->vm->memory[0], &filters->vm->memory[lastfilteraddress], lastfilterlength); if (!execute_filter(a, filter, filters->vm, (size_t)rar->offset)) return 0; lastfilteraddress = filter->filteredblockaddress; lastfilterlength = filter->filteredblocklength; filters->stack = filter->next; filter->next = NULL; delete_filter(filter); } if (filters->stack) { if (filters->stack->blockstartpos < end) return 0; filters->filterstart = filters->stack->blockstartpos; } filters->lastend = end; filters->bytes = &filters->vm->memory[lastfilteraddress]; filters->bytes_ready = lastfilterlength; return 1; } static struct rar_program_code * compile_program(const uint8_t *bytes, size_t length) { struct memory_bit_reader br = { 0 }; struct rar_program_code *prog; // uint32_t instrcount = 0; uint8_t xor; size_t i; xor = 0; for (i = 1; i < length; i++) xor ^= bytes[i]; if (!length || xor != bytes[0]) return NULL; br.bytes = bytes; br.length = length; br.offset = 1; prog = calloc(1, sizeof(*prog)); if (!prog) return NULL; prog->fingerprint = crc32(0, bytes, (unsigned int)length) | ((uint64_t)length << 32); if (membr_bits(&br, 1)) { prog->staticdatalen = membr_next_rarvm_number(&br) + 1; prog->staticdata = malloc(prog->staticdatalen); if (!prog->staticdata) { delete_program_code(prog); return NULL; } for (i = 0; i < prog->staticdatalen; i++) prog->staticdata[i] = (uint8_t)membr_bits(&br, 8); } return prog; } static void delete_filter(struct rar_filter *filter) { while (filter) { struct rar_filter *next = filter->next; free(filter->globaldata); free(filter); filter = next; } } static void clear_filters(struct rar_filters *filters) { delete_filter(filters->stack); delete_program_code(filters->progs); free(filters->vm); } static void delete_program_code(struct rar_program_code *prog) { while (prog) { struct rar_program_code *next = prog->next; free(prog->staticdata); free(prog->globalbackup); free(prog); prog = next; } } static uint32_t membr_next_rarvm_number(struct memory_bit_reader *br) { uint32_t val; switch (membr_bits(br, 2)) { case 0: return membr_bits(br, 4); case 1: val = membr_bits(br, 8); if (val >= 16) return val; return 0xFFFFFF00 | (val << 4) | membr_bits(br, 4); case 2: return membr_bits(br, 16); default: return membr_bits(br, 32); } } static inline uint32_t membr_bits(struct memory_bit_reader *br, int bits) { if (bits > br->available && (br->at_eof || !membr_fill(br, bits))) return 0; return (uint32_t)((br->bits >> (br->available -= bits)) & (((uint64_t)1 << bits) - 1)); } static int membr_fill(struct memory_bit_reader *br, int bits) { while (br->available < bits && br->offset < br->length) { br->bits = (br->bits << 8) | br->bytes[br->offset++]; br->available += 8; } if (bits > br->available) { br->at_eof = 1; return 0; } return 1; } static int read_filter(struct archive_read *a, int64_t *end) { struct rar *rar = (struct rar *)(a->format->data); uint8_t flags, val, *code; uint16_t length, i; if (!rar_decode_byte(a, &flags)) return 0; length = (flags & 0x07) + 1; if (length == 7) { if (!rar_decode_byte(a, &val)) return 0; length = val + 7; } else if (length == 8) { if (!rar_decode_byte(a, &val)) return 0; length = val << 8; if (!rar_decode_byte(a, &val)) return 0; length |= val; } code = malloc(length); if (!code) return 0; for (i = 0; i < length; i++) { if (!rar_decode_byte(a, &code[i])) { free(code); return 0; } } if (!parse_filter(a, code, length, flags)) { free(code); return 0; } free(code); if (rar->filters.filterstart < *end) *end = rar->filters.filterstart; return 1; } static int execute_filter_delta(struct rar_filter *filter, struct rar_virtual_machine *vm) { uint32_t length = filter->initialregisters[4]; uint32_t numchannels = filter->initialregisters[0]; uint8_t *src, *dst; uint32_t i, idx; if (length > PROGRAM_WORK_SIZE / 2) return 0; src = &vm->memory[0]; dst = &vm->memory[length]; for (i = 0; i < numchannels; i++) { uint8_t lastbyte = 0; for (idx = i; idx < length; idx += numchannels) { /* * The src block should not overlap with the dst block. * If so it would be better to consider this archive is broken. */ if (src >= dst) return 0; lastbyte = dst[idx] = lastbyte - *src++; } } filter->filteredblockaddress = length; filter->filteredblocklength = length; return 1; } static int execute_filter_e8(struct rar_filter *filter, struct rar_virtual_machine *vm, size_t pos, int e9also) { uint32_t length = filter->initialregisters[4]; uint32_t filesize = 0x1000000; uint32_t i; if (length > PROGRAM_WORK_SIZE || length <= 4) return 0; for (i = 0; i <= length - 5; i++) { if (vm->memory[i] == 0xE8 || (e9also && vm->memory[i] == 0xE9)) { uint32_t currpos = (uint32_t)pos + i + 1; int32_t address = (int32_t)vm_read_32(vm, i + 1); if (address < 0 && currpos >= (~(uint32_t)address + 1)) vm_write_32(vm, i + 1, address + filesize); else if (address >= 0 && (uint32_t)address < filesize) vm_write_32(vm, i + 1, address - currpos); i += 4; } } filter->filteredblockaddress = 0; filter->filteredblocklength = length; return 1; } static int execute_filter_rgb(struct rar_filter *filter, struct rar_virtual_machine *vm) { uint32_t stride = filter->initialregisters[0]; uint32_t byteoffset = filter->initialregisters[1]; uint32_t blocklength = filter->initialregisters[4]; uint8_t *src, *dst; uint32_t i, j; if (blocklength > PROGRAM_WORK_SIZE / 2 || stride > blocklength || blocklength < 3 || byteoffset > 2) return 0; src = &vm->memory[0]; dst = &vm->memory[blocklength]; for (i = 0; i < 3; i++) { uint8_t byte = 0; uint8_t *prev = dst + i - stride; for (j = i; j < blocklength; j += 3) { /* * The src block should not overlap with the dst block. * If so it would be better to consider this archive is broken. */ if (src >= dst) return 0; if (prev >= dst) { uint32_t delta1 = abs(prev[3] - prev[0]); uint32_t delta2 = abs(byte - prev[0]); uint32_t delta3 = abs(prev[3] - prev[0] + byte - prev[0]); if (delta1 > delta2 || delta1 > delta3) byte = delta2 <= delta3 ? prev[3] : prev[0]; } byte -= *src++; dst[j] = byte; prev += 3; } } for (i = byteoffset; i < blocklength - 2; i += 3) { dst[i] += dst[i + 1]; dst[i + 2] += dst[i + 1]; } filter->filteredblockaddress = blocklength; filter->filteredblocklength = blocklength; return 1; } static int execute_filter_audio(struct rar_filter *filter, struct rar_virtual_machine *vm) { uint32_t length = filter->initialregisters[4]; uint32_t numchannels = filter->initialregisters[0]; uint8_t *src, *dst; uint32_t i, j; if (length > PROGRAM_WORK_SIZE / 2) return 0; src = &vm->memory[0]; dst = &vm->memory[length]; for (i = 0; i < numchannels; i++) { struct audio_state state; memset(&state, 0, sizeof(state)); for (j = i; j < length; j += numchannels) { /* * The src block should not overlap with the dst block. * If so it would be better to consider this archive is broken. */ if (src >= dst) return 0; int8_t delta = (int8_t)*src++; uint8_t predbyte, byte; int prederror; state.delta[2] = state.delta[1]; state.delta[1] = state.lastdelta - state.delta[0]; state.delta[0] = state.lastdelta; predbyte = ((8 * state.lastbyte + state.weight[0] * state.delta[0] + state.weight[1] * state.delta[1] + state.weight[2] * state.delta[2]) >> 3) & 0xFF; byte = (predbyte - delta) & 0xFF; prederror = delta << 3; state.error[0] += abs(prederror); state.error[1] += abs(prederror - state.delta[0]); state.error[2] += abs(prederror + state.delta[0]); state.error[3] += abs(prederror - state.delta[1]); state.error[4] += abs(prederror + state.delta[1]); state.error[5] += abs(prederror - state.delta[2]); state.error[6] += abs(prederror + state.delta[2]); state.lastdelta = (int8_t)(byte - state.lastbyte); dst[j] = state.lastbyte = byte; if (!(state.count++ & 0x1F)) { uint8_t k, idx = 0; for (k = 1; k < 7; k++) { if (state.error[k] < state.error[idx]) idx = k; } memset(state.error, 0, sizeof(state.error)); switch (idx) { case 1: if (state.weight[0] >= -16) state.weight[0]--; break; case 2: if (state.weight[0] < 16) state.weight[0]++; break; case 3: if (state.weight[1] >= -16) state.weight[1]--; break; case 4: if (state.weight[1] < 16) state.weight[1]++; break; case 5: if (state.weight[2] >= -16) state.weight[2]--; break; case 6: if (state.weight[2] < 16) state.weight[2]++; break; } } } } filter->filteredblockaddress = length; filter->filteredblocklength = length; return 1; } static int execute_filter(struct archive_read *a, struct rar_filter *filter, struct rar_virtual_machine *vm, size_t pos) { if (filter->prog->fingerprint == 0x1D0E06077D) return execute_filter_delta(filter, vm); if (filter->prog->fingerprint == 0x35AD576887) return execute_filter_e8(filter, vm, pos, 0); if (filter->prog->fingerprint == 0x393CD7E57E) return execute_filter_e8(filter, vm, pos, 1); if (filter->prog->fingerprint == 0x951C2C5DC8) return execute_filter_rgb(filter, vm); if (filter->prog->fingerprint == 0xD8BC85E701) return execute_filter_audio(filter, vm); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "No support for RAR VM program filter"); return 0; } static int rar_decode_byte(struct archive_read *a, uint8_t *byte) { struct rar *rar = (struct rar *)(a->format->data); struct rar_br *br = &(rar->br); if (!rar_br_read_ahead(a, br, 8)) return 0; *byte = (uint8_t)rar_br_bits(br, 8); rar_br_consume(br, 8); return 1; } static inline void vm_write_32(struct rar_virtual_machine* vm, size_t offset, uint32_t u32) { archive_le32enc(vm->memory + offset, u32); } static inline uint32_t vm_read_32(struct rar_virtual_machine* vm, size_t offset) { return archive_le32dec(vm->memory + offset); } diff --git a/contrib/libarchive/libarchive/archive_read_support_format_rar5.c b/contrib/libarchive/libarchive/archive_read_support_format_rar5.c index 48dde0c2e814..17e501e02e9f 100644 --- a/contrib/libarchive/libarchive/archive_read_support_format_rar5.c +++ b/contrib/libarchive/libarchive/archive_read_support_format_rar5.c @@ -1,4363 +1,4411 @@ /*- * Copyright (c) 2018 Grzegorz Antoniak (http://antoniak.org) * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #include "archive_endian.h" #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_ZLIB_H #include /* crc32 */ #endif #ifdef HAVE_LIMITS_H #include #endif #include "archive.h" #ifndef HAVE_ZLIB_H #include "archive_crc32.h" #endif #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_ppmd7_private.h" #include "archive_entry_private.h" #include "archive_time_private.h" #ifdef HAVE_BLAKE2_H #include #else #include "archive_blake2.h" #endif /*#define CHECK_CRC_ON_SOLID_SKIP*/ /*#define DONT_FAIL_ON_CRC_ERROR*/ /*#define DEBUG*/ #define rar5_min(a, b) (((a) > (b)) ? (b) : (a)) #define rar5_max(a, b) (((a) > (b)) ? (a) : (b)) #define rar5_countof(X) ((const ssize_t) (sizeof(X) / sizeof(*X))) #if defined DEBUG #define DEBUG_CODE if(1) #define LOG(...) do { printf("rar5: " __VA_ARGS__); puts(""); } while(0) #else #define DEBUG_CODE if(0) #endif /* Real RAR5 magic number is: * * 0x52, 0x61, 0x72, 0x21, 0x1a, 0x07, 0x01, 0x00 * "Rar!→•☺·\x00" * * Retrieved with `rar5_signature()` by XOR'ing it with 0xA1, because I don't * want to put this magic sequence in each binary that uses libarchive, so * applications that scan through the file for this marker won't trigger on * this "false" one. * * The array itself is decrypted in `rar5_init` function. */ static unsigned char rar5_signature_xor[] = { 243, 192, 211, 128, 187, 166, 160, 161 }; static const size_t g_unpack_window_size = 0x20000; /* These could have been static const's, but they aren't, because of * Visual Studio. */ #define MAX_NAME_IN_CHARS 2048 #define MAX_NAME_IN_BYTES (4 * MAX_NAME_IN_CHARS) struct file_header { ssize_t bytes_remaining; ssize_t unpacked_size; int64_t last_offset; /* Used in sanity checks. */ int64_t last_size; /* Used in sanity checks. */ uint8_t solid : 1; /* Is this a solid stream? */ uint8_t service : 1; /* Is this file a service data? */ uint8_t eof : 1; /* Did we finish unpacking the file? */ uint8_t dir : 1; /* Is this file entry a directory? */ /* Optional time fields. */ int64_t e_mtime; int64_t e_ctime; int64_t e_atime; uint32_t e_mtime_ns; uint32_t e_ctime_ns; uint32_t e_atime_ns; /* Optional hash fields. */ uint32_t stored_crc32; uint32_t calculated_crc32; uint8_t blake2sp[32]; blake2sp_state b2state; char has_blake2; /* Optional redir fields */ uint64_t redir_type; uint64_t redir_flags; ssize_t solid_window_size; /* Used in file format check. */ }; enum EXTRA { EX_CRYPT = 0x01, EX_HASH = 0x02, EX_HTIME = 0x03, EX_VERSION = 0x04, EX_REDIR = 0x05, EX_UOWNER = 0x06, EX_SUBDATA = 0x07 }; #define REDIR_SYMLINK_IS_DIR 1 enum REDIR_TYPE { REDIR_TYPE_NONE = 0, REDIR_TYPE_UNIXSYMLINK = 1, REDIR_TYPE_WINSYMLINK = 2, REDIR_TYPE_JUNCTION = 3, REDIR_TYPE_HARDLINK = 4, REDIR_TYPE_FILECOPY = 5, }; #define OWNER_USER_NAME 0x01 #define OWNER_GROUP_NAME 0x02 #define OWNER_USER_UID 0x04 #define OWNER_GROUP_GID 0x08 #define OWNER_MAXNAMELEN 256 enum FILTER_TYPE { FILTER_DELTA = 0, /* Generic pattern. */ FILTER_E8 = 1, /* Intel x86 code. */ FILTER_E8E9 = 2, /* Intel x86 code. */ FILTER_ARM = 3, /* ARM code. */ FILTER_AUDIO = 4, /* Audio filter, not used in RARv5. */ FILTER_RGB = 5, /* Color palette, not used in RARv5. */ FILTER_ITANIUM = 6, /* Intel's Itanium, not used in RARv5. */ FILTER_PPM = 7, /* Predictive pattern matching, not used in RARv5. */ FILTER_NONE = 8, }; struct filter_info { int type; int channels; int pos_r; int64_t block_start; ssize_t block_length; uint16_t width; }; struct data_ready { char used; const uint8_t* buf; size_t size; int64_t offset; }; struct cdeque { uint16_t beg_pos; uint16_t end_pos; uint16_t cap_mask; uint16_t size; size_t* arr; }; struct decode_table { uint32_t size; int32_t decode_len[16]; uint32_t decode_pos[16]; uint32_t quick_bits; uint8_t quick_len[1 << 10]; uint16_t quick_num[1 << 10]; uint16_t decode_num[306]; }; struct comp_state { /* Flag used to specify if unpacker needs to reinitialize the uncompression context. */ uint8_t initialized : 1; /* Flag used when applying filters. */ uint8_t all_filters_applied : 1; /* Flag used to skip file context reinitialization, used when unpacker is skipping through different multivolume archives. */ uint8_t switch_multivolume : 1; /* Flag used to specify if unpacker has processed the whole data block or just a part of it. */ uint8_t block_parsing_finished : 1; /* Flag used to indicate that a previous file using this buffer was encrypted, meaning no data in the buffer can be trusted */ uint8_t data_encrypted : 1; signed int notused : 3; int flags; /* Uncompression flags. */ int method; /* Uncompression algorithm method. */ int version; /* Uncompression algorithm version. */ ssize_t window_size; /* Size of window_buf. */ uint8_t* window_buf; /* Circular buffer used during decompression. */ uint8_t* filtered_buf; /* Buffer used when applying filters. */ const uint8_t* block_buf; /* Buffer used when merging blocks. */ ssize_t window_mask; /* Convenience field; window_size - 1. */ int64_t write_ptr; /* This amount of data has been unpacked in the window buffer. */ int64_t last_write_ptr; /* This amount of data has been stored in the output file. */ int64_t last_unstore_ptr; /* Counter of bytes extracted during unstoring. This is separate from last_write_ptr because of how SERVICE base blocks are handled during skipping in solid multiarchive archives. */ int64_t solid_offset; /* Additional offset inside the window buffer, used in unpacking solid archives. */ ssize_t cur_block_size; /* Size of current data block. */ int last_len; /* Flag used in lzss decompression. */ /* Decode tables used during lzss uncompression. */ #define HUFF_BC 20 struct decode_table bd; /* huffman bit lengths */ #define HUFF_NC 306 struct decode_table ld; /* literals */ #define HUFF_DC 64 struct decode_table dd; /* distances */ #define HUFF_LDC 16 struct decode_table ldd; /* lower bits of distances */ #define HUFF_RC 44 struct decode_table rd; /* repeating distances */ #define HUFF_TABLE_SIZE (HUFF_NC + HUFF_DC + HUFF_RC + HUFF_LDC) /* Circular deque for storing filters. */ struct cdeque filters; int64_t last_block_start; /* Used for sanity checking. */ ssize_t last_block_length; /* Used for sanity checking. */ /* Distance cache used during lzss uncompression. */ int dist_cache[4]; /* Data buffer stack. */ struct data_ready dready[2]; }; /* Bit reader state. */ struct bit_reader { int8_t bit_addr; /* Current bit pointer inside current byte. */ int in_addr; /* Current byte pointer. */ }; /* RARv5 block header structure. Use bf_* functions to get values from * block_flags_u8 field. I.e. bf_byte_count, etc. */ struct compressed_block_header { /* block_flags_u8 contain fields encoded in little-endian bitfield: * * - table present flag (shr 7, and 1), * - last block flag (shr 6, and 1), * - byte_count (shr 3, and 7), * - bit_size (shr 0, and 7). */ uint8_t block_flags_u8; uint8_t block_cksum; }; /* RARv5 main header structure. */ struct main_header { /* Does the archive contain solid streams? */ uint8_t solid : 1; /* If this a multi-file archive? */ uint8_t volume : 1; uint8_t endarc : 1; uint8_t notused : 5; unsigned int vol_no; }; struct generic_header { uint8_t split_after : 1; uint8_t split_before : 1; uint8_t padding : 6; int size; int last_header_id; }; struct multivolume { unsigned int expected_vol_no; uint8_t* push_buf; }; /* Main context structure. */ struct rar5 { int header_initialized; /* Set to 1 if current file is positioned AFTER the magic value * of the archive file. This is used in header reading functions. */ int skipped_magic; /* Set to not zero if we're in skip mode (either by calling * rar5_data_skip function or when skipping over solid streams). * Set to 0 when in * extraction mode. This is used during checksum * calculation functions. */ int skip_mode; /* Set to not zero if we're in block merging mode (i.e. when switching * to another file in multivolume archive, last block from 1st archive * needs to be merged with 1st block from 2nd archive). This flag * guards against recursive use of the merging function, which doesn't * support recursive calls. */ int merge_mode; /* An offset to QuickOpen list. This is not supported by this unpacker, * because we're focusing on streaming interface. QuickOpen is designed * to make things quicker for non-stream interfaces, so it's not our * use case. */ uint64_t qlist_offset; /* An offset to additional Recovery data. This is not supported by this * unpacker. Recovery data are additional Reed-Solomon codes that could * be used to calculate bytes that are missing in archive or are * corrupted. */ uint64_t rr_offset; /* Various context variables grouped to different structures. */ struct generic_header generic; struct main_header main; struct comp_state cstate; struct file_header file; struct bit_reader bits; struct multivolume vol; /* The header of currently processed RARv5 block. Used in main * decompression logic loop. */ struct compressed_block_header last_block_hdr; /* * Custom field to denote that this archive contains encrypted entries */ int has_encrypted_entries; int headers_are_encrypted; }; /* Forward function declarations. */ static void rar5_signature(char *buf); static int verify_global_checksums(struct archive_read* a); static int rar5_read_data_skip(struct archive_read *a); static int push_data_ready(struct archive_read* a, struct rar5* rar, const uint8_t* buf, size_t size, int64_t offset); static void clear_data_ready_stack(struct rar5* rar); /* CDE_xxx = Circular Double Ended (Queue) return values. */ enum CDE_RETURN_VALUES { CDE_OK, CDE_ALLOC, CDE_PARAM, CDE_OUT_OF_BOUNDS, }; /* Clears the contents of this circular deque. */ static void cdeque_clear(struct cdeque* d) { d->size = 0; d->beg_pos = 0; d->end_pos = 0; } /* Creates a new circular deque object. Capacity must be power of 2: 8, 16, 32, * 64, 256, etc. When the user will add another item above current capacity, * the circular deque will overwrite the oldest entry. */ static int cdeque_init(struct cdeque* d, int max_capacity_power_of_2) { if(d == NULL || max_capacity_power_of_2 == 0) return CDE_PARAM; d->cap_mask = max_capacity_power_of_2 - 1; d->arr = NULL; if((max_capacity_power_of_2 & d->cap_mask) != 0) return CDE_PARAM; cdeque_clear(d); d->arr = malloc(sizeof(void*) * max_capacity_power_of_2); return d->arr ? CDE_OK : CDE_ALLOC; } /* Return the current size (not capacity) of circular deque `d`. */ static size_t cdeque_size(struct cdeque* d) { return d->size; } /* Returns the first element of current circular deque. Note that this function * doesn't perform any bounds checking. If you need bounds checking, use * `cdeque_front()` function instead. */ static void cdeque_front_fast(struct cdeque* d, void** value) { *value = (void*) d->arr[d->beg_pos]; } /* Returns the first element of current circular deque. This function * performs bounds checking. */ static int cdeque_front(struct cdeque* d, void** value) { if(d->size > 0) { cdeque_front_fast(d, value); return CDE_OK; } else return CDE_OUT_OF_BOUNDS; } /* Pushes a new element into the end of this circular deque object. If current * size will exceed capacity, the oldest element will be overwritten. */ static int cdeque_push_back(struct cdeque* d, void* item) { if(d == NULL) return CDE_PARAM; if(d->size == d->cap_mask + 1) return CDE_OUT_OF_BOUNDS; d->arr[d->end_pos] = (size_t) item; d->end_pos = (d->end_pos + 1) & d->cap_mask; d->size++; return CDE_OK; } /* Pops a front element of this circular deque object and returns its value. * This function doesn't perform any bounds checking. */ static void cdeque_pop_front_fast(struct cdeque* d, void** value) { *value = (void*) d->arr[d->beg_pos]; d->beg_pos = (d->beg_pos + 1) & d->cap_mask; d->size--; } /* Pops a front element of this circular deque object and returns its value. * This function performs bounds checking. */ static int cdeque_pop_front(struct cdeque* d, void** value) { if(!d || !value) return CDE_PARAM; if(d->size == 0) return CDE_OUT_OF_BOUNDS; cdeque_pop_front_fast(d, value); return CDE_OK; } /* Convenience function to cast filter_info** to void **. */ static void** cdeque_filter_p(struct filter_info** f) { return (void**) (size_t) f; } /* Convenience function to cast filter_info* to void *. */ static void* cdeque_filter(struct filter_info* f) { return (void**) (size_t) f; } /* Destroys this circular deque object. Deallocates the memory of the * collection buffer, but doesn't deallocate the memory of any pointer passed * to this deque as a value. */ static void cdeque_free(struct cdeque* d) { if(!d) return; if(!d->arr) return; free(d->arr); d->arr = NULL; d->beg_pos = -1; d->end_pos = -1; d->cap_mask = 0; } static inline uint8_t bf_bit_size(const struct compressed_block_header* hdr) { return hdr->block_flags_u8 & 7; } static inline uint8_t bf_byte_count(const struct compressed_block_header* hdr) { return (hdr->block_flags_u8 >> 3) & 7; } static inline uint8_t bf_is_table_present(const struct compressed_block_header* hdr) { return (hdr->block_flags_u8 >> 7) & 1; } static inline uint8_t bf_is_last_block(const struct compressed_block_header* hdr) { return (hdr->block_flags_u8 >> 6) & 1; } static inline struct rar5* get_context(struct archive_read* a) { return (struct rar5*) a->format->data; } /* Convenience functions used by filter implementations. */ static void circular_memcpy(uint8_t* dst, uint8_t* window, const ssize_t mask, int64_t start, int64_t end) { if((start & mask) > (end & mask)) { ssize_t len1 = mask + 1 - (start & mask); ssize_t len2 = end & mask; memcpy(dst, &window[start & mask], len1); memcpy(dst + len1, window, len2); } else { memcpy(dst, &window[start & mask], (size_t) (end - start)); } } static uint32_t read_filter_data(struct rar5* rar, uint32_t offset) { uint8_t linear_buf[4]; circular_memcpy(linear_buf, rar->cstate.window_buf, rar->cstate.window_mask, offset, offset + 4); return archive_le32dec(linear_buf); } static void write_filter_data(struct rar5* rar, uint32_t offset, uint32_t value) { archive_le32enc(&rar->cstate.filtered_buf[offset], value); } /* Allocates a new filter descriptor and adds it to the filter array. */ static struct filter_info* add_new_filter(struct rar5* rar) { struct filter_info* f = calloc(1, sizeof(*f)); if(!f) { return NULL; } cdeque_push_back(&rar->cstate.filters, cdeque_filter(f)); return f; } static int run_delta_filter(struct rar5* rar, struct filter_info* flt) { int i; ssize_t dest_pos, src_pos = 0; for(i = 0; i < flt->channels; i++) { uint8_t prev_byte = 0; for(dest_pos = i; dest_pos < flt->block_length; dest_pos += flt->channels) { uint8_t byte; byte = rar->cstate.window_buf[ (rar->cstate.solid_offset + flt->block_start + src_pos) & rar->cstate.window_mask]; prev_byte -= byte; rar->cstate.filtered_buf[dest_pos] = prev_byte; src_pos++; } } return ARCHIVE_OK; } static int run_e8e9_filter(struct rar5* rar, struct filter_info* flt, int extended) { const uint32_t file_size = 0x1000000; ssize_t i; circular_memcpy(rar->cstate.filtered_buf, rar->cstate.window_buf, rar->cstate.window_mask, rar->cstate.solid_offset + flt->block_start, rar->cstate.solid_offset + flt->block_start + flt->block_length); for(i = 0; i < flt->block_length - 4;) { uint8_t b = rar->cstate.window_buf[ (rar->cstate.solid_offset + flt->block_start + i++) & rar->cstate.window_mask]; /* * 0xE8 = x86's call (function call) * 0xE9 = x86's jmp (unconditional jump) */ if(b == 0xE8 || (extended && b == 0xE9)) { uint32_t addr; uint32_t offset = (i + flt->block_start) % file_size; addr = read_filter_data(rar, (uint32_t)(rar->cstate.solid_offset + flt->block_start + i) & rar->cstate.window_mask); if(addr & 0x80000000) { if(((addr + offset) & 0x80000000) == 0) { write_filter_data(rar, (uint32_t)i, addr + file_size); } } else { if((addr - file_size) & 0x80000000) { uint32_t naddr = addr - offset; write_filter_data(rar, (uint32_t)i, naddr); } } i += 4; } } return ARCHIVE_OK; } static int run_arm_filter(struct rar5* rar, struct filter_info* flt) { ssize_t i = 0; uint32_t offset; circular_memcpy(rar->cstate.filtered_buf, rar->cstate.window_buf, rar->cstate.window_mask, rar->cstate.solid_offset + flt->block_start, rar->cstate.solid_offset + flt->block_start + flt->block_length); for(i = 0; i < flt->block_length - 3; i += 4) { uint8_t* b = &rar->cstate.window_buf[ (rar->cstate.solid_offset + flt->block_start + i + 3) & rar->cstate.window_mask]; if(*b == 0xEB) { /* 0xEB = ARM's BL (branch + link) instruction. */ offset = read_filter_data(rar, (rar->cstate.solid_offset + flt->block_start + i) & (uint32_t)rar->cstate.window_mask) & 0x00ffffff; offset -= (uint32_t) ((i + flt->block_start) / 4); offset = (offset & 0x00ffffff) | 0xeb000000; write_filter_data(rar, (uint32_t)i, offset); } } return ARCHIVE_OK; } static int run_filter(struct archive_read* a, struct filter_info* flt) { int ret; struct rar5* rar = get_context(a); clear_data_ready_stack(rar); free(rar->cstate.filtered_buf); rar->cstate.filtered_buf = malloc(flt->block_length); if(!rar->cstate.filtered_buf) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for filter data."); return ARCHIVE_FATAL; } switch(flt->type) { case FILTER_DELTA: ret = run_delta_filter(rar, flt); break; case FILTER_E8: /* fallthrough */ case FILTER_E8E9: ret = run_e8e9_filter(rar, flt, flt->type == FILTER_E8E9); break; case FILTER_ARM: ret = run_arm_filter(rar, flt); break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported filter type: 0x%x", (unsigned int)flt->type); return ARCHIVE_FATAL; } if(ret != ARCHIVE_OK) { /* Filter has failed. */ return ret; } if(ARCHIVE_OK != push_data_ready(a, rar, rar->cstate.filtered_buf, flt->block_length, rar->cstate.last_write_ptr)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Stack overflow when submitting unpacked data"); return ARCHIVE_FATAL; } rar->cstate.last_write_ptr += flt->block_length; return ARCHIVE_OK; } /* The `push_data` function submits the selected data range to the user. * Next call of `use_data` will use the pointer, size and offset arguments * that are specified here. These arguments are pushed to the FIFO stack here, * and popped from the stack by the `use_data` function. */ static void push_data(struct archive_read* a, struct rar5* rar, const uint8_t* buf, int64_t idx_begin, int64_t idx_end) { const ssize_t wmask = rar->cstate.window_mask; const ssize_t solid_write_ptr = (rar->cstate.solid_offset + rar->cstate.last_write_ptr) & wmask; idx_begin += rar->cstate.solid_offset; idx_end += rar->cstate.solid_offset; /* Check if our unpacked data is wrapped inside the window circular * buffer. If it's not wrapped, it can be copied out by using * a single memcpy, but when it's wrapped, we need to copy the first * part with one memcpy, and the second part with another memcpy. */ if((idx_begin & wmask) > (idx_end & wmask)) { /* The data is wrapped (begin offset sis bigger than end * offset). */ const ssize_t frag1_size = rar->cstate.window_size - (idx_begin & wmask); const ssize_t frag2_size = idx_end & wmask; /* Copy the first part of the buffer first. */ push_data_ready(a, rar, buf + solid_write_ptr, frag1_size, rar->cstate.last_write_ptr); /* Copy the second part of the buffer. */ push_data_ready(a, rar, buf, frag2_size, rar->cstate.last_write_ptr + frag1_size); rar->cstate.last_write_ptr += frag1_size + frag2_size; } else { /* Data is not wrapped, so we can just use one call to copy the * data. */ push_data_ready(a, rar, buf + solid_write_ptr, (idx_end - idx_begin) & wmask, rar->cstate.last_write_ptr); rar->cstate.last_write_ptr += idx_end - idx_begin; } } /* Convenience function that submits the data to the user. It uses the * unpack window buffer as a source location. */ static void push_window_data(struct archive_read* a, struct rar5* rar, int64_t idx_begin, int64_t idx_end) { push_data(a, rar, rar->cstate.window_buf, idx_begin, idx_end); } static int apply_filters(struct archive_read* a) { struct filter_info* flt; struct rar5* rar = get_context(a); int ret; rar->cstate.all_filters_applied = 0; /* Get the first filter that can be applied to our data. The data * needs to be fully unpacked before the filter can be run. */ if(CDE_OK == cdeque_front(&rar->cstate.filters, cdeque_filter_p(&flt))) { /* Check if our unpacked data fully covers this filter's * range. */ if(rar->cstate.write_ptr > flt->block_start && rar->cstate.write_ptr >= flt->block_start + flt->block_length) { /* Check if we have some data pending to be written * right before the filter's start offset. */ if(rar->cstate.last_write_ptr == flt->block_start) { /* Run the filter specified by descriptor * `flt`. */ ret = run_filter(a, flt); if(ret != ARCHIVE_OK) { /* Filter failure, return error. */ return ret; } /* Filter descriptor won't be needed anymore * after it's used, * so remove it from the * filter list and free its memory. */ (void) cdeque_pop_front(&rar->cstate.filters, cdeque_filter_p(&flt)); free(flt); } else { /* We can't run filters yet, dump the memory * right before the filter. */ push_window_data(a, rar, rar->cstate.last_write_ptr, flt->block_start); } /* Return 'filter applied or not needed' state to the * caller. */ return ARCHIVE_RETRY; } } rar->cstate.all_filters_applied = 1; return ARCHIVE_OK; } static void dist_cache_push(struct rar5* rar, int value) { int* q = rar->cstate.dist_cache; q[3] = q[2]; q[2] = q[1]; q[1] = q[0]; q[0] = value; } static int dist_cache_touch(struct rar5* rar, int idx) { int* q = rar->cstate.dist_cache; int i, dist = q[idx]; for(i = idx; i > 0; i--) q[i] = q[i - 1]; q[0] = dist; return dist; } static void free_filters(struct rar5* rar) { struct cdeque* d = &rar->cstate.filters; /* Free any remaining filters. All filters should be naturally * consumed by the unpacking function, so remaining filters after * unpacking normally mean that unpacking wasn't successful. * But still of course we shouldn't leak memory in such case. */ /* cdeque_size() is a fast operation, so we can use it as a loop * expression. */ while(cdeque_size(d) > 0) { struct filter_info* f = NULL; /* Pop_front will also decrease the collection's size. */ if (CDE_OK == cdeque_pop_front(d, cdeque_filter_p(&f))) free(f); } cdeque_clear(d); /* Also clear out the variables needed for sanity checking. */ rar->cstate.last_block_start = 0; rar->cstate.last_block_length = 0; } static void reset_file_context(struct rar5* rar) { memset(&rar->file, 0, sizeof(rar->file)); blake2sp_init(&rar->file.b2state, 32); if(rar->main.solid) { rar->cstate.solid_offset += rar->cstate.write_ptr; } else { rar->cstate.solid_offset = 0; } rar->cstate.write_ptr = 0; rar->cstate.last_write_ptr = 0; rar->cstate.last_unstore_ptr = 0; rar->file.redir_type = REDIR_TYPE_NONE; rar->file.redir_flags = 0; free_filters(rar); } static inline int get_archive_read(struct archive* a, struct archive_read** ar) { *ar = (struct archive_read*) a; archive_check_magic(a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_rar5"); return ARCHIVE_OK; } static int read_ahead(struct archive_read* a, size_t how_many, const uint8_t** ptr) { ssize_t avail = -1; if(!ptr) return 0; *ptr = __archive_read_ahead(a, how_many, &avail); if(*ptr == NULL) { return 0; } return 1; } static int consume(struct archive_read* a, int64_t how_many) { int ret; ret = how_many == __archive_read_consume(a, how_many) ? ARCHIVE_OK : ARCHIVE_FATAL; return ret; } /** * Read a RAR5 variable sized numeric value. This value will be stored in * `pvalue`. The `pvalue_len` argument points to a variable that will receive * the byte count that was consumed in order to decode the `pvalue` value, plus * one. * * pvalue_len is optional and can be NULL. * * NOTE: if `pvalue_len` is NOT NULL, the caller needs to manually consume * the number of bytes that `pvalue_len` value contains. If the `pvalue_len` * is NULL, this consuming operation is done automatically. * * Returns 1 if *pvalue was successfully read. * Returns 0 if there was an error. In this case, *pvalue contains an * invalid value. */ static int read_var(struct archive_read* a, uint64_t* pvalue, uint64_t* pvalue_len) { uint64_t result = 0; size_t shift, i; const uint8_t* p; uint8_t b; /* We will read maximum of 8 bytes. We don't have to handle the * situation to read the RAR5 variable-sized value stored at the end of * the file, because such situation will never happen. */ if(!read_ahead(a, 8, &p)) return 0; for(shift = 0, i = 0; i < 8; i++, shift += 7) { b = p[i]; /* Strip the MSB from the input byte and add the resulting * number to the `result`. */ result += (b & (uint64_t)0x7F) << shift; /* MSB set to 1 means we need to continue decoding process. * MSB set to 0 means we're done. * * This conditional checks for the second case. */ if((b & 0x80) == 0) { if(pvalue) { *pvalue = result; } /* If the caller has passed the `pvalue_len` pointer, * store the number of consumed bytes in it and do NOT * consume those bytes, since the caller has all the * information it needs to perform */ if(pvalue_len) { *pvalue_len = 1 + i; } else { /* If the caller did not provide the * `pvalue_len` pointer, it will not have the * possibility to advance the file pointer, * because it will not know how many bytes it * needs to consume. This is why we handle * such situation here automatically. */ if(ARCHIVE_OK != consume(a, 1 + i)) { return 0; } } /* End of decoding process, return success. */ return 1; } } /* The decoded value takes the maximum number of 8 bytes. * It's a maximum number of bytes, so end decoding process here * even if the first bit of last byte is 1. */ if(pvalue) { *pvalue = result; } if(pvalue_len) { *pvalue_len = 9; } else { if(ARCHIVE_OK != consume(a, 9)) { return 0; } } return 1; } static int read_var_sized(struct archive_read* a, size_t* pvalue, size_t* pvalue_len) { uint64_t v; uint64_t v_size = 0; const int ret = pvalue_len ? read_var(a, &v, &v_size) : read_var(a, &v, NULL); if(ret == 1 && pvalue) { *pvalue = (size_t) v; } if(pvalue_len) { /* Possible data truncation should be safe. */ *pvalue_len = (size_t) v_size; } return ret; } static int read_bits_32(struct archive_read* a, struct rar5* rar, const uint8_t* p, uint32_t* value) { if(rar->bits.in_addr >= rar->cstate.cur_block_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Premature end of stream during extraction of data (#1)"); return ARCHIVE_FATAL; } uint32_t bits = ((uint32_t) p[rar->bits.in_addr]) << 24; bits |= p[rar->bits.in_addr + 1] << 16; bits |= p[rar->bits.in_addr + 2] << 8; bits |= p[rar->bits.in_addr + 3]; bits <<= rar->bits.bit_addr; bits |= p[rar->bits.in_addr + 4] >> (8 - rar->bits.bit_addr); *value = bits; return ARCHIVE_OK; } static int read_bits_16(struct archive_read* a, struct rar5* rar, const uint8_t* p, uint16_t* value) { if(rar->bits.in_addr >= rar->cstate.cur_block_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Premature end of stream during extraction of data (#2)"); return ARCHIVE_FATAL; } int bits = (int) ((uint32_t) p[rar->bits.in_addr]) << 16; bits |= (int) p[rar->bits.in_addr + 1] << 8; bits |= (int) p[rar->bits.in_addr + 2]; bits >>= (8 - rar->bits.bit_addr); *value = bits & 0xffff; return ARCHIVE_OK; } static void skip_bits(struct rar5* rar, int bits) { const int new_bits = rar->bits.bit_addr + bits; rar->bits.in_addr += new_bits >> 3; rar->bits.bit_addr = new_bits & 7; } /* n = up to 16 */ static int read_consume_bits(struct archive_read* a, struct rar5* rar, const uint8_t* p, int n, int* value) { uint16_t v; int ret, num; if(n == 0 || n > 16) { /* This is a programmer error and should never happen * in runtime. */ return ARCHIVE_FATAL; } ret = read_bits_16(a, rar, p, &v); if(ret != ARCHIVE_OK) return ret; num = (int) v; num >>= 16 - n; skip_bits(rar, n); if(value) *value = num; return ARCHIVE_OK; } static char read_u32(struct archive_read* a, uint32_t* pvalue) { const uint8_t* p; if(!read_ahead(a, 4, &p)) return 0; *pvalue = archive_le32dec(p); return ARCHIVE_OK == consume(a, 4); } static char read_u64(struct archive_read* a, uint64_t* pvalue) { const uint8_t* p; if(!read_ahead(a, 8, &p)) return 0; *pvalue = archive_le64dec(p); return ARCHIVE_OK == consume(a, 8); } static int bid_standard(struct archive_read* a) { const uint8_t* p; char signature[sizeof(rar5_signature_xor)]; rar5_signature(signature); if(!read_ahead(a, sizeof(rar5_signature_xor), &p)) return -1; if(!memcmp(signature, p, sizeof(rar5_signature_xor))) return 30; return -1; } static int bid_sfx(struct archive_read *a) { const char *p; if ((p = __archive_read_ahead(a, 7, NULL)) == NULL) return -1; if ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0) { /* This is a PE file */ char signature[sizeof(rar5_signature_xor)]; ssize_t offset = 0x10000; ssize_t window = 4096; ssize_t bytes_avail; rar5_signature(signature); while (offset + window <= (1024 * 512)) { const char *buff = __archive_read_ahead(a, offset + window, &bytes_avail); if (buff == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < 0x40) return 0; continue; } p = buff + offset; while (p + 8 < buff + bytes_avail) { if (memcmp(p, signature, sizeof(signature)) == 0) return 30; p += 0x10; } offset = p - buff; } } return 0; } static int rar5_bid(struct archive_read* a, int best_bid) { int my_bid; if(best_bid > 30) return -1; my_bid = bid_standard(a); if(my_bid > -1) { return my_bid; } my_bid = bid_sfx(a); if (my_bid > -1) { return my_bid; } return -1; } static int rar5_options(struct archive_read *a, const char *key, const char *val) { (void) a; (void) key; (void) val; /* No options supported in this version. Return the ARCHIVE_WARN code * to signal the options supervisor that the unpacker didn't handle * setting this option. */ return ARCHIVE_WARN; } static void init_header(struct archive_read* a) { a->archive.archive_format = ARCHIVE_FORMAT_RAR_V5; a->archive.archive_format_name = "RAR5"; } static void init_window_mask(struct rar5* rar) { if (rar->cstate.window_size) rar->cstate.window_mask = rar->cstate.window_size - 1; else rar->cstate.window_mask = 0; } enum HEADER_FLAGS { HFL_EXTRA_DATA = 0x0001, HFL_DATA = 0x0002, HFL_SKIP_IF_UNKNOWN = 0x0004, HFL_SPLIT_BEFORE = 0x0008, HFL_SPLIT_AFTER = 0x0010, HFL_CHILD = 0x0020, HFL_INHERITED = 0x0040 }; static int process_main_locator_extra_block(struct archive_read* a, struct rar5* rar) { uint64_t locator_flags; enum LOCATOR_FLAGS { QLIST = 0x01, RECOVERY = 0x02, }; if(!read_var(a, &locator_flags, NULL)) { return ARCHIVE_EOF; } if(locator_flags & QLIST) { if(!read_var(a, &rar->qlist_offset, NULL)) { return ARCHIVE_EOF; } /* qlist is not used */ } if(locator_flags & RECOVERY) { if(!read_var(a, &rar->rr_offset, NULL)) { return ARCHIVE_EOF; } /* rr is not used */ } return ARCHIVE_OK; } static int parse_file_extra_hash(struct archive_read* a, struct rar5* rar, int64_t* extra_data_size) { size_t hash_type = 0; size_t value_len; enum HASH_TYPE { BLAKE2sp = 0x00 }; if(!read_var_sized(a, &hash_type, &value_len)) return ARCHIVE_EOF; *extra_data_size -= value_len; if(ARCHIVE_OK != consume(a, value_len)) { return ARCHIVE_EOF; } /* The file uses BLAKE2sp checksum algorithm instead of plain old * CRC32. */ if(hash_type == BLAKE2sp) { const uint8_t* p; const int hash_size = sizeof(rar->file.blake2sp); if(!read_ahead(a, hash_size, &p)) return ARCHIVE_EOF; rar->file.has_blake2 = 1; memcpy(&rar->file.blake2sp, p, hash_size); if(ARCHIVE_OK != consume(a, hash_size)) { return ARCHIVE_EOF; } *extra_data_size -= hash_size; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported hash type (0x%jx)", (uintmax_t)hash_type); return ARCHIVE_FATAL; } return ARCHIVE_OK; } static int parse_htime_item(struct archive_read* a, char unix_time, int64_t* sec, uint32_t* nsec, int64_t* extra_data_size) { if(unix_time) { uint32_t time_val; if(!read_u32(a, &time_val)) return ARCHIVE_EOF; *extra_data_size -= 4; *sec = (int64_t) time_val; } else { uint64_t windows_time; if(!read_u64(a, &windows_time)) return ARCHIVE_EOF; ntfs_to_unix(windows_time, sec, nsec); *extra_data_size -= 8; } return ARCHIVE_OK; } static int parse_file_extra_version(struct archive_read* a, struct archive_entry* e, int64_t* extra_data_size) { size_t flags = 0; size_t version = 0; size_t value_len = 0; struct archive_string version_string; struct archive_string name_utf8_string; const char* cur_filename; /* Flags are ignored. */ if(!read_var_sized(a, &flags, &value_len)) return ARCHIVE_EOF; *extra_data_size -= value_len; if(ARCHIVE_OK != consume(a, value_len)) return ARCHIVE_EOF; if(!read_var_sized(a, &version, &value_len)) return ARCHIVE_EOF; *extra_data_size -= value_len; if(ARCHIVE_OK != consume(a, value_len)) return ARCHIVE_EOF; /* extra_data_size should be zero here. */ cur_filename = archive_entry_pathname_utf8(e); if(cur_filename == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Version entry without file name"); return ARCHIVE_FATAL; } archive_string_init(&version_string); archive_string_init(&name_utf8_string); /* Prepare a ;123 suffix for the filename, where '123' is the version * value of this file. */ archive_string_sprintf(&version_string, ";%zu", version); /* Build the new filename. */ archive_strcat(&name_utf8_string, cur_filename); archive_strcat(&name_utf8_string, version_string.s); /* Apply the new filename into this file's context. */ archive_entry_update_pathname_utf8(e, name_utf8_string.s); /* Free buffers. */ archive_string_free(&version_string); archive_string_free(&name_utf8_string); return ARCHIVE_OK; } static int parse_file_extra_htime(struct archive_read* a, struct archive_entry* e, struct rar5* rar, int64_t* extra_data_size) { char unix_time, has_unix_ns, has_mtime, has_ctime, has_atime; size_t flags = 0; size_t value_len; enum HTIME_FLAGS { IS_UNIX = 0x01, HAS_MTIME = 0x02, HAS_CTIME = 0x04, HAS_ATIME = 0x08, HAS_UNIX_NS = 0x10, }; if(!read_var_sized(a, &flags, &value_len)) return ARCHIVE_EOF; *extra_data_size -= value_len; if(ARCHIVE_OK != consume(a, value_len)) { return ARCHIVE_EOF; } unix_time = flags & IS_UNIX; has_unix_ns = unix_time && (flags & HAS_UNIX_NS); has_mtime = flags & HAS_MTIME; has_atime = flags & HAS_ATIME; has_ctime = flags & HAS_CTIME; rar->file.e_atime_ns = rar->file.e_ctime_ns = rar->file.e_mtime_ns = 0; if(has_mtime) { parse_htime_item(a, unix_time, &rar->file.e_mtime, &rar->file.e_mtime_ns, extra_data_size); } if(has_ctime) { parse_htime_item(a, unix_time, &rar->file.e_ctime, &rar->file.e_ctime_ns, extra_data_size); } if(has_atime) { parse_htime_item(a, unix_time, &rar->file.e_atime, &rar->file.e_atime_ns, extra_data_size); } if(has_mtime && has_unix_ns) { if(!read_u32(a, &rar->file.e_mtime_ns)) return ARCHIVE_EOF; *extra_data_size -= 4; } if(has_ctime && has_unix_ns) { if(!read_u32(a, &rar->file.e_ctime_ns)) return ARCHIVE_EOF; *extra_data_size -= 4; } if(has_atime && has_unix_ns) { if(!read_u32(a, &rar->file.e_atime_ns)) return ARCHIVE_EOF; *extra_data_size -= 4; } /* The seconds and nanoseconds are either together, or separated in two * fields so we parse them, then set the archive_entry's times. */ if(has_mtime) { archive_entry_set_mtime(e, rar->file.e_mtime, rar->file.e_mtime_ns); } if(has_ctime) { archive_entry_set_ctime(e, rar->file.e_ctime, rar->file.e_ctime_ns); } if(has_atime) { archive_entry_set_atime(e, rar->file.e_atime, rar->file.e_atime_ns); } return ARCHIVE_OK; } static int parse_file_extra_redir(struct archive_read* a, struct archive_entry* e, struct rar5* rar, int64_t* extra_data_size) { uint64_t value_size = 0; size_t target_size = 0; char target_utf8_buf[MAX_NAME_IN_BYTES]; const uint8_t* p; if(!read_var(a, &rar->file.redir_type, &value_size)) return ARCHIVE_EOF; if(ARCHIVE_OK != consume(a, (int64_t)value_size)) return ARCHIVE_EOF; *extra_data_size -= value_size; if(!read_var(a, &rar->file.redir_flags, &value_size)) return ARCHIVE_EOF; if(ARCHIVE_OK != consume(a, (int64_t)value_size)) return ARCHIVE_EOF; *extra_data_size -= value_size; if(!read_var_sized(a, &target_size, NULL)) return ARCHIVE_EOF; *extra_data_size -= target_size + 1; if(target_size > (MAX_NAME_IN_CHARS - 1)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Link target is too long"); return ARCHIVE_FATAL; } if(target_size == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "No link target specified"); return ARCHIVE_FATAL; } if(!read_ahead(a, target_size, &p)) return ARCHIVE_EOF; memcpy(target_utf8_buf, p, target_size); target_utf8_buf[target_size] = 0; if(ARCHIVE_OK != consume(a, (int64_t)target_size)) return ARCHIVE_EOF; switch(rar->file.redir_type) { case REDIR_TYPE_UNIXSYMLINK: case REDIR_TYPE_WINSYMLINK: archive_entry_set_filetype(e, AE_IFLNK); archive_entry_update_symlink_utf8(e, target_utf8_buf); if (rar->file.redir_flags & REDIR_SYMLINK_IS_DIR) { archive_entry_set_symlink_type(e, AE_SYMLINK_TYPE_DIRECTORY); } else { archive_entry_set_symlink_type(e, AE_SYMLINK_TYPE_FILE); } break; case REDIR_TYPE_HARDLINK: archive_entry_set_filetype(e, AE_IFREG); archive_entry_update_hardlink_utf8(e, target_utf8_buf); break; default: /* Unknown redir type, skip it. */ break; } return ARCHIVE_OK; } static int parse_file_extra_owner(struct archive_read* a, struct archive_entry* e, int64_t* extra_data_size) { uint64_t flags = 0; uint64_t value_size = 0; uint64_t id = 0; size_t name_len = 0; size_t name_size = 0; char namebuf[OWNER_MAXNAMELEN]; const uint8_t* p; if(!read_var(a, &flags, &value_size)) return ARCHIVE_EOF; if(ARCHIVE_OK != consume(a, (int64_t)value_size)) return ARCHIVE_EOF; *extra_data_size -= value_size; if ((flags & OWNER_USER_NAME) != 0) { if(!read_var_sized(a, &name_size, NULL)) return ARCHIVE_EOF; *extra_data_size -= name_size + 1; if(!read_ahead(a, name_size, &p)) return ARCHIVE_EOF; if (name_size >= OWNER_MAXNAMELEN) { name_len = OWNER_MAXNAMELEN - 1; } else { name_len = name_size; } memcpy(namebuf, p, name_len); namebuf[name_len] = 0; if(ARCHIVE_OK != consume(a, (int64_t)name_size)) return ARCHIVE_EOF; archive_entry_set_uname(e, namebuf); } if ((flags & OWNER_GROUP_NAME) != 0) { if(!read_var_sized(a, &name_size, NULL)) return ARCHIVE_EOF; *extra_data_size -= name_size + 1; if(!read_ahead(a, name_size, &p)) return ARCHIVE_EOF; if (name_size >= OWNER_MAXNAMELEN) { name_len = OWNER_MAXNAMELEN - 1; } else { name_len = name_size; } memcpy(namebuf, p, name_len); namebuf[name_len] = 0; if(ARCHIVE_OK != consume(a, (int64_t)name_size)) return ARCHIVE_EOF; archive_entry_set_gname(e, namebuf); } if ((flags & OWNER_USER_UID) != 0) { if(!read_var(a, &id, &value_size)) return ARCHIVE_EOF; if(ARCHIVE_OK != consume(a, (int64_t)value_size)) return ARCHIVE_EOF; *extra_data_size -= value_size; archive_entry_set_uid(e, (la_int64_t)id); } if ((flags & OWNER_GROUP_GID) != 0) { if(!read_var(a, &id, &value_size)) return ARCHIVE_EOF; if(ARCHIVE_OK != consume(a, (int64_t)value_size)) return ARCHIVE_EOF; *extra_data_size -= value_size; archive_entry_set_gid(e, (la_int64_t)id); } return ARCHIVE_OK; } static int process_head_file_extra(struct archive_read* a, struct archive_entry* e, struct rar5* rar, int64_t extra_data_size) { uint64_t extra_field_size; uint64_t extra_field_id = 0; - int ret = ARCHIVE_FATAL; uint64_t var_size; while(extra_data_size > 0) { + /* Make sure we won't fail if the file declares only unsupported + attributes. */ + int ret = ARCHIVE_OK; + if(!read_var(a, &extra_field_size, &var_size)) return ARCHIVE_EOF; extra_data_size -= var_size; if(ARCHIVE_OK != consume(a, var_size)) { return ARCHIVE_EOF; } if(!read_var(a, &extra_field_id, &var_size)) return ARCHIVE_EOF; extra_field_size -= var_size; extra_data_size -= var_size; if(ARCHIVE_OK != consume(a, var_size)) { return ARCHIVE_EOF; } switch(extra_field_id) { case EX_HASH: ret = parse_file_extra_hash(a, rar, &extra_data_size); break; case EX_HTIME: ret = parse_file_extra_htime(a, e, rar, &extra_data_size); break; case EX_REDIR: ret = parse_file_extra_redir(a, e, rar, &extra_data_size); break; case EX_UOWNER: ret = parse_file_extra_owner(a, e, &extra_data_size); break; case EX_VERSION: ret = parse_file_extra_version(a, e, &extra_data_size); break; case EX_CRYPT: /* Mark the entry as encrypted */ archive_entry_set_is_data_encrypted(e, 1); rar->has_encrypted_entries = 1; rar->cstate.data_encrypted = 1; /* fallthrough */ case EX_SUBDATA: /* fallthrough */ default: /* Skip unsupported entry. */ extra_data_size -= extra_field_size; if (ARCHIVE_OK != consume(a, extra_field_size)) { return ARCHIVE_EOF; } + + /* Don't fail on unsupported attribute -- we've handled it + by skipping over it. */ + ret = ARCHIVE_OK; + } + + if (ret != ARCHIVE_OK) { + /* Forward any errors signalled by the attribute parsing + functions. */ + return ret; } } - if(ret != ARCHIVE_OK) { - /* Attribute not implemented. */ - return ret; + if (extra_data_size != 0) { + /* We didn't skip everything, or we skipped too much; either way, + there's an error in this parsing function. */ + + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "unsupported structure of file header extra data"); + return ARCHIVE_FATAL; + } + + return ARCHIVE_OK; +} + +static int file_entry_sanity_checks(struct archive_read* a, + size_t block_flags, uint8_t is_dir, uint64_t unpacked_size, + size_t packed_size) +{ + if (is_dir) { + const int declares_data_size = + (int) (unpacked_size != 0 || packed_size != 0); + + /* FILE entries for directories still declare HFL_DATA in block flags, + even though attaching data to such blocks doesn't make much sense. */ + if (declares_data_size) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "directory entries cannot have any data"); + return ARCHIVE_FATAL; + } + } else { + const int declares_hfl_data = (int) ((block_flags & HFL_DATA) != 0); + if (!declares_hfl_data) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "no data found in file/service block"); + return ARCHIVE_FATAL; + } } return ARCHIVE_OK; } static int process_head_file(struct archive_read* a, struct rar5* rar, struct archive_entry* entry, size_t block_flags) { int64_t extra_data_size = 0; size_t data_size = 0; size_t file_flags = 0; size_t file_attr = 0; size_t compression_info = 0; size_t host_os = 0; size_t name_size = 0; uint64_t unpacked_size, window_size; uint32_t mtime = 0, crc = 0; int c_method = 0, c_version = 0; char name_utf8_buf[MAX_NAME_IN_BYTES]; const uint8_t* p; + int sanity_ret; enum FILE_FLAGS { DIRECTORY = 0x0001, UTIME = 0x0002, CRC32 = 0x0004, UNKNOWN_UNPACKED_SIZE = 0x0008, }; enum FILE_ATTRS { ATTR_READONLY = 0x1, ATTR_HIDDEN = 0x2, ATTR_SYSTEM = 0x4, ATTR_DIRECTORY = 0x10, }; enum COMP_INFO_FLAGS { SOLID = 0x0040, }; enum HOST_OS { HOST_WINDOWS = 0, HOST_UNIX = 1, }; archive_entry_clear(entry); /* Do not reset file context if we're switching archives. */ if(!rar->cstate.switch_multivolume) { reset_file_context(rar); } if(block_flags & HFL_EXTRA_DATA) { uint64_t edata_size = 0; if(!read_var(a, &edata_size, NULL)) return ARCHIVE_EOF; /* Intentional type cast from unsigned to signed. */ extra_data_size = (int64_t) edata_size; } if(block_flags & HFL_DATA) { if(!read_var_sized(a, &data_size, NULL)) return ARCHIVE_EOF; rar->file.bytes_remaining = data_size; } else { rar->file.bytes_remaining = 0; - - archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, - "no data found in file/service block"); - return ARCHIVE_FATAL; } if(!read_var_sized(a, &file_flags, NULL)) return ARCHIVE_EOF; if(!read_var(a, &unpacked_size, NULL)) return ARCHIVE_EOF; if(file_flags & UNKNOWN_UNPACKED_SIZE) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Files with unknown unpacked size are not supported"); return ARCHIVE_FATAL; } rar->file.dir = (uint8_t) ((file_flags & DIRECTORY) > 0); + sanity_ret = file_entry_sanity_checks(a, block_flags, rar->file.dir, + unpacked_size, data_size); + + if (sanity_ret != ARCHIVE_OK) { + return sanity_ret; + } + if(!read_var_sized(a, &file_attr, NULL)) return ARCHIVE_EOF; if(file_flags & UTIME) { if(!read_u32(a, &mtime)) return ARCHIVE_EOF; } if(file_flags & CRC32) { if(!read_u32(a, &crc)) return ARCHIVE_EOF; } if(!read_var_sized(a, &compression_info, NULL)) return ARCHIVE_EOF; c_method = (int) (compression_info >> 7) & 0x7; c_version = (int) (compression_info & 0x3f); /* RAR5 seems to limit the dictionary size to 64MB. */ window_size = (rar->file.dir > 0) ? 0 : g_unpack_window_size << ((compression_info >> 10) & 15); rar->cstate.method = c_method; rar->cstate.version = c_version + 50; rar->file.solid = (compression_info & SOLID) > 0; /* Archives which declare solid files without initializing the window * buffer first are invalid, unless previous data was encrypted, in * which case we may never have had the chance */ if(rar->file.solid > 0 && rar->cstate.data_encrypted == 0 && rar->cstate.window_buf == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Declared solid file, but no window buffer " "initialized yet."); return ARCHIVE_FATAL; } /* Check if window_size is a sane value. Also, if the file is not * declared as a directory, disallow window_size == 0. */ if(window_size > (64 * 1024 * 1024) || (rar->file.dir == 0 && window_size == 0)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Declared dictionary size is not supported."); return ARCHIVE_FATAL; } if(rar->file.solid > 0) { /* Re-check if current window size is the same as previous * window size (for solid files only). */ if(rar->file.solid_window_size > 0 && rar->file.solid_window_size != (ssize_t) window_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Window size for this solid file doesn't match " "the window size used in previous solid file. "); return ARCHIVE_FATAL; } } else rar->cstate.data_encrypted = 0; /* Reset for new buffer */ if(rar->cstate.window_size < (ssize_t) window_size && rar->cstate.window_buf) { /* The `data_ready` stack contains pointers to the `window_buf` or * `filtered_buf` buffers. Since we're about to reallocate the first * buffer, some of those pointers could become invalid. Therefore, we * need to dispose of all entries from the stack before attempting the * realloc. */ clear_data_ready_stack(rar); /* If window_buf has been allocated before, reallocate it, so * that its size will match new window_size. */ uint8_t* new_window_buf = realloc(rar->cstate.window_buf, (size_t) window_size); if(!new_window_buf) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Not enough memory when trying to realloc the window " "buffer."); return ARCHIVE_FATAL; } rar->cstate.window_buf = new_window_buf; } /* Values up to 64M should fit into ssize_t on every * architecture. */ rar->cstate.window_size = (ssize_t) window_size; if(rar->file.solid > 0 && rar->file.solid_window_size == 0) { /* Solid files have to have the same window_size across whole archive. Remember the window_size parameter for first solid file found. */ rar->file.solid_window_size = rar->cstate.window_size; } init_window_mask(rar); rar->file.service = 0; if(!read_var_sized(a, &host_os, NULL)) return ARCHIVE_EOF; if(host_os == HOST_WINDOWS) { /* Host OS is Windows */ __LA_MODE_T mode; if(file_attr & ATTR_DIRECTORY) { if (file_attr & ATTR_READONLY) { mode = 0555 | AE_IFDIR; } else { mode = 0755 | AE_IFDIR; } } else { if (file_attr & ATTR_READONLY) { mode = 0444 | AE_IFREG; } else { mode = 0644 | AE_IFREG; } } archive_entry_set_mode(entry, mode); if (file_attr & (ATTR_READONLY | ATTR_HIDDEN | ATTR_SYSTEM)) { char *fflags_text, *ptr; /* allocate for ",rdonly,hidden,system" */ fflags_text = malloc(22 * sizeof(*fflags_text)); if (fflags_text != NULL) { ptr = fflags_text; if (file_attr & ATTR_READONLY) { strcpy(ptr, ",rdonly"); ptr = ptr + 7; } if (file_attr & ATTR_HIDDEN) { strcpy(ptr, ",hidden"); ptr = ptr + 7; } if (file_attr & ATTR_SYSTEM) { strcpy(ptr, ",system"); ptr = ptr + 7; } if (ptr > fflags_text) { archive_entry_copy_fflags_text(entry, fflags_text + 1); } free(fflags_text); } } } else if(host_os == HOST_UNIX) { /* Host OS is Unix */ archive_entry_set_mode(entry, (__LA_MODE_T) file_attr); } else { /* Unknown host OS */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported Host OS: 0x%jx", (uintmax_t)host_os); return ARCHIVE_FATAL; } if(!read_var_sized(a, &name_size, NULL)) return ARCHIVE_EOF; if(name_size > (MAX_NAME_IN_CHARS - 1)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Filename is too long"); return ARCHIVE_FATAL; } if(name_size == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "No filename specified"); return ARCHIVE_FATAL; } if(!read_ahead(a, name_size, &p)) return ARCHIVE_EOF; memcpy(name_utf8_buf, p, name_size); name_utf8_buf[name_size] = 0; if(ARCHIVE_OK != consume(a, name_size)) { return ARCHIVE_EOF; } archive_entry_update_pathname_utf8(entry, name_utf8_buf); if(extra_data_size > 0) { int ret = process_head_file_extra(a, entry, rar, extra_data_size); /* * TODO: rewrite or remove useless sanity check * as extra_data_size is not passed as a pointer * if(extra_data_size < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "File extra data size is not zero"); return ARCHIVE_FATAL; } */ if(ret != ARCHIVE_OK) return ret; } if((file_flags & UNKNOWN_UNPACKED_SIZE) == 0) { rar->file.unpacked_size = (ssize_t) unpacked_size; if(rar->file.redir_type == REDIR_TYPE_NONE) archive_entry_set_size(entry, unpacked_size); } if(file_flags & UTIME) { archive_entry_set_mtime(entry, (time_t) mtime, 0); } if(file_flags & CRC32) { rar->file.stored_crc32 = crc; } if(!rar->cstate.switch_multivolume) { /* Do not reinitialize unpacking state if we're switching * archives. */ rar->cstate.block_parsing_finished = 1; rar->cstate.all_filters_applied = 1; rar->cstate.initialized = 0; } if(rar->generic.split_before > 0) { /* If now we're standing on a header that has a 'split before' * mark, it means we're standing on a 'continuation' file * header. Signal the caller that if it wants to move to * another file, it must call rar5_read_header() function * again. */ return ARCHIVE_RETRY; } else { return ARCHIVE_OK; } } static int process_head_service(struct archive_read* a, struct rar5* rar, struct archive_entry* entry, size_t block_flags) { /* Process this SERVICE block the same way as FILE blocks. */ int ret = process_head_file(a, rar, entry, block_flags); if(ret != ARCHIVE_OK) return ret; rar->file.service = 1; /* But skip the data part automatically. It's no use for the user * anyway. It contains only service data, not even needed to * properly unpack the file. */ ret = rar5_read_data_skip(a); if(ret != ARCHIVE_OK) return ret; /* After skipping, try parsing another block automatically. */ return ARCHIVE_RETRY; } static int process_head_main(struct archive_read* a, struct rar5* rar, struct archive_entry* entry, size_t block_flags) { int ret; uint64_t extra_data_size = 0; size_t extra_field_size = 0; size_t extra_field_id = 0; size_t archive_flags = 0; enum MAIN_FLAGS { VOLUME = 0x0001, /* multi-volume archive */ VOLUME_NUMBER = 0x0002, /* volume number, first vol doesn't * have it */ SOLID = 0x0004, /* solid archive */ PROTECT = 0x0008, /* contains Recovery info */ LOCK = 0x0010, /* readonly flag, not used */ }; enum MAIN_EXTRA { // Just one attribute here. LOCATOR = 0x01, }; (void) entry; if(block_flags & HFL_EXTRA_DATA) { if(!read_var(a, &extra_data_size, NULL)) return ARCHIVE_EOF; } else { extra_data_size = 0; } if(!read_var_sized(a, &archive_flags, NULL)) { return ARCHIVE_EOF; } rar->main.volume = (archive_flags & VOLUME) > 0; rar->main.solid = (archive_flags & SOLID) > 0; if(archive_flags & VOLUME_NUMBER) { size_t v = 0; if(!read_var_sized(a, &v, NULL)) { return ARCHIVE_EOF; } if (v > UINT_MAX) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid volume number"); return ARCHIVE_FATAL; } rar->main.vol_no = (unsigned int) v; } else { rar->main.vol_no = 0; } if(rar->vol.expected_vol_no > 0 && rar->main.vol_no != rar->vol.expected_vol_no) { /* Returning EOF instead of FATAL because of strange * libarchive behavior. When opening multiple files via * archive_read_open_filenames(), after reading up the whole * last file, the __archive_read_ahead function wraps up to * the first archive instead of returning EOF. */ return ARCHIVE_EOF; } if(extra_data_size == 0) { /* Early return. */ return ARCHIVE_OK; } if(!read_var_sized(a, &extra_field_size, NULL)) { return ARCHIVE_EOF; } if(!read_var_sized(a, &extra_field_id, NULL)) { return ARCHIVE_EOF; } if(extra_field_size == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid extra field size"); return ARCHIVE_FATAL; } switch(extra_field_id) { case LOCATOR: ret = process_main_locator_extra_block(a, rar); if(ret != ARCHIVE_OK) { /* Error while parsing main locator extra * block. */ return ret; } break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported extra type (0x%jx)", (uintmax_t)extra_field_id); return ARCHIVE_FATAL; } return ARCHIVE_OK; } static int skip_unprocessed_bytes(struct archive_read* a) { struct rar5* rar = get_context(a); int ret; if(rar->file.bytes_remaining) { /* Use different skipping method in block merging mode than in * normal mode. If merge mode is active, rar5_read_data_skip * can't be used, because it could allow recursive use of * merge_block() * function, and this function doesn't support * recursive use. */ if(rar->merge_mode) { /* Discard whole merged block. This is valid in solid * mode as well, because the code will discard blocks * only if those blocks are safe to discard (i.e. * they're not FILE blocks). */ ret = consume(a, rar->file.bytes_remaining); if(ret != ARCHIVE_OK) { return ret; } rar->file.bytes_remaining = 0; } else { /* If we're not in merge mode, use safe skipping code. * This will ensure we'll handle solid archives * properly. */ ret = rar5_read_data_skip(a); if(ret != ARCHIVE_OK) { return ret; } } } return ARCHIVE_OK; } static int scan_for_signature(struct archive_read* a); /* Base block processing function. A 'base block' is a RARv5 header block * that tells the reader what kind of data is stored inside the block. * * From the birds-eye view a RAR file looks file this: * * ... * * There are a few types of base blocks. Those types are specified inside * the 'switch' statement in this function. For example purposes, I'll write * how a standard RARv5 file could look like here: * *
* * The structure above could describe an archive file with 3 files in it, * one service "QuickOpen" block (that is ignored by this parser), and an * end of file base block marker. * * If the file is stored in multiple archive files ("multiarchive"), it might * look like this: * * .part01.rar:
* .part02.rar:
* .part03.rar:
* * This example could describe 3 RAR files that contain ONE archived file. * Or it could describe 3 RAR files that contain 3 different files. Or 3 * RAR files than contain 2 files. It all depends what metadata is stored in * the headers of blocks. * * Each block contains info about its size, the name of the file it's * storing inside, and whether this FILE block is a continuation block of * previous archive ('split before'), and is this FILE block should be * continued in another archive ('split after'). By parsing the 'split before' * and 'split after' flags, we're able to tell if multiple base blocks * are describing one file, or multiple files (with the same filename, for * example). * * One thing to note is that if we're parsing the first block, and * we see 'split after' flag, then we need to jump over to another * block to be able to decompress rest of the data. To do this, we need * to skip the block, then switch to another file, then skip the * block,
block, and then we're standing on the proper * block. */ static int process_base_block(struct archive_read* a, struct archive_entry* entry) { const size_t SMALLEST_RAR5_BLOCK_SIZE = 3; struct rar5* rar = get_context(a); uint32_t hdr_crc, computed_crc; size_t raw_hdr_size = 0, hdr_size_len, hdr_size; size_t header_id = 0; size_t header_flags = 0; const uint8_t* p; int ret; enum HEADER_TYPE { HEAD_MARK = 0x00, HEAD_MAIN = 0x01, HEAD_FILE = 0x02, HEAD_SERVICE = 0x03, HEAD_CRYPT = 0x04, HEAD_ENDARC = 0x05, HEAD_UNKNOWN = 0xff, }; /* Skip any unprocessed data for this file. */ ret = skip_unprocessed_bytes(a); if(ret != ARCHIVE_OK) return ret; /* Read the expected CRC32 checksum. */ if(!read_u32(a, &hdr_crc)) { return ARCHIVE_EOF; } /* Read header size. */ if(!read_var_sized(a, &raw_hdr_size, &hdr_size_len)) { return ARCHIVE_EOF; } hdr_size = raw_hdr_size + hdr_size_len; /* Sanity check, maximum header size for RAR5 is 2MB. */ if(hdr_size > (2 * 1024 * 1024)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Base block header is too large"); return ARCHIVE_FATAL; } /* Additional sanity checks to weed out invalid files. */ if(raw_hdr_size == 0 || hdr_size_len == 0 || hdr_size < SMALLEST_RAR5_BLOCK_SIZE) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Too small block encountered (%zu bytes)", raw_hdr_size); return ARCHIVE_FATAL; } /* Read the whole header data into memory, maximum memory use here is * 2MB. */ if(!read_ahead(a, hdr_size, &p)) { return ARCHIVE_EOF; } /* Verify the CRC32 of the header data. */ computed_crc = (uint32_t) crc32(0, p, (int) hdr_size); if(computed_crc != hdr_crc) { #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Header CRC error"); return ARCHIVE_FATAL; #endif } /* If the checksum is OK, we proceed with parsing. */ if(ARCHIVE_OK != consume(a, hdr_size_len)) { return ARCHIVE_EOF; } if(!read_var_sized(a, &header_id, NULL)) return ARCHIVE_EOF; if(!read_var_sized(a, &header_flags, NULL)) return ARCHIVE_EOF; rar->generic.split_after = (header_flags & HFL_SPLIT_AFTER) > 0; rar->generic.split_before = (header_flags & HFL_SPLIT_BEFORE) > 0; rar->generic.size = (int)hdr_size; rar->generic.last_header_id = (int)header_id; rar->main.endarc = 0; /* Those are possible header ids in RARv5. */ switch(header_id) { case HEAD_MAIN: ret = process_head_main(a, rar, entry, header_flags); /* Main header doesn't have any files in it, so it's * pointless to return to the caller. Retry to next * header, which should be HEAD_FILE/HEAD_SERVICE. */ if(ret == ARCHIVE_OK) return ARCHIVE_RETRY; return ret; case HEAD_SERVICE: ret = process_head_service(a, rar, entry, header_flags); return ret; case HEAD_FILE: ret = process_head_file(a, rar, entry, header_flags); return ret; case HEAD_CRYPT: archive_entry_set_is_metadata_encrypted(entry, 1); archive_entry_set_is_data_encrypted(entry, 1); rar->has_encrypted_entries = 1; rar->headers_are_encrypted = 1; archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Encryption is not supported"); return ARCHIVE_FATAL; case HEAD_ENDARC: rar->main.endarc = 1; /* After encountering an end of file marker, we need * to take into consideration if this archive is * continued in another file (i.e. is it part01.rar: * is there a part02.rar?) */ if(rar->main.volume) { /* In case there is part02.rar, position the * read pointer in a proper place, so we can * resume parsing. */ ret = scan_for_signature(a); if(ret == ARCHIVE_FATAL) { return ARCHIVE_EOF; } else { if(rar->vol.expected_vol_no == UINT_MAX) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Header error"); return ARCHIVE_FATAL; } rar->vol.expected_vol_no = rar->main.vol_no + 1; return ARCHIVE_OK; } } else { return ARCHIVE_EOF; } case HEAD_MARK: return ARCHIVE_EOF; default: if((header_flags & HFL_SKIP_IF_UNKNOWN) == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Header type error"); return ARCHIVE_FATAL; } else { /* If the block is marked as 'skip if unknown', * do as the flag says: skip the block * instead on failing on it. */ return ARCHIVE_RETRY; } } #if !defined WIN32 // Not reached. archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Internal unpacker error"); return ARCHIVE_FATAL; #endif } static int skip_base_block(struct archive_read* a) { int ret; struct rar5* rar = get_context(a); /* Create a new local archive_entry structure that will be operated on * by header reader; operations on this archive_entry will be discarded. */ struct archive_entry* entry = archive_entry_new(); ret = process_base_block(a, entry); /* Discard operations on this archive_entry structure. */ archive_entry_free(entry); if(ret == ARCHIVE_FATAL) return ret; if(rar->generic.last_header_id == 2 && rar->generic.split_before > 0) return ARCHIVE_OK; if(ret == ARCHIVE_OK) return ARCHIVE_RETRY; else return ret; } static int try_skip_sfx(struct archive_read *a) { const char *p; if ((p = __archive_read_ahead(a, 7, NULL)) == NULL) return ARCHIVE_EOF; if ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0) { char signature[sizeof(rar5_signature_xor)]; const void *h; const char *q; size_t skip, total = 0; ssize_t bytes, window = 4096; rar5_signature(signature); while (total + window <= (1024 * 512)) { h = __archive_read_ahead(a, window, &bytes); if (h == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < 0x40) goto fatal; continue; } if (bytes < 0x40) goto fatal; p = h; q = p + bytes; /* * Scan ahead until we find something that looks * like the RAR header. */ while (p + 8 < q) { if (memcmp(p, signature, sizeof(signature)) == 0) { skip = p - (const char *)h; __archive_read_consume(a, skip); return (ARCHIVE_OK); } p += 0x10; } skip = p - (const char *)h; __archive_read_consume(a, skip); total += skip; } } return ARCHIVE_OK; fatal: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Couldn't find out RAR header"); return (ARCHIVE_FATAL); } static int rar5_read_header(struct archive_read *a, struct archive_entry *entry) { struct rar5* rar = get_context(a); int ret; /* * It should be sufficient to call archive_read_next_header() for * a reader to determine if an entry is encrypted or not. */ if (rar->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { rar->has_encrypted_entries = 0; } if(rar->header_initialized == 0) { init_header(a); if ((ret = try_skip_sfx(a)) < ARCHIVE_WARN) return ret; rar->header_initialized = 1; } if(rar->skipped_magic == 0) { if(ARCHIVE_OK != consume(a, sizeof(rar5_signature_xor))) { return ARCHIVE_EOF; } rar->skipped_magic = 1; } do { ret = process_base_block(a, entry); } while(ret == ARCHIVE_RETRY || (rar->main.endarc > 0 && ret == ARCHIVE_OK)); return ret; } static void init_unpack(struct rar5* rar) { rar->file.calculated_crc32 = 0; init_window_mask(rar); free(rar->cstate.window_buf); free(rar->cstate.filtered_buf); if(rar->cstate.window_size > 0) { rar->cstate.window_buf = calloc(1, rar->cstate.window_size); rar->cstate.filtered_buf = calloc(1, rar->cstate.window_size); } else { rar->cstate.window_buf = NULL; rar->cstate.filtered_buf = NULL; } clear_data_ready_stack(rar); rar->cstate.write_ptr = 0; rar->cstate.last_write_ptr = 0; memset(&rar->cstate.bd, 0, sizeof(rar->cstate.bd)); memset(&rar->cstate.ld, 0, sizeof(rar->cstate.ld)); memset(&rar->cstate.dd, 0, sizeof(rar->cstate.dd)); memset(&rar->cstate.ldd, 0, sizeof(rar->cstate.ldd)); memset(&rar->cstate.rd, 0, sizeof(rar->cstate.rd)); } static void update_crc(struct rar5* rar, const uint8_t* p, size_t to_read) { int verify_crc; if(rar->skip_mode) { #if defined CHECK_CRC_ON_SOLID_SKIP verify_crc = 1; #else verify_crc = 0; #endif } else verify_crc = 1; if(verify_crc) { /* Don't update CRC32 if the file doesn't have the * `stored_crc32` info filled in. */ if(rar->file.stored_crc32 > 0) { rar->file.calculated_crc32 = crc32(rar->file.calculated_crc32, p, (unsigned int)to_read); } /* Check if the file uses an optional BLAKE2sp checksum * algorithm. */ if(rar->file.has_blake2 > 0) { /* Return value of the `update` function is always 0, * so we can explicitly ignore it here. */ (void) blake2sp_update(&rar->file.b2state, p, to_read); } } } static int create_decode_tables(uint8_t* bit_length, struct decode_table* table, int size) { int code, upper_limit = 0, i, lc[16]; uint32_t decode_pos_clone[rar5_countof(table->decode_pos)]; ssize_t cur_len, quick_data_size; memset(&lc, 0, sizeof(lc)); memset(table->decode_num, 0, sizeof(table->decode_num)); table->size = size; table->quick_bits = size == HUFF_NC ? 10 : 7; for(i = 0; i < size; i++) { lc[bit_length[i] & 15]++; } lc[0] = 0; table->decode_pos[0] = 0; table->decode_len[0] = 0; for(i = 1; i < 16; i++) { upper_limit += lc[i]; table->decode_len[i] = upper_limit << (16 - i); table->decode_pos[i] = table->decode_pos[i - 1] + lc[i - 1]; upper_limit <<= 1; } memcpy(decode_pos_clone, table->decode_pos, sizeof(decode_pos_clone)); for(i = 0; i < size; i++) { uint8_t clen = bit_length[i] & 15; if(clen > 0) { int last_pos = decode_pos_clone[clen]; table->decode_num[last_pos] = i; decode_pos_clone[clen]++; } } quick_data_size = (int64_t)1 << table->quick_bits; cur_len = 1; for(code = 0; code < quick_data_size; code++) { int bit_field = code << (16 - table->quick_bits); int dist, pos; while(cur_len < rar5_countof(table->decode_len) && bit_field >= table->decode_len[cur_len]) { cur_len++; } table->quick_len[code] = (uint8_t) cur_len; dist = bit_field - table->decode_len[cur_len - 1]; dist >>= (16 - cur_len); pos = table->decode_pos[cur_len & 15] + dist; if(cur_len < rar5_countof(table->decode_pos) && pos < size) { table->quick_num[code] = table->decode_num[pos]; } else { table->quick_num[code] = 0; } } return ARCHIVE_OK; } static int decode_number(struct archive_read* a, struct decode_table* table, const uint8_t* p, uint16_t* num) { int i, bits, dist, ret; uint16_t bitfield; uint32_t pos; struct rar5* rar = get_context(a); if(ARCHIVE_OK != (ret = read_bits_16(a, rar, p, &bitfield))) { return ret; } bitfield &= 0xfffe; if(bitfield < table->decode_len[table->quick_bits]) { int code = bitfield >> (16 - table->quick_bits); skip_bits(rar, table->quick_len[code]); *num = table->quick_num[code]; return ARCHIVE_OK; } bits = 15; for(i = table->quick_bits + 1; i < 15; i++) { if(bitfield < table->decode_len[i]) { bits = i; break; } } skip_bits(rar, bits); dist = bitfield - table->decode_len[bits - 1]; dist >>= (16 - bits); pos = table->decode_pos[bits] + dist; if(pos >= table->size) pos = 0; *num = table->decode_num[pos]; return ARCHIVE_OK; } /* Reads and parses Huffman tables from the beginning of the block. */ static int parse_tables(struct archive_read* a, struct rar5* rar, const uint8_t* p) { int ret, value, i, w, idx = 0; uint8_t bit_length[HUFF_BC], table[HUFF_TABLE_SIZE], nibble_mask = 0xF0, nibble_shift = 4; enum { ESCAPE = 15 }; /* The data for table generation is compressed using a simple RLE-like * algorithm when storing zeroes, so we need to unpack it first. */ for(w = 0, i = 0; w < HUFF_BC;) { if(i >= rar->cstate.cur_block_size) { /* Truncated data, can't continue. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated data in huffman tables"); return ARCHIVE_FATAL; } value = (p[i] & nibble_mask) >> nibble_shift; if(nibble_mask == 0x0F) ++i; nibble_mask ^= 0xFF; nibble_shift ^= 4; /* Values smaller than 15 is data, so we write it directly. * Value 15 is a flag telling us that we need to unpack more * bytes. */ if(value == ESCAPE) { value = (p[i] & nibble_mask) >> nibble_shift; if(nibble_mask == 0x0F) ++i; nibble_mask ^= 0xFF; nibble_shift ^= 4; if(value == 0) { /* We sometimes need to write the actual value * of 15, so this case handles that. */ bit_length[w++] = ESCAPE; } else { int k; /* Fill zeroes. */ for(k = 0; (k < value + 2) && (w < HUFF_BC); k++) { bit_length[w++] = 0; } } } else { bit_length[w++] = value; } } rar->bits.in_addr = i; rar->bits.bit_addr = nibble_shift ^ 4; ret = create_decode_tables(bit_length, &rar->cstate.bd, HUFF_BC); if(ret != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Decoding huffman tables failed"); return ARCHIVE_FATAL; } for(i = 0; i < HUFF_TABLE_SIZE;) { uint16_t num; ret = decode_number(a, &rar->cstate.bd, p, &num); if(ret != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Decoding huffman tables failed"); return ARCHIVE_FATAL; } if(num < 16) { /* 0..15: store directly */ table[i] = (uint8_t) num; i++; } else if(num < 18) { /* 16..17: repeat previous code */ uint16_t n; if(ARCHIVE_OK != (ret = read_bits_16(a, rar, p, &n))) return ret; if(num == 16) { n >>= 13; n += 3; skip_bits(rar, 3); } else { n >>= 9; n += 11; skip_bits(rar, 7); } if(i > 0) { while(n-- > 0 && i < HUFF_TABLE_SIZE) { table[i] = table[i - 1]; i++; } } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unexpected error when decoding " "huffman tables"); return ARCHIVE_FATAL; } } else { /* other codes: fill with zeroes `n` times */ uint16_t n; if(ARCHIVE_OK != (ret = read_bits_16(a, rar, p, &n))) return ret; if(num == 18) { n >>= 13; n += 3; skip_bits(rar, 3); } else { n >>= 9; n += 11; skip_bits(rar, 7); } while(n-- > 0 && i < HUFF_TABLE_SIZE) table[i++] = 0; } } ret = create_decode_tables(&table[idx], &rar->cstate.ld, HUFF_NC); if(ret != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Failed to create literal table"); return ARCHIVE_FATAL; } idx += HUFF_NC; ret = create_decode_tables(&table[idx], &rar->cstate.dd, HUFF_DC); if(ret != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Failed to create distance table"); return ARCHIVE_FATAL; } idx += HUFF_DC; ret = create_decode_tables(&table[idx], &rar->cstate.ldd, HUFF_LDC); if(ret != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Failed to create lower bits of distances table"); return ARCHIVE_FATAL; } idx += HUFF_LDC; ret = create_decode_tables(&table[idx], &rar->cstate.rd, HUFF_RC); if(ret != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Failed to create repeating distances table"); return ARCHIVE_FATAL; } return ARCHIVE_OK; } /* Parses the block header, verifies its CRC byte, and saves the header * fields inside the `hdr` pointer. */ static int parse_block_header(struct archive_read* a, const uint8_t* p, ssize_t* block_size, struct compressed_block_header* hdr) { uint8_t calculated_cksum; memcpy(hdr, p, sizeof(struct compressed_block_header)); if(bf_byte_count(hdr) > 2) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported block header size (was %d, max is 2)", bf_byte_count(hdr)); return ARCHIVE_FATAL; } /* This should probably use bit reader interface in order to be more * future-proof. */ *block_size = 0; switch(bf_byte_count(hdr)) { /* 1-byte block size */ case 0: *block_size = *(const uint8_t*) &p[2]; break; /* 2-byte block size */ case 1: *block_size = archive_le16dec(&p[2]); break; /* 3-byte block size */ case 2: *block_size = archive_le32dec(&p[2]); *block_size &= 0x00FFFFFF; break; /* Other block sizes are not supported. This case is not * reached, because we have an 'if' guard before the switch * that makes sure of it. */ default: return ARCHIVE_FATAL; } /* Verify the block header checksum. 0x5A is a magic value and is * always * constant. */ calculated_cksum = 0x5A ^ (uint8_t) hdr->block_flags_u8 ^ (uint8_t) *block_size ^ (uint8_t) (*block_size >> 8) ^ (uint8_t) (*block_size >> 16); if(calculated_cksum != hdr->block_cksum) { #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Block checksum error: got 0x%x, expected 0x%x", hdr->block_cksum, calculated_cksum); return ARCHIVE_FATAL; #endif } return ARCHIVE_OK; } /* Convenience function used during filter processing. */ static int parse_filter_data(struct archive_read* a, struct rar5* rar, const uint8_t* p, uint32_t* filter_data) { int i, bytes, ret; uint32_t data = 0; if(ARCHIVE_OK != (ret = read_consume_bits(a, rar, p, 2, &bytes))) return ret; bytes++; for(i = 0; i < bytes; i++) { uint16_t byte; if(ARCHIVE_OK != (ret = read_bits_16(a, rar, p, &byte))) { return ret; } /* Cast to uint32_t will ensure the shift operation will not * produce undefined result. */ data += ((uint32_t) byte >> 8) << (i * 8); skip_bits(rar, 8); } *filter_data = data; return ARCHIVE_OK; } /* Function is used during sanity checking. */ static int is_valid_filter_block_start(struct rar5* rar, uint32_t start) { const int64_t block_start = (ssize_t) start + rar->cstate.write_ptr; const int64_t last_bs = rar->cstate.last_block_start; const ssize_t last_bl = rar->cstate.last_block_length; if(last_bs == 0 || last_bl == 0) { /* We didn't have any filters yet, so accept this offset. */ return 1; } if(block_start >= last_bs + last_bl) { /* Current offset is bigger than last block's end offset, so * accept current offset. */ return 1; } /* Any other case is not a normal situation and we should fail. */ return 0; } /* The function will create a new filter, read its parameters from the input * stream and add it to the filter collection. */ static int parse_filter(struct archive_read* ar, const uint8_t* p) { uint32_t block_start, block_length; uint16_t filter_type; struct filter_info* filt = NULL; struct rar5* rar = get_context(ar); int ret; /* Read the parameters from the input stream. */ if(ARCHIVE_OK != (ret = parse_filter_data(ar, rar, p, &block_start))) return ret; if(ARCHIVE_OK != (ret = parse_filter_data(ar, rar, p, &block_length))) return ret; if(ARCHIVE_OK != (ret = read_bits_16(ar, rar, p, &filter_type))) return ret; filter_type >>= 13; skip_bits(rar, 3); /* Perform some sanity checks on this filter parameters. Note that we * allow only DELTA, E8/E9 and ARM filters here, because rest of * filters are not used in RARv5. */ if(block_length < 4 || block_length > 0x400000 || filter_type > FILTER_ARM || !is_valid_filter_block_start(rar, block_start)) { archive_set_error(&ar->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid filter encountered"); return ARCHIVE_FATAL; } /* Allocate a new filter. */ filt = add_new_filter(rar); if(filt == NULL) { archive_set_error(&ar->archive, ENOMEM, "Can't allocate memory for a filter descriptor."); return ARCHIVE_FATAL; } filt->type = filter_type; filt->block_start = rar->cstate.write_ptr + block_start; filt->block_length = block_length; rar->cstate.last_block_start = filt->block_start; rar->cstate.last_block_length = filt->block_length; /* Read some more data in case this is a DELTA filter. Other filter * types don't require any additional data over what was already * read. */ if(filter_type == FILTER_DELTA) { int channels; if(ARCHIVE_OK != (ret = read_consume_bits(ar, rar, p, 5, &channels))) return ret; filt->channels = channels + 1; } return ARCHIVE_OK; } static int decode_code_length(struct archive_read* a, struct rar5* rar, const uint8_t* p, uint16_t code) { int lbits, length = 2; if(code < 8) { lbits = 0; length += code; } else { lbits = code / 4 - 1; length += (4 | (code & 3)) << lbits; } if(lbits > 0) { int add; if(ARCHIVE_OK != read_consume_bits(a, rar, p, lbits, &add)) return -1; length += add; } return length; } static int copy_string(struct archive_read* a, int len, int dist) { struct rar5* rar = get_context(a); const ssize_t cmask = rar->cstate.window_mask; const uint64_t write_ptr = rar->cstate.write_ptr + rar->cstate.solid_offset; int i; if (rar->cstate.window_buf == NULL) return ARCHIVE_FATAL; /* The unpacker spends most of the time in this function. It would be * a good idea to introduce some optimizations here. * * Just remember that this loop treats buffers that overlap differently * than buffers that do not overlap. This is why a simple memcpy(3) * call will not be enough. */ for(i = 0; i < len; i++) { const ssize_t write_idx = (write_ptr + i) & cmask; const ssize_t read_idx = (write_ptr + i - dist) & cmask; rar->cstate.window_buf[write_idx] = rar->cstate.window_buf[read_idx]; } rar->cstate.write_ptr += len; return ARCHIVE_OK; } static int do_uncompress_block(struct archive_read* a, const uint8_t* p) { struct rar5* rar = get_context(a); uint16_t num; int ret; const uint64_t cmask = rar->cstate.window_mask; const struct compressed_block_header* hdr = &rar->last_block_hdr; const uint8_t bit_size = 1 + bf_bit_size(hdr); while(1) { if(rar->cstate.write_ptr - rar->cstate.last_write_ptr > (rar->cstate.window_size >> 1)) { /* Don't allow growing data by more than half of the * window size at a time. In such case, break the loop; * next call to this function will continue processing * from this moment. */ break; } if(rar->bits.in_addr > rar->cstate.cur_block_size - 1 || (rar->bits.in_addr == rar->cstate.cur_block_size - 1 && rar->bits.bit_addr >= bit_size)) { /* If the program counter is here, it means the * function has finished processing the block. */ rar->cstate.block_parsing_finished = 1; break; } /* Decode the next literal. */ if(ARCHIVE_OK != decode_number(a, &rar->cstate.ld, p, &num)) { return ARCHIVE_EOF; } /* Num holds a decompression literal, or 'command code'. * * - Values lower than 256 are just bytes. Those codes * can be stored in the output buffer directly. * * - Code 256 defines a new filter, which is later used to * transform the data block accordingly to the filter type. * The data block needs to be fully uncompressed first. * * - Code bigger than 257 and smaller than 262 define * a repetition pattern that should be copied from * an already uncompressed chunk of data. */ if(num < 256) { /* Directly store the byte. */ int64_t write_idx = rar->cstate.solid_offset + rar->cstate.write_ptr++; rar->cstate.window_buf[write_idx & cmask] = (uint8_t) num; continue; } else if(num >= 262) { uint16_t dist_slot; int len = decode_code_length(a, rar, p, num - 262), dbits, dist = 1; if(len == -1) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Failed to decode the code length"); return ARCHIVE_FATAL; } if(ARCHIVE_OK != decode_number(a, &rar->cstate.dd, p, &dist_slot)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Failed to decode the distance slot"); return ARCHIVE_FATAL; } if(dist_slot < 4) { dbits = 0; dist += dist_slot; } else { dbits = dist_slot / 2 - 1; /* Cast to uint32_t will make sure the shift * left operation won't produce undefined * result. Then, the uint32_t type will * be implicitly casted to int. */ dist += (uint32_t) (2 | (dist_slot & 1)) << dbits; } if(dbits > 0) { if(dbits >= 4) { uint32_t add = 0; uint16_t low_dist; if(dbits > 4) { if(ARCHIVE_OK != (ret = read_bits_32( a, rar, p, &add))) { /* Return EOF if we * can't read more * data. */ return ret; } skip_bits(rar, dbits - 4); add = (add >> ( 36 - dbits)) << 4; dist += add; } if(ARCHIVE_OK != decode_number(a, &rar->cstate.ldd, p, &low_dist)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Failed to decode the " "distance slot"); return ARCHIVE_FATAL; } if(dist >= INT_MAX - low_dist - 1) { /* This only happens in * invalid archives. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Distance pointer " "overflow"); return ARCHIVE_FATAL; } dist += low_dist; } else { /* dbits is one of [0,1,2,3] */ int add; if(ARCHIVE_OK != (ret = read_consume_bits(a, rar, p, dbits, &add))) { /* Return EOF if we can't read * more data. */ return ret; } dist += add; } } if(dist > 0x100) { len++; if(dist > 0x2000) { len++; if(dist > 0x40000) { len++; } } } dist_cache_push(rar, dist); rar->cstate.last_len = len; if(ARCHIVE_OK != copy_string(a, len, dist)) return ARCHIVE_FATAL; continue; } else if(num == 256) { /* Create a filter. */ ret = parse_filter(a, p); if(ret != ARCHIVE_OK) return ret; continue; } else if(num == 257) { if(rar->cstate.last_len != 0) { if(ARCHIVE_OK != copy_string(a, rar->cstate.last_len, rar->cstate.dist_cache[0])) { return ARCHIVE_FATAL; } } continue; } else { /* num < 262 */ const int idx = num - 258; const int dist = dist_cache_touch(rar, idx); uint16_t len_slot; int len; if(ARCHIVE_OK != decode_number(a, &rar->cstate.rd, p, &len_slot)) { return ARCHIVE_FATAL; } len = decode_code_length(a, rar, p, len_slot); if (len == -1) { return ARCHIVE_FATAL; } rar->cstate.last_len = len; if(ARCHIVE_OK != copy_string(a, len, dist)) return ARCHIVE_FATAL; continue; } } return ARCHIVE_OK; } /* Binary search for the RARv5 signature. */ static int scan_for_signature(struct archive_read* a) { const uint8_t* p; const int chunk_size = 512; ssize_t i; char signature[sizeof(rar5_signature_xor)]; /* If we're here, it means we're on an 'unknown territory' data. * There's no indication what kind of data we're reading here. * It could be some text comment, any kind of binary data, * digital sign, dragons, etc. * * We want to find a valid RARv5 magic header inside this unknown * data. */ /* Is it possible in libarchive to just skip everything until the * end of the file? If so, it would be a better approach than the * current implementation of this function. */ rar5_signature(signature); while(1) { if(!read_ahead(a, chunk_size, &p)) return ARCHIVE_EOF; for(i = 0; i < chunk_size - (int)sizeof(rar5_signature_xor); i++) { if(memcmp(&p[i], signature, sizeof(rar5_signature_xor)) == 0) { /* Consume the number of bytes we've used to * search for the signature, as well as the * number of bytes used by the signature * itself. After this we should be standing * on a valid base block header. */ (void) consume(a, i + sizeof(rar5_signature_xor)); return ARCHIVE_OK; } } consume(a, chunk_size); } return ARCHIVE_FATAL; } /* This function will switch the multivolume archive file to another file, * i.e. from part03 to part 04. */ static int advance_multivolume(struct archive_read* a) { int lret; struct rar5* rar = get_context(a); /* A small state machine that will skip unnecessary data, needed to * switch from one multivolume to another. Such skipping is needed if * we want to be an stream-oriented (instead of file-oriented) * unpacker. * * The state machine starts with `rar->main.endarc` == 0. It also * assumes that current stream pointer points to some base block * header. * * The `endarc` field is being set when the base block parsing * function encounters the 'end of archive' marker. */ while(1) { if(rar->main.endarc == 1) { int looping = 1; rar->main.endarc = 0; while(looping) { lret = skip_base_block(a); switch(lret) { case ARCHIVE_RETRY: /* Continue looping. */ break; case ARCHIVE_OK: /* Break loop. */ looping = 0; break; default: /* Forward any errors to the * caller. */ return lret; } } break; } else { /* Skip current base block. In order to properly skip * it, we really need to simply parse it and discard * the results. */ lret = skip_base_block(a); if(lret == ARCHIVE_FATAL || lret == ARCHIVE_FAILED) return lret; /* The `skip_base_block` function tells us if we * should continue with skipping, or we should stop * skipping. We're trying to skip everything up to * a base FILE block. */ if(lret != ARCHIVE_RETRY) { /* If there was an error during skipping, or we * have just skipped a FILE base block... */ if(rar->main.endarc == 0) { return lret; } else { continue; } } } } return ARCHIVE_OK; } /* Merges the partial block from the first multivolume archive file, and * partial block from the second multivolume archive file. The result is * a chunk of memory containing the whole block, and the stream pointer * is advanced to the next block in the second multivolume archive file. */ static int merge_block(struct archive_read* a, ssize_t block_size, const uint8_t** p) { struct rar5* rar = get_context(a); ssize_t cur_block_size, partial_offset = 0; const uint8_t* lp; int ret; if(rar->merge_mode) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Recursive merge is not allowed"); return ARCHIVE_FATAL; } /* Set a flag that we're in the switching mode. */ rar->cstate.switch_multivolume = 1; /* Reallocate the memory which will hold the whole block. */ if(rar->vol.push_buf) free((void*) rar->vol.push_buf); /* Increasing the allocation block by 8 is due to bit reading functions, * which are using additional 2 or 4 bytes. Allocating the block size * by exact value would make bit reader perform reads from invalid * memory block when reading the last byte from the buffer. */ rar->vol.push_buf = malloc(block_size + 8); if(!rar->vol.push_buf) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for a merge block buffer."); return ARCHIVE_FATAL; } /* Valgrind complains if the extension block for bit reader is not * initialized, so initialize it. */ memset(&rar->vol.push_buf[block_size], 0, 8); /* A single block can span across multiple multivolume archive files, * so we use a loop here. This loop will consume enough multivolume * archive files until the whole block is read. */ while(1) { /* Get the size of current block chunk in this multivolume * archive file and read it. */ cur_block_size = rar5_min(rar->file.bytes_remaining, block_size - partial_offset); if(cur_block_size == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Encountered block size == 0 during block merge"); return ARCHIVE_FATAL; } if(!read_ahead(a, cur_block_size, &lp)) return ARCHIVE_EOF; /* Sanity check; there should never be a situation where this * function reads more data than the block's size. */ if(partial_offset + cur_block_size > block_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Consumed too much data when merging blocks."); return ARCHIVE_FATAL; } /* Merge previous block chunk with current block chunk, * or create first block chunk if this is our first * iteration. */ memcpy(&rar->vol.push_buf[partial_offset], lp, cur_block_size); /* Advance the stream read pointer by this block chunk size. */ if(ARCHIVE_OK != consume(a, cur_block_size)) return ARCHIVE_EOF; /* Update the pointers. `partial_offset` contains information * about the sum of merged block chunks. */ partial_offset += cur_block_size; rar->file.bytes_remaining -= cur_block_size; /* If `partial_offset` is the same as `block_size`, this means * we've merged all block chunks and we have a valid full * block. */ if(partial_offset == block_size) { break; } /* If we don't have any bytes to read, this means we should * switch to another multivolume archive file. */ if(rar->file.bytes_remaining == 0) { rar->merge_mode++; ret = advance_multivolume(a); rar->merge_mode--; if(ret != ARCHIVE_OK) { return ret; } } } *p = rar->vol.push_buf; /* If we're here, we can resume unpacking by processing the block * pointed to by the `*p` memory pointer. */ return ARCHIVE_OK; } static int process_block(struct archive_read* a) { const uint8_t* p; struct rar5* rar = get_context(a); int ret; /* If we don't have any data to be processed, this most probably means * we need to switch to the next volume. */ if(rar->main.volume && rar->file.bytes_remaining == 0) { ret = advance_multivolume(a); if(ret != ARCHIVE_OK) return ret; } if(rar->cstate.block_parsing_finished) { ssize_t block_size; ssize_t to_skip; ssize_t cur_block_size; /* The header size won't be bigger than 6 bytes. */ if(!read_ahead(a, 6, &p)) { /* Failed to prefetch data block header. */ return ARCHIVE_EOF; } /* * Read block_size by parsing block header. Validate the header * by calculating CRC byte stored inside the header. Size of * the header is not constant (block size can be stored either * in 1 or 2 bytes), that's why block size is left out from the * `compressed_block_header` structure and returned by * `parse_block_header` as the second argument. */ ret = parse_block_header(a, p, &block_size, &rar->last_block_hdr); if(ret != ARCHIVE_OK) { return ret; } /* Skip block header. Next data is huffman tables, * if present. */ to_skip = sizeof(struct compressed_block_header) + bf_byte_count(&rar->last_block_hdr) + 1; if(ARCHIVE_OK != consume(a, to_skip)) return ARCHIVE_EOF; rar->file.bytes_remaining -= to_skip; /* The block size gives information about the whole block size, * but the block could be stored in split form when using * multi-volume archives. In this case, the block size will be * bigger than the actual data stored in this file. Remaining * part of the data will be in another file. */ cur_block_size = rar5_min(rar->file.bytes_remaining, block_size); if(block_size > rar->file.bytes_remaining) { /* If current blocks' size is bigger than our data * size, this means we have a multivolume archive. * In this case, skip all base headers until the end * of the file, proceed to next "partXXX.rar" volume, * find its signature, skip all headers up to the first * FILE base header, and continue from there. * * Note that `merge_block` will update the `rar` * context structure quite extensively. */ ret = merge_block(a, block_size, &p); if(ret != ARCHIVE_OK) { return ret; } cur_block_size = block_size; /* Current stream pointer should be now directly * *after* the block that spanned through multiple * archive files. `p` pointer should have the data of * the *whole* block (merged from partial blocks * stored in multiple archives files). */ } else { rar->cstate.switch_multivolume = 0; /* Read the whole block size into memory. This can take * up to 8 megabytes of memory in theoretical cases. * Might be worth to optimize this and use a standard * chunk of 4kb's. */ if(!read_ahead(a, 4 + cur_block_size, &p)) { /* Failed to prefetch block data. */ return ARCHIVE_EOF; } } rar->cstate.block_buf = p; rar->cstate.cur_block_size = cur_block_size; rar->cstate.block_parsing_finished = 0; rar->bits.in_addr = 0; rar->bits.bit_addr = 0; if(bf_is_table_present(&rar->last_block_hdr)) { /* Load Huffman tables. */ ret = parse_tables(a, rar, p); if(ret != ARCHIVE_OK) { /* Error during decompression of Huffman * tables. */ return ret; } } } else { /* Block parsing not finished, reuse previous memory buffer. */ p = rar->cstate.block_buf; } /* Uncompress the block, or a part of it, depending on how many bytes * will be generated by uncompressing the block. * * In case too many bytes will be generated, calling this function * again will resume the uncompression operation. */ ret = do_uncompress_block(a, p); if(ret != ARCHIVE_OK) { return ret; } if(rar->cstate.block_parsing_finished && rar->cstate.switch_multivolume == 0 && rar->cstate.cur_block_size > 0) { /* If we're processing a normal block, consume the whole * block. We can do this because we've already read the whole * block to memory. */ if(ARCHIVE_OK != consume(a, rar->cstate.cur_block_size)) return ARCHIVE_FATAL; rar->file.bytes_remaining -= rar->cstate.cur_block_size; } else if(rar->cstate.switch_multivolume) { /* Don't consume the block if we're doing multivolume * processing. The volume switching function will consume * the proper count of bytes instead. */ rar->cstate.switch_multivolume = 0; } return ARCHIVE_OK; } /* Pops the `buf`, `size` and `offset` from the "data ready" stack. * * Returns ARCHIVE_OK when those arguments can be used, ARCHIVE_RETRY * when there is no data on the stack. */ static int use_data(struct rar5* rar, const void** buf, size_t* size, int64_t* offset) { int i; for(i = 0; i < rar5_countof(rar->cstate.dready); i++) { struct data_ready *d = &rar->cstate.dready[i]; if(d->used) { if(buf) *buf = d->buf; if(size) *size = d->size; if(offset) *offset = d->offset; d->used = 0; return ARCHIVE_OK; } } return ARCHIVE_RETRY; } static void clear_data_ready_stack(struct rar5* rar) { memset(&rar->cstate.dready, 0, sizeof(rar->cstate.dready)); } /* Pushes the `buf`, `size` and `offset` arguments to the rar->cstate.dready * FIFO stack. Those values will be popped from this stack by the `use_data` * function. */ static int push_data_ready(struct archive_read* a, struct rar5* rar, const uint8_t* buf, size_t size, int64_t offset) { int i; /* Don't push if we're in skip mode. This is needed because solid * streams need full processing even if we're skipping data. After * fully processing the stream, we need to discard the generated bytes, * because we're interested only in the side effect: building up the * internal window circular buffer. This window buffer will be used * later during unpacking of requested data. */ if(rar->skip_mode) return ARCHIVE_OK; /* Sanity check. */ if(offset != rar->file.last_offset + rar->file.last_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Sanity check error: output stream is not continuous"); return ARCHIVE_FATAL; } for(i = 0; i < rar5_countof(rar->cstate.dready); i++) { struct data_ready* d = &rar->cstate.dready[i]; if(!d->used) { d->used = 1; d->buf = buf; d->size = size; d->offset = offset; /* These fields are used only in sanity checking. */ rar->file.last_offset = offset; rar->file.last_size = size; /* Calculate the checksum of this new block before * submitting data to libarchive's engine. */ update_crc(rar, d->buf, d->size); return ARCHIVE_OK; } } /* Program counter will reach this code if the `rar->cstate.data_ready` * stack will be filled up so that no new entries will be allowed. The * code shouldn't allow such situation to occur. So we treat this case * as an internal error. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Error: premature end of data_ready stack"); return ARCHIVE_FATAL; } /* This function uncompresses the data that is stored in the base * block. * * The FILE base block looks like this: * *
... * * The
is a block header, that is parsed in parse_block_header(). * It's a "compressed_block_header" structure, containing metadata needed * to know when we should stop looking for more blocks. * * contain data needed to set up the huffman tables, needed * for the actual decompression. * * Each consists of series of literals: * * ... * * Those literals generate the uncompression data. They operate on a circular * buffer, sometimes writing raw data into it, sometimes referencing * some previous data inside this buffer, and sometimes declaring a filter * that will need to be executed on the data stored in the circular buffer. * It all depends on the literal that is used. * * Sometimes blocks produce output data, sometimes they don't. For example, for * some huge files that use lots of filters, sometimes a block is filled with * only filter declaration literals. Such blocks won't produce any data in the * circular buffer. * * Sometimes blocks will produce 4 bytes of data, and sometimes 1 megabyte, * because a literal can reference previously decompressed data. For example, * there can be a literal that says: 'append a byte 0xFE here', and after * it another literal can say 'append 1 megabyte of data from circular buffer * offset 0x12345'. This is how RAR format handles compressing repeated * patterns. * * The RAR compressor creates those literals and the actual efficiency of * compression depends on what those literals are. The literals can also * be seen as a kind of a non-turing-complete virtual machine that simply * tells the decompressor what it should do. * */ static int do_uncompress_file(struct archive_read* a) { struct rar5* rar = get_context(a); int ret; int64_t max_end_pos; if(!rar->cstate.initialized) { /* Don't perform full context reinitialization if we're * processing a solid archive. */ if(!rar->main.solid || !rar->cstate.window_buf) { init_unpack(rar); } rar->cstate.initialized = 1; } /* Don't allow extraction if window_size is invalid. */ if(rar->cstate.window_size == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid window size declaration in this file"); /* This should never happen in valid files. */ return ARCHIVE_FATAL; } if(rar->cstate.all_filters_applied == 1) { /* We use while(1) here, but standard case allows for just 1 * iteration. The loop will iterate if process_block() didn't * generate any data at all. This can happen if the block * contains only filter definitions (this is common in big * files). */ while(1) { ret = process_block(a); if(ret == ARCHIVE_EOF || ret == ARCHIVE_FATAL) return ret; if(rar->cstate.last_write_ptr == rar->cstate.write_ptr) { /* The block didn't generate any new data, * so just process a new block if this one * wasn't the last block in the file. */ if (bf_is_last_block(&rar->last_block_hdr)) { return ARCHIVE_EOF; } continue; } /* The block has generated some new data, so break * the loop. */ break; } } /* Try to run filters. If filters won't be applied, it means that * insufficient data was generated. */ ret = apply_filters(a); if(ret == ARCHIVE_RETRY) { return ARCHIVE_OK; } else if(ret == ARCHIVE_FATAL) { return ARCHIVE_FATAL; } /* If apply_filters() will return ARCHIVE_OK, we can continue here. */ if(cdeque_size(&rar->cstate.filters) > 0) { /* Check if we can write something before hitting first * filter. */ struct filter_info* flt; /* Get the block_start offset from the first filter. */ if(CDE_OK != cdeque_front(&rar->cstate.filters, cdeque_filter_p(&flt))) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Can't read first filter"); return ARCHIVE_FATAL; } max_end_pos = rar5_min(flt->block_start, rar->cstate.write_ptr); } else { /* There are no filters defined, or all filters were applied. * This means we can just store the data without any * postprocessing. */ max_end_pos = rar->cstate.write_ptr; } if(max_end_pos == rar->cstate.last_write_ptr) { /* We can't write anything yet. The block uncompression * function did not generate enough data, and no filter can be * applied. At the same time we don't have any data that can be * stored without filter postprocessing. This means we need to * wait for more data to be generated, so we can apply the * filters. * * Signal the caller that we need more data to be able to do * anything. */ return ARCHIVE_RETRY; } else { /* We can write the data before hitting the first filter. * So let's do it. The push_window_data() function will * effectively return the selected data block to the user * application. */ push_window_data(a, rar, rar->cstate.last_write_ptr, max_end_pos); rar->cstate.last_write_ptr = max_end_pos; } return ARCHIVE_OK; } static int uncompress_file(struct archive_read* a) { int ret; while(1) { /* Sometimes the uncompression function will return a * 'retry' signal. If this will happen, we have to retry * the function. */ ret = do_uncompress_file(a); if(ret != ARCHIVE_RETRY) return ret; } } static int do_unstore_file(struct archive_read* a, struct rar5* rar, const void** buf, size_t* size, int64_t* offset) { size_t to_read; const uint8_t* p; if(rar->file.bytes_remaining == 0 && rar->main.volume > 0 && rar->generic.split_after > 0) { int ret; rar->cstate.switch_multivolume = 1; ret = advance_multivolume(a); rar->cstate.switch_multivolume = 0; if(ret != ARCHIVE_OK) { /* Failed to advance to next multivolume archive * file. */ return ret; } } to_read = rar5_min(rar->file.bytes_remaining, 64 * 1024); if(to_read == 0) { return ARCHIVE_EOF; } if(!read_ahead(a, to_read, &p)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "I/O error when unstoring file"); return ARCHIVE_FATAL; } if(ARCHIVE_OK != consume(a, to_read)) { return ARCHIVE_EOF; } if(buf) *buf = p; if(size) *size = to_read; if(offset) *offset = rar->cstate.last_unstore_ptr; rar->file.bytes_remaining -= to_read; rar->cstate.last_unstore_ptr += to_read; update_crc(rar, p, to_read); return ARCHIVE_OK; } static int do_unpack(struct archive_read* a, struct rar5* rar, const void** buf, size_t* size, int64_t* offset) { enum COMPRESSION_METHOD { STORE = 0, FASTEST = 1, FAST = 2, NORMAL = 3, GOOD = 4, BEST = 5 }; if(rar->file.service > 0) { return do_unstore_file(a, rar, buf, size, offset); } else { switch(rar->cstate.method) { case STORE: return do_unstore_file(a, rar, buf, size, offset); case FASTEST: /* fallthrough */ case FAST: /* fallthrough */ case NORMAL: /* fallthrough */ case GOOD: /* fallthrough */ case BEST: /* No data is returned here. But because a sparse-file aware * caller (like archive_read_data_into_fd) may treat zero-size * as a sparse file block, we need to update the offset * accordingly. At this point the decoder doesn't have any * pending uncompressed data blocks, so the current position in * the output file should be last_write_ptr. */ if (offset) *offset = rar->cstate.last_write_ptr; return uncompress_file(a); default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Compression method not supported: 0x%x", (unsigned int)rar->cstate.method); return ARCHIVE_FATAL; } } #if !defined WIN32 /* Not reached. */ return ARCHIVE_OK; #endif } static int verify_checksums(struct archive_read* a) { int verify_crc; struct rar5* rar = get_context(a); /* Check checksums only when actually unpacking the data. There's no * need to calculate checksum when we're skipping data in solid archives * (skipping in solid archives is the same thing as unpacking compressed * data and discarding the result). */ if(!rar->skip_mode) { /* Always check checksums if we're not in skip mode */ verify_crc = 1; } else { /* We can override the logic above with a compile-time option * NO_CRC_ON_SOLID_SKIP. This option is used during debugging, * and it will check checksums of unpacked data even when * we're skipping it. */ #if defined CHECK_CRC_ON_SOLID_SKIP /* Debug case */ verify_crc = 1; #else /* Normal case */ verify_crc = 0; #endif } if(verify_crc) { /* During unpacking, on each unpacked block we're calling the * update_crc() function. Since we are here, the unpacking * process is already over and we can check if calculated * checksum (CRC32 or BLAKE2sp) is the same as what is stored * in the archive. */ if(rar->file.stored_crc32 > 0) { /* Check CRC32 only when the file contains a CRC32 * value for this file. */ if(rar->file.calculated_crc32 != rar->file.stored_crc32) { /* Checksums do not match; the unpacked file * is corrupted. */ DEBUG_CODE { printf("Checksum error: CRC32 " "(was: %08" PRIx32 ", expected: %08" PRIx32 ")\n", rar->file.calculated_crc32, rar->file.stored_crc32); } #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Checksum error: CRC32"); return ARCHIVE_FATAL; #endif } else { DEBUG_CODE { printf("Checksum OK: CRC32 " "(%08" PRIx32 "/%08" PRIx32 ")\n", rar->file.stored_crc32, rar->file.calculated_crc32); } } } if(rar->file.has_blake2 > 0) { /* BLAKE2sp is an optional checksum algorithm that is * added to RARv5 archives when using the `-htb` switch * during creation of archive. * * We now finalize the hash calculation by calling the * `final` function. This will generate the final hash * value we can use to compare it with the BLAKE2sp * checksum that is stored in the archive. * * The return value of this `final` function is not * very helpful, as it guards only against improper use. * This is why we're explicitly ignoring it. */ uint8_t b2_buf[32]; (void) blake2sp_final(&rar->file.b2state, b2_buf, 32); if(memcmp(&rar->file.blake2sp, b2_buf, 32) != 0) { #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Checksum error: BLAKE2"); return ARCHIVE_FATAL; #endif } } } /* Finalization for this file has been successfully completed. */ return ARCHIVE_OK; } static int verify_global_checksums(struct archive_read* a) { return verify_checksums(a); } /* * Decryption function for the magic signature pattern. Check the comment near * the `rar5_signature_xor` symbol to read the rationale behind this. */ static void rar5_signature(char *buf) { size_t i; for(i = 0; i < sizeof(rar5_signature_xor); i++) { buf[i] = rar5_signature_xor[i] ^ 0xA1; } } static int rar5_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { int ret; struct rar5* rar = get_context(a); if (size) *size = 0; if (rar->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { rar->has_encrypted_entries = 0; } if (rar->headers_are_encrypted || rar->cstate.data_encrypted) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Reading encrypted data is not currently supported"); return ARCHIVE_FATAL; } if(rar->file.dir > 0) { /* Don't process any data if this file entry was declared * as a directory. This is needed, because entries marked as * directory doesn't have any dictionary buffer allocated, so * it's impossible to perform any decompression. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't decompress an entry marked as a directory"); - return ARCHIVE_FAILED; + return ARCHIVE_FATAL; } if(!rar->skip_mode && (rar->cstate.last_write_ptr > rar->file.unpacked_size)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "Unpacker has written too many bytes"); return ARCHIVE_FATAL; } ret = use_data(rar, buff, size, offset); if(ret == ARCHIVE_OK) { return ret; } if(rar->file.eof == 1) { return ARCHIVE_EOF; } ret = do_unpack(a, rar, buff, size, offset); if(ret != ARCHIVE_OK) { return ret; } if(rar->file.bytes_remaining == 0 && rar->cstate.last_write_ptr == rar->file.unpacked_size) { /* If all bytes of current file were processed, run * finalization. * * Finalization will check checksum against proper values. If * some of the checksums will not match, we'll return an error * value in the last `archive_read_data` call to signal an error * to the user. */ rar->file.eof = 1; return verify_global_checksums(a); } return ARCHIVE_OK; } static int rar5_read_data_skip(struct archive_read *a) { struct rar5* rar = get_context(a); if(rar->main.solid && (rar->cstate.data_encrypted == 0)) { /* In solid archives, instead of skipping the data, we need to * extract it, and dispose the result. The side effect of this * operation will be setting up the initial window buffer state * needed to be able to extract the selected file. Note that * this is only possible when data withing this solid block is * not encrypted, in which case we'll skip and fail if the user * tries to read data. */ int ret; /* Make sure to process all blocks in the compressed stream. */ while(rar->file.bytes_remaining > 0) { /* Setting the "skip mode" will allow us to skip * checksum checks during data skipping. Checking the * checksum of skipped data isn't really necessary and * it's only slowing things down. * * This is incremented instead of setting to 1 because * this data skipping function can be called * recursively. */ rar->skip_mode++; /* We're disposing 1 block of data, so we use triple * NULLs in arguments. */ ret = rar5_read_data(a, NULL, NULL, NULL); /* Turn off "skip mode". */ rar->skip_mode--; if(ret < 0 || ret == ARCHIVE_EOF) { /* Propagate any potential error conditions * to the caller. */ return ret; } } } else { /* In standard archives, we can just jump over the compressed * stream. Each file in non-solid archives starts from an empty * window buffer. */ if(ARCHIVE_OK != consume(a, rar->file.bytes_remaining)) { return ARCHIVE_FATAL; } rar->file.bytes_remaining = 0; } return ARCHIVE_OK; } static int64_t rar5_seek_data(struct archive_read *a, int64_t offset, int whence) { (void) a; (void) offset; (void) whence; /* We're a streaming unpacker, and we don't support seeking. */ return ARCHIVE_FATAL; } static int rar5_cleanup(struct archive_read *a) { struct rar5* rar = get_context(a); free(rar->cstate.window_buf); free(rar->cstate.filtered_buf); clear_data_ready_stack(rar); free(rar->vol.push_buf); free_filters(rar); cdeque_free(&rar->cstate.filters); free(rar); a->format->data = NULL; return ARCHIVE_OK; } static int rar5_capabilities(struct archive_read * a) { (void) a; return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); } static int rar5_has_encrypted_entries(struct archive_read *_a) { if (_a && _a->format) { struct rar5 *rar = (struct rar5 *)_a->format->data; if (rar) { return rar->has_encrypted_entries; } } return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; } static int rar5_init(struct rar5* rar) { memset(rar, 0, sizeof(struct rar5)); if(CDE_OK != cdeque_init(&rar->cstate.filters, 8192)) return ARCHIVE_FATAL; /* * Until enough data has been read, we cannot tell about * any encrypted entries yet. */ rar->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; return ARCHIVE_OK; } int archive_read_support_format_rar5(struct archive *_a) { struct archive_read* ar; int ret; struct rar5* rar; if(ARCHIVE_OK != (ret = get_archive_read(_a, &ar))) return ret; rar = malloc(sizeof(*rar)); if(rar == NULL) { archive_set_error(&ar->archive, ENOMEM, "Can't allocate rar5 data"); return ARCHIVE_FATAL; } if(ARCHIVE_OK != rar5_init(rar)) { archive_set_error(&ar->archive, ENOMEM, "Can't allocate rar5 filter buffer"); free(rar); return ARCHIVE_FATAL; } ret = __archive_read_register_format(ar, rar, "rar5", rar5_bid, rar5_options, rar5_read_header, rar5_read_data, rar5_read_data_skip, rar5_seek_data, rar5_cleanup, rar5_capabilities, rar5_has_encrypted_entries); if(ret != ARCHIVE_OK) { (void) rar5_cleanup(ar); } return ret; } diff --git a/contrib/libarchive/libarchive/archive_read_support_format_tar.c b/contrib/libarchive/libarchive/archive_read_support_format_tar.c index 0c87bc6d732f..eeb2c725f6eb 100644 --- a/contrib/libarchive/libarchive/archive_read_support_format_tar.c +++ b/contrib/libarchive/libarchive/archive_read_support_format_tar.c @@ -1,3691 +1,3721 @@ /*- * Copyright (c) 2003-2023 Tim Kientzle * Copyright (c) 2011-2012 Michihiro NAKAJIMA * Copyright (c) 2016 Martin Matuska * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_acl_private.h" /* For ACL parsing routines. */ #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_read_private.h" #define tar_min(a,b) ((a) < (b) ? (a) : (b)) /* * Layout of POSIX 'ustar' tar header. */ struct archive_entry_header_ustar { char name[100]; char mode[8]; char uid[8]; char gid[8]; char size[12]; char mtime[12]; char checksum[8]; char typeflag[1]; char linkname[100]; /* "old format" header ends here */ char magic[6]; /* For POSIX: "ustar\0" */ char version[2]; /* For POSIX: "00" */ char uname[32]; char gname[32]; char rdevmajor[8]; char rdevminor[8]; char prefix[155]; }; /* * Structure of GNU tar header */ struct gnu_sparse { char offset[12]; char numbytes[12]; }; struct archive_entry_header_gnutar { char name[100]; char mode[8]; char uid[8]; char gid[8]; char size[12]; char mtime[12]; char checksum[8]; char typeflag[1]; char linkname[100]; char magic[8]; /* "ustar \0" (note blank/blank/null at end) */ char uname[32]; char gname[32]; char rdevmajor[8]; char rdevminor[8]; char atime[12]; char ctime[12]; char offset[12]; char longnames[4]; char unused[1]; struct gnu_sparse sparse[4]; char isextended[1]; char realsize[12]; /* * Old GNU format doesn't use POSIX 'prefix' field; they use * the 'L' (longname) entry instead. */ }; /* * Data specific to this format. */ struct sparse_block { struct sparse_block *next; int64_t offset; int64_t remaining; int hole; }; struct tar { struct archive_string entry_pathname; /* For "GNU.sparse.name" and other similar path extensions. */ struct archive_string entry_pathname_override; struct archive_string entry_uname; struct archive_string entry_gname; struct archive_string entry_linkpath; struct archive_string line; int pax_hdrcharset_utf8; int64_t entry_bytes_remaining; int64_t entry_offset; int64_t entry_padding; int64_t entry_bytes_unconsumed; int64_t disk_size; int64_t GNU_sparse_realsize; int64_t GNU_sparse_size; int64_t SCHILY_sparse_realsize; int64_t pax_size; struct sparse_block *sparse_list; struct sparse_block *sparse_last; int64_t sparse_offset; int64_t sparse_numbytes; int sparse_gnu_major; int sparse_gnu_minor; char sparse_gnu_attributes_seen; char filetype; char size_fields; /* Bits defined below */ struct archive_string localname; struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv; struct archive_string_conv *sconv_acl; struct archive_string_conv *sconv_default; int init_default_conversion; int compat_2x; int process_mac_extensions; int read_concatenated_archives; }; /* Track which size fields were present in the headers */ #define TAR_SIZE_PAX_SIZE 1 #define TAR_SIZE_GNU_SPARSE_REALSIZE 2 #define TAR_SIZE_GNU_SPARSE_SIZE 4 #define TAR_SIZE_SCHILY_SPARSE_REALSIZE 8 static int archive_block_is_null(const char *p); static char *base64_decode(const char *, size_t, size_t *); static int gnu_add_sparse_entry(struct archive_read *, struct tar *, int64_t offset, int64_t remaining); static void gnu_clear_sparse_list(struct tar *); static int gnu_sparse_old_read(struct archive_read *, struct tar *, const struct archive_entry_header_gnutar *header, int64_t *); static int gnu_sparse_old_parse(struct archive_read *, struct tar *, const struct gnu_sparse *sparse, int length); static int gnu_sparse_01_parse(struct archive_read *, struct tar *, const char *, size_t); static int64_t gnu_sparse_10_read(struct archive_read *, struct tar *, int64_t *); static int header_Solaris_ACL(struct archive_read *, struct tar *, struct archive_entry *, const void *, int64_t *); static int header_common(struct archive_read *, struct tar *, struct archive_entry *, const void *); static int header_old_tar(struct archive_read *, struct tar *, struct archive_entry *, const void *); static int header_pax_extension(struct archive_read *, struct tar *, struct archive_entry *, const void *, int64_t *); static int header_pax_global(struct archive_read *, struct tar *, struct archive_entry *, const void *h, int64_t *); static int header_gnu_longlink(struct archive_read *, struct tar *, struct archive_entry *, const void *h, int64_t *); static int header_gnu_longname(struct archive_read *, struct tar *, struct archive_entry *, const void *h, int64_t *); static int is_mac_metadata_entry(struct archive_entry *entry); static int read_mac_metadata_blob(struct archive_read *, struct archive_entry *, int64_t *); static int header_volume(struct archive_read *, struct tar *, struct archive_entry *, const void *h, int64_t *); static int header_ustar(struct archive_read *, struct tar *, struct archive_entry *, const void *h); static int header_gnutar(struct archive_read *, struct tar *, struct archive_entry *, const void *h, int64_t *); static int archive_read_format_tar_bid(struct archive_read *, int); static int archive_read_format_tar_options(struct archive_read *, const char *, const char *); static int archive_read_format_tar_cleanup(struct archive_read *); static int archive_read_format_tar_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset); static int archive_read_format_tar_skip(struct archive_read *a); static int archive_read_format_tar_read_header(struct archive_read *, struct archive_entry *); static int checksum(struct archive_read *, const void *); static int pax_attribute(struct archive_read *, struct tar *, struct archive_entry *, const char *key, size_t key_length, size_t value_length, int64_t *unconsumed); static int pax_attribute_LIBARCHIVE_xattr(struct archive_entry *, const char *, size_t, const char *, size_t); static int pax_attribute_SCHILY_acl(struct archive_read *, struct tar *, struct archive_entry *, size_t, int); static int pax_attribute_SUN_holesdata(struct archive_read *, struct tar *, struct archive_entry *, const char *, size_t); static void pax_time(const char *, size_t, int64_t *sec, long *nanos); static ssize_t readline(struct archive_read *, struct tar *, const char **, ssize_t limit, int64_t *); static int read_body_to_string(struct archive_read *, struct tar *, struct archive_string *, const void *h, int64_t *); static int read_bytes_to_string(struct archive_read *, struct archive_string *, size_t, int64_t *); static int64_t tar_atol(const char *, size_t); static int64_t tar_atol10(const char *, size_t); static int64_t tar_atol256(const char *, size_t); static int64_t tar_atol8(const char *, size_t); static int tar_read_header(struct archive_read *, struct tar *, struct archive_entry *, int64_t *); static int tohex(int c); static char *url_decode(const char *, size_t); -static void tar_flush_unconsumed(struct archive_read *, int64_t *); +static int tar_flush_unconsumed(struct archive_read *, int64_t *); /* Sanity limits: These numbers should be low enough to * prevent a maliciously-crafted archive from forcing us to * allocate extreme amounts of memory. But of course, they * need to be high enough for any correct value. These * will likely need some adjustment as we get more experience. */ static const size_t guname_limit = 65536; /* Longest uname or gname: 64kiB */ static const size_t pathname_limit = 1048576; /* Longest path name: 1MiB */ static const size_t sparse_map_limit = 8 * 1048576; /* Longest sparse map: 8MiB */ static const size_t xattr_limit = 16 * 1048576; /* Longest xattr: 16MiB */ static const size_t fflags_limit = 512; /* Longest fflags */ static const size_t acl_limit = 131072; /* Longest textual ACL: 128kiB */ static const int64_t entry_limit = 0xfffffffffffffffLL; /* 2^60 bytes = 1 ExbiByte */ int archive_read_support_format_gnutar(struct archive *a) { archive_check_magic(a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_gnutar"); return (archive_read_support_format_tar(a)); } int archive_read_support_format_tar(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct tar *tar; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_tar"); tar = calloc(1, sizeof(*tar)); if (tar == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate tar data"); return (ARCHIVE_FATAL); } #ifdef HAVE_COPYFILE_H /* Set this by default on Mac OS. */ tar->process_mac_extensions = 1; #endif r = __archive_read_register_format(a, tar, "tar", archive_read_format_tar_bid, archive_read_format_tar_options, archive_read_format_tar_read_header, archive_read_format_tar_read_data, archive_read_format_tar_skip, NULL, archive_read_format_tar_cleanup, NULL, NULL); if (r != ARCHIVE_OK) free(tar); return (ARCHIVE_OK); } static int archive_read_format_tar_cleanup(struct archive_read *a) { struct tar *tar; tar = (struct tar *)(a->format->data); gnu_clear_sparse_list(tar); archive_string_free(&tar->entry_pathname); archive_string_free(&tar->entry_pathname_override); archive_string_free(&tar->entry_uname); archive_string_free(&tar->entry_gname); archive_string_free(&tar->entry_linkpath); archive_string_free(&tar->line); archive_string_free(&tar->localname); free(tar); (a->format->data) = NULL; return (ARCHIVE_OK); } /* * Validate number field * * This has to be pretty lenient in order to accommodate the enormous * variety of tar writers in the world: * = POSIX (IEEE Std 1003.1-1988) ustar requires octal values with leading * zeros and allows fields to be terminated with space or null characters * = Many writers use different termination (in particular, libarchive * omits terminator bytes to squeeze one or two more digits) * = Many writers pad with space and omit leading zeros * = GNU tar and star write base-256 values if numbers are too * big to be represented in octal * * Examples of specific tar headers that we should support: * = Perl Archive::Tar terminates uid, gid, devminor and devmajor with two * null bytes, pads size with spaces and other numeric fields with zeroes * = plexus-archiver prior to 2.6.3 (before switching to commons-compress) * may have uid and gid fields filled with spaces without any octal digits * at all and pads all numeric fields with spaces * * This should tolerate all variants in use. It will reject a field * where the writer just left garbage after a trailing NUL. */ static int validate_number_field(const char* p_field, size_t i_size) { unsigned char marker = (unsigned char)p_field[0]; if (marker == 128 || marker == 255 || marker == 0) { /* Base-256 marker, there's nothing we can check. */ return 1; } else { /* Must be octal */ size_t i = 0; /* Skip any leading spaces */ while (i < i_size && p_field[i] == ' ') { ++i; } /* Skip octal digits. */ while (i < i_size && p_field[i] >= '0' && p_field[i] <= '7') { ++i; } /* Any remaining characters must be space or NUL padding. */ while (i < i_size) { if (p_field[i] != ' ' && p_field[i] != 0) { return 0; } ++i; } return 1; } } static int archive_read_format_tar_bid(struct archive_read *a, int best_bid) { int bid; const char *h; const struct archive_entry_header_ustar *header; (void)best_bid; /* UNUSED */ bid = 0; /* Now let's look at the actual header and see if it matches. */ h = __archive_read_ahead(a, 512, NULL); if (h == NULL) return (-1); /* If it's an end-of-archive mark, we can handle it. */ if (h[0] == 0 && archive_block_is_null(h)) { /* * Usually, I bid the number of bits verified, but * in this case, 4096 seems excessive so I picked 10 as * an arbitrary but reasonable-seeming value. */ return (10); } /* If it's not an end-of-archive mark, it must have a valid checksum.*/ if (!checksum(a, h)) return (0); bid += 48; /* Checksum is usually 6 octal digits. */ header = (const struct archive_entry_header_ustar *)h; /* Recognize POSIX formats. */ if ((memcmp(header->magic, "ustar\0", 6) == 0) && (memcmp(header->version, "00", 2) == 0)) bid += 56; /* Recognize GNU tar format. */ if ((memcmp(header->magic, "ustar ", 6) == 0) && (memcmp(header->version, " \0", 2) == 0)) bid += 56; /* Type flag must be null, digit or A-Z, a-z. */ if (header->typeflag[0] != 0 && !( header->typeflag[0] >= '0' && header->typeflag[0] <= '9') && !( header->typeflag[0] >= 'A' && header->typeflag[0] <= 'Z') && !( header->typeflag[0] >= 'a' && header->typeflag[0] <= 'z') ) return (0); bid += 2; /* 6 bits of variation in an 8-bit field leaves 2 bits. */ /* * Check format of mode/uid/gid/mtime/size/rdevmajor/rdevminor fields. */ if (validate_number_field(header->mode, sizeof(header->mode)) == 0 || validate_number_field(header->uid, sizeof(header->uid)) == 0 || validate_number_field(header->gid, sizeof(header->gid)) == 0 || validate_number_field(header->mtime, sizeof(header->mtime)) == 0 || validate_number_field(header->size, sizeof(header->size)) == 0 || validate_number_field(header->rdevmajor, sizeof(header->rdevmajor)) == 0 || validate_number_field(header->rdevminor, sizeof(header->rdevminor)) == 0) { bid = 0; } return (bid); } static int archive_read_format_tar_options(struct archive_read *a, const char *key, const char *val) { struct tar *tar; int ret = ARCHIVE_FAILED; tar = (struct tar *)(a->format->data); if (strcmp(key, "compat-2x") == 0) { /* Handle UTF-8 filenames as libarchive 2.x */ tar->compat_2x = (val != NULL && val[0] != 0); tar->init_default_conversion = tar->compat_2x; return (ARCHIVE_OK); } else if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "tar: hdrcharset option needs a character-set name"); else { tar->opt_sconv = archive_string_conversion_from_charset( &a->archive, val, 0); if (tar->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } else if (strcmp(key, "mac-ext") == 0) { tar->process_mac_extensions = (val != NULL && val[0] != 0); return (ARCHIVE_OK); } else if (strcmp(key, "read_concatenated_archives") == 0) { tar->read_concatenated_archives = (val != NULL && val[0] != 0); return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } /* utility function- this exists to centralize the logic of tracking * how much unconsumed data we have floating around, and to consume * anything outstanding since we're going to do read_aheads */ -static void +static int tar_flush_unconsumed(struct archive_read *a, int64_t *unconsumed) { if (*unconsumed) { /* void *data = (void *)__archive_read_ahead(a, *unconsumed, NULL); * this block of code is to poison claimed unconsumed space, ensuring * things break if it is in use still. * currently it WILL break things, so enable it only for debugging this issue if (data) { memset(data, 0xff, *unconsumed); } */ - __archive_read_consume(a, *unconsumed); + int64_t consumed = __archive_read_consume(a, *unconsumed); + if (consumed != *unconsumed) { + return (ARCHIVE_FATAL); + } *unconsumed = 0; } + return (ARCHIVE_OK); } /* * The function invoked by archive_read_next_header(). This * just sets up a few things and then calls the internal * tar_read_header() function below. */ static int archive_read_format_tar_read_header(struct archive_read *a, struct archive_entry *entry) { /* * When converting tar archives to cpio archives, it is * essential that each distinct file have a distinct inode * number. To simplify this, we keep a static count here to * assign fake dev/inode numbers to each tar entry. Note that * pax format archives may overwrite this with something more * useful. * * Ideally, we would track every file read from the archive so * that we could assign the same dev/ino pair to hardlinks, * but the memory required to store a complete lookup table is * probably not worthwhile just to support the relatively * obscure tar->cpio conversion case. */ /* TODO: Move this into `struct tar` to avoid conflicts * when reading multiple archives */ static int default_inode; static int default_dev; struct tar *tar; const char *p; const wchar_t *wp; int r; size_t l; int64_t unconsumed = 0; /* Assign default device/inode values. */ archive_entry_set_dev(entry, 1 + default_dev); /* Don't use zero. */ archive_entry_set_ino(entry, ++default_inode); /* Don't use zero. */ /* Limit generated st_ino number to 16 bits. */ if (default_inode >= 0xffff) { ++default_dev; default_inode = 0; } tar = (struct tar *)(a->format->data); tar->entry_offset = 0; gnu_clear_sparse_list(tar); tar->size_fields = 0; /* We don't have any size info yet */ /* Setup default string conversion. */ tar->sconv = tar->opt_sconv; if (tar->sconv == NULL) { if (!tar->init_default_conversion) { tar->sconv_default = archive_string_default_conversion_for_read(&(a->archive)); tar->init_default_conversion = 1; } tar->sconv = tar->sconv_default; } r = tar_read_header(a, tar, entry, &unconsumed); tar_flush_unconsumed(a, &unconsumed); /* * "non-sparse" files are really just sparse files with * a single block. */ if (tar->sparse_list == NULL) { if (gnu_add_sparse_entry(a, tar, 0, tar->entry_bytes_remaining) != ARCHIVE_OK) return (ARCHIVE_FATAL); } else { struct sparse_block *sb; for (sb = tar->sparse_list; sb != NULL; sb = sb->next) { if (!sb->hole) archive_entry_sparse_add_entry(entry, sb->offset, sb->remaining); } } if (r == ARCHIVE_OK && archive_entry_filetype(entry) == AE_IFREG) { /* * "Regular" entry with trailing '/' is really * directory: This is needed for certain old tar * variants and even for some broken newer ones. */ if ((wp = archive_entry_pathname_w(entry)) != NULL) { l = wcslen(wp); if (l > 0 && wp[l - 1] == L'/') { archive_entry_set_filetype(entry, AE_IFDIR); tar->entry_bytes_remaining = 0; tar->entry_padding = 0; } } else if ((p = archive_entry_pathname(entry)) != NULL) { l = strlen(p); if (l > 0 && p[l - 1] == '/') { archive_entry_set_filetype(entry, AE_IFDIR); tar->entry_bytes_remaining = 0; tar->entry_padding = 0; } } } return (r); } static int archive_read_format_tar_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { ssize_t bytes_read; struct tar *tar; struct sparse_block *p; tar = (struct tar *)(a->format->data); for (;;) { /* Remove exhausted entries from sparse list. */ while (tar->sparse_list != NULL && tar->sparse_list->remaining == 0) { p = tar->sparse_list; tar->sparse_list = p->next; free(p); } if (tar->entry_bytes_unconsumed) { __archive_read_consume(a, tar->entry_bytes_unconsumed); tar->entry_bytes_unconsumed = 0; } /* If we're at end of file, return EOF. */ if (tar->sparse_list == NULL || tar->entry_bytes_remaining == 0) { int64_t request = tar->entry_bytes_remaining + tar->entry_padding; if (__archive_read_consume(a, request) != request) return (ARCHIVE_FATAL); tar->entry_padding = 0; *buff = NULL; *size = 0; *offset = tar->disk_size; return (ARCHIVE_EOF); } *buff = __archive_read_ahead(a, 1, &bytes_read); if (*buff == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Truncated tar archive" " detected while reading data"); return (ARCHIVE_FATAL); } if (bytes_read > tar->entry_bytes_remaining) bytes_read = (ssize_t)tar->entry_bytes_remaining; /* Don't read more than is available in the * current sparse block. */ if (tar->sparse_list->remaining < bytes_read) bytes_read = (ssize_t)tar->sparse_list->remaining; *size = bytes_read; *offset = tar->sparse_list->offset; tar->sparse_list->remaining -= bytes_read; tar->sparse_list->offset += bytes_read; tar->entry_bytes_remaining -= bytes_read; tar->entry_bytes_unconsumed = bytes_read; if (!tar->sparse_list->hole) return (ARCHIVE_OK); /* Current is hole data and skip this. */ } } static int archive_read_format_tar_skip(struct archive_read *a) { int64_t request; struct tar* tar; tar = (struct tar *)(a->format->data); request = tar->entry_bytes_remaining + tar->entry_padding + tar->entry_bytes_unconsumed; if (__archive_read_consume(a, request) != request) return (ARCHIVE_FATAL); tar->entry_bytes_remaining = 0; tar->entry_bytes_unconsumed = 0; tar->entry_padding = 0; /* Free the sparse list. */ gnu_clear_sparse_list(tar); return (ARCHIVE_OK); } /* * This function resets the accumulated state while reading * a header. */ static void tar_reset_header_state(struct tar *tar) { tar->pax_hdrcharset_utf8 = 1; tar->sparse_gnu_attributes_seen = 0; archive_string_empty(&(tar->entry_gname)); archive_string_empty(&(tar->entry_pathname)); archive_string_empty(&(tar->entry_pathname_override)); archive_string_empty(&(tar->entry_uname)); archive_string_empty(&tar->entry_linkpath); } /* * This function reads and interprets all of the headers associated * with a single entry. */ static int tar_read_header(struct archive_read *a, struct tar *tar, struct archive_entry *entry, int64_t *unconsumed) { ssize_t bytes; int err = ARCHIVE_OK, err2; int eof_fatal = 0; /* EOF is okay at some points... */ const char *h; const struct archive_entry_header_ustar *header; const struct archive_entry_header_gnutar *gnuheader; /* Bitmask of what header types we've seen. */ int32_t seen_headers = 0; static const int32_t seen_A_header = 1; static const int32_t seen_g_header = 2; static const int32_t seen_K_header = 4; static const int32_t seen_L_header = 8; static const int32_t seen_V_header = 16; static const int32_t seen_x_header = 32; /* Also X */ static const int32_t seen_mac_metadata = 512; tar_reset_header_state(tar); /* Ensure format is set. */ if (a->archive.archive_format_name == NULL) { a->archive.archive_format = ARCHIVE_FORMAT_TAR; a->archive.archive_format_name = "tar"; } /* * TODO: Write global/default pax options into * 'entry' struct here before overwriting with * file-specific options. */ /* Loop over all the headers needed for the next entry */ for (;;) { /* Find the next valid header record. */ while (1) { - tar_flush_unconsumed(a, unconsumed); + if (tar_flush_unconsumed(a, unconsumed) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } /* Read 512-byte header record */ h = __archive_read_ahead(a, 512, &bytes); if (bytes == 0) { /* EOF at a block boundary. */ if (eof_fatal) { /* We've read a special header already; * if there's no regular header, then this is * a premature EOF. */ archive_set_error(&a->archive, EINVAL, "Damaged tar archive (end-of-archive within a sequence of headers)"); return (ARCHIVE_FATAL); } else { return (ARCHIVE_EOF); } } if (h == NULL) { /* Short block at EOF; this is bad. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated tar archive" " detected while reading next header"); return (ARCHIVE_FATAL); } *unconsumed += 512; if (h[0] == 0 && archive_block_is_null(h)) { /* We found a NULL block which indicates end-of-archive */ if (tar->read_concatenated_archives) { /* We're ignoring NULL blocks, so keep going. */ continue; } /* Try to consume a second all-null record, as well. */ /* If we can't, that's okay. */ tar_flush_unconsumed(a, unconsumed); h = __archive_read_ahead(a, 512, NULL); if (h != NULL && h[0] == 0 && archive_block_is_null(h)) __archive_read_consume(a, 512); archive_clear_error(&a->archive); return (ARCHIVE_EOF); } /* This is NOT a null block, so it must be a valid header. */ if (!checksum(a, h)) { - tar_flush_unconsumed(a, unconsumed); + if (tar_flush_unconsumed(a, unconsumed) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } archive_set_error(&a->archive, EINVAL, "Damaged tar archive (bad header checksum)"); /* If we've read some critical information (pax headers, etc) * and _then_ see a bad header, we can't really recover. */ if (eof_fatal) { return (ARCHIVE_FATAL); } else { return (ARCHIVE_RETRY); } } break; } /* Determine the format variant. */ header = (const struct archive_entry_header_ustar *)h; switch(header->typeflag[0]) { case 'A': /* Solaris tar ACL */ if (seen_headers & seen_A_header) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Redundant 'A' header"); return (ARCHIVE_FATAL); } seen_headers |= seen_A_header; a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "Solaris tar"; err2 = header_Solaris_ACL(a, tar, entry, h, unconsumed); break; case 'g': /* POSIX-standard 'g' header. */ if (seen_headers & seen_g_header) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Redundant 'g' header"); return (ARCHIVE_FATAL); } seen_headers |= seen_g_header; a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "POSIX pax interchange format"; err2 = header_pax_global(a, tar, entry, h, unconsumed); break; case 'K': /* Long link name (GNU tar, others) */ if (seen_headers & seen_K_header) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Damaged archive: Redundant 'K' headers may cause linknames to be incorrect"); err = err_combine(err, ARCHIVE_WARN); } seen_headers |= seen_K_header; a->archive.archive_format = ARCHIVE_FORMAT_TAR_GNUTAR; a->archive.archive_format_name = "GNU tar format"; err2 = header_gnu_longlink(a, tar, entry, h, unconsumed); break; case 'L': /* Long filename (GNU tar, others) */ if (seen_headers & seen_L_header) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Damaged archive: Redundant 'L' headers may cause filenames to be incorrect"); err = err_combine(err, ARCHIVE_WARN); } seen_headers |= seen_L_header; a->archive.archive_format = ARCHIVE_FORMAT_TAR_GNUTAR; a->archive.archive_format_name = "GNU tar format"; err2 = header_gnu_longname(a, tar, entry, h, unconsumed); break; case 'V': /* GNU volume header */ if (seen_headers & seen_V_header) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Redundant 'V' header"); err = err_combine(err, ARCHIVE_WARN); } seen_headers |= seen_V_header; a->archive.archive_format = ARCHIVE_FORMAT_TAR_GNUTAR; a->archive.archive_format_name = "GNU tar format"; err2 = header_volume(a, tar, entry, h, unconsumed); break; case 'X': /* Used by SUN tar; same as 'x'. */ if (seen_headers & seen_x_header) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Redundant 'X'/'x' header"); return (ARCHIVE_FATAL); } seen_headers |= seen_x_header; a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "POSIX pax interchange format (Sun variant)"; err2 = header_pax_extension(a, tar, entry, h, unconsumed); break; case 'x': /* POSIX-standard 'x' header. */ if (seen_headers & seen_x_header) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Redundant 'x' header"); return (ARCHIVE_FATAL); } seen_headers |= seen_x_header; a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "POSIX pax interchange format"; err2 = header_pax_extension(a, tar, entry, h, unconsumed); break; default: /* Regular header: Legacy tar, GNU tar, or ustar */ gnuheader = (const struct archive_entry_header_gnutar *)h; if (memcmp(gnuheader->magic, "ustar \0", 8) == 0) { a->archive.archive_format = ARCHIVE_FORMAT_TAR_GNUTAR; a->archive.archive_format_name = "GNU tar format"; err2 = header_gnutar(a, tar, entry, h, unconsumed); } else if (memcmp(header->magic, "ustar", 5) == 0) { if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE) { a->archive.archive_format = ARCHIVE_FORMAT_TAR_USTAR; a->archive.archive_format_name = "POSIX ustar format"; } err2 = header_ustar(a, tar, entry, h); } else { a->archive.archive_format = ARCHIVE_FORMAT_TAR; a->archive.archive_format_name = "tar (non-POSIX)"; err2 = header_old_tar(a, tar, entry, h); } err = err_combine(err, err2); /* We return warnings or success as-is. Anything else is fatal. */ if (err < ARCHIVE_WARN) { return (ARCHIVE_FATAL); } /* Filename of the form `._filename` is an AppleDouble * extension entry. The body is the macOS metadata blob; * this is followed by another entry with the actual * regular file data. * This design has two drawbacks: * = it's brittle; you might just have a file with such a name * = it duplicates any long pathname extensions * * TODO: This probably shouldn't be here at all. Consider * just returning the contents as a regular entry here and * then dealing with it when we write data to disk. */ if (tar->process_mac_extensions && ((seen_headers & seen_mac_metadata) == 0) && is_mac_metadata_entry(entry)) { err2 = read_mac_metadata_blob(a, entry, unconsumed); if (err2 < ARCHIVE_WARN) { return (ARCHIVE_FATAL); } err = err_combine(err, err2); /* Note: Other headers can appear again. */ seen_headers = seen_mac_metadata; tar_reset_header_state(tar); break; } /* Reconcile GNU sparse attributes */ if (tar->sparse_gnu_attributes_seen) { /* Only 'S' (GNU sparse) and ustar '0' regular files can be sparse */ if (tar->filetype != 'S' && tar->filetype != '0') { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Non-regular file cannot be sparse"); return (ARCHIVE_WARN); } else if (tar->sparse_gnu_major == 0 && tar->sparse_gnu_minor == 0) { /* Sparse map already parsed from 'x' header */ } else if (tar->sparse_gnu_major == 0 && tar->sparse_gnu_minor == 1) { /* Sparse map already parsed from 'x' header */ } else if (tar->sparse_gnu_major == 1 && tar->sparse_gnu_minor == 0) { /* Sparse map is prepended to file contents */ ssize_t bytes_read; bytes_read = gnu_sparse_10_read(a, tar, unconsumed); if (bytes_read < 0) return ((int)bytes_read); tar->entry_bytes_remaining -= bytes_read; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unrecognized GNU sparse file format"); return (ARCHIVE_WARN); } } return (err); } /* We're between headers ... */ err = err_combine(err, err2); if (err == ARCHIVE_FATAL) return (err); /* The GNU volume header and the pax `g` global header * are both allowed to be the only header in an * archive. If we've seen any other header, a * following EOF is fatal. */ if ((seen_headers & ~seen_V_header & ~seen_g_header) != 0) { eof_fatal = 1; } } } /* * Return true if block checksum is correct. */ static int checksum(struct archive_read *a, const void *h) { const unsigned char *bytes; const struct archive_entry_header_ustar *header; int check, sum; size_t i; (void)a; /* UNUSED */ bytes = (const unsigned char *)h; header = (const struct archive_entry_header_ustar *)h; /* Checksum field must hold an octal number */ for (i = 0; i < sizeof(header->checksum); ++i) { char c = header->checksum[i]; if (c != ' ' && c != '\0' && (c < '0' || c > '7')) return 0; } /* * Test the checksum. Note that POSIX specifies _unsigned_ * bytes for this calculation. */ sum = (int)tar_atol(header->checksum, sizeof(header->checksum)); check = 0; for (i = 0; i < 148; i++) check += (unsigned char)bytes[i]; for (; i < 156; i++) check += 32; for (; i < 512; i++) check += (unsigned char)bytes[i]; if (sum == check) return (1); /* * Repeat test with _signed_ bytes, just in case this archive * was created by an old BSD, Solaris, or HP-UX tar with a * broken checksum calculation. */ check = 0; for (i = 0; i < 148; i++) check += (signed char)bytes[i]; for (; i < 156; i++) check += 32; for (; i < 512; i++) check += (signed char)bytes[i]; if (sum == check) return (1); #if DONT_FAIL_ON_CRC_ERROR /* Speed up fuzzing by pretending the checksum is always right. */ return (1); #else return (0); #endif } /* * Return true if this block contains only nulls. */ static int archive_block_is_null(const char *p) { unsigned i; for (i = 0; i < 512; i++) if (*p++) return (0); return (1); } /* * Interpret 'A' Solaris ACL header */ static int header_Solaris_ACL(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, int64_t *unconsumed) { const struct archive_entry_header_ustar *header; struct archive_string acl_text; size_t size; int err, acl_type; uint64_t type; char *acl, *p; header = (const struct archive_entry_header_ustar *)h; size = (size_t)tar_atol(header->size, sizeof(header->size)); archive_string_init(&acl_text); err = read_body_to_string(a, tar, &acl_text, h, unconsumed); if (err != ARCHIVE_OK) { archive_string_free(&acl_text); return (err); } /* TODO: Examine the first characters to see if this * is an AIX ACL descriptor. We'll likely never support * them, but it would be polite to recognize and warn when * we do see them. */ /* Leading octal number indicates ACL type and number of entries. */ p = acl = acl_text.s; type = 0; while (*p != '\0' && p < acl + size) { if (*p < '0' || *p > '7') { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (invalid digit)"); archive_string_free(&acl_text); return(ARCHIVE_WARN); } type <<= 3; type += *p - '0'; if (type > 077777777) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (count too large)"); archive_string_free(&acl_text); return (ARCHIVE_WARN); } p++; } switch (type & ~0777777) { case 01000000: /* POSIX.1e ACL */ acl_type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; break; case 03000000: /* NFSv4 ACL */ acl_type = ARCHIVE_ENTRY_ACL_TYPE_NFS4; break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (unsupported type %llu)", (unsigned long long)type); archive_string_free(&acl_text); return (ARCHIVE_WARN); } p++; if (p >= acl + size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (body overflow)"); archive_string_free(&acl_text); return(ARCHIVE_WARN); } /* ACL text is null-terminated; find the end. */ size -= (p - acl); acl = p; while (*p != '\0' && p < acl + size) p++; if (tar->sconv_acl == NULL) { tar->sconv_acl = archive_string_conversion_from_charset( &(a->archive), "UTF-8", 1); if (tar->sconv_acl == NULL) { archive_string_free(&acl_text); return (ARCHIVE_FATAL); } } archive_strncpy(&(tar->localname), acl, p - acl); err = archive_acl_from_text_l(archive_entry_acl(entry), tar->localname.s, acl_type, tar->sconv_acl); /* Workaround: Force perm_is_set() to be correct */ /* If this bit were stored in the ACL, this wouldn't be needed */ archive_entry_set_perm(entry, archive_entry_perm(entry)); if (err != ARCHIVE_OK) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for ACL"); } else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (unparsable)"); } archive_string_free(&acl_text); return (err); } /* * Interpret 'K' long linkname header. */ static int header_gnu_longlink(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, int64_t *unconsumed) { int err; struct archive_string linkpath; archive_string_init(&linkpath); err = read_body_to_string(a, tar, &linkpath, h, unconsumed); if (err == ARCHIVE_OK) { archive_entry_set_link(entry, linkpath.s); } archive_string_free(&linkpath); return (err); } static int set_conversion_failed_error(struct archive_read *a, struct archive_string_conv *sconv, const char *name) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for %s", name); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "%s can't be converted from %s to current locale.", name, archive_string_conversion_charset_name(sconv)); return (ARCHIVE_WARN); } /* * Interpret 'L' long filename header. */ static int header_gnu_longname(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, int64_t *unconsumed) { int err; struct archive_string longname; archive_string_init(&longname); err = read_body_to_string(a, tar, &longname, h, unconsumed); if (err == ARCHIVE_OK) { if (archive_entry_copy_pathname_l(entry, longname.s, archive_strlen(&longname), tar->sconv) != 0) err = set_conversion_failed_error(a, tar->sconv, "Pathname"); } archive_string_free(&longname); return (err); } /* * Interpret 'V' GNU tar volume header. */ static int header_volume(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, int64_t *unconsumed) { const struct archive_entry_header_ustar *header; int64_t size, to_consume; (void)a; /* UNUSED */ (void)tar; /* UNUSED */ (void)entry; /* UNUSED */ header = (const struct archive_entry_header_ustar *)h; size = tar_atol(header->size, sizeof(header->size)); - if (size > (int64_t)pathname_limit) { + if (size < 0 || size > (int64_t)pathname_limit) { return (ARCHIVE_FATAL); } to_consume = ((size + 511) & ~511); *unconsumed += to_consume; return (ARCHIVE_OK); } /* * Read the next `size` bytes into the provided string. * Null-terminate the string. */ static int read_bytes_to_string(struct archive_read *a, struct archive_string *as, size_t size, int64_t *unconsumed) { const void *src; /* Fail if we can't make our buffer big enough. */ - if (archive_string_ensure(as, (size_t)size+1) == NULL) { + if (archive_string_ensure(as, size + 1) == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory"); return (ARCHIVE_FATAL); } - tar_flush_unconsumed(a, unconsumed); + if (tar_flush_unconsumed(a, unconsumed) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } /* Read the body into the string. */ src = __archive_read_ahead(a, size, NULL); if (src == NULL) { archive_set_error(&a->archive, EINVAL, "Truncated archive" " detected while reading metadata"); *unconsumed = 0; return (ARCHIVE_FATAL); } - memcpy(as->s, src, (size_t)size); + memcpy(as->s, src, size); as->s[size] = '\0'; - as->length = (size_t)size; + as->length = size; *unconsumed += size; return (ARCHIVE_OK); } /* * Read body of an archive entry into an archive_string object. */ static int read_body_to_string(struct archive_read *a, struct tar *tar, struct archive_string *as, const void *h, int64_t *unconsumed) { int64_t size; const struct archive_entry_header_ustar *header; int r; (void)tar; /* UNUSED */ header = (const struct archive_entry_header_ustar *)h; size = tar_atol(header->size, sizeof(header->size)); if (size < 0 || size > entry_limit) { archive_set_error(&a->archive, EINVAL, "Special header has invalid size: %lld", (long long)size); return (ARCHIVE_FATAL); } if (size > (int64_t)pathname_limit) { archive_string_empty(as); int64_t to_consume = ((size + 511) & ~511); if (to_consume != __archive_read_consume(a, to_consume)) { return (ARCHIVE_FATAL); } archive_set_error(&a->archive, EINVAL, "Special header too large: %lld > 1MiB", (long long)size); return (ARCHIVE_WARN); } r = read_bytes_to_string(a, as, size, unconsumed); *unconsumed += 0x1ff & (-size); return(r); } /* * Parse out common header elements. * * This would be the same as header_old_tar, except that the * filename is handled slightly differently for old and POSIX * entries (POSIX entries support a 'prefix'). This factoring * allows header_old_tar and header_ustar * to handle filenames differently, while still putting most of the * common parsing into one place. * * This is called _after_ ustar, GNU tar, Schily, etc, special * fields have already been parsed into the `tar` structure. * So we can make final decisions here about how to reconcile * size, mode, etc, information. */ static int header_common(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h) { const struct archive_entry_header_ustar *header; const char *existing_linkpath; const wchar_t *existing_wcs_linkpath; int err = ARCHIVE_OK; header = (const struct archive_entry_header_ustar *)h; /* Parse out the numeric fields (all are octal) */ /* Split mode handling: Set filetype always, perm only if not already set */ archive_entry_set_filetype(entry, (mode_t)tar_atol(header->mode, sizeof(header->mode))); if (!archive_entry_perm_is_set(entry)) { archive_entry_set_perm(entry, (mode_t)tar_atol(header->mode, sizeof(header->mode))); } /* Set uid, gid, mtime if not already set */ if (!archive_entry_uid_is_set(entry)) { archive_entry_set_uid(entry, tar_atol(header->uid, sizeof(header->uid))); } if (!archive_entry_gid_is_set(entry)) { archive_entry_set_gid(entry, tar_atol(header->gid, sizeof(header->gid))); } if (!archive_entry_mtime_is_set(entry)) { archive_entry_set_mtime(entry, tar_atol(header->mtime, sizeof(header->mtime)), 0); } /* Reconcile the size info. */ /* First, how big is the file on disk? */ if ((tar->size_fields & TAR_SIZE_GNU_SPARSE_REALSIZE) != 0) { /* GNU sparse format 1.0 uses `GNU.sparse.realsize` * to hold the size of the file on disk. */ tar->disk_size = tar->GNU_sparse_realsize; } else if ((tar->size_fields & TAR_SIZE_GNU_SPARSE_SIZE) != 0 && (tar->sparse_gnu_major == 0)) { /* GNU sparse format 0.0 and 0.1 use `GNU.sparse.size` * to hold the size of the file on disk. */ tar->disk_size = tar->GNU_sparse_size; } else if ((tar->size_fields & TAR_SIZE_SCHILY_SPARSE_REALSIZE) != 0) { tar->disk_size = tar->SCHILY_sparse_realsize; } else if ((tar->size_fields & TAR_SIZE_PAX_SIZE) != 0) { tar->disk_size = tar->pax_size; } else { /* There wasn't a suitable pax header, so use the ustar info */ tar->disk_size = tar_atol(header->size, sizeof(header->size)); } if (tar->disk_size < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Tar entry has negative file size"); return (ARCHIVE_FATAL); } else if (tar->disk_size > entry_limit) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Tar entry size overflow"); return (ARCHIVE_FATAL); } else { archive_entry_set_size(entry, tar->disk_size); } /* Second, how big is the data in the archive? */ if ((tar->size_fields & TAR_SIZE_GNU_SPARSE_SIZE) != 0 && (tar->sparse_gnu_major == 1)) { /* GNU sparse format 1.0 uses `GNU.sparse.size` * to hold the size of the data in the archive. */ tar->entry_bytes_remaining = tar->GNU_sparse_size; } else if ((tar->size_fields & TAR_SIZE_PAX_SIZE) != 0) { tar->entry_bytes_remaining = tar->pax_size; } else { tar->entry_bytes_remaining = tar_atol(header->size, sizeof(header->size)); } if (tar->entry_bytes_remaining < 0) { tar->entry_bytes_remaining = 0; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Tar entry has negative size"); return (ARCHIVE_FATAL); } else if (tar->entry_bytes_remaining > entry_limit) { tar->entry_bytes_remaining = 0; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Tar entry size overflow"); return (ARCHIVE_FATAL); } /* Handle the tar type flag appropriately. */ tar->filetype = header->typeflag[0]; /* * TODO: If the linkpath came from Pax extension header, then * we should obey the hdrcharset_utf8 flag when converting these. */ switch (tar->filetype) { case '1': /* Hard link */ archive_entry_set_link_to_hardlink(entry); existing_wcs_linkpath = archive_entry_hardlink_w(entry); existing_linkpath = archive_entry_hardlink(entry); if ((existing_linkpath == NULL || existing_linkpath[0] == '\0') && (existing_wcs_linkpath == NULL || existing_wcs_linkpath[0] == '\0')) { struct archive_string linkpath; archive_string_init(&linkpath); archive_strncpy(&linkpath, header->linkname, sizeof(header->linkname)); if (archive_entry_copy_hardlink_l(entry, linkpath.s, archive_strlen(&linkpath), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Linkname"); if (err == ARCHIVE_FATAL) { archive_string_free(&linkpath); return (err); } } archive_string_free(&linkpath); } /* * The following may seem odd, but: Technically, tar * does not store the file type for a "hard link" * entry, only the fact that it is a hard link. So, I * leave the type zero normally. But, pax interchange * format allows hard links to have data, which * implies that the underlying entry is a regular * file. */ if (archive_entry_size(entry) > 0) archive_entry_set_filetype(entry, AE_IFREG); /* * A tricky point: Traditionally, tar readers have * ignored the size field when reading hardlink * entries, and some writers put non-zero sizes even * though the body is empty. POSIX blessed this * convention in the 1988 standard, but broke with * this tradition in 2001 by permitting hardlink * entries to store valid bodies in pax interchange * format, but not in ustar format. Since there is no * hard and fast way to distinguish pax interchange * from earlier archives (the 'x' and 'g' entries are * optional, after all), we need a heuristic. */ if (archive_entry_size(entry) == 0) { /* If the size is already zero, we're done. */ } else if (a->archive.archive_format == ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE) { /* Definitely pax extended; must obey hardlink size. */ } else if (a->archive.archive_format == ARCHIVE_FORMAT_TAR || a->archive.archive_format == ARCHIVE_FORMAT_TAR_GNUTAR) { /* Old-style or GNU tar: we must ignore the size. */ archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; } else if (archive_read_format_tar_bid(a, 50) > 50) { /* * We don't know if it's pax: If the bid * function sees a valid ustar header * immediately following, then let's ignore * the hardlink size. */ archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; } /* * TODO: There are still two cases I'd like to handle: * = a ustar non-pax archive with a hardlink entry at * end-of-archive. (Look for block of nulls following?) * = a pax archive that has not seen any pax headers * and has an entry which is a hardlink entry storing * a body containing an uncompressed tar archive. * The first is worth addressing; I don't see any reliable * way to deal with the second possibility. */ break; case '2': /* Symlink */ archive_entry_set_link_to_symlink(entry); existing_wcs_linkpath = archive_entry_symlink_w(entry); existing_linkpath = archive_entry_symlink(entry); if ((existing_linkpath == NULL || existing_linkpath[0] == '\0') && (existing_wcs_linkpath == NULL || existing_wcs_linkpath[0] == '\0')) { struct archive_string linkpath; archive_string_init(&linkpath); archive_strncpy(&linkpath, header->linkname, sizeof(header->linkname)); if (archive_entry_copy_symlink_l(entry, linkpath.s, archive_strlen(&linkpath), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Linkname"); if (err == ARCHIVE_FATAL) { archive_string_free(&linkpath); return (err); } } archive_string_free(&linkpath); } archive_entry_set_filetype(entry, AE_IFLNK); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case '3': /* Character device */ archive_entry_set_filetype(entry, AE_IFCHR); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case '4': /* Block device */ archive_entry_set_filetype(entry, AE_IFBLK); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case '5': /* Dir */ archive_entry_set_filetype(entry, AE_IFDIR); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case '6': /* FIFO device */ archive_entry_set_filetype(entry, AE_IFIFO); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case 'D': /* GNU incremental directory type */ /* * No special handling is actually required here. * It might be nice someday to preprocess the file list and * provide it to the client, though. */ archive_entry_set_filetype(entry, AE_IFDIR); break; case 'M': /* GNU "Multi-volume" (remainder of file from last archive)*/ /* * As far as I can tell, this is just like a regular file * entry, except that the contents should be _appended_ to * the indicated file at the indicated offset. This may * require some API work to fully support. */ break; case 'N': /* Old GNU "long filename" entry. */ /* The body of this entry is a script for renaming * previously-extracted entries. Ugh. It will never * be supported by libarchive. */ archive_entry_set_filetype(entry, AE_IFREG); break; case 'S': /* GNU sparse files */ /* * Sparse files are really just regular files with * sparse information in the extended area. */ /* FALLTHROUGH */ case '0': /* ustar "regular" file */ /* FALLTHROUGH */ default: /* Non-standard file types */ /* * Per POSIX: non-recognized types should always be * treated as regular files. */ archive_entry_set_filetype(entry, AE_IFREG); break; } return (err); } /* * Parse out header elements for "old-style" tar archives. */ static int header_old_tar(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h) { const struct archive_entry_header_ustar *header; int err = ARCHIVE_OK, err2; /* * Copy filename over (to ensure null termination). * Skip if pathname was already set e.g. by header_gnu_longname() */ header = (const struct archive_entry_header_ustar *)h; const char *existing_pathname = archive_entry_pathname(entry); const wchar_t *existing_wcs_pathname = archive_entry_pathname_w(entry); if ((existing_pathname == NULL || existing_pathname[0] == '\0') && (existing_wcs_pathname == NULL || existing_wcs_pathname[0] == '\0') && archive_entry_copy_pathname_l(entry, header->name, sizeof(header->name), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Pathname"); if (err == ARCHIVE_FATAL) return (err); } /* Grab rest of common fields */ err2 = header_common(a, tar, entry, h); if (err > err2) err = err2; tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); return (err); } /* * Is this likely an AppleDouble extension? */ static int is_mac_metadata_entry(struct archive_entry *entry) { const char *p, *name; const wchar_t *wp, *wname; wname = wp = archive_entry_pathname_w(entry); if (wp != NULL) { /* Find the last path element. */ for (; *wp != L'\0'; ++wp) { if (wp[0] == '/' && wp[1] != L'\0') wname = wp + 1; } /* * If last path element starts with "._", then * this is a Mac extension. */ if (wname[0] == L'.' && wname[1] == L'_' && wname[2] != L'\0') return 1; } else { /* Find the last path element. */ name = p = archive_entry_pathname(entry); if (p == NULL) return (ARCHIVE_FAILED); for (; *p != '\0'; ++p) { if (p[0] == '/' && p[1] != '\0') name = p + 1; } /* * If last path element starts with "._", then * this is a Mac extension. */ if (name[0] == '.' && name[1] == '_' && name[2] != '\0') return 1; } /* Not a mac extension */ return 0; } /* * Read a Mac AppleDouble-encoded blob of file metadata, * if there is one. * * TODO: In Libarchive 4, we should consider ripping this * out -- instead, return a file starting with `._` as * a regular file and let the client (or archive_write logic) * handle it. */ static int read_mac_metadata_blob(struct archive_read *a, struct archive_entry *entry, int64_t *unconsumed) { int64_t size; size_t msize; const void *data; /* Read the body as a Mac OS metadata blob. */ size = archive_entry_size(entry); msize = (size_t)size; if (size < 0 || (uintmax_t)msize != (uintmax_t)size) { *unconsumed = 0; return (ARCHIVE_FATAL); } /* TODO: Should this merely skip the overlarge entry and * WARN? Or is xattr_limit sufficiently large that we can * safely assume anything larger is malicious? */ if (size > (int64_t)xattr_limit) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Oversized AppleDouble extension has size %llu > %llu", (unsigned long long)size, (unsigned long long)xattr_limit); return (ARCHIVE_FATAL); } /* * TODO: Look beyond the body here to peek at the next header. * If it's a regular header (not an extension header) * that has the wrong name, just return the current * entry as-is, without consuming the body here. * That would reduce the risk of us mis-identifying * an ordinary file that just happened to have * a name starting with "._". * * Q: Is the above idea really possible? Even * when there are GNU or pax extension entries? */ - tar_flush_unconsumed(a, unconsumed); + if (tar_flush_unconsumed(a, unconsumed) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } data = __archive_read_ahead(a, msize, NULL); if (data == NULL) { archive_set_error(&a->archive, EINVAL, "Truncated archive" " detected while reading macOS metadata"); *unconsumed = 0; return (ARCHIVE_FATAL); } archive_entry_clear(entry); archive_entry_copy_mac_metadata(entry, data, msize); *unconsumed = (msize + 511) & ~ 511; return (ARCHIVE_OK); } /* * Parse a file header for a pax extended archive entry. */ static int header_pax_global(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, int64_t *unconsumed) { const struct archive_entry_header_ustar *header; int64_t size, to_consume; (void)a; /* UNUSED */ (void)tar; /* UNUSED */ (void)entry; /* UNUSED */ header = (const struct archive_entry_header_ustar *)h; size = tar_atol(header->size, sizeof(header->size)); if (size < 0 || size > entry_limit) { archive_set_error(&a->archive, EINVAL, "Special header has invalid size: %lld", (long long)size); return (ARCHIVE_FATAL); } to_consume = ((size + 511) & ~511); *unconsumed += to_consume; return (ARCHIVE_OK); } /* * Parse a file header for a Posix "ustar" archive entry. This also * handles "pax" or "extended ustar" entries. * * In order to correctly handle pax attributes (which precede this), * we have to skip parsing any field for which the entry already has * contents. */ static int header_ustar(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h) { const struct archive_entry_header_ustar *header; struct archive_string as; int err = ARCHIVE_OK, r; header = (const struct archive_entry_header_ustar *)h; /* Copy name into an internal buffer to ensure null-termination. */ const char *existing_pathname = archive_entry_pathname(entry); const wchar_t *existing_wcs_pathname = archive_entry_pathname_w(entry); if ((existing_pathname == NULL || existing_pathname[0] == '\0') && (existing_wcs_pathname == NULL || existing_wcs_pathname[0] == '\0')) { archive_string_init(&as); if (header->prefix[0]) { archive_strncpy(&as, header->prefix, sizeof(header->prefix)); if (as.s[archive_strlen(&as) - 1] != '/') archive_strappend_char(&as, '/'); archive_strncat(&as, header->name, sizeof(header->name)); } else { archive_strncpy(&as, header->name, sizeof(header->name)); } if (archive_entry_copy_pathname_l(entry, as.s, archive_strlen(&as), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Pathname"); if (err == ARCHIVE_FATAL) return (err); } archive_string_free(&as); } /* Handle rest of common fields. */ r = header_common(a, tar, entry, h); if (r == ARCHIVE_FATAL) return (r); if (r < err) err = r; /* Handle POSIX ustar fields. */ const char *existing_uname = archive_entry_uname(entry); if (existing_uname == NULL || existing_uname[0] == '\0') { if (archive_entry_copy_uname_l(entry, header->uname, sizeof(header->uname), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Uname"); if (err == ARCHIVE_FATAL) return (err); } } const char *existing_gname = archive_entry_gname(entry); if (existing_gname == NULL || existing_gname[0] == '\0') { if (archive_entry_copy_gname_l(entry, header->gname, sizeof(header->gname), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Gname"); if (err == ARCHIVE_FATAL) return (err); } } /* Parse out device numbers only for char and block specials. */ if (header->typeflag[0] == '3' || header->typeflag[0] == '4') { if (!archive_entry_rdev_is_set(entry)) { archive_entry_set_rdevmajor(entry, (dev_t) tar_atol(header->rdevmajor, sizeof(header->rdevmajor))); archive_entry_set_rdevminor(entry, (dev_t) tar_atol(header->rdevminor, sizeof(header->rdevminor))); } } else { archive_entry_set_rdev(entry, 0); } tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); return (err); } static int header_pax_extension(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, int64_t *unconsumed) { /* Sanity checks: The largest `x` body I've ever heard of was * a little over 4MB. So I doubt there has ever been a * well-formed archive with an `x` body over 1GiB. Similarly, * it seems plausible that no single attribute has ever been * larger than 100MB. So if we see a larger value here, it's * almost certainly a sign of a corrupted/malicious archive. */ /* Maximum sane size for extension body: 1 GiB */ /* This cannot be raised to larger than 8GiB without * exceeding the maximum size for a standard ustar * entry. */ const int64_t ext_size_limit = 1024 * 1024 * (int64_t)1024; /* Maximum size for a single line/attr: 100 million characters */ /* This cannot be raised to more than 2GiB without exceeding * a `size_t` on 32-bit platforms. */ const size_t max_parsed_line_length = 99999999ULL; /* Largest attribute prolog: size + name. */ const size_t max_size_name = 512; /* Size and padding of the full extension body */ int64_t ext_size, ext_padding; size_t line_length, value_length, name_length; ssize_t to_read, did_read; const struct archive_entry_header_ustar *header; const char *p, *attr_start, *name_start; struct archive_string_conv *sconv; struct archive_string *pas = NULL; struct archive_string attr_name; int err = ARCHIVE_OK, r; header = (const struct archive_entry_header_ustar *)h; ext_size = tar_atol(header->size, sizeof(header->size)); if (ext_size > entry_limit) { return (ARCHIVE_FATAL); } if (ext_size < 0) { archive_set_error(&a->archive, EINVAL, "pax extension header has invalid size: %lld", (long long)ext_size); return (ARCHIVE_FATAL); } ext_padding = 0x1ff & (-ext_size); if (ext_size > ext_size_limit) { /* Consume the pax extension body and return an error */ if (ext_size + ext_padding != __archive_read_consume(a, ext_size + ext_padding)) { return (ARCHIVE_FATAL); } archive_set_error(&a->archive, EINVAL, "Ignoring oversized pax extensions: %lld > %lld", (long long)ext_size, (long long)ext_size_limit); return (ARCHIVE_WARN); } - tar_flush_unconsumed(a, unconsumed); + if (tar_flush_unconsumed(a, unconsumed) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } /* Parse the size/name of each pax attribute in the body */ archive_string_init(&attr_name); while (ext_size > 0) { /* Read enough bytes to parse the size/name of the next attribute */ to_read = max_size_name; if (to_read > ext_size) { to_read = ext_size; } p = __archive_read_ahead(a, to_read, &did_read); if (p == NULL) { /* EOF */ archive_set_error(&a->archive, EINVAL, "Truncated tar archive" " detected while reading pax attribute name"); return (ARCHIVE_FATAL); } if (did_read > ext_size) { did_read = ext_size; } /* Parse size of attribute */ line_length = 0; attr_start = p; while (1) { if (p >= attr_start + did_read) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring malformed pax attributes: overlarge attribute size field"); *unconsumed += ext_size + ext_padding; return (ARCHIVE_WARN); } if (*p == ' ') { p++; break; } if (*p < '0' || *p > '9') { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring malformed pax attributes: malformed attribute size field"); *unconsumed += ext_size + ext_padding; return (ARCHIVE_WARN); } line_length *= 10; line_length += *p - '0'; if (line_length > max_parsed_line_length) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring malformed pax attribute: size > %lld", (long long)max_parsed_line_length); *unconsumed += ext_size + ext_padding; return (ARCHIVE_WARN); } p++; } if ((int64_t)line_length > ext_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring malformed pax attribute: %lld > %lld", (long long)line_length, (long long)ext_size); *unconsumed += ext_size + ext_padding; return (ARCHIVE_WARN); } /* Parse name of attribute */ if (p >= attr_start + did_read || p >= attr_start + line_length || *p == '=') { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring malformed pax attributes: empty name found"); *unconsumed += ext_size + ext_padding; return (ARCHIVE_WARN); } name_start = p; while (1) { if (p >= attr_start + did_read || p >= attr_start + line_length) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring malformed pax attributes: overlarge attribute name"); *unconsumed += ext_size + ext_padding; return (ARCHIVE_WARN); } if (*p == '=') { break; } p++; } name_length = p - name_start; p++; // Skip '=' // Save the name before we consume it archive_strncpy(&attr_name, name_start, name_length); ext_size -= p - attr_start; value_length = line_length - (p - attr_start); /* Consume size, name, and `=` */ *unconsumed += p - attr_start; - tar_flush_unconsumed(a, unconsumed); + if (tar_flush_unconsumed(a, unconsumed) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } if (value_length == 0) { archive_set_error(&a->archive, EINVAL, "Malformed pax attributes"); *unconsumed += ext_size + ext_padding; return (ARCHIVE_WARN); } /* pax_attribute will consume value_length - 1 */ r = pax_attribute(a, tar, entry, attr_name.s, archive_strlen(&attr_name), value_length - 1, unconsumed); ext_size -= value_length - 1; // Release the allocated attr_name (either here or before every return in this function) archive_string_free(&attr_name); if (r < ARCHIVE_WARN) { *unconsumed += ext_size + ext_padding; return (r); } err = err_combine(err, r); /* Consume the `\n` that follows the pax attribute value. */ - tar_flush_unconsumed(a, unconsumed); + if (tar_flush_unconsumed(a, unconsumed) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } p = __archive_read_ahead(a, 1, &did_read); if (p == NULL) { archive_set_error(&a->archive, EINVAL, "Truncated tar archive" " detected while completing pax attribute"); return (ARCHIVE_FATAL); } if (p[0] != '\n') { archive_set_error(&a->archive, EINVAL, "Malformed pax attributes"); *unconsumed += ext_size + ext_padding; return (ARCHIVE_WARN); } ext_size -= 1; *unconsumed += 1; - tar_flush_unconsumed(a, unconsumed); + if (tar_flush_unconsumed(a, unconsumed) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } } *unconsumed += ext_size + ext_padding; /* * Some PAX values -- pathname, linkpath, uname, gname -- * can't be copied into the entry until we know the character * set to use: */ if (!tar->pax_hdrcharset_utf8) /* PAX specified "BINARY", so use the default charset */ sconv = tar->opt_sconv; else { /* PAX default UTF-8 */ sconv = archive_string_conversion_from_charset( &(a->archive), "UTF-8", 1); if (sconv == NULL) return (ARCHIVE_FATAL); if (tar->compat_2x) archive_string_conversion_set_opt(sconv, SCONV_SET_OPT_UTF8_LIBARCHIVE2X); } /* Pathname */ pas = NULL; if (archive_strlen(&(tar->entry_pathname_override)) > 0) { /* Prefer GNU.sparse.name attribute if present */ /* GNU sparse files store a fake name under the standard * "pathname" key. */ pas = &(tar->entry_pathname_override); } else if (archive_strlen(&(tar->entry_pathname)) > 0) { /* Use standard "pathname" PAX extension */ pas = &(tar->entry_pathname); } if (pas != NULL) { if (archive_entry_copy_pathname_l(entry, pas->s, archive_strlen(pas), sconv) != 0) { err = set_conversion_failed_error(a, sconv, "Pathname"); if (err == ARCHIVE_FATAL) return (err); /* Use raw name without conversion */ archive_entry_copy_pathname(entry, pas->s); } } /* Uname */ if (archive_strlen(&(tar->entry_uname)) > 0) { if (archive_entry_copy_uname_l(entry, tar->entry_uname.s, archive_strlen(&(tar->entry_uname)), sconv) != 0) { err = set_conversion_failed_error(a, sconv, "Uname"); if (err == ARCHIVE_FATAL) return (err); /* Use raw name without conversion */ archive_entry_copy_uname(entry, tar->entry_uname.s); } } /* Gname */ if (archive_strlen(&(tar->entry_gname)) > 0) { if (archive_entry_copy_gname_l(entry, tar->entry_gname.s, archive_strlen(&(tar->entry_gname)), sconv) != 0) { err = set_conversion_failed_error(a, sconv, "Gname"); if (err == ARCHIVE_FATAL) return (err); /* Use raw name without conversion */ archive_entry_copy_gname(entry, tar->entry_gname.s); } } /* Linkpath */ if (archive_strlen(&(tar->entry_linkpath)) > 0) { if (archive_entry_copy_link_l(entry, tar->entry_linkpath.s, archive_strlen(&(tar->entry_linkpath)), sconv) != 0) { err = set_conversion_failed_error(a, sconv, "Linkpath"); if (err == ARCHIVE_FATAL) return (err); /* Use raw name without conversion */ archive_entry_copy_link(entry, tar->entry_linkpath.s); } } /* Extension may have given us a corrected `entry_bytes_remaining` for * the main entry; update the padding appropriately. */ tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); return (err); } static int pax_attribute_LIBARCHIVE_xattr(struct archive_entry *entry, const char *name, size_t name_length, const char *value, size_t value_length) { char *name_decoded; void *value_decoded; size_t value_len; if (name_length < 1) return 3; /* URL-decode name */ name_decoded = url_decode(name, name_length); if (name_decoded == NULL) return 2; /* Base-64 decode value */ value_decoded = base64_decode(value, value_length, &value_len); if (value_decoded == NULL) { free(name_decoded); return 1; } archive_entry_xattr_add_entry(entry, name_decoded, value_decoded, value_len); free(name_decoded); free(value_decoded); return 0; } static int pax_attribute_SCHILY_xattr(struct archive_entry *entry, const char *name, size_t name_length, const char *value, size_t value_length) { if (name_length < 1 || name_length > 128) { return 1; } char * null_terminated_name = malloc(name_length + 1); if (null_terminated_name != NULL) { memcpy(null_terminated_name, name, name_length); null_terminated_name[name_length] = '\0'; archive_entry_xattr_add_entry(entry, null_terminated_name, value, value_length); free(null_terminated_name); } return 0; } static int pax_attribute_RHT_security_selinux(struct archive_entry *entry, const char *value, size_t value_length) { archive_entry_xattr_add_entry(entry, "security.selinux", value, value_length); return 0; } static int pax_attribute_SCHILY_acl(struct archive_read *a, struct tar *tar, struct archive_entry *entry, size_t value_length, int type) { int r; const char *p; const char* errstr; switch (type) { case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: errstr = "SCHILY.acl.access"; break; case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: errstr = "SCHILY.acl.default"; break; case ARCHIVE_ENTRY_ACL_TYPE_NFS4: errstr = "SCHILY.acl.ace"; break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unknown ACL type: %d", type); return(ARCHIVE_FATAL); } if (tar->sconv_acl == NULL) { tar->sconv_acl = archive_string_conversion_from_charset( &(a->archive), "UTF-8", 1); if (tar->sconv_acl == NULL) return (ARCHIVE_FATAL); } if (value_length > acl_limit) { __archive_read_consume(a, value_length); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unreasonably large ACL: %llu > %llu", (unsigned long long)value_length, (unsigned long long)acl_limit); return (ARCHIVE_WARN); } p = __archive_read_ahead(a, value_length, NULL); if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated tar archive " "detected while reading ACL data"); return (ARCHIVE_FATAL); } r = archive_acl_from_text_nl(archive_entry_acl(entry), p, value_length, type, tar->sconv_acl); __archive_read_consume(a, value_length); /* Workaround: Force perm_is_set() to be correct */ /* If this bit were stored in the ACL, this wouldn't be needed */ archive_entry_set_perm(entry, archive_entry_perm(entry)); if (r != ARCHIVE_OK) { if (r == ARCHIVE_FATAL) { archive_set_error(&a->archive, ENOMEM, "%s %s", "Can't allocate memory for ", errstr); return (r); } archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s %s", "Parse error: ", errstr); } return (r); } static int pax_attribute_read_time(struct archive_read *a, size_t value_length, int64_t *ps, long *pn, int64_t *unconsumed) { struct archive_string as; int r; if (value_length > 128) { __archive_read_consume(a, value_length); *ps = 0; *pn = 0; return (ARCHIVE_FATAL); } archive_string_init(&as); r = read_bytes_to_string(a, &as, value_length, unconsumed); if (r < ARCHIVE_OK) { archive_string_free(&as); *ps = 0; *pn = 0; return (r); } pax_time(as.s, archive_strlen(&as), ps, pn); archive_string_free(&as); if (*ps == INT64_MIN) { *ps = 0; *pn = 0; return (ARCHIVE_WARN); } return (ARCHIVE_OK); } static int pax_attribute_read_number(struct archive_read *a, size_t value_length, int64_t *result) { struct archive_string as; int64_t unconsumed = 0; int r; if (value_length > 64) { __archive_read_consume(a, value_length); *result = 0; return (ARCHIVE_FATAL); } archive_string_init(&as); r = read_bytes_to_string(a, &as, value_length, &unconsumed); - tar_flush_unconsumed(a, &unconsumed); + if (tar_flush_unconsumed(a, &unconsumed) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } if (r < ARCHIVE_OK) { archive_string_free(&as); *result = 0; return (r); } *result = tar_atol10(as.s, archive_strlen(&as)); archive_string_free(&as); if (*result < 0 || *result == INT64_MAX) { *result = INT64_MAX; return (ARCHIVE_WARN); } return (ARCHIVE_OK); } /* * Parse a single key=value attribute. * * POSIX reserves all-lowercase keywords. Vendor-specific extensions * should always have keywords of the form "VENDOR.attribute" In * particular, it's quite feasible to support many different vendor * extensions here. I'm using "LIBARCHIVE" for extensions unique to * this library. * * TODO: Investigate other vendor-specific extensions and see if * any of them look useful. */ static int pax_attribute(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const char *key, size_t key_length, size_t value_length, int64_t *unconsumed) { int64_t t; long n; const char *p; ssize_t bytes_read; int err = ARCHIVE_OK; switch (key[0]) { case 'G': /* GNU.* extensions */ if (key_length > 4 && memcmp(key, "GNU.", 4) == 0) { key += 4; key_length -= 4; /* GNU.sparse marks the existence of GNU sparse information */ if (key_length == 6 && memcmp(key, "sparse", 6) == 0) { tar->sparse_gnu_attributes_seen = 1; } /* GNU.sparse.* extensions */ else if (key_length > 7 && memcmp(key, "sparse.", 7) == 0) { tar->sparse_gnu_attributes_seen = 1; key += 7; key_length -= 7; /* GNU "0.0" sparse pax format. */ if (key_length == 9 && memcmp(key, "numblocks", 9) == 0) { /* GNU.sparse.numblocks */ tar->sparse_offset = -1; tar->sparse_numbytes = -1; tar->sparse_gnu_major = 0; tar->sparse_gnu_minor = 0; } else if (key_length == 6 && memcmp(key, "offset", 6) == 0) { /* GNU.sparse.offset */ if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { tar->sparse_offset = t; if (tar->sparse_numbytes != -1) { if (gnu_add_sparse_entry(a, tar, tar->sparse_offset, tar->sparse_numbytes) != ARCHIVE_OK) return (ARCHIVE_FATAL); tar->sparse_offset = -1; tar->sparse_numbytes = -1; } } return (err); } else if (key_length == 8 && memcmp(key, "numbytes", 8) == 0) { /* GNU.sparse.numbytes */ if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { tar->sparse_numbytes = t; if (tar->sparse_offset != -1) { if (gnu_add_sparse_entry(a, tar, tar->sparse_offset, tar->sparse_numbytes) != ARCHIVE_OK) return (ARCHIVE_FATAL); tar->sparse_offset = -1; tar->sparse_numbytes = -1; } } return (err); } else if (key_length == 4 && memcmp(key, "size", 4) == 0) { /* GNU.sparse.size */ /* This is either the size of stored entry OR the size of data on disk, * depending on which GNU sparse format version is in use. * Since pax attributes can be in any order, we may not actually * know at this point how to interpret this. */ if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { tar->GNU_sparse_size = t; tar->size_fields |= TAR_SIZE_GNU_SPARSE_SIZE; } return (err); } /* GNU "0.1" sparse pax format. */ else if (key_length == 3 && memcmp(key, "map", 3) == 0) { /* GNU.sparse.map */ tar->sparse_gnu_major = 0; tar->sparse_gnu_minor = 1; if (value_length > sparse_map_limit) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unreasonably large sparse map: %llu > %llu", (unsigned long long)value_length, (unsigned long long)sparse_map_limit); err = ARCHIVE_FAILED; } else { p = __archive_read_ahead(a, value_length, &bytes_read); if (p == NULL) { archive_set_error(&a->archive, EINVAL, "Truncated archive" " detected while reading GNU sparse data"); return (ARCHIVE_FATAL); } if (gnu_sparse_01_parse(a, tar, p, value_length) != ARCHIVE_OK) { err = ARCHIVE_WARN; } } __archive_read_consume(a, value_length); return (err); } /* GNU "1.0" sparse pax format */ else if (key_length == 5 && memcmp(key, "major", 5) == 0) { /* GNU.sparse.major */ if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK && t >= 0 && t <= 10) { tar->sparse_gnu_major = (int)t; } return (err); } else if (key_length == 5 && memcmp(key, "minor", 5) == 0) { /* GNU.sparse.minor */ if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK && t >= 0 && t <= 10) { tar->sparse_gnu_minor = (int)t; } return (err); } else if (key_length == 4 && memcmp(key, "name", 4) == 0) { /* GNU.sparse.name */ /* * The real filename; when storing sparse * files, GNU tar puts a synthesized name into * the regular 'path' attribute in an attempt * to limit confusion. ;-) */ if (value_length > pathname_limit) { *unconsumed += value_length; err = ARCHIVE_WARN; } else { err = read_bytes_to_string(a, &(tar->entry_pathname_override), value_length, unconsumed); } return (err); } else if (key_length == 8 && memcmp(key, "realsize", 8) == 0) { /* GNU.sparse.realsize = size of file on disk */ if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { tar->GNU_sparse_realsize = t; tar->size_fields |= TAR_SIZE_GNU_SPARSE_REALSIZE; } return (err); } } } break; case 'L': /* LIBARCHIVE extensions */ if (key_length > 11 && memcmp(key, "LIBARCHIVE.", 11) == 0) { key_length -= 11; key += 11; /* TODO: Handle arbitrary extended attributes... */ /* if (strcmp(key, "LIBARCHIVE.xxxxxxx") == 0) archive_entry_set_xxxxxx(entry, value); */ if (key_length == 12 && memcmp(key, "creationtime", 12) == 0) { /* LIBARCHIVE.creationtime */ if ((err = pax_attribute_read_time(a, value_length, &t, &n, unconsumed)) == ARCHIVE_OK) { archive_entry_set_birthtime(entry, t, n); } return (err); } else if (key_length == 11 && memcmp(key, "symlinktype", 11) == 0) { /* LIBARCHIVE.symlinktype */ if (value_length < 16) { p = __archive_read_ahead(a, value_length, &bytes_read); if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated tar archive " "detected while reading `symlinktype` attribute"); return (ARCHIVE_FATAL); } if (value_length == 4 && memcmp(p, "file", 4) == 0) { archive_entry_set_symlink_type(entry, AE_SYMLINK_TYPE_FILE); } else if (value_length == 3 && memcmp(p, "dir", 3) == 0) { archive_entry_set_symlink_type(entry, AE_SYMLINK_TYPE_DIRECTORY); } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unrecognized symlink type"); err = ARCHIVE_WARN; } } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "symlink type is very long" "(longest recognized value is 4 bytes, this is %llu)", (unsigned long long)value_length); err = ARCHIVE_WARN; } __archive_read_consume(a, value_length); return (err); } else if (key_length > 6 && memcmp(key, "xattr.", 6) == 0) { key_length -= 6; key += 6; if (value_length > xattr_limit) { err = ARCHIVE_WARN; } else { p = __archive_read_ahead(a, value_length, &bytes_read); if (p == NULL) { archive_set_error(&a->archive, EINVAL, "Truncated archive" " detected while reading xattr information"); return (ARCHIVE_FATAL); } if (pax_attribute_LIBARCHIVE_xattr(entry, key, key_length, p, value_length)) { /* TODO: Unable to parse xattr */ err = ARCHIVE_WARN; } } __archive_read_consume(a, value_length); return (err); } } break; case 'R': /* GNU tar uses RHT.security header to store SELinux xattrs * SCHILY.xattr.security.selinux == RHT.security.selinux */ if (key_length == 20 && memcmp(key, "RHT.security.selinux", 20) == 0) { if (value_length > xattr_limit) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring unreasonably large security.selinux attribute:" " %llu > %llu", (unsigned long long)value_length, (unsigned long long)xattr_limit); /* TODO: Should this be FAILED instead? */ err = ARCHIVE_WARN; } else { p = __archive_read_ahead(a, value_length, &bytes_read); if (p == NULL) { archive_set_error(&a->archive, EINVAL, "Truncated archive" " detected while reading selinux data"); return (ARCHIVE_FATAL); } if (pax_attribute_RHT_security_selinux(entry, p, value_length)) { /* TODO: Unable to parse xattr */ err = ARCHIVE_WARN; } } __archive_read_consume(a, value_length); return (err); } break; case 'S': /* SCHILY.* extensions used by "star" archiver */ if (key_length > 7 && memcmp(key, "SCHILY.", 7) == 0) { key_length -= 7; key += 7; if (key_length == 10 && memcmp(key, "acl.access", 10) == 0) { err = pax_attribute_SCHILY_acl(a, tar, entry, value_length, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); // TODO: Mark mode as set return (err); } else if (key_length == 11 && memcmp(key, "acl.default", 11) == 0) { err = pax_attribute_SCHILY_acl(a, tar, entry, value_length, ARCHIVE_ENTRY_ACL_TYPE_DEFAULT); return (err); } else if (key_length == 7 && memcmp(key, "acl.ace", 7) == 0) { err = pax_attribute_SCHILY_acl(a, tar, entry, value_length, ARCHIVE_ENTRY_ACL_TYPE_NFS4); // TODO: Mark mode as set return (err); } else if (key_length == 8 && memcmp(key, "devmajor", 8) == 0) { if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { archive_entry_set_rdevmajor(entry, (dev_t)t); } return (err); } else if (key_length == 8 && memcmp(key, "devminor", 8) == 0) { if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { archive_entry_set_rdevminor(entry, (dev_t)t); } return (err); } else if (key_length == 6 && memcmp(key, "fflags", 6) == 0) { if (value_length < fflags_limit) { p = __archive_read_ahead(a, value_length, &bytes_read); if (p == NULL) { /* Truncated archive */ archive_set_error(&a->archive, EINVAL, "Truncated archive" " detected while reading SCHILY.fflags"); return (ARCHIVE_FATAL); } archive_entry_copy_fflags_text_len(entry, p, value_length); err = ARCHIVE_OK; } else { /* Overlong fflags field */ err = ARCHIVE_WARN; } __archive_read_consume(a, value_length); return (err); } else if (key_length == 3 && memcmp(key, "dev", 3) == 0) { if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { archive_entry_set_dev(entry, (dev_t)t); } return (err); } else if (key_length == 3 && memcmp(key, "ino", 3) == 0) { if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { archive_entry_set_ino(entry, t); } return (err); } else if (key_length == 5 && memcmp(key, "nlink", 5) == 0) { if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { archive_entry_set_nlink(entry, (unsigned int)t); } return (err); } else if (key_length == 8 && memcmp(key, "realsize", 8) == 0) { if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { tar->SCHILY_sparse_realsize = t; tar->size_fields |= TAR_SIZE_SCHILY_SPARSE_REALSIZE; } return (err); } /* TODO: Is there a SCHILY.sparse.size similar to GNU.sparse.size ? */ else if (key_length > 6 && memcmp(key, "xattr.", 6) == 0) { key_length -= 6; key += 6; if (value_length < xattr_limit) { p = __archive_read_ahead(a, value_length, &bytes_read); if (p == NULL) { archive_set_error(&a->archive, EINVAL, "Truncated archive" " detected while reading SCHILY.xattr"); return (ARCHIVE_FATAL); } if (pax_attribute_SCHILY_xattr(entry, key, key_length, p, value_length)) { /* TODO: Unable to parse xattr */ err = ARCHIVE_WARN; } } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unreasonably large xattr: %llu > %llu", (unsigned long long)value_length, (unsigned long long)xattr_limit); err = ARCHIVE_WARN; } __archive_read_consume(a, value_length); return (err); } } /* SUN.* extensions from Solaris tar */ if (key_length > 4 && memcmp(key, "SUN.", 4) == 0) { key_length -= 4; key += 4; if (key_length == 9 && memcmp(key, "holesdata", 9) == 0) { /* SUN.holesdata */ if (value_length < sparse_map_limit) { p = __archive_read_ahead(a, value_length, &bytes_read); if (p == NULL) { archive_set_error(&a->archive, EINVAL, "Truncated archive" " detected while reading SUN.holesdata"); return (ARCHIVE_FATAL); } err = pax_attribute_SUN_holesdata(a, tar, entry, p, value_length); if (err < ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Parse error: SUN.holesdata"); } } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unreasonably large sparse map: %llu > %llu", (unsigned long long)value_length, (unsigned long long)sparse_map_limit); err = ARCHIVE_FAILED; } __archive_read_consume(a, value_length); return (err); } } break; case 'a': if (key_length == 5 && memcmp(key, "atime", 5) == 0) { if ((err = pax_attribute_read_time(a, value_length, &t, &n, unconsumed)) == ARCHIVE_OK) { archive_entry_set_atime(entry, t, n); } return (err); } break; case 'c': if (key_length == 5 && memcmp(key, "ctime", 5) == 0) { if ((err = pax_attribute_read_time(a, value_length, &t, &n, unconsumed)) == ARCHIVE_OK) { archive_entry_set_ctime(entry, t, n); } return (err); } else if (key_length == 7 && memcmp(key, "charset", 7) == 0) { /* TODO: Publish charset information in entry. */ } else if (key_length == 7 && memcmp(key, "comment", 7) == 0) { /* TODO: Publish comment in entry. */ } break; case 'g': if (key_length == 3 && memcmp(key, "gid", 3) == 0) { if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { archive_entry_set_gid(entry, t); } return (err); } else if (key_length == 5 && memcmp(key, "gname", 5) == 0) { if (value_length > guname_limit) { *unconsumed += value_length; err = ARCHIVE_WARN; } else { err = read_bytes_to_string(a, &(tar->entry_gname), value_length, unconsumed); } return (err); } break; case 'h': if (key_length == 10 && memcmp(key, "hdrcharset", 10) == 0) { if (value_length < 64) { p = __archive_read_ahead(a, value_length, &bytes_read); if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated tar archive " "detected while reading hdrcharset attribute"); return (ARCHIVE_FATAL); } if (value_length == 6 && memcmp(p, "BINARY", 6) == 0) { /* Binary mode. */ tar->pax_hdrcharset_utf8 = 0; err = ARCHIVE_OK; } else if (value_length == 23 && memcmp(p, "ISO-IR 10646 2000 UTF-8", 23) == 0) { tar->pax_hdrcharset_utf8 = 1; err = ARCHIVE_OK; } else { /* TODO: Unrecognized character set */ err = ARCHIVE_WARN; } } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "hdrcharset attribute is unreasonably large (%llu bytes)", (unsigned long long)value_length); err = ARCHIVE_WARN; } __archive_read_consume(a, value_length); return (err); } break; case 'l': /* pax interchange doesn't distinguish hardlink vs. symlink. */ if (key_length == 8 && memcmp(key, "linkpath", 8) == 0) { if (value_length > pathname_limit) { *unconsumed += value_length; err = ARCHIVE_WARN; } else { err = read_bytes_to_string(a, &tar->entry_linkpath, value_length, unconsumed); } return (err); } break; case 'm': if (key_length == 5 && memcmp(key, "mtime", 5) == 0) { if ((err = pax_attribute_read_time(a, value_length, &t, &n, unconsumed)) == ARCHIVE_OK) { archive_entry_set_mtime(entry, t, n); } return (err); } break; case 'p': if (key_length == 4 && memcmp(key, "path", 4) == 0) { if (value_length > pathname_limit) { *unconsumed += value_length; err = ARCHIVE_WARN; } else { err = read_bytes_to_string(a, &(tar->entry_pathname), value_length, unconsumed); } return (err); } break; case 'r': /* POSIX has reserved 'realtime.*' */ break; case 's': /* POSIX has reserved 'security.*' */ /* Someday: if (strcmp(key, "security.acl") == 0) { ... } */ if (key_length == 4 && memcmp(key, "size", 4) == 0) { /* "size" is the size of the data in the entry. */ if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { tar->pax_size = t; tar->size_fields |= TAR_SIZE_PAX_SIZE; } else if (t == INT64_MAX) { /* Note: pax_attr_read_number returns INT64_MAX on overflow or < 0 */ tar->entry_bytes_remaining = 0; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Tar size attribute overflow"); return (ARCHIVE_FATAL); } return (err); } break; case 'u': if (key_length == 3 && memcmp(key, "uid", 3) == 0) { if ((err = pax_attribute_read_number(a, value_length, &t)) == ARCHIVE_OK) { archive_entry_set_uid(entry, t); } return (err); } else if (key_length == 5 && memcmp(key, "uname", 5) == 0) { if (value_length > guname_limit) { *unconsumed += value_length; err = ARCHIVE_WARN; } else { err = read_bytes_to_string(a, &(tar->entry_uname), value_length, unconsumed); } return (err); } break; } /* Unrecognized key, just skip the entire value. */ __archive_read_consume(a, value_length); return (err); } /* * Parse a decimal time value, which may include a fractional portion * * Sets ps to INT64_MIN on error. */ static void pax_time(const char *p, size_t length, int64_t *ps, long *pn) { char digit; int64_t s; unsigned long l; int sign; int64_t limit, last_digit_limit; limit = INT64_MAX / 10; last_digit_limit = INT64_MAX % 10; if (length <= 0) { *ps = 0; *pn = 0; return; } s = 0; sign = 1; if (*p == '-') { sign = -1; p++; length--; } while (length > 0 && *p >= '0' && *p <= '9') { digit = *p - '0'; if (s > limit || (s == limit && digit > last_digit_limit)) { *ps = INT64_MIN; *pn = 0; return; } s = (s * 10) + digit; ++p; --length; } *ps = s * sign; /* Calculate nanoseconds. */ *pn = 0; if (length <= 0 || *p != '.') return; l = 100000000UL; do { ++p; --length; if (length > 0 && *p >= '0' && *p <= '9') *pn += (*p - '0') * l; else break; } while (l /= 10); } /* * Parse GNU tar header */ static int header_gnutar(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, int64_t *unconsumed) { const struct archive_entry_header_gnutar *header; int64_t t; int err = ARCHIVE_OK; /* * GNU header is like POSIX ustar, except 'prefix' is * replaced with some other fields. This also means the * filename is stored as in old-style archives. */ /* Copy filename over (to ensure null termination). */ header = (const struct archive_entry_header_gnutar *)h; const char *existing_pathname = archive_entry_pathname(entry); - if (existing_pathname == NULL || existing_pathname[0] == '\0') { + const wchar_t *existing_wcs_pathname = archive_entry_pathname_w(entry); + if ((existing_pathname == NULL || existing_pathname[0] == '\0') + && (existing_wcs_pathname == NULL || existing_wcs_pathname[0] == L'\0')) { if (archive_entry_copy_pathname_l(entry, header->name, sizeof(header->name), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Pathname"); if (err == ARCHIVE_FATAL) return (err); } } /* Fields common to ustar and GNU */ /* XXX Can the following be factored out since it's common * to ustar and gnu tar? Is it okay to move it down into * header_common, perhaps? */ const char *existing_uname = archive_entry_uname(entry); if (existing_uname == NULL || existing_uname[0] == '\0') { if (archive_entry_copy_uname_l(entry, header->uname, sizeof(header->uname), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Uname"); if (err == ARCHIVE_FATAL) return (err); } } const char *existing_gname = archive_entry_gname(entry); if (existing_gname == NULL || existing_gname[0] == '\0') { if (archive_entry_copy_gname_l(entry, header->gname, sizeof(header->gname), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Gname"); if (err == ARCHIVE_FATAL) return (err); } } /* Parse out device numbers only for char and block specials */ if (header->typeflag[0] == '3' || header->typeflag[0] == '4') { if (!archive_entry_rdev_is_set(entry)) { archive_entry_set_rdevmajor(entry, (dev_t) tar_atol(header->rdevmajor, sizeof(header->rdevmajor))); archive_entry_set_rdevminor(entry, (dev_t) tar_atol(header->rdevminor, sizeof(header->rdevminor))); } } else { archive_entry_set_rdev(entry, 0); } /* Grab GNU-specific fields. */ if (!archive_entry_atime_is_set(entry)) { t = tar_atol(header->atime, sizeof(header->atime)); if (t > 0) archive_entry_set_atime(entry, t, 0); } if (!archive_entry_ctime_is_set(entry)) { t = tar_atol(header->ctime, sizeof(header->ctime)); if (t > 0) archive_entry_set_ctime(entry, t, 0); } if (header->realsize[0] != 0) { /* Treat as a synonym for the pax GNU.sparse.realsize attr */ tar->GNU_sparse_realsize = tar_atol(header->realsize, sizeof(header->realsize)); tar->size_fields |= TAR_SIZE_GNU_SPARSE_REALSIZE; } if (header->sparse[0].offset[0] != 0) { if (gnu_sparse_old_read(a, tar, header, unconsumed) != ARCHIVE_OK) return (ARCHIVE_FATAL); } else { if (header->isextended[0] != 0) { /* XXX WTF? XXX */ } } /* Grab fields common to all tar variants. */ err = header_common(a, tar, entry, h); if (err == ARCHIVE_FATAL) return (err); tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); return (err); } static int gnu_add_sparse_entry(struct archive_read *a, struct tar *tar, int64_t offset, int64_t remaining) { struct sparse_block *p; p = calloc(1, sizeof(*p)); if (p == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } if (tar->sparse_last != NULL) tar->sparse_last->next = p; else tar->sparse_list = p; tar->sparse_last = p; if (remaining < 0 || offset < 0 || offset > INT64_MAX - remaining) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed sparse map data"); return (ARCHIVE_FATAL); } p->offset = offset; p->remaining = remaining; return (ARCHIVE_OK); } static void gnu_clear_sparse_list(struct tar *tar) { struct sparse_block *p; while (tar->sparse_list != NULL) { p = tar->sparse_list; tar->sparse_list = p->next; free(p); } tar->sparse_last = NULL; } /* * GNU tar old-format sparse data. * * GNU old-format sparse data is stored in a fixed-field * format. Offset/size values are 11-byte octal fields (same * format as 'size' field in ustart header). These are * stored in the header, allocating subsequent header blocks * as needed. Extending the header in this way is a pretty * severe POSIX violation; this design has earned GNU tar a * lot of criticism. */ static int gnu_sparse_old_read(struct archive_read *a, struct tar *tar, const struct archive_entry_header_gnutar *header, int64_t *unconsumed) { ssize_t bytes_read; const void *data; struct extended { struct gnu_sparse sparse[21]; char isextended[1]; char padding[7]; }; const struct extended *ext; if (gnu_sparse_old_parse(a, tar, header->sparse, 4) != ARCHIVE_OK) return (ARCHIVE_FATAL); if (header->isextended[0] == 0) return (ARCHIVE_OK); do { - tar_flush_unconsumed(a, unconsumed); + if (tar_flush_unconsumed(a, unconsumed) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } data = __archive_read_ahead(a, 512, &bytes_read); if (data == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated tar archive " "detected while reading sparse file data"); return (ARCHIVE_FATAL); } *unconsumed = 512; ext = (const struct extended *)data; if (gnu_sparse_old_parse(a, tar, ext->sparse, 21) != ARCHIVE_OK) return (ARCHIVE_FATAL); } while (ext->isextended[0] != 0); if (tar->sparse_list != NULL) tar->entry_offset = tar->sparse_list->offset; return (ARCHIVE_OK); } static int gnu_sparse_old_parse(struct archive_read *a, struct tar *tar, const struct gnu_sparse *sparse, int length) { while (length > 0 && sparse->offset[0] != 0) { if (gnu_add_sparse_entry(a, tar, tar_atol(sparse->offset, sizeof(sparse->offset)), tar_atol(sparse->numbytes, sizeof(sparse->numbytes))) != ARCHIVE_OK) return (ARCHIVE_FATAL); sparse++; length--; } return (ARCHIVE_OK); } /* * GNU tar sparse format 0.0 * * Beginning with GNU tar 1.15, sparse files are stored using * information in the pax extended header. The GNU tar maintainers * have gone through a number of variations in the process of working * out this scheme; fortunately, they're all numbered. * * Sparse format 0.0 uses attribute GNU.sparse.numblocks to store the * number of blocks, and GNU.sparse.offset/GNU.sparse.numbytes to * store offset/size for each block. The repeated instances of these * latter fields violate the pax specification (which frowns on * duplicate keys), so this format was quickly replaced. */ /* * GNU tar sparse format 0.1 * * This version replaced the offset/numbytes attributes with * a single "map" attribute that stored a list of integers. This * format had two problems: First, the "map" attribute could be very * long, which caused problems for some implementations. More * importantly, the sparse data was lost when extracted by archivers * that didn't recognize this extension. */ static int gnu_sparse_01_parse(struct archive_read *a, struct tar *tar, const char *p, size_t length) { const char *e; int64_t offset = -1, size = -1; for (;;) { e = p; while (length > 0 && *e != ',') { if (*e < '0' || *e > '9') return (ARCHIVE_WARN); e++; length--; } if (offset < 0) { offset = tar_atol10(p, e - p); if (offset < 0) return (ARCHIVE_WARN); } else { size = tar_atol10(p, e - p); if (size < 0) return (ARCHIVE_WARN); if (gnu_add_sparse_entry(a, tar, offset, size) != ARCHIVE_OK) return (ARCHIVE_FATAL); offset = -1; } if (length == 0) return (ARCHIVE_OK); p = e + 1; length--; } } /* * GNU tar sparse format 1.0 * * The idea: The offset/size data is stored as a series of base-10 * ASCII numbers prepended to the file data, so that dearchivers that * don't support this format will extract the block map along with the * data and a separate post-process can restore the sparseness. * * Unfortunately, GNU tar 1.16 had a bug that added unnecessary * padding to the body of the file when using this format. GNU tar * 1.17 corrected this bug without bumping the version number, so * it's not possible to support both variants. This code supports * the later variant at the expense of not supporting the former. * * This variant also introduced the GNU.sparse.major/GNU.sparse.minor attributes. */ /* * Read the next line from the input, and parse it as a decimal * integer followed by '\n'. Returns positive integer value or * negative on error. */ static int64_t gnu_sparse_10_atol(struct archive_read *a, struct tar *tar, int64_t *remaining, int64_t *unconsumed) { int64_t l, limit, last_digit_limit; const char *p; ssize_t bytes_read; int base, digit; base = 10; limit = INT64_MAX / base; last_digit_limit = INT64_MAX % base; /* * Skip any lines starting with '#'; GNU tar specs * don't require this, but they should. */ do { bytes_read = readline(a, tar, &p, (ssize_t)tar_min(*remaining, 100), unconsumed); if (bytes_read <= 0) return (ARCHIVE_FATAL); *remaining -= bytes_read; } while (p[0] == '#'); l = 0; while (bytes_read > 0) { if (*p == '\n') return (l); if (*p < '0' || *p >= '0' + base) return (ARCHIVE_WARN); digit = *p - '0'; if (l > limit || (l == limit && digit > last_digit_limit)) l = INT64_MAX; /* Truncate on overflow. */ else l = (l * base) + digit; p++; bytes_read--; } /* TODO: Error message. */ return (ARCHIVE_WARN); } /* * Returns length (in bytes) of the sparse data description * that was read. */ static int64_t gnu_sparse_10_read(struct archive_read *a, struct tar *tar, int64_t *unconsumed) { int64_t bytes_read, entries, offset, size, to_skip, remaining; /* Clear out the existing sparse list. */ gnu_clear_sparse_list(tar); remaining = tar->entry_bytes_remaining; /* Parse entries. */ entries = gnu_sparse_10_atol(a, tar, &remaining, unconsumed); if (entries < 0) return (ARCHIVE_FATAL); /* Parse the individual entries. */ while (entries-- > 0) { /* Parse offset/size */ offset = gnu_sparse_10_atol(a, tar, &remaining, unconsumed); if (offset < 0) return (ARCHIVE_FATAL); size = gnu_sparse_10_atol(a, tar, &remaining, unconsumed); if (size < 0) return (ARCHIVE_FATAL); /* Add a new sparse entry. */ if (gnu_add_sparse_entry(a, tar, offset, size) != ARCHIVE_OK) return (ARCHIVE_FATAL); } /* Skip rest of block... */ - tar_flush_unconsumed(a, unconsumed); + if (tar_flush_unconsumed(a, unconsumed) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } bytes_read = tar->entry_bytes_remaining - remaining; to_skip = 0x1ff & -bytes_read; /* Fail if tar->entry_bytes_remaing would get negative */ if (to_skip > remaining) return (ARCHIVE_FATAL); if (to_skip != __archive_read_consume(a, to_skip)) return (ARCHIVE_FATAL); return (bytes_read + to_skip); } /* * Solaris pax extension for a sparse file. This is recorded with the * data and hole pairs. The way recording sparse information by Solaris' * pax simply indicates where data and sparse are, so the stored contents * consist of both data and hole. */ static int pax_attribute_SUN_holesdata(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const char *p, size_t length) { const char *e; int64_t start, end; int hole = 1; (void)entry; /* UNUSED */ end = 0; if (length <= 0) return (ARCHIVE_WARN); if (*p == ' ') { p++; length--; } else { return (ARCHIVE_WARN); } for (;;) { e = p; while (length > 0 && *e != ' ') { if (*e < '0' || *e > '9') return (ARCHIVE_WARN); e++; length--; } start = end; end = tar_atol10(p, e - p); if (end < 0) return (ARCHIVE_WARN); if (start < end) { if (gnu_add_sparse_entry(a, tar, start, end - start) != ARCHIVE_OK) return (ARCHIVE_FATAL); tar->sparse_last->hole = hole; } if (length == 0 || *e == '\n') { if (length == 0 && *e == '\n') { return (ARCHIVE_OK); } else { return (ARCHIVE_WARN); } } p = e + 1; length--; hole = hole == 0; } } /*- * Convert text->integer. * * Traditional tar formats (including POSIX) specify base-8 for * all of the standard numeric fields. This is a significant limitation * in practice: * = file size is limited to 8GB * = rdevmajor and rdevminor are limited to 21 bits * = uid/gid are limited to 21 bits * * There are two workarounds for this: * = pax extended headers, which use variable-length string fields * = GNU tar and STAR both allow either base-8 or base-256 in * most fields. The high bit is set to indicate base-256. * * On read, this implementation supports both extensions. */ static int64_t tar_atol(const char *p, size_t char_cnt) { /* * Technically, GNU tar considers a field to be in base-256 * only if the first byte is 0xff or 0x80. */ if (*p & 0x80) return (tar_atol256(p, char_cnt)); return (tar_atol8(p, char_cnt)); } /* * Note that this implementation does not (and should not!) obey * locale settings; you cannot simply substitute strtol here, since * it does obey locale. */ static int64_t tar_atol_base_n(const char *p, size_t char_cnt, int base) { int64_t l, maxval, limit, last_digit_limit; int digit, sign; maxval = INT64_MAX; limit = INT64_MAX / base; last_digit_limit = INT64_MAX % base; /* the pointer will not be dereferenced if char_cnt is zero * due to the way the && operator is evaluated. */ while (char_cnt != 0 && (*p == ' ' || *p == '\t')) { p++; char_cnt--; } sign = 1; if (char_cnt != 0 && *p == '-') { sign = -1; p++; char_cnt--; maxval = INT64_MIN; limit = -(INT64_MIN / base); last_digit_limit = -(INT64_MIN % base); } l = 0; if (char_cnt != 0) { digit = *p - '0'; while (digit >= 0 && digit < base && char_cnt != 0) { if (l>limit || (l == limit && digit >= last_digit_limit)) { return maxval; /* Truncate on overflow. */ } l = (l * base) + digit; digit = *++p - '0'; char_cnt--; } } return (sign < 0) ? -l : l; } static int64_t tar_atol8(const char *p, size_t char_cnt) { return tar_atol_base_n(p, char_cnt, 8); } static int64_t tar_atol10(const char *p, size_t char_cnt) { return tar_atol_base_n(p, char_cnt, 10); } /* * Parse a base-256 integer. This is just a variable-length * twos-complement signed binary value in big-endian order, except * that the high-order bit is ignored. The values here can be up to * 12 bytes, so we need to be careful about overflowing 64-bit * (8-byte) integers. * * This code unashamedly assumes that the local machine uses 8-bit * bytes and twos-complement arithmetic. */ static int64_t tar_atol256(const char *_p, size_t char_cnt) { uint64_t l; const unsigned char *p = (const unsigned char *)_p; unsigned char c, neg; /* Extend 7-bit 2s-comp to 8-bit 2s-comp, decide sign. */ c = *p; if (c & 0x40) { neg = 0xff; c |= 0x80; l = ~ARCHIVE_LITERAL_ULL(0); } else { neg = 0; c &= 0x7f; l = 0; } /* If more than 8 bytes, check that we can ignore * high-order bits without overflow. */ while (char_cnt > sizeof(int64_t)) { --char_cnt; if (c != neg) return neg ? INT64_MIN : INT64_MAX; c = *++p; } /* c is first byte that fits; if sign mismatch, return overflow */ if ((c ^ neg) & 0x80) { return neg ? INT64_MIN : INT64_MAX; } /* Accumulate remaining bytes. */ while (--char_cnt > 0) { l = (l << 8) | c; c = *++p; } l = (l << 8) | c; /* Return signed twos-complement value. */ return (int64_t)(l); } /* * Returns length of line (including trailing newline) * or negative on error. 'start' argument is updated to * point to first character of line. This avoids copying * when possible. */ static ssize_t readline(struct archive_read *a, struct tar *tar, const char **start, ssize_t limit, int64_t *unconsumed) { ssize_t bytes_read; ssize_t total_size = 0; const void *t; const char *s; void *p; - tar_flush_unconsumed(a, unconsumed); + if (tar_flush_unconsumed(a, unconsumed) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } t = __archive_read_ahead(a, 1, &bytes_read); if (bytes_read <= 0 || t == NULL) return (ARCHIVE_FATAL); s = t; /* Start of line? */ p = memchr(t, '\n', bytes_read); /* If we found '\n' in the read buffer, return pointer to that. */ if (p != NULL) { bytes_read = 1 + ((const char *)p) - s; if (bytes_read > limit) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Line too long"); return (ARCHIVE_FATAL); } *unconsumed = bytes_read; *start = s; return (bytes_read); } *unconsumed = bytes_read; /* Otherwise, we need to accumulate in a line buffer. */ for (;;) { if (total_size + bytes_read > limit) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Line too long"); return (ARCHIVE_FATAL); } if (archive_string_ensure(&tar->line, total_size + bytes_read) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate working buffer"); return (ARCHIVE_FATAL); } memcpy(tar->line.s + total_size, t, bytes_read); tar_flush_unconsumed(a, unconsumed); total_size += bytes_read; /* If we found '\n', clean up and return. */ if (p != NULL) { *start = tar->line.s; return (total_size); } /* Read some more. */ t = __archive_read_ahead(a, 1, &bytes_read); if (bytes_read <= 0 || t == NULL) return (ARCHIVE_FATAL); s = t; /* Start of line? */ p = memchr(t, '\n', bytes_read); /* If we found '\n', trim the read. */ if (p != NULL) { bytes_read = 1 + ((const char *)p) - s; } *unconsumed = bytes_read; } } /* * base64_decode - Base64 decode * * This accepts most variations of base-64 encoding, including: * * with or without line breaks * * with or without the final group padded with '=' or '_' characters * (The most economical Base-64 variant does not pad the last group and * omits line breaks; RFC1341 used for MIME requires both.) */ static char * base64_decode(const char *s, size_t len, size_t *out_len) { static const unsigned char digits[64] = { 'A','B','C','D','E','F','G','H','I','J','K','L','M','N', 'O','P','Q','R','S','T','U','V','W','X','Y','Z','a','b', 'c','d','e','f','g','h','i','j','k','l','m','n','o','p', 'q','r','s','t','u','v','w','x','y','z','0','1','2','3', '4','5','6','7','8','9','+','/' }; static unsigned char decode_table[128]; char *out, *d; const unsigned char *src = (const unsigned char *)s; /* If the decode table is not yet initialized, prepare it. */ if (decode_table[digits[1]] != 1) { unsigned i; memset(decode_table, 0xff, sizeof(decode_table)); for (i = 0; i < sizeof(digits); i++) decode_table[digits[i]] = i; } /* Allocate enough space to hold the entire output. */ /* Note that we may not use all of this... */ out = malloc(len - len / 4 + 1); if (out == NULL) { *out_len = 0; return (NULL); } d = out; while (len > 0) { /* Collect the next group of (up to) four characters. */ int v = 0; int group_size = 0; while (group_size < 4 && len > 0) { /* '=' or '_' padding indicates final group. */ if (*src == '=' || *src == '_') { len = 0; break; } /* Skip illegal characters (including line breaks) */ if (*src > 127 || *src < 32 || decode_table[*src] == 0xff) { len--; src++; continue; } v <<= 6; v |= decode_table[*src++]; len --; group_size++; } /* Align a short group properly. */ v <<= 6 * (4 - group_size); /* Unpack the group we just collected. */ switch (group_size) { case 4: d[2] = v & 0xff; /* FALLTHROUGH */ case 3: d[1] = (v >> 8) & 0xff; /* FALLTHROUGH */ case 2: d[0] = (v >> 16) & 0xff; break; case 1: /* this is invalid! */ break; } d += group_size * 3 / 4; } *out_len = d - out; return (out); } static char * url_decode(const char *in, size_t length) { char *out, *d; const char *s; out = malloc(length + 1); if (out == NULL) return (NULL); for (s = in, d = out; length > 0 && *s != '\0'; ) { if (s[0] == '%' && length > 2) { /* Try to convert % escape */ int digit1 = tohex(s[1]); int digit2 = tohex(s[2]); if (digit1 >= 0 && digit2 >= 0) { /* Looks good, consume three chars */ s += 3; length -= 3; /* Convert output */ *d++ = ((digit1 << 4) | digit2); continue; } /* Else fall through and treat '%' as normal char */ } *d++ = *s++; --length; } *d = '\0'; return (out); } static int tohex(int c) { if (c >= '0' && c <= '9') return (c - '0'); else if (c >= 'A' && c <= 'F') return (c - 'A' + 10); else if (c >= 'a' && c <= 'f') return (c - 'a' + 10); else return (-1); } diff --git a/contrib/libarchive/libarchive/archive_read_support_format_warc.c b/contrib/libarchive/libarchive/archive_read_support_format_warc.c index 696f959c341d..d8f188cf0b44 100644 --- a/contrib/libarchive/libarchive/archive_read_support_format_warc.c +++ b/contrib/libarchive/libarchive/archive_read_support_format_warc.c @@ -1,858 +1,858 @@ /*- * Copyright (c) 2014 Sebastian Freundt * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" /** * WARC is standardised by ISO TC46/SC4/WG12 and currently available as * ISO 28500:2009. * For the purposes of this file we used the final draft from: * http://bibnum.bnf.fr/warc/WARC_ISO_28500_version1_latestdraft.pdf * * Todo: * [ ] real-world warcs can contain resources at endpoints ending in / * e.g. http://bibnum.bnf.fr/warc/ * if you're lucky their response contains a Content-Location: header * pointing to a unix-compliant filename, in the example above it's * Content-Location: http://bibnum.bnf.fr/warc/index.html * however, that's not mandated and github for example doesn't follow * this convention. * We need a set of archive options to control what to do with * entries like these, at the moment care is taken to skip them. * **/ #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_CTYPE_H #include #endif #ifdef HAVE_TIME_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_read_private.h" typedef enum { WT_NONE, /* warcinfo */ WT_INFO, /* metadata */ WT_META, /* resource */ WT_RSRC, /* request, unsupported */ WT_REQ, /* response, unsupported */ WT_RSP, /* revisit, unsupported */ WT_RVIS, /* conversion, unsupported */ WT_CONV, /* continuation, unsupported at the moment */ WT_CONT, /* invalid type */ LAST_WT } warc_type_t; typedef struct { size_t len; const char *str; } warc_string_t; typedef struct { size_t len; char *str; } warc_strbuf_t; struct warc_s { /* content length ahead */ size_t cntlen; /* and how much we've processed so far */ size_t cntoff; /* and how much we need to consume between calls */ size_t unconsumed; /* string pool */ warc_strbuf_t pool; /* previous version */ unsigned int pver; /* stringified format name */ struct archive_string sver; }; static int _warc_bid(struct archive_read *a, int); static int _warc_cleanup(struct archive_read *a); static int _warc_read(struct archive_read*, const void**, size_t*, int64_t*); static int _warc_skip(struct archive_read *a); static int _warc_rdhdr(struct archive_read *a, struct archive_entry *e); /* private routines */ static unsigned int _warc_rdver(const char *buf, size_t bsz); static unsigned int _warc_rdtyp(const char *buf, size_t bsz); static warc_string_t _warc_rduri(const char *buf, size_t bsz); static ssize_t _warc_rdlen(const char *buf, size_t bsz); static time_t _warc_rdrtm(const char *buf, size_t bsz); static time_t _warc_rdmtm(const char *buf, size_t bsz); static const char *_warc_find_eoh(const char *buf, size_t bsz); static const char *_warc_find_eol(const char *buf, size_t bsz); int archive_read_support_format_warc(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct warc_s *w; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_warc"); if ((w = calloc(1, sizeof(*w))) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate warc data"); return (ARCHIVE_FATAL); } r = __archive_read_register_format( a, w, "warc", _warc_bid, NULL, _warc_rdhdr, _warc_read, _warc_skip, NULL, _warc_cleanup, NULL, NULL); if (r != ARCHIVE_OK) { free(w); return (r); } return (ARCHIVE_OK); } static int _warc_cleanup(struct archive_read *a) { struct warc_s *w = a->format->data; if (w->pool.len > 0U) { free(w->pool.str); } archive_string_free(&w->sver); free(w); a->format->data = NULL; return (ARCHIVE_OK); } static int _warc_bid(struct archive_read *a, int best_bid) { const char *hdr; ssize_t nrd; unsigned int ver; (void)best_bid; /* UNUSED */ /* check first line of file, it should be a record already */ if ((hdr = __archive_read_ahead(a, 12U, &nrd)) == NULL) { /* no idea what to do */ return -1; } else if (nrd < 12) { /* nah, not for us, our magic cookie is at least 12 bytes */ return -1; } /* otherwise snarf the record's version number */ ver = _warc_rdver(hdr, nrd); if (ver < 1200U || ver > 10000U) { /* we only support WARC 0.12 to 1.0 */ return -1; } /* otherwise be confident */ return (64); } static int _warc_rdhdr(struct archive_read *a, struct archive_entry *entry) { #define HDR_PROBE_LEN (12U) struct warc_s *w = a->format->data; unsigned int ver; const char *buf; ssize_t nrd; const char *eoh; char *tmp; /* for the file name, saves some strndup()'ing */ warc_string_t fnam; /* warc record type, not that we really use it a lot */ warc_type_t ftyp; /* content-length+error monad */ ssize_t cntlen; /* record time is the WARC-Date time we reinterpret it as ctime */ time_t rtime; /* mtime is the Last-Modified time which will be the entry's mtime */ time_t mtime; start_over: /* just use read_ahead() they keep track of unconsumed * bits and bobs for us; no need to put an extra shift in * and reproduce that functionality here */ buf = __archive_read_ahead(a, HDR_PROBE_LEN, &nrd); if (nrd < 0) { /* no good */ archive_set_error( &a->archive, ARCHIVE_ERRNO_MISC, "Bad record header"); return (ARCHIVE_FATAL); } else if (buf == NULL) { /* there should be room for at least WARC/bla\r\n * must be EOF therefore */ return (ARCHIVE_EOF); } /* looks good so far, try and find the end of the header now */ eoh = _warc_find_eoh(buf, nrd); if (eoh == NULL) { /* still no good, the header end might be beyond the * probe we've requested, but then again who'd cram * so much stuff into the header *and* be 28500-compliant */ archive_set_error( &a->archive, ARCHIVE_ERRNO_MISC, "Bad record header"); return (ARCHIVE_FATAL); } ver = _warc_rdver(buf, eoh - buf); /* we currently support WARC 0.12 to 1.0 */ if (ver == 0U) { archive_set_error( &a->archive, ARCHIVE_ERRNO_MISC, "Invalid record version"); return (ARCHIVE_FATAL); } else if (ver < 1200U || ver > 10000U) { archive_set_error( &a->archive, ARCHIVE_ERRNO_MISC, "Unsupported record version: %u.%u", ver / 10000, (ver % 10000) / 100); return (ARCHIVE_FATAL); } cntlen = _warc_rdlen(buf, eoh - buf); if (cntlen < 0) { /* nightmare! the specs say content-length is mandatory * so I don't feel overly bad stopping the reader here */ archive_set_error( &a->archive, EINVAL, "Bad content length"); return (ARCHIVE_FATAL); } rtime = _warc_rdrtm(buf, eoh - buf); if (rtime == (time_t)-1) { /* record time is mandatory as per WARC/1.0, * so just barf here, fast and loud */ archive_set_error( &a->archive, EINVAL, "Bad record time"); return (ARCHIVE_FATAL); } /* let the world know we're a WARC archive */ a->archive.archive_format = ARCHIVE_FORMAT_WARC; if (ver != w->pver) { /* stringify this entry's version */ archive_string_sprintf(&w->sver, "WARC/%u.%u", ver / 10000, (ver % 10000) / 100); /* remember the version */ w->pver = ver; } /* start off with the type */ ftyp = _warc_rdtyp(buf, eoh - buf); /* and let future calls know about the content */ w->cntlen = cntlen; w->cntoff = 0U; mtime = 0;/* Avoid compiling error on some platform. */ switch (ftyp) { case WT_RSRC: case WT_RSP: /* only try and read the filename in the cases that are * guaranteed to have one */ fnam = _warc_rduri(buf, eoh - buf); /* check the last character in the URI to avoid creating * directory endpoints as files, see Todo above */ if (fnam.len == 0 || fnam.str[fnam.len - 1] == '/') { /* break here for now */ fnam.len = 0U; fnam.str = NULL; break; } /* bang to our string pool, so we save a * malloc()+free() roundtrip */ if (fnam.len + 1U > w->pool.len) { w->pool.len = ((fnam.len + 64U) / 64U) * 64U; tmp = realloc(w->pool.str, w->pool.len); if (tmp == NULL) { archive_set_error( &a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } w->pool.str = tmp; } memcpy(w->pool.str, fnam.str, fnam.len); w->pool.str[fnam.len] = '\0'; /* let no one else know about the pool, it's a secret, shhh */ fnam.str = w->pool.str; /* snarf mtime or deduce from rtime * this is a custom header added by our writer, it's quite * hard to believe anyone else would go through with it * (apart from being part of some http responses of course) */ if ((mtime = _warc_rdmtm(buf, eoh - buf)) == (time_t)-1) { mtime = rtime; } break; case WT_NONE: case WT_INFO: case WT_META: case WT_REQ: case WT_RVIS: case WT_CONV: case WT_CONT: case LAST_WT: default: fnam.len = 0U; fnam.str = NULL; break; } /* now eat some of those delicious buffer bits */ __archive_read_consume(a, eoh - buf); switch (ftyp) { case WT_RSRC: case WT_RSP: if (fnam.len > 0U) { /* populate entry object */ archive_entry_set_filetype(entry, AE_IFREG); archive_entry_copy_pathname(entry, fnam.str); archive_entry_set_size(entry, cntlen); archive_entry_set_perm(entry, 0644); /* rtime is the new ctime, mtime stays mtime */ archive_entry_set_ctime(entry, rtime, 0L); archive_entry_set_mtime(entry, mtime, 0L); break; } /* FALLTHROUGH */ case WT_NONE: case WT_INFO: case WT_META: case WT_REQ: case WT_RVIS: case WT_CONV: case WT_CONT: case LAST_WT: default: /* consume the content and start over */ if (_warc_skip(a) < 0) return (ARCHIVE_FATAL); goto start_over; } return (ARCHIVE_OK); } static int _warc_read(struct archive_read *a, const void **buf, size_t *bsz, int64_t *off) { struct warc_s *w = a->format->data; const char *rab; ssize_t nrd; if (w->cntoff >= w->cntlen) { eof: /* it's our lucky day, no work, we can leave early */ *buf = NULL; *bsz = 0U; - *off = w->cntoff + 4U/*for \r\n\r\n separator*/; + *off = w->cntoff; w->unconsumed = 0U; return (ARCHIVE_EOF); } if (w->unconsumed) { __archive_read_consume(a, w->unconsumed); w->unconsumed = 0U; } rab = __archive_read_ahead(a, 1U, &nrd); if (nrd < 0) { *bsz = 0U; /* big catastrophe */ return (int)nrd; } else if (nrd == 0) { goto eof; } else if ((size_t)nrd > w->cntlen - w->cntoff) { /* clamp to content-length */ nrd = w->cntlen - w->cntoff; } *off = w->cntoff; *bsz = nrd; *buf = rab; w->cntoff += nrd; w->unconsumed = (size_t)nrd; return (ARCHIVE_OK); } static int _warc_skip(struct archive_read *a) { struct warc_s *w = a->format->data; if (__archive_read_consume(a, w->cntlen) < 0 || __archive_read_consume(a, 4U/*\r\n\r\n separator*/) < 0) return (ARCHIVE_FATAL); w->cntlen = 0U; w->cntoff = 0U; return (ARCHIVE_OK); } /* private routines */ static void* deconst(const void *c) { return (void *)(uintptr_t)c; } static char* xmemmem(const char *hay, const size_t haysize, const char *needle, const size_t needlesize) { const char *const eoh = hay + haysize; const char *const eon = needle + needlesize; const char *hp; const char *np; const char *cand; unsigned int hsum; unsigned int nsum; unsigned int eqp; /* trivial checks first * a 0-sized needle is defined to be found anywhere in haystack * then run strchr() to find a candidate in HAYSTACK (i.e. a portion * that happens to begin with *NEEDLE) */ if (needlesize == 0UL) { return deconst(hay); } else if ((hay = memchr(hay, *needle, haysize)) == NULL) { /* trivial */ return NULL; } /* First characters of haystack and needle are the same now. Both are * guaranteed to be at least one character long. Now computes the sum * of characters values of needle together with the sum of the first * needle_len characters of haystack. */ for (hp = hay + 1U, np = needle + 1U, hsum = *hay, nsum = *hay, eqp = 1U; hp < eoh && np < eon; hsum ^= *hp, nsum ^= *np, eqp &= *hp == *np, hp++, np++); /* HP now references the (NEEDLESIZE + 1)-th character. */ if (np < eon) { /* haystack is smaller than needle, :O */ return NULL; } else if (eqp) { /* found a match */ return deconst(hay); } /* now loop through the rest of haystack, * updating the sum iteratively */ for (cand = hay; hp < eoh; hp++) { hsum ^= *cand++; hsum ^= *hp; /* Since the sum of the characters is already known to be * equal at that point, it is enough to check just NEEDLESIZE - 1 * characters for equality, * also CAND is by design < HP, so no need for range checks */ if (hsum == nsum && memcmp(cand, needle, needlesize - 1U) == 0) { return deconst(cand); } } return NULL; } static int strtoi_lim(const char *str, const char **ep, int llim, int ulim) { int res = 0; const char *sp; /* we keep track of the number of digits via rulim */ int rulim; for (sp = str, rulim = ulim > 10 ? ulim : 10; res * 10 <= ulim && rulim && *sp >= '0' && *sp <= '9'; sp++, rulim /= 10) { res *= 10; res += *sp - '0'; } if (sp == str) { res = -1; } else if (res < llim || res > ulim) { res = -2; } *ep = (const char*)sp; return res; } static time_t time_from_tm(struct tm *t) { #if HAVE__MKGMTIME return _mkgmtime(t); #elif HAVE_TIMEGM /* Use platform timegm() if available. */ return (timegm(t)); #else /* Else use direct calculation using POSIX assumptions. */ /* First, fix up tm_yday based on the year/month/day. */ if (mktime(t) == (time_t)-1) return ((time_t)-1); /* Then we can compute timegm() from first principles. */ return (t->tm_sec + t->tm_min * 60 + t->tm_hour * 3600 + t->tm_yday * 86400 + (t->tm_year - 70) * 31536000 + ((t->tm_year - 69) / 4) * 86400 - ((t->tm_year - 1) / 100) * 86400 + ((t->tm_year + 299) / 400) * 86400); #endif } static time_t xstrpisotime(const char *s, char **endptr) { /** like strptime() but strictly for ISO 8601 Zulu strings */ struct tm tm; time_t res = (time_t)-1; /* make sure tm is clean */ memset(&tm, 0, sizeof(tm)); /* as a courtesy to our callers, and since this is a non-standard * routine, we skip leading whitespace */ while (*s == ' ' || *s == '\t') ++s; /* read year */ if ((tm.tm_year = strtoi_lim(s, &s, 1583, 4095)) < 0 || *s++ != '-') { goto out; } /* read month */ if ((tm.tm_mon = strtoi_lim(s, &s, 1, 12)) < 0 || *s++ != '-') { goto out; } /* read day-of-month */ if ((tm.tm_mday = strtoi_lim(s, &s, 1, 31)) < 0 || *s++ != 'T') { goto out; } /* read hour */ if ((tm.tm_hour = strtoi_lim(s, &s, 0, 23)) < 0 || *s++ != ':') { goto out; } /* read minute */ if ((tm.tm_min = strtoi_lim(s, &s, 0, 59)) < 0 || *s++ != ':') { goto out; } /* read second */ if ((tm.tm_sec = strtoi_lim(s, &s, 0, 60)) < 0 || *s++ != 'Z') { goto out; } /* massage TM to fulfill some of POSIX' constraints */ tm.tm_year -= 1900; tm.tm_mon--; /* now convert our custom tm struct to a unix stamp using UTC */ res = time_from_tm(&tm); out: if (endptr != NULL) { *endptr = deconst(s); } return res; } static unsigned int _warc_rdver(const char *buf, size_t bsz) { static const char magic[] = "WARC/"; const char *c; unsigned int ver = 0U; unsigned int end = 0U; if (bsz < 12 || memcmp(buf, magic, sizeof(magic) - 1U) != 0) { /* buffer too small or invalid magic */ return ver; } /* looks good so far, read the version number for a laugh */ buf += sizeof(magic) - 1U; if (isdigit((unsigned char)buf[0U]) && (buf[1U] == '.') && isdigit((unsigned char)buf[2U])) { /* we support a maximum of 2 digits in the minor version */ if (isdigit((unsigned char)buf[3U])) end = 1U; /* set up major version */ ver = (buf[0U] - '0') * 10000U; /* set up minor version */ if (end == 1U) { ver += (buf[2U] - '0') * 1000U; ver += (buf[3U] - '0') * 100U; } else ver += (buf[2U] - '0') * 100U; /* * WARC below version 0.12 has a space-separated header * WARC 0.12 and above terminates the version with a CRLF */ c = buf + 3U + end; if (ver >= 1200U) { if (memcmp(c, "\r\n", 2U) != 0) ver = 0U; } else { /* ver < 1200U */ if (*c != ' ' && *c != '\t') ver = 0U; } } return ver; } static unsigned int _warc_rdtyp(const char *buf, size_t bsz) { static const char _key[] = "\r\nWARC-Type:"; const char *val, *eol; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return WT_NONE; } val += sizeof(_key) - 1U; if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL) { /* no end of line */ return WT_NONE; } /* overread whitespace */ while (val < eol && (*val == ' ' || *val == '\t')) ++val; if (val + 8U == eol) { if (memcmp(val, "resource", 8U) == 0) return WT_RSRC; else if (memcmp(val, "response", 8U) == 0) return WT_RSP; } return WT_NONE; } static warc_string_t _warc_rduri(const char *buf, size_t bsz) { static const char _key[] = "\r\nWARC-Target-URI:"; const char *val, *uri, *eol, *p; warc_string_t res = {0U, NULL}; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return res; } /* overread whitespace */ val += sizeof(_key) - 1U; if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL) { /* no end of line */ return res; } while (val < eol && (*val == ' ' || *val == '\t')) ++val; /* overread URL designators */ if ((uri = xmemmem(val, eol - val, "://", 3U)) == NULL) { /* not touching that! */ return res; } /* spaces inside uri are not allowed, CRLF should follow */ for (p = val; p < eol; p++) { if (isspace((unsigned char)*p)) return res; } /* there must be at least space for ftp */ if (uri < (val + 3U)) return res; /* move uri to point to after :// */ uri += 3U; /* now then, inspect the URI */ if (memcmp(val, "file", 4U) == 0) { /* perfect, nothing left to do here */ } else if (memcmp(val, "http", 4U) == 0 || memcmp(val, "ftp", 3U) == 0) { /* overread domain, and the first / */ while (uri < eol && *uri++ != '/'); } else { /* not sure what to do? best to bugger off */ return res; } res.str = uri; res.len = eol - uri; return res; } static ssize_t _warc_rdlen(const char *buf, size_t bsz) { static const char _key[] = "\r\nContent-Length:"; const char *val, *eol; char *on = NULL; long int len; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return -1; } val += sizeof(_key) - 1U; if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL) { /* no end of line */ return -1; } /* skip leading whitespace */ while (val < eol && (*val == ' ' || *val == '\t')) val++; /* there must be at least one digit */ if (!isdigit((unsigned char)*val)) return -1; errno = 0; len = strtol(val, &on, 10); if (errno != 0 || on != eol) { /* line must end here */ return -1; } return (size_t)len; } static time_t _warc_rdrtm(const char *buf, size_t bsz) { static const char _key[] = "\r\nWARC-Date:"; const char *val, *eol; char *on = NULL; time_t res; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return (time_t)-1; } val += sizeof(_key) - 1U; if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL ) { /* no end of line */ return -1; } /* xstrpisotime() kindly overreads whitespace for us, so use that */ res = xstrpisotime(val, &on); if (on != eol) { /* line must end here */ return -1; } return res; } static time_t _warc_rdmtm(const char *buf, size_t bsz) { static const char _key[] = "\r\nLast-Modified:"; const char *val, *eol; char *on = NULL; time_t res; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return (time_t)-1; } val += sizeof(_key) - 1U; if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL ) { /* no end of line */ return -1; } /* xstrpisotime() kindly overreads whitespace for us, so use that */ res = xstrpisotime(val, &on); if (on != eol) { /* line must end here */ return -1; } return res; } static const char* _warc_find_eoh(const char *buf, size_t bsz) { static const char _marker[] = "\r\n\r\n"; const char *hit = xmemmem(buf, bsz, _marker, sizeof(_marker) - 1U); if (hit != NULL) { hit += sizeof(_marker) - 1U; } return hit; } static const char* _warc_find_eol(const char *buf, size_t bsz) { static const char _marker[] = "\r\n"; const char *hit = xmemmem(buf, bsz, _marker, sizeof(_marker) - 1U); return hit; } /* archive_read_support_format_warc.c ends here */ diff --git a/contrib/libarchive/libarchive/archive_read_support_format_xar.c b/contrib/libarchive/libarchive/archive_read_support_format_xar.c index b4e1192ef771..36b5ab3ae04c 100644 --- a/contrib/libarchive/libarchive/archive_read_support_format_xar.c +++ b/contrib/libarchive/libarchive/archive_read_support_format_xar.c @@ -1,3667 +1,3667 @@ /*- * Copyright (c) 2009 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #if HAVE_LIBXML_XMLREADER_H #include #elif HAVE_BSDXML_H #include #elif HAVE_EXPAT_H #include #elif HAVE_XMLLITE_H #include #include #include #endif #ifdef HAVE_BZLIB_H #include #endif #if HAVE_LZMA_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #include "archive.h" #include "archive_digest_private.h" #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_read_private.h" #if (!defined(HAVE_LIBXML_XMLREADER_H) && \ !defined(HAVE_BSDXML_H) && !defined(HAVE_EXPAT_H) && \ !defined(HAVE_XMLLITE_H)) ||\ !defined(HAVE_ZLIB_H) || \ !defined(ARCHIVE_HAS_MD5) || !defined(ARCHIVE_HAS_SHA1) /* * xar needs several external libraries. * o libxml2, expat or (Windows only) xmllite --- XML parser * o openssl or MD5/SHA1 hash function * o zlib * o bzlib2 (option) * o liblzma (option) */ int archive_read_support_format_xar(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_xar"); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Xar not supported on this platform"); return (ARCHIVE_WARN); } #else /* Support xar format */ /* #define DEBUG 1 */ /* #define DEBUG_PRINT_TOC 1 */ #if DEBUG_PRINT_TOC #define PRINT_TOC(d, outbytes) do { \ unsigned char *x = (unsigned char *)(uintptr_t)d; \ unsigned char c = x[outbytes-1]; \ x[outbytes - 1] = 0; \ fprintf(stderr, "%s", x); \ fprintf(stderr, "%c", c); \ x[outbytes - 1] = c; \ } while (0) #else #define PRINT_TOC(d, outbytes) #endif #define HEADER_MAGIC 0x78617221 #define HEADER_SIZE 28 #define HEADER_VERSION 1 #define CKSUM_NONE 0 #define CKSUM_SHA1 1 #define CKSUM_MD5 2 #define MD5_SIZE 16 #define SHA1_SIZE 20 #define MAX_SUM_SIZE 20 enum enctype { NONE, GZIP, BZIP2, LZMA, XZ, }; struct chksumval { int alg; size_t len; unsigned char val[MAX_SUM_SIZE]; }; struct chksumwork { int alg; #ifdef ARCHIVE_HAS_MD5 archive_md5_ctx md5ctx; #endif #ifdef ARCHIVE_HAS_SHA1 archive_sha1_ctx sha1ctx; #endif }; struct xattr { struct xattr *next; struct archive_string name; uint64_t id; uint64_t length; uint64_t offset; uint64_t size; enum enctype encoding; struct chksumval a_sum; struct chksumval e_sum; struct archive_string fstype; }; struct xar_file { struct xar_file *next; struct xar_file *hdnext; struct xar_file *parent; int subdirs; unsigned int has; #define HAS_DATA 0x00001 #define HAS_PATHNAME 0x00002 #define HAS_SYMLINK 0x00004 #define HAS_TIME 0x00008 #define HAS_UID 0x00010 #define HAS_GID 0x00020 #define HAS_MODE 0x00040 #define HAS_TYPE 0x00080 #define HAS_DEV 0x00100 #define HAS_DEVMAJOR 0x00200 #define HAS_DEVMINOR 0x00400 #define HAS_INO 0x00800 #define HAS_FFLAGS 0x01000 #define HAS_XATTR 0x02000 #define HAS_ACL 0x04000 #define HAS_CTIME 0x08000 #define HAS_MTIME 0x10000 #define HAS_ATIME 0x20000 uint64_t id; uint64_t length; uint64_t offset; uint64_t size; enum enctype encoding; struct chksumval a_sum; struct chksumval e_sum; struct archive_string pathname; struct archive_string symlink; time_t ctime; time_t mtime; time_t atime; struct archive_string uname; int64_t uid; struct archive_string gname; int64_t gid; mode_t mode; dev_t dev; dev_t devmajor; dev_t devminor; int64_t ino64; struct archive_string fflags_text; unsigned int link; unsigned int nlink; struct archive_string hardlink; struct xattr *xattr_list; }; struct hdlink { struct hdlink *next; unsigned int id; int cnt; struct xar_file *files; }; struct heap_queue { struct xar_file **files; int allocated; int used; }; enum xmlstatus { INIT, XAR, TOC, TOC_CREATION_TIME, TOC_CHECKSUM, TOC_CHECKSUM_OFFSET, TOC_CHECKSUM_SIZE, TOC_FILE, FILE_DATA, FILE_DATA_LENGTH, FILE_DATA_OFFSET, FILE_DATA_SIZE, FILE_DATA_ENCODING, FILE_DATA_A_CHECKSUM, FILE_DATA_E_CHECKSUM, FILE_DATA_CONTENT, FILE_EA, FILE_EA_LENGTH, FILE_EA_OFFSET, FILE_EA_SIZE, FILE_EA_ENCODING, FILE_EA_A_CHECKSUM, FILE_EA_E_CHECKSUM, FILE_EA_NAME, FILE_EA_FSTYPE, FILE_CTIME, FILE_MTIME, FILE_ATIME, FILE_GROUP, FILE_GID, FILE_USER, FILE_UID, FILE_MODE, FILE_DEVICE, FILE_DEVICE_MAJOR, FILE_DEVICE_MINOR, FILE_DEVICENO, FILE_INODE, FILE_LINK, FILE_TYPE, FILE_NAME, FILE_ACL, FILE_ACL_DEFAULT, FILE_ACL_ACCESS, FILE_ACL_APPLEEXTENDED, /* BSD file flags. */ FILE_FLAGS, FILE_FLAGS_USER_NODUMP, FILE_FLAGS_USER_IMMUTABLE, FILE_FLAGS_USER_APPEND, FILE_FLAGS_USER_OPAQUE, FILE_FLAGS_USER_NOUNLINK, FILE_FLAGS_SYS_ARCHIVED, FILE_FLAGS_SYS_IMMUTABLE, FILE_FLAGS_SYS_APPEND, FILE_FLAGS_SYS_NOUNLINK, FILE_FLAGS_SYS_SNAPSHOT, /* Linux file flags. */ FILE_EXT2, FILE_EXT2_SecureDeletion, FILE_EXT2_Undelete, FILE_EXT2_Compress, FILE_EXT2_Synchronous, FILE_EXT2_Immutable, FILE_EXT2_AppendOnly, FILE_EXT2_NoDump, FILE_EXT2_NoAtime, FILE_EXT2_CompDirty, FILE_EXT2_CompBlock, FILE_EXT2_NoCompBlock, FILE_EXT2_CompError, FILE_EXT2_BTree, FILE_EXT2_HashIndexed, FILE_EXT2_iMagic, FILE_EXT2_Journaled, FILE_EXT2_NoTail, FILE_EXT2_DirSync, FILE_EXT2_TopDir, FILE_EXT2_Reserved, UNKNOWN, }; struct unknown_tag { struct unknown_tag *next; struct archive_string name; }; struct xar { uint64_t offset; /* Current position in the file. */ int64_t total; uint64_t h_base; int end_of_file; #define OUTBUFF_SIZE (1024 * 64) unsigned char *outbuff; enum xmlstatus xmlsts; enum xmlstatus xmlsts_unknown; struct unknown_tag *unknowntags; int base64text; /* * TOC */ uint64_t toc_remaining; uint64_t toc_total; uint64_t toc_chksum_offset; uint64_t toc_chksum_size; /* * For Decoding data. */ enum enctype rd_encoding; z_stream stream; int stream_valid; #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) bz_stream bzstream; int bzstream_valid; #endif #if HAVE_LZMA_H && HAVE_LIBLZMA lzma_stream lzstream; int lzstream_valid; #endif /* * For Checksum data. */ struct chksumwork a_sumwrk; struct chksumwork e_sumwrk; struct xar_file *file; /* current reading file. */ struct xattr *xattr; /* current reading extended attribute. */ struct heap_queue file_queue; struct xar_file *hdlink_orgs; struct hdlink *hdlink_list; int entry_init; uint64_t entry_total; uint64_t entry_remaining; size_t entry_unconsumed; uint64_t entry_size; enum enctype entry_encoding; struct chksumval entry_a_sum; struct chksumval entry_e_sum; struct archive_string_conv *sconv; }; struct xmlattr { struct xmlattr *next; char *name; char *value; }; struct xmlattr_list { struct xmlattr *first; struct xmlattr **last; }; static int xar_bid(struct archive_read *, int); static int xar_read_header(struct archive_read *, struct archive_entry *); static int xar_read_data(struct archive_read *, const void **, size_t *, int64_t *); static int xar_read_data_skip(struct archive_read *); static int xar_cleanup(struct archive_read *); static int move_reading_point(struct archive_read *, uint64_t); static int rd_contents_init(struct archive_read *, enum enctype, int, int); static int rd_contents(struct archive_read *, const void **, size_t *, size_t *, uint64_t); static uint64_t atol10(const char *, size_t); static int64_t atol8(const char *, size_t); static size_t atohex(unsigned char *, size_t, const char *, size_t); static time_t parse_time(const char *p, size_t n); static int heap_add_entry(struct archive_read *a, struct heap_queue *, struct xar_file *); static struct xar_file *heap_get_entry(struct heap_queue *); static int add_link(struct archive_read *, struct xar *, struct xar_file *); static void checksum_init(struct archive_read *, int, int); static void checksum_update(struct archive_read *, const void *, size_t, const void *, size_t); static int checksum_final(struct archive_read *, const void *, size_t, const void *, size_t); static void checksum_cleanup(struct archive_read *); static int decompression_init(struct archive_read *, enum enctype); static int decompress(struct archive_read *, const void **, size_t *, const void *, size_t *); static int decompression_cleanup(struct archive_read *); static void xmlattr_cleanup(struct xmlattr_list *); static int file_new(struct archive_read *, struct xar *, struct xmlattr_list *); static void file_free(struct xar_file *); static int xattr_new(struct archive_read *, struct xar *, struct xmlattr_list *); static void xattr_free(struct xattr *); static int getencoding(struct xmlattr_list *); static int getsumalgorithm(struct xmlattr_list *); static int unknowntag_start(struct archive_read *, struct xar *, const char *); static void unknowntag_end(struct xar *, const char *); static int xml_start(struct archive_read *, const char *, struct xmlattr_list *); static void xml_end(void *, const char *); static void xml_data(void *, const char *, size_t); static int xml_parse_file_flags(struct xar *, const char *); static int xml_parse_file_ext2(struct xar *, const char *); #if defined(HAVE_LIBXML_XMLREADER_H) static int xml2_xmlattr_setup(struct archive_read *, struct xmlattr_list *, xmlTextReaderPtr); static int xml2_read_cb(void *, char *, int); static int xml2_close_cb(void *); static void xml2_error_hdr(void *, const char *, xmlParserSeverities, xmlTextReaderLocatorPtr); static int xml2_read_toc(struct archive_read *); #elif defined(HAVE_BSDXML_H) || defined(HAVE_EXPAT_H) struct expat_userData { int state; struct archive_read *archive; }; static int expat_xmlattr_setup(struct archive_read *, struct xmlattr_list *, const XML_Char **); static void expat_start_cb(void *, const XML_Char *, const XML_Char **); static void expat_end_cb(void *, const XML_Char *); static void expat_data_cb(void *, const XML_Char *, int); static int expat_read_toc(struct archive_read *); #elif defined(HAVE_XMLLITE_H) static int xmllite_read_toc(struct archive_read *); #endif int archive_read_support_format_xar(struct archive *_a) { struct xar *xar; struct archive_read *a = (struct archive_read *)_a; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_xar"); xar = calloc(1, sizeof(*xar)); if (xar == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate xar data"); return (ARCHIVE_FATAL); } /* initialize xar->file_queue */ xar->file_queue.allocated = 0; xar->file_queue.used = 0; xar->file_queue.files = NULL; r = __archive_read_register_format(a, xar, "xar", xar_bid, NULL, xar_read_header, xar_read_data, xar_read_data_skip, NULL, xar_cleanup, NULL, NULL); if (r != ARCHIVE_OK) free(xar); return (r); } static int xar_bid(struct archive_read *a, int best_bid) { const unsigned char *b; int bid; (void)best_bid; /* UNUSED */ b = __archive_read_ahead(a, HEADER_SIZE, NULL); if (b == NULL) return (-1); bid = 0; /* * Verify magic code */ if (archive_be32dec(b) != HEADER_MAGIC) return (0); bid += 32; /* * Verify header size */ if (archive_be16dec(b+4) != HEADER_SIZE) return (0); bid += 16; /* * Verify header version */ if (archive_be16dec(b+6) != HEADER_VERSION) return (0); bid += 16; /* * Verify type of checksum */ switch (archive_be32dec(b+24)) { case CKSUM_NONE: case CKSUM_SHA1: case CKSUM_MD5: bid += 32; break; default: return (0); } return (bid); } static int read_toc(struct archive_read *a) { struct xar *xar; struct xar_file *file; const unsigned char *b; uint64_t toc_compressed_size; uint64_t toc_uncompressed_size; uint32_t toc_chksum_alg; ssize_t bytes; int r; xar = (struct xar *)(a->format->data); /* * Read xar header. */ b = __archive_read_ahead(a, HEADER_SIZE, &bytes); if (bytes < 0) return ((int)bytes); if (bytes < HEADER_SIZE) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated archive header"); return (ARCHIVE_FATAL); } if (archive_be32dec(b) != HEADER_MAGIC) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header magic"); return (ARCHIVE_FATAL); } if (archive_be16dec(b+6) != HEADER_VERSION) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported header version(%d)", archive_be16dec(b+6)); return (ARCHIVE_FATAL); } toc_compressed_size = archive_be64dec(b+8); xar->toc_remaining = toc_compressed_size; toc_uncompressed_size = archive_be64dec(b+16); toc_chksum_alg = archive_be32dec(b+24); __archive_read_consume(a, HEADER_SIZE); xar->offset += HEADER_SIZE; xar->toc_total = 0; /* * Read TOC(Table of Contents). */ /* Initialize reading contents. */ r = move_reading_point(a, HEADER_SIZE); if (r != ARCHIVE_OK) return (r); r = rd_contents_init(a, GZIP, toc_chksum_alg, CKSUM_NONE); if (r != ARCHIVE_OK) return (r); #ifdef HAVE_LIBXML_XMLREADER_H r = xml2_read_toc(a); #elif defined(HAVE_BSDXML_H) || defined(HAVE_EXPAT_H) r = expat_read_toc(a); #elif defined(HAVE_XMLLITE_H) r = xmllite_read_toc(a); #endif if (r != ARCHIVE_OK) return (r); /* Set 'The HEAP' base. */ xar->h_base = xar->offset; if (xar->toc_total != toc_uncompressed_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "TOC uncompressed size error"); return (ARCHIVE_FATAL); } /* * Checksum TOC */ if (toc_chksum_alg != CKSUM_NONE) { r = move_reading_point(a, xar->toc_chksum_offset); if (r != ARCHIVE_OK) return (r); b = __archive_read_ahead(a, (size_t)xar->toc_chksum_size, &bytes); if (bytes < 0) return ((int)bytes); if ((uint64_t)bytes < xar->toc_chksum_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated archive file"); return (ARCHIVE_FATAL); } r = checksum_final(a, b, (size_t)xar->toc_chksum_size, NULL, 0); __archive_read_consume(a, xar->toc_chksum_size); xar->offset += xar->toc_chksum_size; #ifndef DONT_FAIL_ON_CRC_ERROR if (r != ARCHIVE_OK) return (ARCHIVE_FATAL); #endif } /* * Connect hardlinked files. */ for (file = xar->hdlink_orgs; file != NULL; file = file->hdnext) { struct hdlink **hdlink; for (hdlink = &(xar->hdlink_list); *hdlink != NULL; hdlink = &((*hdlink)->next)) { if ((*hdlink)->id == file->id) { struct hdlink *hltmp; struct xar_file *f2; int nlink = (*hdlink)->cnt + 1; file->nlink = nlink; for (f2 = (*hdlink)->files; f2 != NULL; f2 = f2->hdnext) { f2->nlink = nlink; archive_string_copy( &(f2->hardlink), &(file->pathname)); } /* Remove resolved files from hdlist_list. */ hltmp = *hdlink; *hdlink = hltmp->next; free(hltmp); break; } } } a->archive.archive_format = ARCHIVE_FORMAT_XAR; a->archive.archive_format_name = "xar"; return (ARCHIVE_OK); } static int xar_read_header(struct archive_read *a, struct archive_entry *entry) { struct xar *xar; struct xar_file *file; struct xattr *xattr; int r; xar = (struct xar *)(a->format->data); r = ARCHIVE_OK; if (xar->offset == 0) { /* Create a character conversion object. */ if (xar->sconv == NULL) { xar->sconv = archive_string_conversion_from_charset( &(a->archive), "UTF-8", 1); if (xar->sconv == NULL) return (ARCHIVE_FATAL); } /* Read TOC. */ r = read_toc(a); if (r != ARCHIVE_OK) return (r); } for (;;) { file = xar->file = heap_get_entry(&(xar->file_queue)); if (file == NULL) { xar->end_of_file = 1; return (ARCHIVE_EOF); } if ((file->mode & AE_IFMT) != AE_IFDIR) break; if (file->has != (HAS_PATHNAME | HAS_TYPE)) break; /* * If a file type is a directory and it does not have * any metadata, do not export. */ file_free(file); } if (file->has & HAS_ATIME) { archive_entry_set_atime(entry, file->atime, 0); } if (file->has & HAS_CTIME) { archive_entry_set_ctime(entry, file->ctime, 0); } if (file->has & HAS_MTIME) { archive_entry_set_mtime(entry, file->mtime, 0); } archive_entry_set_gid(entry, file->gid); if (file->gname.length > 0 && archive_entry_copy_gname_l(entry, file->gname.s, archive_strlen(&(file->gname)), xar->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Gname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Gname cannot be converted from %s to current locale.", archive_string_conversion_charset_name(xar->sconv)); r = ARCHIVE_WARN; } archive_entry_set_uid(entry, file->uid); if (file->uname.length > 0 && archive_entry_copy_uname_l(entry, file->uname.s, archive_strlen(&(file->uname)), xar->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Uname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Uname cannot be converted from %s to current locale.", archive_string_conversion_charset_name(xar->sconv)); r = ARCHIVE_WARN; } archive_entry_set_mode(entry, file->mode); if (archive_entry_copy_pathname_l(entry, file->pathname.s, archive_strlen(&(file->pathname)), xar->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Pathname cannot be converted from %s to current locale.", archive_string_conversion_charset_name(xar->sconv)); r = ARCHIVE_WARN; } if (file->symlink.length > 0 && archive_entry_copy_symlink_l(entry, file->symlink.s, archive_strlen(&(file->symlink)), xar->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Linkname cannot be converted from %s to current locale.", archive_string_conversion_charset_name(xar->sconv)); r = ARCHIVE_WARN; } /* Set proper nlink. */ if ((file->mode & AE_IFMT) == AE_IFDIR) archive_entry_set_nlink(entry, file->subdirs + 2); else archive_entry_set_nlink(entry, file->nlink); archive_entry_set_size(entry, file->size); if (archive_strlen(&(file->hardlink)) > 0) archive_entry_set_hardlink(entry, file->hardlink.s); archive_entry_set_ino64(entry, file->ino64); if (file->has & HAS_DEV) archive_entry_set_dev(entry, file->dev); if (file->has & HAS_DEVMAJOR) archive_entry_set_devmajor(entry, file->devmajor); if (file->has & HAS_DEVMINOR) archive_entry_set_devminor(entry, file->devminor); if (archive_strlen(&(file->fflags_text)) > 0) archive_entry_copy_fflags_text(entry, file->fflags_text.s); xar->entry_init = 1; xar->entry_total = 0; xar->entry_remaining = file->length; xar->entry_size = file->size; xar->entry_encoding = file->encoding; xar->entry_a_sum = file->a_sum; xar->entry_e_sum = file->e_sum; /* * Read extended attributes. */ xattr = file->xattr_list; while (xattr != NULL) { const void *d; size_t outbytes = 0; size_t used = 0; r = move_reading_point(a, xattr->offset); if (r != ARCHIVE_OK) break; r = rd_contents_init(a, xattr->encoding, xattr->a_sum.alg, xattr->e_sum.alg); if (r != ARCHIVE_OK) break; d = NULL; r = rd_contents(a, &d, &outbytes, &used, xattr->length); if (r != ARCHIVE_OK) break; if (outbytes != xattr->size) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Decompressed size error"); r = ARCHIVE_FATAL; break; } r = checksum_final(a, xattr->a_sum.val, xattr->a_sum.len, xattr->e_sum.val, xattr->e_sum.len); if (r != ARCHIVE_OK) { #ifndef DONT_FAIL_ON_CRC_ERROR archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Xattr checksum error"); r = ARCHIVE_WARN; break; #endif } if (xattr->name.s == NULL) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Xattr name error"); r = ARCHIVE_WARN; break; } archive_entry_xattr_add_entry(entry, xattr->name.s, d, outbytes); xattr = xattr->next; } if (r != ARCHIVE_OK) { file_free(file); return (r); } if (xar->entry_remaining > 0) /* Move reading point to the beginning of current * file contents. */ r = move_reading_point(a, file->offset); else r = ARCHIVE_OK; file_free(file); return (r); } static int xar_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct xar *xar; size_t used = 0; int r; xar = (struct xar *)(a->format->data); if (xar->entry_unconsumed) { __archive_read_consume(a, xar->entry_unconsumed); xar->entry_unconsumed = 0; } if (xar->end_of_file || xar->entry_remaining <= 0) { r = ARCHIVE_EOF; goto abort_read_data; } if (xar->entry_init) { r = rd_contents_init(a, xar->entry_encoding, xar->entry_a_sum.alg, xar->entry_e_sum.alg); if (r != ARCHIVE_OK) { xar->entry_remaining = 0; return (r); } xar->entry_init = 0; } *buff = NULL; r = rd_contents(a, buff, size, &used, xar->entry_remaining); if (r != ARCHIVE_OK) goto abort_read_data; *offset = xar->entry_total; xar->entry_total += *size; xar->total += *size; xar->offset += used; xar->entry_remaining -= used; xar->entry_unconsumed = used; if (xar->entry_remaining == 0) { if (xar->entry_total != xar->entry_size) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Decompressed size error"); r = ARCHIVE_FATAL; goto abort_read_data; } r = checksum_final(a, xar->entry_a_sum.val, xar->entry_a_sum.len, xar->entry_e_sum.val, xar->entry_e_sum.len); if (r != ARCHIVE_OK) goto abort_read_data; } return (ARCHIVE_OK); abort_read_data: *buff = NULL; *size = 0; - *offset = xar->total; + *offset = (int64_t)xar->entry_total; return (r); } static int xar_read_data_skip(struct archive_read *a) { struct xar *xar; int64_t bytes_skipped; xar = (struct xar *)(a->format->data); if (xar->end_of_file) return (ARCHIVE_EOF); bytes_skipped = __archive_read_consume(a, xar->entry_remaining + xar->entry_unconsumed); if (bytes_skipped < 0) return (ARCHIVE_FATAL); xar->offset += bytes_skipped; xar->entry_unconsumed = 0; return (ARCHIVE_OK); } static int xar_cleanup(struct archive_read *a) { struct xar *xar; struct hdlink *hdlink; int i; int r; xar = (struct xar *)(a->format->data); checksum_cleanup(a); r = decompression_cleanup(a); hdlink = xar->hdlink_list; while (hdlink != NULL) { struct hdlink *next = hdlink->next; free(hdlink); hdlink = next; } for (i = 0; i < xar->file_queue.used; i++) file_free(xar->file_queue.files[i]); free(xar->file_queue.files); while (xar->unknowntags != NULL) { struct unknown_tag *tag; tag = xar->unknowntags; xar->unknowntags = tag->next; archive_string_free(&(tag->name)); free(tag); } free(xar->outbuff); free(xar); a->format->data = NULL; return (r); } static int move_reading_point(struct archive_read *a, uint64_t offset) { struct xar *xar; xar = (struct xar *)(a->format->data); if (xar->offset - xar->h_base != offset) { /* Seek forward to the start of file contents. */ int64_t step; step = offset - (xar->offset - xar->h_base); if (step > 0) { step = __archive_read_consume(a, step); if (step < 0) return ((int)step); xar->offset += step; } else { int64_t pos = __archive_read_seek(a, xar->h_base + offset, SEEK_SET); if (pos == ARCHIVE_FAILED) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Cannot seek."); return (ARCHIVE_FAILED); } xar->offset = pos; } } return (ARCHIVE_OK); } static int rd_contents_init(struct archive_read *a, enum enctype encoding, int a_sum_alg, int e_sum_alg) { int r; /* Init decompress library. */ if ((r = decompression_init(a, encoding)) != ARCHIVE_OK) return (r); /* Init checksum library. */ checksum_init(a, a_sum_alg, e_sum_alg); return (ARCHIVE_OK); } static int rd_contents(struct archive_read *a, const void **buff, size_t *size, size_t *used, uint64_t remaining) { const unsigned char *b; ssize_t bytes; /* Get whatever bytes are immediately available. */ b = __archive_read_ahead(a, 1, &bytes); if (bytes < 0) return ((int)bytes); if (bytes == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Truncated archive file"); return (ARCHIVE_FATAL); } if ((uint64_t)bytes > remaining) bytes = (ssize_t)remaining; /* * Decompress contents of file. */ *used = bytes; if (decompress(a, buff, size, b, used) != ARCHIVE_OK) return (ARCHIVE_FATAL); /* * Update checksum of a compressed data and a extracted data. */ checksum_update(a, b, *used, *buff, *size); return (ARCHIVE_OK); } /* * Note that this implementation does not (and should not!) obey * locale settings; you cannot simply substitute strtol here, since * it does obey locale. */ static uint64_t atol10(const char *p, size_t char_cnt) { uint64_t l; int digit; if (char_cnt == 0) return (0); l = 0; digit = *p - '0'; while (digit >= 0 && digit < 10 && char_cnt-- > 0) { l = (l * 10) + digit; digit = *++p - '0'; } return (l); } static int64_t atol8(const char *p, size_t char_cnt) { int64_t l; int digit; if (char_cnt == 0) return (0); l = 0; while (char_cnt-- > 0) { if (*p >= '0' && *p <= '7') digit = *p - '0'; else break; p++; l <<= 3; l |= digit; } return (l); } static size_t atohex(unsigned char *b, size_t bsize, const char *p, size_t psize) { size_t fbsize = bsize; while (bsize && psize > 1) { unsigned char x; if (p[0] >= 'a' && p[0] <= 'f') x = (p[0] - 'a' + 0x0a) << 4; else if (p[0] >= 'A' && p[0] <= 'F') x = (p[0] - 'A' + 0x0a) << 4; else if (p[0] >= '0' && p[0] <= '9') x = (p[0] - '0') << 4; else return (-1); if (p[1] >= 'a' && p[1] <= 'f') x |= p[1] - 'a' + 0x0a; else if (p[1] >= 'A' && p[1] <= 'F') x |= p[1] - 'A' + 0x0a; else if (p[1] >= '0' && p[1] <= '9') x |= p[1] - '0'; else return (-1); *b++ = x; bsize--; p += 2; psize -= 2; } return (fbsize - bsize); } static time_t time_from_tm(struct tm *t) { #if HAVE__MKGMTIME return _mkgmtime(t); #elif HAVE_TIMEGM /* Use platform timegm() if available. */ return (timegm(t)); #else /* Else use direct calculation using POSIX assumptions. */ /* First, fix up tm_yday based on the year/month/day. */ mktime(t); /* Then we can compute timegm() from first principles. */ return (t->tm_sec + t->tm_min * 60 + t->tm_hour * 3600 + t->tm_yday * 86400 + (t->tm_year - 70) * 31536000 + ((t->tm_year - 69) / 4) * 86400 - ((t->tm_year - 1) / 100) * 86400 + ((t->tm_year + 299) / 400) * 86400); #endif } static time_t parse_time(const char *p, size_t n) { struct tm tm; time_t t = 0; int64_t data; memset(&tm, 0, sizeof(tm)); if (n != 20) return (t); data = atol10(p, 4); if (data < 1900) return (t); tm.tm_year = (int)data - 1900; p += 4; if (*p++ != '-') return (t); data = atol10(p, 2); if (data < 1 || data > 12) return (t); tm.tm_mon = (int)data -1; p += 2; if (*p++ != '-') return (t); data = atol10(p, 2); if (data < 1 || data > 31) return (t); tm.tm_mday = (int)data; p += 2; if (*p++ != 'T') return (t); data = atol10(p, 2); if (data < 0 || data > 23) return (t); tm.tm_hour = (int)data; p += 2; if (*p++ != ':') return (t); data = atol10(p, 2); if (data < 0 || data > 59) return (t); tm.tm_min = (int)data; p += 2; if (*p++ != ':') return (t); data = atol10(p, 2); if (data < 0 || data > 60) return (t); tm.tm_sec = (int)data; #if 0 p += 2; if (*p != 'Z') return (t); #endif t = time_from_tm(&tm); return (t); } static int heap_add_entry(struct archive_read *a, struct heap_queue *heap, struct xar_file *file) { uint64_t file_id, parent_id; int hole, parent; /* Expand our pending files list as necessary. */ if (heap->used >= heap->allocated) { struct xar_file **new_pending_files; int new_size; if (heap->allocated < 1024) new_size = 1024; else new_size = heap->allocated * 2; /* Overflow might keep us from growing the list. */ if (new_size <= heap->allocated) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } new_pending_files = (struct xar_file **) calloc(new_size, sizeof(new_pending_files[0])); if (new_pending_files == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } if (heap->allocated) { memcpy(new_pending_files, heap->files, heap->allocated * sizeof(new_pending_files[0])); free(heap->files); } heap->files = new_pending_files; heap->allocated = new_size; } file_id = file->id; /* * Start with hole at end, walk it up tree to find insertion point. */ hole = heap->used++; while (hole > 0) { parent = (hole - 1)/2; parent_id = heap->files[parent]->id; if (file_id >= parent_id) { heap->files[hole] = file; return (ARCHIVE_OK); } /* Move parent into hole <==> move hole up tree. */ heap->files[hole] = heap->files[parent]; hole = parent; } heap->files[0] = file; return (ARCHIVE_OK); } static struct xar_file * heap_get_entry(struct heap_queue *heap) { uint64_t a_id, b_id, c_id; int a, b, c; struct xar_file *r, *tmp; if (heap->used < 1) return (NULL); /* * The first file in the list is the earliest; we'll return this. */ r = heap->files[0]; /* * Move the last item in the heap to the root of the tree */ heap->files[0] = heap->files[--(heap->used)]; /* * Rebalance the heap. */ a = 0; /* Starting element and its heap key */ a_id = heap->files[a]->id; for (;;) { b = a + a + 1; /* First child */ if (b >= heap->used) return (r); b_id = heap->files[b]->id; c = b + 1; /* Use second child if it is smaller. */ if (c < heap->used) { c_id = heap->files[c]->id; if (c_id < b_id) { b = c; b_id = c_id; } } if (a_id <= b_id) return (r); tmp = heap->files[a]; heap->files[a] = heap->files[b]; heap->files[b] = tmp; a = b; } } static int add_link(struct archive_read *a, struct xar *xar, struct xar_file *file) { struct hdlink *hdlink; for (hdlink = xar->hdlink_list; hdlink != NULL; hdlink = hdlink->next) { if (hdlink->id == file->link) { file->hdnext = hdlink->files; hdlink->cnt++; hdlink->files = file; return (ARCHIVE_OK); } } hdlink = malloc(sizeof(*hdlink)); if (hdlink == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } file->hdnext = NULL; hdlink->id = file->link; hdlink->cnt = 1; hdlink->files = file; hdlink->next = xar->hdlink_list; xar->hdlink_list = hdlink; return (ARCHIVE_OK); } static void _checksum_init(struct chksumwork *sumwrk, int sum_alg) { sumwrk->alg = sum_alg; switch (sum_alg) { case CKSUM_NONE: break; case CKSUM_SHA1: archive_sha1_init(&(sumwrk->sha1ctx)); break; case CKSUM_MD5: archive_md5_init(&(sumwrk->md5ctx)); break; } } static void _checksum_update(struct chksumwork *sumwrk, const void *buff, size_t size) { switch (sumwrk->alg) { case CKSUM_NONE: break; case CKSUM_SHA1: archive_sha1_update(&(sumwrk->sha1ctx), buff, size); break; case CKSUM_MD5: archive_md5_update(&(sumwrk->md5ctx), buff, size); break; } } static int _checksum_final(struct chksumwork *sumwrk, const void *val, size_t len) { unsigned char sum[MAX_SUM_SIZE]; int r = ARCHIVE_OK; switch (sumwrk->alg) { case CKSUM_NONE: break; case CKSUM_SHA1: archive_sha1_final(&(sumwrk->sha1ctx), sum); if (len != SHA1_SIZE || memcmp(val, sum, SHA1_SIZE) != 0) r = ARCHIVE_FAILED; break; case CKSUM_MD5: archive_md5_final(&(sumwrk->md5ctx), sum); if (len != MD5_SIZE || memcmp(val, sum, MD5_SIZE) != 0) r = ARCHIVE_FAILED; break; } return (r); } static void checksum_init(struct archive_read *a, int a_sum_alg, int e_sum_alg) { struct xar *xar; xar = (struct xar *)(a->format->data); _checksum_init(&(xar->a_sumwrk), a_sum_alg); _checksum_init(&(xar->e_sumwrk), e_sum_alg); } static void checksum_update(struct archive_read *a, const void *abuff, size_t asize, const void *ebuff, size_t esize) { struct xar *xar; xar = (struct xar *)(a->format->data); _checksum_update(&(xar->a_sumwrk), abuff, asize); _checksum_update(&(xar->e_sumwrk), ebuff, esize); } static int checksum_final(struct archive_read *a, const void *a_sum_val, size_t a_sum_len, const void *e_sum_val, size_t e_sum_len) { struct xar *xar; int r; xar = (struct xar *)(a->format->data); r = _checksum_final(&(xar->a_sumwrk), a_sum_val, a_sum_len); if (r == ARCHIVE_OK) r = _checksum_final(&(xar->e_sumwrk), e_sum_val, e_sum_len); if (r != ARCHIVE_OK) archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Sumcheck error"); return (r); } static int decompression_init(struct archive_read *a, enum enctype encoding) { struct xar *xar; const char *detail; int r; xar = (struct xar *)(a->format->data); xar->rd_encoding = encoding; switch (encoding) { case NONE: break; case GZIP: if (xar->stream_valid) r = inflateReset(&(xar->stream)); else r = inflateInit(&(xar->stream)); if (r != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Couldn't initialize zlib stream."); return (ARCHIVE_FATAL); } xar->stream_valid = 1; xar->stream.total_in = 0; xar->stream.total_out = 0; break; #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) case BZIP2: if (xar->bzstream_valid) { BZ2_bzDecompressEnd(&(xar->bzstream)); xar->bzstream_valid = 0; } r = BZ2_bzDecompressInit(&(xar->bzstream), 0, 0); if (r == BZ_MEM_ERROR) r = BZ2_bzDecompressInit(&(xar->bzstream), 0, 1); if (r != BZ_OK) { int err = ARCHIVE_ERRNO_MISC; detail = NULL; switch (r) { case BZ_PARAM_ERROR: detail = "invalid setup parameter"; break; case BZ_MEM_ERROR: err = ENOMEM; detail = "out of memory"; break; case BZ_CONFIG_ERROR: detail = "mis-compiled library"; break; } archive_set_error(&a->archive, err, "Internal error initializing decompressor: %s", detail == NULL ? "??" : detail); xar->bzstream_valid = 0; return (ARCHIVE_FATAL); } xar->bzstream_valid = 1; xar->bzstream.total_in_lo32 = 0; xar->bzstream.total_in_hi32 = 0; xar->bzstream.total_out_lo32 = 0; xar->bzstream.total_out_hi32 = 0; break; #endif #if defined(HAVE_LZMA_H) && defined(HAVE_LIBLZMA) #if LZMA_VERSION_MAJOR >= 5 /* Effectively disable the limiter. */ #define LZMA_MEMLIMIT UINT64_MAX #else /* NOTE: This needs to check memory size which running system has. */ #define LZMA_MEMLIMIT (1U << 30) #endif case XZ: case LZMA: if (xar->lzstream_valid) { lzma_end(&(xar->lzstream)); xar->lzstream_valid = 0; } if (xar->entry_encoding == XZ) r = lzma_stream_decoder(&(xar->lzstream), LZMA_MEMLIMIT,/* memlimit */ LZMA_CONCATENATED); else r = lzma_alone_decoder(&(xar->lzstream), LZMA_MEMLIMIT);/* memlimit */ if (r != LZMA_OK) { switch (r) { case LZMA_MEM_ERROR: archive_set_error(&a->archive, ENOMEM, "Internal error initializing " "compression library: " "Cannot allocate memory"); break; case LZMA_OPTIONS_ERROR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Internal error initializing " "compression library: " "Invalid or unsupported options"); break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Internal error initializing " "lzma library"); break; } return (ARCHIVE_FATAL); } xar->lzstream_valid = 1; xar->lzstream.total_in = 0; xar->lzstream.total_out = 0; break; #endif /* * Unsupported compression. */ default: #if !defined(HAVE_BZLIB_H) || !defined(BZ_CONFIG_ERROR) case BZIP2: #endif #if !defined(HAVE_LZMA_H) || !defined(HAVE_LIBLZMA) case LZMA: case XZ: #endif switch (xar->entry_encoding) { case BZIP2: detail = "bzip2"; break; case LZMA: detail = "lzma"; break; case XZ: detail = "xz"; break; default: detail = "??"; break; } archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s compression not supported on this platform", detail); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } static int decompress(struct archive_read *a, const void **buff, size_t *outbytes, const void *b, size_t *used) { struct xar *xar; void *outbuff; size_t avail_in, avail_out; int r; xar = (struct xar *)(a->format->data); avail_in = *used; outbuff = (void *)(uintptr_t)*buff; if (outbuff == NULL) { if (xar->outbuff == NULL) { xar->outbuff = malloc(OUTBUFF_SIZE); if (xar->outbuff == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory for out buffer"); return (ARCHIVE_FATAL); } } outbuff = xar->outbuff; *buff = outbuff; avail_out = OUTBUFF_SIZE; } else avail_out = *outbytes; switch (xar->rd_encoding) { case GZIP: xar->stream.next_in = (Bytef *)(uintptr_t)b; xar->stream.avail_in = (uInt)avail_in; xar->stream.next_out = (unsigned char *)outbuff; xar->stream.avail_out = (uInt)avail_out; r = inflate(&(xar->stream), 0); switch (r) { case Z_OK: /* Decompressor made some progress.*/ case Z_STREAM_END: /* Found end of stream. */ break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "File decompression failed (%d)", r); return (ARCHIVE_FATAL); } *used = avail_in - xar->stream.avail_in; *outbytes = avail_out - xar->stream.avail_out; break; #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) case BZIP2: xar->bzstream.next_in = (char *)(uintptr_t)b; xar->bzstream.avail_in = (unsigned int)avail_in; xar->bzstream.next_out = (char *)outbuff; xar->bzstream.avail_out = (unsigned int)avail_out; r = BZ2_bzDecompress(&(xar->bzstream)); switch (r) { case BZ_STREAM_END: /* Found end of stream. */ switch (BZ2_bzDecompressEnd(&(xar->bzstream))) { case BZ_OK: break; default: archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Failed to clean up decompressor"); return (ARCHIVE_FATAL); } xar->bzstream_valid = 0; /* FALLTHROUGH */ case BZ_OK: /* Decompressor made some progress. */ break; default: archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "bzip decompression failed"); return (ARCHIVE_FATAL); } *used = avail_in - xar->bzstream.avail_in; *outbytes = avail_out - xar->bzstream.avail_out; break; #endif #if defined(HAVE_LZMA_H) && defined(HAVE_LIBLZMA) case LZMA: case XZ: xar->lzstream.next_in = b; xar->lzstream.avail_in = avail_in; xar->lzstream.next_out = (unsigned char *)outbuff; xar->lzstream.avail_out = avail_out; r = lzma_code(&(xar->lzstream), LZMA_RUN); switch (r) { case LZMA_STREAM_END: /* Found end of stream. */ lzma_end(&(xar->lzstream)); xar->lzstream_valid = 0; /* FALLTHROUGH */ case LZMA_OK: /* Decompressor made some progress. */ break; default: archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "%s decompression failed(%d)", (xar->entry_encoding == XZ)?"xz":"lzma", r); return (ARCHIVE_FATAL); } *used = avail_in - xar->lzstream.avail_in; *outbytes = avail_out - xar->lzstream.avail_out; break; #endif #if !defined(HAVE_BZLIB_H) || !defined(BZ_CONFIG_ERROR) case BZIP2: #endif #if !defined(HAVE_LZMA_H) || !defined(HAVE_LIBLZMA) case LZMA: case XZ: #endif case NONE: default: if (outbuff == xar->outbuff) { *buff = b; *used = avail_in; *outbytes = avail_in; } else { if (avail_out > avail_in) avail_out = avail_in; memcpy(outbuff, b, avail_out); *used = avail_out; *outbytes = avail_out; } break; } return (ARCHIVE_OK); } static int decompression_cleanup(struct archive_read *a) { struct xar *xar; int r; xar = (struct xar *)(a->format->data); r = ARCHIVE_OK; if (xar->stream_valid) { if (inflateEnd(&(xar->stream)) != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to clean up zlib decompressor"); r = ARCHIVE_FATAL; } } #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) if (xar->bzstream_valid) { if (BZ2_bzDecompressEnd(&(xar->bzstream)) != BZ_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to clean up bzip2 decompressor"); r = ARCHIVE_FATAL; } } #endif #if defined(HAVE_LZMA_H) && defined(HAVE_LIBLZMA) if (xar->lzstream_valid) lzma_end(&(xar->lzstream)); #endif return (r); } static void checksum_cleanup(struct archive_read *a) { struct xar *xar; xar = (struct xar *)(a->format->data); _checksum_final(&(xar->a_sumwrk), NULL, 0); _checksum_final(&(xar->e_sumwrk), NULL, 0); } static void xmlattr_cleanup(struct xmlattr_list *list) { struct xmlattr *attr, *next; attr = list->first; while (attr != NULL) { next = attr->next; free(attr->name); free(attr->value); free(attr); attr = next; } list->first = NULL; list->last = &(list->first); } static int file_new(struct archive_read *a, struct xar *xar, struct xmlattr_list *list) { struct xar_file *file; struct xmlattr *attr; file = calloc(1, sizeof(*file)); if (file == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } file->parent = xar->file; file->mode = 0777 | AE_IFREG; file->atime = 0; file->mtime = 0; xar->file = file; xar->xattr = NULL; for (attr = list->first; attr != NULL; attr = attr->next) { if (strcmp(attr->name, "id") == 0) file->id = atol10(attr->value, strlen(attr->value)); } file->nlink = 1; if (heap_add_entry(a, &(xar->file_queue), file) != ARCHIVE_OK) return (ARCHIVE_FATAL); return (ARCHIVE_OK); } static void file_free(struct xar_file *file) { struct xattr *xattr; archive_string_free(&(file->pathname)); archive_string_free(&(file->symlink)); archive_string_free(&(file->uname)); archive_string_free(&(file->gname)); archive_string_free(&(file->hardlink)); xattr = file->xattr_list; while (xattr != NULL) { struct xattr *next; next = xattr->next; xattr_free(xattr); xattr = next; } free(file); } static int xattr_new(struct archive_read *a, struct xar *xar, struct xmlattr_list *list) { struct xattr *xattr, **nx; struct xmlattr *attr; xattr = calloc(1, sizeof(*xattr)); if (xattr == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } xar->xattr = xattr; for (attr = list->first; attr != NULL; attr = attr->next) { if (strcmp(attr->name, "id") == 0) xattr->id = atol10(attr->value, strlen(attr->value)); } /* Chain to xattr list. */ for (nx = &(xar->file->xattr_list); *nx != NULL; nx = &((*nx)->next)) { if (xattr->id < (*nx)->id) break; } xattr->next = *nx; *nx = xattr; return (ARCHIVE_OK); } static void xattr_free(struct xattr *xattr) { archive_string_free(&(xattr->name)); free(xattr); } static int getencoding(struct xmlattr_list *list) { struct xmlattr *attr; enum enctype encoding = NONE; for (attr = list->first; attr != NULL; attr = attr->next) { if (strcmp(attr->name, "style") == 0) { if (strcmp(attr->value, "application/octet-stream") == 0) encoding = NONE; else if (strcmp(attr->value, "application/x-gzip") == 0) encoding = GZIP; else if (strcmp(attr->value, "application/x-bzip2") == 0) encoding = BZIP2; else if (strcmp(attr->value, "application/x-lzma") == 0) encoding = LZMA; else if (strcmp(attr->value, "application/x-xz") == 0) encoding = XZ; } } return (encoding); } static int getsumalgorithm(struct xmlattr_list *list) { struct xmlattr *attr; int alg = CKSUM_NONE; for (attr = list->first; attr != NULL; attr = attr->next) { if (strcmp(attr->name, "style") == 0) { const char *v = attr->value; if ((v[0] == 'S' || v[0] == 's') && (v[1] == 'H' || v[1] == 'h') && (v[2] == 'A' || v[2] == 'a') && v[3] == '1' && v[4] == '\0') alg = CKSUM_SHA1; if ((v[0] == 'M' || v[0] == 'm') && (v[1] == 'D' || v[1] == 'd') && v[2] == '5' && v[3] == '\0') alg = CKSUM_MD5; } } return (alg); } static int unknowntag_start(struct archive_read *a, struct xar *xar, const char *name) { struct unknown_tag *tag; tag = malloc(sizeof(*tag)); if (tag == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } tag->next = xar->unknowntags; archive_string_init(&(tag->name)); archive_strcpy(&(tag->name), name); if (xar->unknowntags == NULL) { #if DEBUG fprintf(stderr, "UNKNOWNTAG_START:%s\n", name); #endif xar->xmlsts_unknown = xar->xmlsts; xar->xmlsts = UNKNOWN; } xar->unknowntags = tag; return (ARCHIVE_OK); } static void unknowntag_end(struct xar *xar, const char *name) { struct unknown_tag *tag; tag = xar->unknowntags; if (tag == NULL || name == NULL) return; if (strcmp(tag->name.s, name) == 0) { xar->unknowntags = tag->next; archive_string_free(&(tag->name)); free(tag); if (xar->unknowntags == NULL) { #if DEBUG fprintf(stderr, "UNKNOWNTAG_END:%s\n", name); #endif xar->xmlsts = xar->xmlsts_unknown; } } } static int xml_start(struct archive_read *a, const char *name, struct xmlattr_list *list) { struct xar *xar; struct xmlattr *attr; xar = (struct xar *)(a->format->data); #if DEBUG fprintf(stderr, "xml_sta:[%s]\n", name); for (attr = list->first; attr != NULL; attr = attr->next) fprintf(stderr, " attr:\"%s\"=\"%s\"\n", attr->name, attr->value); #endif xar->base64text = 0; switch (xar->xmlsts) { case INIT: if (strcmp(name, "xar") == 0) xar->xmlsts = XAR; else if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; case XAR: if (strcmp(name, "toc") == 0) xar->xmlsts = TOC; else if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; case TOC: if (strcmp(name, "creation-time") == 0) xar->xmlsts = TOC_CREATION_TIME; else if (strcmp(name, "checksum") == 0) xar->xmlsts = TOC_CHECKSUM; else if (strcmp(name, "file") == 0) { if (file_new(a, xar, list) != ARCHIVE_OK) return (ARCHIVE_FATAL); xar->xmlsts = TOC_FILE; } else if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; case TOC_CHECKSUM: if (strcmp(name, "offset") == 0) xar->xmlsts = TOC_CHECKSUM_OFFSET; else if (strcmp(name, "size") == 0) xar->xmlsts = TOC_CHECKSUM_SIZE; else if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; case TOC_FILE: if (strcmp(name, "file") == 0) { if (file_new(a, xar, list) != ARCHIVE_OK) return (ARCHIVE_FATAL); } else if (strcmp(name, "data") == 0) xar->xmlsts = FILE_DATA; else if (strcmp(name, "ea") == 0) { if (xattr_new(a, xar, list) != ARCHIVE_OK) return (ARCHIVE_FATAL); xar->xmlsts = FILE_EA; } else if (strcmp(name, "ctime") == 0) xar->xmlsts = FILE_CTIME; else if (strcmp(name, "mtime") == 0) xar->xmlsts = FILE_MTIME; else if (strcmp(name, "atime") == 0) xar->xmlsts = FILE_ATIME; else if (strcmp(name, "group") == 0) xar->xmlsts = FILE_GROUP; else if (strcmp(name, "gid") == 0) xar->xmlsts = FILE_GID; else if (strcmp(name, "user") == 0) xar->xmlsts = FILE_USER; else if (strcmp(name, "uid") == 0) xar->xmlsts = FILE_UID; else if (strcmp(name, "mode") == 0) xar->xmlsts = FILE_MODE; else if (strcmp(name, "device") == 0) xar->xmlsts = FILE_DEVICE; else if (strcmp(name, "deviceno") == 0) xar->xmlsts = FILE_DEVICENO; else if (strcmp(name, "inode") == 0) xar->xmlsts = FILE_INODE; else if (strcmp(name, "link") == 0) xar->xmlsts = FILE_LINK; else if (strcmp(name, "type") == 0) { xar->xmlsts = FILE_TYPE; for (attr = list->first; attr != NULL; attr = attr->next) { if (strcmp(attr->name, "link") != 0) continue; if (xar->file->hdnext != NULL || xar->file->link != 0 || xar->file == xar->hdlink_orgs) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "File with multiple link attributes"); return (ARCHIVE_FATAL); } if (strcmp(attr->value, "original") == 0) { xar->file->hdnext = xar->hdlink_orgs; xar->hdlink_orgs = xar->file; } else { xar->file->link = (unsigned)atol10(attr->value, strlen(attr->value)); if (xar->file->link > 0) if (add_link(a, xar, xar->file) != ARCHIVE_OK) { return (ARCHIVE_FATAL); } } } } else if (strcmp(name, "name") == 0) { xar->xmlsts = FILE_NAME; for (attr = list->first; attr != NULL; attr = attr->next) { if (strcmp(attr->name, "enctype") == 0 && strcmp(attr->value, "base64") == 0) xar->base64text = 1; } } else if (strcmp(name, "acl") == 0) xar->xmlsts = FILE_ACL; else if (strcmp(name, "flags") == 0) xar->xmlsts = FILE_FLAGS; else if (strcmp(name, "ext2") == 0) xar->xmlsts = FILE_EXT2; else if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; case FILE_DATA: if (strcmp(name, "length") == 0) xar->xmlsts = FILE_DATA_LENGTH; else if (strcmp(name, "offset") == 0) xar->xmlsts = FILE_DATA_OFFSET; else if (strcmp(name, "size") == 0) xar->xmlsts = FILE_DATA_SIZE; else if (strcmp(name, "encoding") == 0) { xar->xmlsts = FILE_DATA_ENCODING; xar->file->encoding = getencoding(list); } else if (strcmp(name, "archived-checksum") == 0) { xar->xmlsts = FILE_DATA_A_CHECKSUM; xar->file->a_sum.alg = getsumalgorithm(list); } else if (strcmp(name, "extracted-checksum") == 0) { xar->xmlsts = FILE_DATA_E_CHECKSUM; xar->file->e_sum.alg = getsumalgorithm(list); } else if (strcmp(name, "content") == 0) xar->xmlsts = FILE_DATA_CONTENT; else if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; case FILE_DEVICE: if (strcmp(name, "major") == 0) xar->xmlsts = FILE_DEVICE_MAJOR; else if (strcmp(name, "minor") == 0) xar->xmlsts = FILE_DEVICE_MINOR; else if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; case FILE_DATA_CONTENT: if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; case FILE_EA: if (strcmp(name, "length") == 0) xar->xmlsts = FILE_EA_LENGTH; else if (strcmp(name, "offset") == 0) xar->xmlsts = FILE_EA_OFFSET; else if (strcmp(name, "size") == 0) xar->xmlsts = FILE_EA_SIZE; else if (strcmp(name, "encoding") == 0) { xar->xmlsts = FILE_EA_ENCODING; xar->xattr->encoding = getencoding(list); } else if (strcmp(name, "archived-checksum") == 0) xar->xmlsts = FILE_EA_A_CHECKSUM; else if (strcmp(name, "extracted-checksum") == 0) xar->xmlsts = FILE_EA_E_CHECKSUM; else if (strcmp(name, "name") == 0) xar->xmlsts = FILE_EA_NAME; else if (strcmp(name, "fstype") == 0) xar->xmlsts = FILE_EA_FSTYPE; else if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; case FILE_ACL: if (strcmp(name, "appleextended") == 0) xar->xmlsts = FILE_ACL_APPLEEXTENDED; else if (strcmp(name, "default") == 0) xar->xmlsts = FILE_ACL_DEFAULT; else if (strcmp(name, "access") == 0) xar->xmlsts = FILE_ACL_ACCESS; else if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; case FILE_FLAGS: if (!xml_parse_file_flags(xar, name)) if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; case FILE_EXT2: if (!xml_parse_file_ext2(xar, name)) if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; case TOC_CREATION_TIME: case TOC_CHECKSUM_OFFSET: case TOC_CHECKSUM_SIZE: case FILE_DATA_LENGTH: case FILE_DATA_OFFSET: case FILE_DATA_SIZE: case FILE_DATA_ENCODING: case FILE_DATA_A_CHECKSUM: case FILE_DATA_E_CHECKSUM: case FILE_EA_LENGTH: case FILE_EA_OFFSET: case FILE_EA_SIZE: case FILE_EA_ENCODING: case FILE_EA_A_CHECKSUM: case FILE_EA_E_CHECKSUM: case FILE_EA_NAME: case FILE_EA_FSTYPE: case FILE_CTIME: case FILE_MTIME: case FILE_ATIME: case FILE_GROUP: case FILE_GID: case FILE_USER: case FILE_UID: case FILE_INODE: case FILE_DEVICE_MAJOR: case FILE_DEVICE_MINOR: case FILE_DEVICENO: case FILE_MODE: case FILE_TYPE: case FILE_LINK: case FILE_NAME: case FILE_ACL_DEFAULT: case FILE_ACL_ACCESS: case FILE_ACL_APPLEEXTENDED: case FILE_FLAGS_USER_NODUMP: case FILE_FLAGS_USER_IMMUTABLE: case FILE_FLAGS_USER_APPEND: case FILE_FLAGS_USER_OPAQUE: case FILE_FLAGS_USER_NOUNLINK: case FILE_FLAGS_SYS_ARCHIVED: case FILE_FLAGS_SYS_IMMUTABLE: case FILE_FLAGS_SYS_APPEND: case FILE_FLAGS_SYS_NOUNLINK: case FILE_FLAGS_SYS_SNAPSHOT: case FILE_EXT2_SecureDeletion: case FILE_EXT2_Undelete: case FILE_EXT2_Compress: case FILE_EXT2_Synchronous: case FILE_EXT2_Immutable: case FILE_EXT2_AppendOnly: case FILE_EXT2_NoDump: case FILE_EXT2_NoAtime: case FILE_EXT2_CompDirty: case FILE_EXT2_CompBlock: case FILE_EXT2_NoCompBlock: case FILE_EXT2_CompError: case FILE_EXT2_BTree: case FILE_EXT2_HashIndexed: case FILE_EXT2_iMagic: case FILE_EXT2_Journaled: case FILE_EXT2_NoTail: case FILE_EXT2_DirSync: case FILE_EXT2_TopDir: case FILE_EXT2_Reserved: case UNKNOWN: if (unknowntag_start(a, xar, name) != ARCHIVE_OK) return (ARCHIVE_FATAL); break; } return (ARCHIVE_OK); } static void xml_end(void *userData, const char *name) { struct archive_read *a; struct xar *xar; a = (struct archive_read *)userData; xar = (struct xar *)(a->format->data); #if DEBUG fprintf(stderr, "xml_end:[%s]\n", name); #endif switch (xar->xmlsts) { case INIT: break; case XAR: if (strcmp(name, "xar") == 0) xar->xmlsts = INIT; break; case TOC: if (strcmp(name, "toc") == 0) xar->xmlsts = XAR; break; case TOC_CREATION_TIME: if (strcmp(name, "creation-time") == 0) xar->xmlsts = TOC; break; case TOC_CHECKSUM: if (strcmp(name, "checksum") == 0) xar->xmlsts = TOC; break; case TOC_CHECKSUM_OFFSET: if (strcmp(name, "offset") == 0) xar->xmlsts = TOC_CHECKSUM; break; case TOC_CHECKSUM_SIZE: if (strcmp(name, "size") == 0) xar->xmlsts = TOC_CHECKSUM; break; case TOC_FILE: if (strcmp(name, "file") == 0) { if (xar->file->parent != NULL && ((xar->file->mode & AE_IFMT) == AE_IFDIR)) xar->file->parent->subdirs++; xar->file = xar->file->parent; if (xar->file == NULL) xar->xmlsts = TOC; } break; case FILE_DATA: if (strcmp(name, "data") == 0) xar->xmlsts = TOC_FILE; break; case FILE_DATA_LENGTH: if (strcmp(name, "length") == 0) xar->xmlsts = FILE_DATA; break; case FILE_DATA_OFFSET: if (strcmp(name, "offset") == 0) xar->xmlsts = FILE_DATA; break; case FILE_DATA_SIZE: if (strcmp(name, "size") == 0) xar->xmlsts = FILE_DATA; break; case FILE_DATA_ENCODING: if (strcmp(name, "encoding") == 0) xar->xmlsts = FILE_DATA; break; case FILE_DATA_A_CHECKSUM: if (strcmp(name, "archived-checksum") == 0) xar->xmlsts = FILE_DATA; break; case FILE_DATA_E_CHECKSUM: if (strcmp(name, "extracted-checksum") == 0) xar->xmlsts = FILE_DATA; break; case FILE_DATA_CONTENT: if (strcmp(name, "content") == 0) xar->xmlsts = FILE_DATA; break; case FILE_EA: if (strcmp(name, "ea") == 0) { xar->xmlsts = TOC_FILE; xar->xattr = NULL; } break; case FILE_EA_LENGTH: if (strcmp(name, "length") == 0) xar->xmlsts = FILE_EA; break; case FILE_EA_OFFSET: if (strcmp(name, "offset") == 0) xar->xmlsts = FILE_EA; break; case FILE_EA_SIZE: if (strcmp(name, "size") == 0) xar->xmlsts = FILE_EA; break; case FILE_EA_ENCODING: if (strcmp(name, "encoding") == 0) xar->xmlsts = FILE_EA; break; case FILE_EA_A_CHECKSUM: if (strcmp(name, "archived-checksum") == 0) xar->xmlsts = FILE_EA; break; case FILE_EA_E_CHECKSUM: if (strcmp(name, "extracted-checksum") == 0) xar->xmlsts = FILE_EA; break; case FILE_EA_NAME: if (strcmp(name, "name") == 0) xar->xmlsts = FILE_EA; break; case FILE_EA_FSTYPE: if (strcmp(name, "fstype") == 0) xar->xmlsts = FILE_EA; break; case FILE_CTIME: if (strcmp(name, "ctime") == 0) xar->xmlsts = TOC_FILE; break; case FILE_MTIME: if (strcmp(name, "mtime") == 0) xar->xmlsts = TOC_FILE; break; case FILE_ATIME: if (strcmp(name, "atime") == 0) xar->xmlsts = TOC_FILE; break; case FILE_GROUP: if (strcmp(name, "group") == 0) xar->xmlsts = TOC_FILE; break; case FILE_GID: if (strcmp(name, "gid") == 0) xar->xmlsts = TOC_FILE; break; case FILE_USER: if (strcmp(name, "user") == 0) xar->xmlsts = TOC_FILE; break; case FILE_UID: if (strcmp(name, "uid") == 0) xar->xmlsts = TOC_FILE; break; case FILE_MODE: if (strcmp(name, "mode") == 0) xar->xmlsts = TOC_FILE; break; case FILE_DEVICE: if (strcmp(name, "device") == 0) xar->xmlsts = TOC_FILE; break; case FILE_DEVICE_MAJOR: if (strcmp(name, "major") == 0) xar->xmlsts = FILE_DEVICE; break; case FILE_DEVICE_MINOR: if (strcmp(name, "minor") == 0) xar->xmlsts = FILE_DEVICE; break; case FILE_DEVICENO: if (strcmp(name, "deviceno") == 0) xar->xmlsts = TOC_FILE; break; case FILE_INODE: if (strcmp(name, "inode") == 0) xar->xmlsts = TOC_FILE; break; case FILE_LINK: if (strcmp(name, "link") == 0) xar->xmlsts = TOC_FILE; break; case FILE_TYPE: if (strcmp(name, "type") == 0) xar->xmlsts = TOC_FILE; break; case FILE_NAME: if (strcmp(name, "name") == 0) xar->xmlsts = TOC_FILE; break; case FILE_ACL: if (strcmp(name, "acl") == 0) xar->xmlsts = TOC_FILE; break; case FILE_ACL_DEFAULT: if (strcmp(name, "default") == 0) xar->xmlsts = FILE_ACL; break; case FILE_ACL_ACCESS: if (strcmp(name, "access") == 0) xar->xmlsts = FILE_ACL; break; case FILE_ACL_APPLEEXTENDED: if (strcmp(name, "appleextended") == 0) xar->xmlsts = FILE_ACL; break; case FILE_FLAGS: if (strcmp(name, "flags") == 0) xar->xmlsts = TOC_FILE; break; case FILE_FLAGS_USER_NODUMP: if (strcmp(name, "UserNoDump") == 0) xar->xmlsts = FILE_FLAGS; break; case FILE_FLAGS_USER_IMMUTABLE: if (strcmp(name, "UserImmutable") == 0) xar->xmlsts = FILE_FLAGS; break; case FILE_FLAGS_USER_APPEND: if (strcmp(name, "UserAppend") == 0) xar->xmlsts = FILE_FLAGS; break; case FILE_FLAGS_USER_OPAQUE: if (strcmp(name, "UserOpaque") == 0) xar->xmlsts = FILE_FLAGS; break; case FILE_FLAGS_USER_NOUNLINK: if (strcmp(name, "UserNoUnlink") == 0) xar->xmlsts = FILE_FLAGS; break; case FILE_FLAGS_SYS_ARCHIVED: if (strcmp(name, "SystemArchived") == 0) xar->xmlsts = FILE_FLAGS; break; case FILE_FLAGS_SYS_IMMUTABLE: if (strcmp(name, "SystemImmutable") == 0) xar->xmlsts = FILE_FLAGS; break; case FILE_FLAGS_SYS_APPEND: if (strcmp(name, "SystemAppend") == 0) xar->xmlsts = FILE_FLAGS; break; case FILE_FLAGS_SYS_NOUNLINK: if (strcmp(name, "SystemNoUnlink") == 0) xar->xmlsts = FILE_FLAGS; break; case FILE_FLAGS_SYS_SNAPSHOT: if (strcmp(name, "SystemSnapshot") == 0) xar->xmlsts = FILE_FLAGS; break; case FILE_EXT2: if (strcmp(name, "ext2") == 0) xar->xmlsts = TOC_FILE; break; case FILE_EXT2_SecureDeletion: if (strcmp(name, "SecureDeletion") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_Undelete: if (strcmp(name, "Undelete") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_Compress: if (strcmp(name, "Compress") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_Synchronous: if (strcmp(name, "Synchronous") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_Immutable: if (strcmp(name, "Immutable") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_AppendOnly: if (strcmp(name, "AppendOnly") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_NoDump: if (strcmp(name, "NoDump") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_NoAtime: if (strcmp(name, "NoAtime") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_CompDirty: if (strcmp(name, "CompDirty") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_CompBlock: if (strcmp(name, "CompBlock") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_NoCompBlock: if (strcmp(name, "NoCompBlock") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_CompError: if (strcmp(name, "CompError") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_BTree: if (strcmp(name, "BTree") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_HashIndexed: if (strcmp(name, "HashIndexed") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_iMagic: if (strcmp(name, "iMagic") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_Journaled: if (strcmp(name, "Journaled") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_NoTail: if (strcmp(name, "NoTail") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_DirSync: if (strcmp(name, "DirSync") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_TopDir: if (strcmp(name, "TopDir") == 0) xar->xmlsts = FILE_EXT2; break; case FILE_EXT2_Reserved: if (strcmp(name, "Reserved") == 0) xar->xmlsts = FILE_EXT2; break; case UNKNOWN: unknowntag_end(xar, name); break; } } static const int base64[256] = { -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* 00 - 0F */ -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* 10 - 1F */ -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 62, -1, -1, -1, 63, /* 20 - 2F */ 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, -1, -1, -1, -1, -1, -1, /* 30 - 3F */ -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, /* 40 - 4F */ 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, -1, -1, -1, -1, -1, /* 50 - 5F */ -1, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, /* 60 - 6F */ 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, -1, -1, -1, -1, -1, /* 70 - 7F */ -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* 80 - 8F */ -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* 90 - 9F */ -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* A0 - AF */ -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* B0 - BF */ -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* C0 - CF */ -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* D0 - DF */ -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* E0 - EF */ -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* F0 - FF */ }; static void strappend_base64(struct xar *xar, struct archive_string *as, const char *s, size_t l) { unsigned char buff[256]; unsigned char *out; const unsigned char *b; size_t len; (void)xar; /* UNUSED */ len = 0; out = buff; b = (const unsigned char *)s; while (l > 0) { int n = 0; if (base64[b[0]] < 0 || base64[b[1]] < 0) break; n = base64[*b++] << 18; n |= base64[*b++] << 12; *out++ = n >> 16; len++; l -= 2; if (l > 0) { if (base64[*b] < 0) break; n |= base64[*b++] << 6; *out++ = (n >> 8) & 0xFF; len++; --l; } if (l > 0) { if (base64[*b] < 0) break; n |= base64[*b++]; *out++ = n & 0xFF; len++; --l; } if (len+3 >= sizeof(buff)) { archive_strncat(as, (const char *)buff, len); len = 0; out = buff; } } if (len > 0) archive_strncat(as, (const char *)buff, len); } static int is_string(const char *known, const char *data, size_t len) { if (strlen(known) != len) return -1; return memcmp(data, known, len); } static void xml_data(void *userData, const char *s, size_t len) { struct archive_read *a; struct xar *xar; a = (struct archive_read *)userData; xar = (struct xar *)(a->format->data); #if DEBUG { char buff[1024]; if (len > (int)(sizeof(buff)-1)) len = (int)(sizeof(buff)-1); strncpy(buff, s, len); buff[len] = 0; fprintf(stderr, "\tlen=%d:\"%s\"\n", len, buff); } #endif switch (xar->xmlsts) { case TOC_CHECKSUM_OFFSET: xar->toc_chksum_offset = atol10(s, len); break; case TOC_CHECKSUM_SIZE: xar->toc_chksum_size = atol10(s, len); break; default: break; } if (xar->file == NULL) return; switch (xar->xmlsts) { case FILE_NAME: if (xar->file->has & HAS_PATHNAME) break; if (xar->file->parent != NULL) { archive_string_concat(&(xar->file->pathname), &(xar->file->parent->pathname)); archive_strappend_char(&(xar->file->pathname), '/'); } xar->file->has |= HAS_PATHNAME; if (xar->base64text) { strappend_base64(xar, &(xar->file->pathname), s, len); } else archive_strncat(&(xar->file->pathname), s, len); break; case FILE_LINK: xar->file->has |= HAS_SYMLINK; archive_strncpy(&(xar->file->symlink), s, len); break; case FILE_TYPE: if (is_string("file", s, len) == 0 || is_string("hardlink", s, len) == 0) xar->file->mode = (xar->file->mode & ~AE_IFMT) | AE_IFREG; if (is_string("directory", s, len) == 0) xar->file->mode = (xar->file->mode & ~AE_IFMT) | AE_IFDIR; if (is_string("symlink", s, len) == 0) xar->file->mode = (xar->file->mode & ~AE_IFMT) | AE_IFLNK; if (is_string("character special", s, len) == 0) xar->file->mode = (xar->file->mode & ~AE_IFMT) | AE_IFCHR; if (is_string("block special", s, len) == 0) xar->file->mode = (xar->file->mode & ~AE_IFMT) | AE_IFBLK; if (is_string("socket", s, len) == 0) xar->file->mode = (xar->file->mode & ~AE_IFMT) | AE_IFSOCK; if (is_string("fifo", s, len) == 0) xar->file->mode = (xar->file->mode & ~AE_IFMT) | AE_IFIFO; xar->file->has |= HAS_TYPE; break; case FILE_INODE: xar->file->has |= HAS_INO; xar->file->ino64 = atol10(s, len); break; case FILE_DEVICE_MAJOR: xar->file->has |= HAS_DEVMAJOR; xar->file->devmajor = (dev_t)atol10(s, len); break; case FILE_DEVICE_MINOR: xar->file->has |= HAS_DEVMINOR; xar->file->devminor = (dev_t)atol10(s, len); break; case FILE_DEVICENO: xar->file->has |= HAS_DEV; xar->file->dev = (dev_t)atol10(s, len); break; case FILE_MODE: xar->file->has |= HAS_MODE; xar->file->mode = (xar->file->mode & AE_IFMT) | ((mode_t)(atol8(s, len)) & ~AE_IFMT); break; case FILE_GROUP: xar->file->has |= HAS_GID; archive_strncpy(&(xar->file->gname), s, len); break; case FILE_GID: xar->file->has |= HAS_GID; xar->file->gid = atol10(s, len); break; case FILE_USER: xar->file->has |= HAS_UID; archive_strncpy(&(xar->file->uname), s, len); break; case FILE_UID: xar->file->has |= HAS_UID; xar->file->uid = atol10(s, len); break; case FILE_CTIME: xar->file->has |= HAS_TIME | HAS_CTIME; xar->file->ctime = parse_time(s, len); break; case FILE_MTIME: xar->file->has |= HAS_TIME | HAS_MTIME; xar->file->mtime = parse_time(s, len); break; case FILE_ATIME: xar->file->has |= HAS_TIME | HAS_ATIME; xar->file->atime = parse_time(s, len); break; case FILE_DATA_LENGTH: xar->file->has |= HAS_DATA; xar->file->length = atol10(s, len); break; case FILE_DATA_OFFSET: xar->file->has |= HAS_DATA; xar->file->offset = atol10(s, len); break; case FILE_DATA_SIZE: xar->file->has |= HAS_DATA; xar->file->size = atol10(s, len); break; case FILE_DATA_A_CHECKSUM: xar->file->a_sum.len = atohex(xar->file->a_sum.val, sizeof(xar->file->a_sum.val), s, len); break; case FILE_DATA_E_CHECKSUM: xar->file->e_sum.len = atohex(xar->file->e_sum.val, sizeof(xar->file->e_sum.val), s, len); break; case FILE_EA_LENGTH: xar->file->has |= HAS_XATTR; xar->xattr->length = atol10(s, len); break; case FILE_EA_OFFSET: xar->file->has |= HAS_XATTR; xar->xattr->offset = atol10(s, len); break; case FILE_EA_SIZE: xar->file->has |= HAS_XATTR; xar->xattr->size = atol10(s, len); break; case FILE_EA_A_CHECKSUM: xar->file->has |= HAS_XATTR; xar->xattr->a_sum.len = atohex(xar->xattr->a_sum.val, sizeof(xar->xattr->a_sum.val), s, len); break; case FILE_EA_E_CHECKSUM: xar->file->has |= HAS_XATTR; xar->xattr->e_sum.len = atohex(xar->xattr->e_sum.val, sizeof(xar->xattr->e_sum.val), s, len); break; case FILE_EA_NAME: xar->file->has |= HAS_XATTR; archive_strncpy(&(xar->xattr->name), s, len); break; case FILE_EA_FSTYPE: xar->file->has |= HAS_XATTR; archive_strncpy(&(xar->xattr->fstype), s, len); break; case FILE_ACL_DEFAULT: case FILE_ACL_ACCESS: case FILE_ACL_APPLEEXTENDED: xar->file->has |= HAS_ACL; /* TODO */ break; case INIT: case XAR: case TOC: case TOC_CREATION_TIME: case TOC_CHECKSUM: case TOC_CHECKSUM_OFFSET: case TOC_CHECKSUM_SIZE: case TOC_FILE: case FILE_DATA: case FILE_DATA_ENCODING: case FILE_DATA_CONTENT: case FILE_DEVICE: case FILE_EA: case FILE_EA_ENCODING: case FILE_ACL: case FILE_FLAGS: case FILE_FLAGS_USER_NODUMP: case FILE_FLAGS_USER_IMMUTABLE: case FILE_FLAGS_USER_APPEND: case FILE_FLAGS_USER_OPAQUE: case FILE_FLAGS_USER_NOUNLINK: case FILE_FLAGS_SYS_ARCHIVED: case FILE_FLAGS_SYS_IMMUTABLE: case FILE_FLAGS_SYS_APPEND: case FILE_FLAGS_SYS_NOUNLINK: case FILE_FLAGS_SYS_SNAPSHOT: case FILE_EXT2: case FILE_EXT2_SecureDeletion: case FILE_EXT2_Undelete: case FILE_EXT2_Compress: case FILE_EXT2_Synchronous: case FILE_EXT2_Immutable: case FILE_EXT2_AppendOnly: case FILE_EXT2_NoDump: case FILE_EXT2_NoAtime: case FILE_EXT2_CompDirty: case FILE_EXT2_CompBlock: case FILE_EXT2_NoCompBlock: case FILE_EXT2_CompError: case FILE_EXT2_BTree: case FILE_EXT2_HashIndexed: case FILE_EXT2_iMagic: case FILE_EXT2_Journaled: case FILE_EXT2_NoTail: case FILE_EXT2_DirSync: case FILE_EXT2_TopDir: case FILE_EXT2_Reserved: case UNKNOWN: break; } } /* * BSD file flags. */ static int xml_parse_file_flags(struct xar *xar, const char *name) { const char *flag = NULL; if (strcmp(name, "UserNoDump") == 0) { xar->xmlsts = FILE_FLAGS_USER_NODUMP; flag = "nodump"; } else if (strcmp(name, "UserImmutable") == 0) { xar->xmlsts = FILE_FLAGS_USER_IMMUTABLE; flag = "uimmutable"; } else if (strcmp(name, "UserAppend") == 0) { xar->xmlsts = FILE_FLAGS_USER_APPEND; flag = "uappend"; } else if (strcmp(name, "UserOpaque") == 0) { xar->xmlsts = FILE_FLAGS_USER_OPAQUE; flag = "opaque"; } else if (strcmp(name, "UserNoUnlink") == 0) { xar->xmlsts = FILE_FLAGS_USER_NOUNLINK; flag = "nouunlink"; } else if (strcmp(name, "SystemArchived") == 0) { xar->xmlsts = FILE_FLAGS_SYS_ARCHIVED; flag = "archived"; } else if (strcmp(name, "SystemImmutable") == 0) { xar->xmlsts = FILE_FLAGS_SYS_IMMUTABLE; flag = "simmutable"; } else if (strcmp(name, "SystemAppend") == 0) { xar->xmlsts = FILE_FLAGS_SYS_APPEND; flag = "sappend"; } else if (strcmp(name, "SystemNoUnlink") == 0) { xar->xmlsts = FILE_FLAGS_SYS_NOUNLINK; flag = "nosunlink"; } else if (strcmp(name, "SystemSnapshot") == 0) { xar->xmlsts = FILE_FLAGS_SYS_SNAPSHOT; flag = "snapshot"; } if (flag == NULL) return (0); xar->file->has |= HAS_FFLAGS; if (archive_strlen(&(xar->file->fflags_text)) > 0) archive_strappend_char(&(xar->file->fflags_text), ','); archive_strcat(&(xar->file->fflags_text), flag); return (1); } /* * Linux file flags. */ static int xml_parse_file_ext2(struct xar *xar, const char *name) { const char *flag = NULL; if (strcmp(name, "SecureDeletion") == 0) { xar->xmlsts = FILE_EXT2_SecureDeletion; flag = "securedeletion"; } else if (strcmp(name, "Undelete") == 0) { xar->xmlsts = FILE_EXT2_Undelete; flag = "nouunlink"; } else if (strcmp(name, "Compress") == 0) { xar->xmlsts = FILE_EXT2_Compress; flag = "compress"; } else if (strcmp(name, "Synchronous") == 0) { xar->xmlsts = FILE_EXT2_Synchronous; flag = "sync"; } else if (strcmp(name, "Immutable") == 0) { xar->xmlsts = FILE_EXT2_Immutable; flag = "simmutable"; } else if (strcmp(name, "AppendOnly") == 0) { xar->xmlsts = FILE_EXT2_AppendOnly; flag = "sappend"; } else if (strcmp(name, "NoDump") == 0) { xar->xmlsts = FILE_EXT2_NoDump; flag = "nodump"; } else if (strcmp(name, "NoAtime") == 0) { xar->xmlsts = FILE_EXT2_NoAtime; flag = "noatime"; } else if (strcmp(name, "CompDirty") == 0) { xar->xmlsts = FILE_EXT2_CompDirty; flag = "compdirty"; } else if (strcmp(name, "CompBlock") == 0) { xar->xmlsts = FILE_EXT2_CompBlock; flag = "comprblk"; } else if (strcmp(name, "NoCompBlock") == 0) { xar->xmlsts = FILE_EXT2_NoCompBlock; flag = "nocomprblk"; } else if (strcmp(name, "CompError") == 0) { xar->xmlsts = FILE_EXT2_CompError; flag = "comperr"; } else if (strcmp(name, "BTree") == 0) { xar->xmlsts = FILE_EXT2_BTree; flag = "btree"; } else if (strcmp(name, "HashIndexed") == 0) { xar->xmlsts = FILE_EXT2_HashIndexed; flag = "hashidx"; } else if (strcmp(name, "iMagic") == 0) { xar->xmlsts = FILE_EXT2_iMagic; flag = "imagic"; } else if (strcmp(name, "Journaled") == 0) { xar->xmlsts = FILE_EXT2_Journaled; flag = "journal"; } else if (strcmp(name, "NoTail") == 0) { xar->xmlsts = FILE_EXT2_NoTail; flag = "notail"; } else if (strcmp(name, "DirSync") == 0) { xar->xmlsts = FILE_EXT2_DirSync; flag = "dirsync"; } else if (strcmp(name, "TopDir") == 0) { xar->xmlsts = FILE_EXT2_TopDir; flag = "topdir"; } else if (strcmp(name, "Reserved") == 0) { xar->xmlsts = FILE_EXT2_Reserved; flag = "reserved"; } if (flag == NULL) return (0); if (archive_strlen(&(xar->file->fflags_text)) > 0) archive_strappend_char(&(xar->file->fflags_text), ','); archive_strcat(&(xar->file->fflags_text), flag); return (1); } #ifdef HAVE_LIBXML_XMLREADER_H static int xml2_xmlattr_setup(struct archive_read *a, struct xmlattr_list *list, xmlTextReaderPtr reader) { struct xmlattr *attr; int r; list->first = NULL; list->last = &(list->first); r = xmlTextReaderMoveToFirstAttribute(reader); while (r == 1) { attr = malloc(sizeof*(attr)); if (attr == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } attr->name = strdup( (const char *)xmlTextReaderConstLocalName(reader)); if (attr->name == NULL) { free(attr); archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } attr->value = strdup( (const char *)xmlTextReaderConstValue(reader)); if (attr->value == NULL) { free(attr->name); free(attr); archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } attr->next = NULL; *list->last = attr; list->last = &(attr->next); r = xmlTextReaderMoveToNextAttribute(reader); } return (r); } static int xml2_read_cb(void *context, char *buffer, int len) { struct archive_read *a; struct xar *xar; const void *d; size_t outbytes; size_t used = 0; int r; a = (struct archive_read *)context; xar = (struct xar *)(a->format->data); if (xar->toc_remaining <= 0) return (0); d = buffer; outbytes = len; r = rd_contents(a, &d, &outbytes, &used, xar->toc_remaining); if (r != ARCHIVE_OK) return (r); __archive_read_consume(a, used); xar->toc_remaining -= used; xar->offset += used; xar->toc_total += outbytes; PRINT_TOC(buffer, len); return ((int)outbytes); } static int xml2_close_cb(void *context) { (void)context; /* UNUSED */ return (0); } static void xml2_error_hdr(void *arg, const char *msg, xmlParserSeverities severity, xmlTextReaderLocatorPtr locator) { struct archive_read *a; (void)locator; /* UNUSED */ a = (struct archive_read *)arg; switch (severity) { case XML_PARSER_SEVERITY_VALIDITY_WARNING: case XML_PARSER_SEVERITY_WARNING: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "XML Parsing error: %s", msg); break; case XML_PARSER_SEVERITY_VALIDITY_ERROR: case XML_PARSER_SEVERITY_ERROR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "XML Parsing error: %s", msg); break; } } static int xml2_read_toc(struct archive_read *a) { xmlTextReaderPtr reader; struct xmlattr_list list; int r; reader = xmlReaderForIO(xml2_read_cb, xml2_close_cb, a, NULL, NULL, 0); if (reader == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory for xml parser"); return (ARCHIVE_FATAL); } xmlTextReaderSetErrorHandler(reader, xml2_error_hdr, a); while ((r = xmlTextReaderRead(reader)) == 1) { const char *name, *value; int type, empty; type = xmlTextReaderNodeType(reader); name = (const char *)xmlTextReaderConstLocalName(reader); switch (type) { case XML_READER_TYPE_ELEMENT: empty = xmlTextReaderIsEmptyElement(reader); r = xml2_xmlattr_setup(a, &list, reader); if (r == ARCHIVE_OK) r = xml_start(a, name, &list); xmlattr_cleanup(&list); if (r != ARCHIVE_OK) { xmlFreeTextReader(reader); xmlCleanupParser(); return (r); } if (empty) xml_end(a, name); break; case XML_READER_TYPE_END_ELEMENT: xml_end(a, name); break; case XML_READER_TYPE_TEXT: value = (const char *)xmlTextReaderConstValue(reader); xml_data(a, value, strlen(value)); break; case XML_READER_TYPE_SIGNIFICANT_WHITESPACE: default: break; } if (r < 0) break; } xmlFreeTextReader(reader); xmlCleanupParser(); return ((r == 0)?ARCHIVE_OK:ARCHIVE_FATAL); } #elif defined(HAVE_BSDXML_H) || defined(HAVE_EXPAT_H) static int expat_xmlattr_setup(struct archive_read *a, struct xmlattr_list *list, const XML_Char **atts) { struct xmlattr *attr; char *name, *value; list->first = NULL; list->last = &(list->first); if (atts == NULL) return (ARCHIVE_OK); while (atts[0] != NULL && atts[1] != NULL) { attr = malloc(sizeof*(attr)); name = strdup(atts[0]); value = strdup(atts[1]); if (attr == NULL || name == NULL || value == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); free(attr); free(name); free(value); return (ARCHIVE_FATAL); } attr->name = name; attr->value = value; attr->next = NULL; *list->last = attr; list->last = &(attr->next); atts += 2; } return (ARCHIVE_OK); } static void expat_start_cb(void *userData, const XML_Char *name, const XML_Char **atts) { struct expat_userData *ud = (struct expat_userData *)userData; struct archive_read *a = ud->archive; struct xmlattr_list list; int r; if (ud->state != ARCHIVE_OK) return; r = expat_xmlattr_setup(a, &list, atts); if (r == ARCHIVE_OK) r = xml_start(a, (const char *)name, &list); xmlattr_cleanup(&list); ud->state = r; } static void expat_end_cb(void *userData, const XML_Char *name) { struct expat_userData *ud = (struct expat_userData *)userData; xml_end(ud->archive, (const char *)name); } static void expat_data_cb(void *userData, const XML_Char *s, int len) { struct expat_userData *ud = (struct expat_userData *)userData; xml_data(ud->archive, s, (size_t)len); } static int expat_read_toc(struct archive_read *a) { struct xar *xar; XML_Parser parser; struct expat_userData ud; ud.state = ARCHIVE_OK; ud.archive = a; xar = (struct xar *)(a->format->data); /* Initialize XML Parser library. */ parser = XML_ParserCreate(NULL); if (parser == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory for xml parser"); return (ARCHIVE_FATAL); } XML_SetUserData(parser, &ud); XML_SetElementHandler(parser, expat_start_cb, expat_end_cb); XML_SetCharacterDataHandler(parser, expat_data_cb); xar->xmlsts = INIT; while (xar->toc_remaining && ud.state == ARCHIVE_OK) { enum XML_Status xr; const void *d; size_t outbytes; size_t used; int r; d = NULL; r = rd_contents(a, &d, &outbytes, &used, xar->toc_remaining); if (r != ARCHIVE_OK) { XML_ParserFree(parser); return (r); } xar->toc_remaining -= used; xar->offset += used; xar->toc_total += outbytes; PRINT_TOC(d, outbytes); xr = XML_Parse(parser, d, (int)outbytes, xar->toc_remaining == 0); __archive_read_consume(a, used); if (xr == XML_STATUS_ERROR) { XML_ParserFree(parser); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "XML Parsing failed"); return (ARCHIVE_FATAL); } } XML_ParserFree(parser); return (ud.state); } #elif defined(HAVE_XMLLITE_H) struct ArchiveStreamAdapter { const ISequentialStreamVtbl *lpVtbl; /* see asaStaticVtable */ struct archive_read *a; }; static HRESULT STDMETHODCALLTYPE asaQueryInterface(ISequentialStream *this, REFIID riid, void **ppv) { if (!IsEqualIID(riid, &IID_ISequentialStream)) { *ppv = NULL; return E_NOINTERFACE; } *ppv = this; return S_OK; } /* * We can dispense with reference counting as we tightly manage the lifetime * of an ArchiveStreamAdapter. */ static ULONG STDMETHODCALLTYPE asaAddRef(ISequentialStream *this) { (void)this; /* UNUSED */ return ULONG_MAX; } static ULONG STDMETHODCALLTYPE asaRelease(ISequentialStream *this) { (void)this; /* UNUSED */ return ULONG_MAX; } static HRESULT STDMETHODCALLTYPE asaRead(ISequentialStream *this, void *pv, ULONG cb, ULONG *pcbRead) { struct ArchiveStreamAdapter *asa = (struct ArchiveStreamAdapter *)this; struct archive_read *a; struct xar *xar; const void *d = pv; size_t outbytes = cb; size_t used = 0; int r; a = asa->a; xar = (struct xar *)(a->format->data); *pcbRead = 0; if (xar->toc_remaining <= 0) return cb != 0 ? S_FALSE : S_OK; r = rd_contents(a, &d, &outbytes, &used, xar->toc_remaining); if (r != ARCHIVE_OK) return E_FAIL; __archive_read_consume(a, used); xar->toc_remaining -= used; xar->offset += used; xar->toc_total += outbytes; PRINT_TOC(pv, outbytes); *pcbRead = (ULONG)outbytes; return outbytes < cb ? S_FALSE : S_OK; } static HRESULT STDMETHODCALLTYPE asaWrite(ISequentialStream *this, const void *pv, ULONG cb, ULONG *pcbWritten) { (void)this; /* UNUSED */ (void)pv; /* UNUSED */ (void)cb; /* UNUSED */ if (!pcbWritten) return E_INVALIDARG; *pcbWritten = 0; return E_NOTIMPL; } static const ISequentialStreamVtbl asaStaticVtable = { .QueryInterface = asaQueryInterface, .AddRef = asaAddRef, .Release = asaRelease, .Read = asaRead, .Write = asaWrite, }; static int xmllite_create_stream_adapter(struct archive_read *a, struct ArchiveStreamAdapter **pasa) { struct ArchiveStreamAdapter *asa = calloc(1, sizeof(struct ArchiveStreamAdapter)); if (!asa) { archive_set_error(&(a->archive), ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } asa->lpVtbl = &asaStaticVtable; asa->a = a; *pasa = asa; return (ARCHIVE_OK); } typedef HRESULT(STDMETHODCALLTYPE *xmllite_wstr_func)(IXmlReader *, LPCWSTR *, UINT *); /* * Returns an narrow-char archive_string in *as after calling * the wide-char COM API callee() on the XmlReader reader. * Sets an appropriate error on the archive if it fails. */ static int xmllite_call_return_as(struct archive_read *a, struct archive_string *as, IXmlReader *reader, xmllite_wstr_func callee) { LPCWSTR wcs; UINT wlen; if (FAILED(callee(reader, &wcs, &wlen))) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Failed to read XML data"); return (ARCHIVE_FATAL); } archive_string_init(as); if (archive_string_append_from_wcs(as, wcs, (size_t)wlen) < 0) { archive_string_free(as); archive_set_error(&(a->archive), ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } static char * xmllite_call_return_mbs(struct archive_read *a, IXmlReader *reader, xmllite_wstr_func callee) { char *ret; struct archive_string as; if (xmllite_call_return_as(a, &as, reader, callee) < 0) { return NULL; } ret = strdup(as.s); archive_string_free(&as); if (ret == NULL) { archive_set_error(&(a->archive), ENOMEM, "Out of memory"); return NULL; } return ret; } static int xmllite_xmlattr_setup(struct archive_read *a, struct xmlattr_list *list, IXmlReader *reader) { struct xmlattr *attr; HRESULT hr; list->first = NULL; list->last = &(list->first); hr = reader->lpVtbl->MoveToFirstAttribute(reader); /* Contrary to other checks, we're not using SUCCEEDED/FAILED * because MoveToNextAttribute returns *S_FALSE* (success!) * when it runs out of attributes. */ while (hr == S_OK) { /* Attributes implied as being default by the DTD are ignored */ if (reader->lpVtbl->IsDefault(reader)) continue; attr = malloc(sizeof*(attr)); if (attr == NULL) { archive_set_error(&(a->archive), ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } attr->name = xmllite_call_return_mbs(a, reader, reader->lpVtbl->GetLocalName); if (attr->name == NULL) { free(attr); /* xmllite_call_return_mbs sets an appropriate error */ return (ARCHIVE_FATAL); } attr->value = xmllite_call_return_mbs(a, reader, reader->lpVtbl->GetValue); if (attr->value == NULL) { free(attr->name); free(attr); /* xmllite_call_return_mbs sets an appropriate error */ return (ARCHIVE_FATAL); } attr->next = NULL; *list->last = attr; list->last = &(attr->next); hr = reader->lpVtbl->MoveToNextAttribute(reader); } if (FAILED(hr)) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_FILE_FORMAT, "Failed to parse XML document"); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } static int xmllite_read_toc(struct archive_read *a) { struct ArchiveStreamAdapter *asa = NULL; char *name; struct archive_string as; BOOL empty; XmlNodeType type; struct xmlattr_list list; IXmlReader *reader = NULL; int r = ARCHIVE_OK; if ((r = xmllite_create_stream_adapter(a, &asa)) < 0) { goto out; } if (FAILED(CreateXmlReader(&IID_IXmlReader, (void **)&reader, NULL))) { r = ARCHIVE_FATAL; goto out; } if (FAILED(reader->lpVtbl->SetInput(reader, (IUnknown *)asa))) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Failed to prepare XML stream"); r = ARCHIVE_FATAL; goto out; } while (!reader->lpVtbl->IsEOF(reader)) { if (FAILED(reader->lpVtbl->Read(reader, &type))) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Failed to read XML stream"); r = ARCHIVE_FATAL; goto out; } switch (type) { case XmlNodeType_Element: empty = reader->lpVtbl->IsEmptyElement(reader); name = xmllite_call_return_mbs(a, reader, reader->lpVtbl->GetLocalName); if (name == NULL) { /* xmllite_call_return_mbs sets an appropriate error */ r = ARCHIVE_FATAL; goto out; } r = xmllite_xmlattr_setup(a, &list, reader); if (r == ARCHIVE_OK) { r = xml_start(a, name, &list); } xmlattr_cleanup(&list); if (r == ARCHIVE_OK && empty) { xml_end(a, name); } free(name); if (r != ARCHIVE_OK) { goto out; } break; case XmlNodeType_EndElement: name = xmllite_call_return_mbs(a, reader, reader->lpVtbl->GetLocalName); if (name == NULL) { /* xmllite_call_return_mbs sets an appropriate error */ r = ARCHIVE_FATAL; goto out; } xml_end(a, name); free(name); break; case XmlNodeType_Text: r = xmllite_call_return_as(a, &as, reader, reader->lpVtbl->GetValue); if (r != ARCHIVE_OK) { /* xmllite_call_return_as sets an appropriate error */ goto out; } xml_data(a, as.s, (int)archive_strlen(&as)); archive_string_free(&as); case XmlNodeType_None: case XmlNodeType_Attribute: case XmlNodeType_CDATA: case XmlNodeType_ProcessingInstruction: case XmlNodeType_Comment: case XmlNodeType_DocumentType: case XmlNodeType_Whitespace: case XmlNodeType_XmlDeclaration: default: break; } } out: if (reader) reader->lpVtbl->Release(reader); free(asa); return r; } #endif /* defined(XMLLITE) */ #endif /* Support xar format */ diff --git a/contrib/libarchive/libarchive/archive_read_support_format_zip.c b/contrib/libarchive/libarchive/archive_read_support_format_zip.c index daf51933d687..9abd55709e3f 100644 --- a/contrib/libarchive/libarchive/archive_read_support_format_zip.c +++ b/contrib/libarchive/libarchive/archive_read_support_format_zip.c @@ -1,4406 +1,4406 @@ /*- * Copyright (c) 2004-2013 Tim Kientzle * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA * Copyright (c) 2013 Konrad Kleine * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" /* * The definitive documentation of the Zip file format is: * http://www.pkware.com/documents/casestudies/APPNOTE.TXT * * The Info-Zip project has pioneered various extensions to better * support Zip on Unix, including the 0x5455 "UT", 0x5855 "UX", 0x7855 * "Ux", and 0x7875 "ux" extensions for time and ownership * information. * * History of this code: The streaming Zip reader was first added to * libarchive in January 2005. Support for seekable input sources was * added in Nov 2011. Zip64 support (including a significant code * refactoring) was added in 2014. */ #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #ifdef HAVE_BZLIB_H #include #endif #ifdef HAVE_LZMA_H #include #endif #ifdef HAVE_ZSTD_H #include #endif #include "archive.h" #include "archive_digest_private.h" #include "archive_cryptor_private.h" #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_hmac_private.h" #include "archive_private.h" #include "archive_rb.h" #include "archive_read_private.h" #include "archive_time_private.h" #include "archive_ppmd8_private.h" #ifndef HAVE_ZLIB_H #include "archive_crc32.h" #endif struct zip_entry { struct archive_rb_node node; struct zip_entry *next; int64_t local_header_offset; int64_t compressed_size; int64_t uncompressed_size; int64_t gid; int64_t uid; struct archive_string rsrcname; time_t mtime; time_t atime; time_t ctime; uint32_t crc32; uint16_t mode; uint16_t zip_flags; /* From GP Flags Field */ unsigned char compression; unsigned char system; /* From "version written by" */ unsigned char flags; /* Our extra markers. */ unsigned char decdat;/* Used for Decryption check */ /* WinZip AES encryption extra field should be available * when compression is 99. */ struct { /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */ unsigned vendor; #define AES_VENDOR_AE_1 0x0001 #define AES_VENDOR_AE_2 0x0002 /* AES encryption strength: * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */ unsigned strength; /* Actual compression method. */ unsigned char compression; } aes_extra; }; struct trad_enc_ctx { uint32_t keys[3]; }; /* Bits used in zip_flags. */ #define ZIP_ENCRYPTED (1 << 0) #define ZIP_LENGTH_AT_END (1 << 3) /* Also called "Streaming bit" */ #define ZIP_STRONG_ENCRYPTED (1 << 6) #define ZIP_UTF8_NAME (1 << 11) /* See "7.2 Single Password Symmetric Encryption Method" in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */ #define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13) /* Bits used in flags. */ #define LA_USED_ZIP64 (1 << 0) #define LA_FROM_CENTRAL_DIRECTORY (1 << 1) /* * See "WinZip - AES Encryption Information" * http://www.winzip.com/aes_info.htm */ /* Value used in compression method. */ #define WINZIP_AES_ENCRYPTION 99 /* Authentication code size. */ #define AUTH_CODE_SIZE 10 /**/ #define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) struct zip { /* Structural information about the archive. */ struct archive_string format_name; int64_t central_directory_offset; int64_t central_directory_offset_adjusted; size_t central_directory_entries_total; size_t central_directory_entries_on_this_disk; int has_encrypted_entries; /* List of entries (seekable Zip only) */ struct zip_entry *zip_entries; struct archive_rb_tree tree; struct archive_rb_tree tree_rsrc; /* Bytes read but not yet consumed via __archive_read_consume() */ size_t unconsumed; /* Information about entry we're currently reading. */ struct zip_entry *entry; int64_t entry_bytes_remaining; /* These count the number of bytes actually read for the entry. */ int64_t entry_compressed_bytes_read; int64_t entry_uncompressed_bytes_read; /* Running CRC32 of the decompressed and decrypted data */ unsigned long computed_crc32; unsigned long (*crc32func)(unsigned long, const void *, size_t); char ignore_crc32; /* Flags to mark progress of decompression. */ char decompress_init; char end_of_entry; unsigned char *uncompressed_buffer; size_t uncompressed_buffer_size; #ifdef HAVE_ZLIB_H z_stream stream; char stream_valid; #endif #if HAVE_LZMA_H && HAVE_LIBLZMA lzma_stream zipx_lzma_stream; char zipx_lzma_valid; #endif #ifdef HAVE_BZLIB_H bz_stream bzstream; char bzstream_valid; #endif #if HAVE_ZSTD_H && HAVE_LIBZSTD ZSTD_DStream *zstdstream; char zstdstream_valid; #endif IByteIn zipx_ppmd_stream; ssize_t zipx_ppmd_read_compressed; CPpmd8 ppmd8; char ppmd8_valid; char ppmd8_stream_failed; struct archive_string_conv *sconv; struct archive_string_conv *sconv_default; struct archive_string_conv *sconv_utf8; int init_default_conversion; int process_mac_extensions; char init_decryption; /* Decryption buffer. */ /* * The decrypted data starts at decrypted_ptr and * extends for decrypted_bytes_remaining. Decryption * adds new data to the end of this block, data is returned * to clients from the beginning. When the block hits the * end of decrypted_buffer, it has to be shuffled back to * the beginning of the buffer. */ unsigned char *decrypted_buffer; unsigned char *decrypted_ptr; size_t decrypted_buffer_size; size_t decrypted_bytes_remaining; size_t decrypted_unconsumed_bytes; /* Traditional PKWARE decryption. */ struct trad_enc_ctx tctx; char tctx_valid; /* WinZip AES decryption. */ /* Contexts used for AES decryption. */ archive_crypto_ctx cctx; char cctx_valid; archive_hmac_sha1_ctx hctx; char hctx_valid; /* Strong encryption's decryption header information. */ unsigned iv_size; unsigned alg_id; unsigned bit_len; unsigned flags; unsigned erd_size; unsigned v_size; unsigned v_crc32; uint8_t *iv; uint8_t *erd; uint8_t *v_data; }; /* Many systems define min or MIN, but not all. */ #define zipmin(a,b) ((a) < (b) ? (a) : (b)) #ifdef HAVE_ZLIB_H static int zip_read_data_deflate(struct archive_read *a, const void **buff, size_t *size, int64_t *offset); #endif #if HAVE_LZMA_H && HAVE_LIBLZMA static int zip_read_data_zipx_lzma_alone(struct archive_read *a, const void **buff, size_t *size, int64_t *offset); #endif /* This function is used by Ppmd8_DecodeSymbol during decompression of Ppmd8 * streams inside ZIP files. It has 2 purposes: one is to fetch the next * compressed byte from the stream, second one is to increase the counter how * many compressed bytes were read. */ static Byte ppmd_read(void* p) { /* Get the handle to current decompression context. */ struct archive_read *a = ((IByteIn*)p)->a; struct zip *zip = (struct zip*) a->format->data; ssize_t bytes_avail = 0; /* Fetch next byte. */ const uint8_t* data = __archive_read_ahead(a, 1, &bytes_avail); if(bytes_avail < 1) { zip->ppmd8_stream_failed = 1; return 0; } __archive_read_consume(a, 1); /* Increment the counter. */ ++zip->zipx_ppmd_read_compressed; /* Return the next compressed byte. */ return data[0]; } /* ------------------------------------------------------------------------ */ /* Traditional PKWARE Decryption functions. */ static void trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) { uint8_t t; #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) ctx->keys[0] = CRC32(ctx->keys[0], c); ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; t = (ctx->keys[1] >> 24) & 0xff; ctx->keys[2] = CRC32(ctx->keys[2], t); #undef CRC32 } static uint8_t trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) { unsigned temp = ctx->keys[2] | 2; return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; } static void trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, size_t in_len, uint8_t *out, size_t out_len) { unsigned i, max; max = (unsigned)((in_len < out_len)? in_len: out_len); for (i = 0; i < max; i++) { uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx); out[i] = t; trad_enc_update_keys(ctx, t); } } static int trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len, const uint8_t *key, size_t key_len, uint8_t *crcchk) { uint8_t header[12]; if (key_len < 12) { *crcchk = 0xff; return -1; } ctx->keys[0] = 305419896L; ctx->keys[1] = 591751049L; ctx->keys[2] = 878082192L; for (;pw_len; --pw_len) trad_enc_update_keys(ctx, *pw++); trad_enc_decrypt_update(ctx, key, 12, header, 12); /* Return the last byte for CRC check. */ *crcchk = header[11]; return 0; } #if 0 static void crypt_derive_key_sha1(const void *p, int size, unsigned char *key, int key_size) { #define MD_SIZE 20 archive_sha1_ctx ctx; unsigned char md1[MD_SIZE]; unsigned char md2[MD_SIZE * 2]; unsigned char mkb[64]; int i; archive_sha1_init(&ctx); archive_sha1_update(&ctx, p, size); archive_sha1_final(&ctx, md1); memset(mkb, 0x36, sizeof(mkb)); for (i = 0; i < MD_SIZE; i++) mkb[i] ^= md1[i]; archive_sha1_init(&ctx); archive_sha1_update(&ctx, mkb, sizeof(mkb)); archive_sha1_final(&ctx, md2); memset(mkb, 0x5C, sizeof(mkb)); for (i = 0; i < MD_SIZE; i++) mkb[i] ^= md1[i]; archive_sha1_init(&ctx); archive_sha1_update(&ctx, mkb, sizeof(mkb)); archive_sha1_final(&ctx, md2 + MD_SIZE); if (key_size > 32) key_size = 32; memcpy(key, md2, key_size); #undef MD_SIZE } #endif /* * Common code for streaming or seeking modes. * * Includes code to read local file headers, decompress data * from entry bodies, and common API. */ static unsigned long real_crc32(unsigned long crc, const void *buff, size_t len) { return crc32(crc, buff, (unsigned int)len); } /* Used by "ignorecrc32" option to speed up tests. */ static unsigned long fake_crc32(unsigned long crc, const void *buff, size_t len) { (void)crc; /* UNUSED */ (void)buff; /* UNUSED */ (void)len; /* UNUSED */ return 0; } static const struct { int id; const char * name; } compression_methods[] = { {0, "uncompressed"}, /* The file is stored (no compression) */ {1, "shrinking"}, /* The file is Shrunk */ {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */ {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */ {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */ {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */ {6, "imploded"}, /* The file is Imploded */ {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */ {8, "deflation"}, /* The file is Deflated */ {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */ {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding * (old IBM TERSE) */ {11, "reserved"}, /* Reserved by PKWARE */ {12, "bzip"}, /* File is compressed using BZIP2 algorithm */ {13, "reserved"}, /* Reserved by PKWARE */ {14, "lzma"}, /* LZMA (EFS) */ {15, "reserved"}, /* Reserved by PKWARE */ {16, "reserved"}, /* Reserved by PKWARE */ {17, "reserved"}, /* Reserved by PKWARE */ {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */ {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */ {93, "zstd"}, /* Zstandard (zstd) Compression */ {95, "xz"}, /* XZ compressed data */ {96, "jpeg"}, /* JPEG compressed data */ {97, "wav-pack"}, /* WavPack compressed data */ {98, "ppmd-1"}, /* PPMd version I, Rev 1 */ {99, "aes"} /* WinZip AES encryption */ }; static const char * compression_name(const int compression) { static const int num_compression_methods = sizeof(compression_methods)/sizeof(compression_methods[0]); int i=0; while(compression >= 0 && i < num_compression_methods) { if (compression_methods[i].id == compression) return compression_methods[i].name; i++; } return "??"; } /* * The extra data is stored as a list of * id1+size1+data1 + id2+size2+data2 ... * triplets. id and size are 2 bytes each. */ static int process_extra(struct archive_read *a, struct archive_entry *entry, const char *p, size_t extra_length, struct zip_entry* zip_entry) { unsigned offset = 0; struct zip *zip = (struct zip *)(a->format->data); if (extra_length == 0) { return ARCHIVE_OK; } if (extra_length < 4) { size_t i = 0; /* Some ZIP files may have trailing 0 bytes. Let's check they * are all 0 and ignore them instead of returning an error. * * This is not technically correct, but some ZIP files look * like this and other tools support those files - so let's * also support them. */ for (; i < extra_length; i++) { if (p[i] != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Too-small extra data: " "Need at least 4 bytes, " "but only found %d bytes", (int)extra_length); return ARCHIVE_FAILED; } } return ARCHIVE_OK; } while (offset <= extra_length - 4) { unsigned short headerid = archive_le16dec(p + offset); unsigned short datasize = archive_le16dec(p + offset + 2); offset += 4; if (offset + datasize > extra_length) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Extra data overflow: " "Need %d bytes but only found %d bytes", (int)datasize, (int)(extra_length - offset)); return ARCHIVE_FAILED; } #ifdef DEBUG fprintf(stderr, "Header id 0x%04x, length %d\n", headerid, datasize); #endif switch (headerid) { case 0x0001: /* Zip64 extended information extra field. */ zip_entry->flags |= LA_USED_ZIP64; if (zip_entry->uncompressed_size == 0xffffffff) { uint64_t t = 0; if (datasize < 8 || (t = archive_le64dec(p + offset)) > INT64_MAX) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Malformed 64-bit " "uncompressed size"); return ARCHIVE_FAILED; } zip_entry->uncompressed_size = t; offset += 8; datasize -= 8; } if (zip_entry->compressed_size == 0xffffffff) { uint64_t t = 0; if (datasize < 8 || (t = archive_le64dec(p + offset)) > INT64_MAX) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Malformed 64-bit " "compressed size"); return ARCHIVE_FAILED; } zip_entry->compressed_size = t; offset += 8; datasize -= 8; } if (zip_entry->local_header_offset == 0xffffffff) { uint64_t t = 0; if (datasize < 8 || (t = archive_le64dec(p + offset)) > INT64_MAX) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Malformed 64-bit " "local header offset"); return ARCHIVE_FAILED; } zip_entry->local_header_offset = t; offset += 8; datasize -= 8; } /* archive_le32dec(p + offset) gives disk * on which file starts, but we don't handle * multi-volume Zip files. */ break; #ifdef DEBUG case 0x0017: { /* Strong encryption field. */ if (archive_le16dec(p + offset) == 2) { unsigned algId = archive_le16dec(p + offset + 2); unsigned bitLen = archive_le16dec(p + offset + 4); int flags = archive_le16dec(p + offset + 6); fprintf(stderr, "algId=0x%04x, bitLen=%u, " "flgas=%d\n", algId, bitLen,flags); } break; } #endif case 0x5455: { /* Extended time field "UT". */ int flags; if (datasize == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Incomplete extended time field"); return ARCHIVE_FAILED; } flags = p[offset]; offset++; datasize--; /* Flag bits indicate which dates are present. */ if (flags & 0x01) { #ifdef DEBUG fprintf(stderr, "mtime: %lld -> %d\n", (long long)zip_entry->mtime, archive_le32dec(p + offset)); #endif if (datasize < 4) break; zip_entry->mtime = archive_le32dec(p + offset); offset += 4; datasize -= 4; } if (flags & 0x02) { if (datasize < 4) break; zip_entry->atime = archive_le32dec(p + offset); offset += 4; datasize -= 4; } if (flags & 0x04) { if (datasize < 4) break; zip_entry->ctime = archive_le32dec(p + offset); offset += 4; datasize -= 4; } break; } case 0x5855: { /* Info-ZIP Unix Extra Field (old version) "UX". */ if (datasize >= 8) { zip_entry->atime = archive_le32dec(p + offset); zip_entry->mtime = archive_le32dec(p + offset + 4); } if (datasize >= 12) { zip_entry->uid = archive_le16dec(p + offset + 8); zip_entry->gid = archive_le16dec(p + offset + 10); } break; } case 0x6c78: { /* Experimental 'xl' field */ /* * Introduced Dec 2013 to provide a way to * include external file attributes (and other * fields that ordinarily appear only in * central directory) in local file header. * This provides file type and permission * information necessary to support full * streaming extraction. Currently being * discussed with other Zip developers * ... subject to change. * * Format: * The field starts with a bitmap that specifies * which additional fields are included. The * bitmap is variable length and can be extended in * the future. * * n bytes - feature bitmap: first byte has low-order * 7 bits. If high-order bit is set, a subsequent * byte holds the next 7 bits, etc. * * if bitmap & 1, 2 byte "version made by" * if bitmap & 2, 2 byte "internal file attributes" * if bitmap & 4, 4 byte "external file attributes" * if bitmap & 8, 2 byte comment length + n byte * comment */ int bitmap, bitmap_last; if (datasize < 1) break; bitmap_last = bitmap = 0xff & p[offset]; offset += 1; datasize -= 1; /* We only support first 7 bits of bitmap; skip rest. */ while ((bitmap_last & 0x80) != 0 && datasize >= 1) { bitmap_last = p[offset]; offset += 1; datasize -= 1; } if (bitmap & 1) { /* 2 byte "version made by" */ if (datasize < 2) break; zip_entry->system = archive_le16dec(p + offset) >> 8; offset += 2; datasize -= 2; } if (bitmap & 2) { /* 2 byte "internal file attributes" */ uint32_t internal_attributes; if (datasize < 2) break; internal_attributes = archive_le16dec(p + offset); /* Not used by libarchive at present. */ (void)internal_attributes; /* UNUSED */ offset += 2; datasize -= 2; } if (bitmap & 4) { /* 4 byte "external file attributes" */ uint32_t external_attributes; if (datasize < 4) break; external_attributes = archive_le32dec(p + offset); if (zip_entry->system == 3) { zip_entry->mode = external_attributes >> 16; } else if (zip_entry->system == 0) { // Interpret MSDOS directory bit if (0x10 == (external_attributes & 0x10)) { zip_entry->mode = AE_IFDIR | 0775; } else { zip_entry->mode = AE_IFREG | 0664; } if (0x01 == (external_attributes & 0x01)) { /* Read-only bit; * strip write permissions */ zip_entry->mode &= 0555; } } else { zip_entry->mode = 0; } offset += 4; datasize -= 4; } if (bitmap & 8) { /* 2 byte comment length + comment */ uint32_t comment_length; if (datasize < 2) break; comment_length = archive_le16dec(p + offset); offset += 2; datasize -= 2; if (datasize < comment_length) break; /* Comment is not supported by libarchive */ offset += comment_length; datasize -= comment_length; } break; } case 0x7075: { /* Info-ZIP Unicode Path Extra Field. */ if (datasize < 5 || entry == NULL) break; offset += 5; datasize -= 5; /* The path name in this field is always encoded * in UTF-8. */ if (zip->sconv_utf8 == NULL) { zip->sconv_utf8 = archive_string_conversion_from_charset( &a->archive, "UTF-8", 1); /* If the converter from UTF-8 is not * available, then the path name from the main * field will more likely be correct. */ if (zip->sconv_utf8 == NULL) break; } /* Make sure the CRC32 of the filename matches. */ if (!zip->ignore_crc32) { const char *cp = archive_entry_pathname(entry); if (cp) { unsigned long file_crc = zip->crc32func(0, cp, strlen(cp)); unsigned long utf_crc = archive_le32dec(p + offset - 4); if (file_crc != utf_crc) { #ifdef DEBUG fprintf(stderr, "CRC filename mismatch; " "CDE is %lx, but UTF8 " "is outdated with %lx\n", file_crc, utf_crc); #endif break; } } } if (archive_entry_copy_pathname_l(entry, p + offset, datasize, zip->sconv_utf8) != 0) { /* Ignore the error, and fallback to the path * name from the main field. */ #ifdef DEBUG fprintf(stderr, "Failed to read the ZIP " "0x7075 extra field path.\n"); #endif } break; } case 0x7855: /* Info-ZIP Unix Extra Field (type 2) "Ux". */ #ifdef DEBUG fprintf(stderr, "uid %d gid %d\n", archive_le16dec(p + offset), archive_le16dec(p + offset + 2)); #endif if (datasize >= 2) zip_entry->uid = archive_le16dec(p + offset); if (datasize >= 4) zip_entry->gid = archive_le16dec(p + offset + 2); break; case 0x7875: { /* Info-Zip Unix Extra Field (type 3) "ux". */ int uidsize = 0, gidsize = 0; /* TODO: support arbitrary uidsize/gidsize. */ if (datasize >= 1 && p[offset] == 1) {/* version=1 */ if (datasize >= 4) { /* get a uid size. */ uidsize = 0xff & (int)p[offset+1]; if (uidsize == 2) zip_entry->uid = archive_le16dec( p + offset + 2); else if (uidsize == 4 && datasize >= 6) zip_entry->uid = archive_le32dec( p + offset + 2); } if (datasize >= (2 + uidsize + 3)) { /* get a gid size. */ gidsize = 0xff & (int)p[offset+2+uidsize]; if (gidsize == 2) zip_entry->gid = archive_le16dec( p+offset+2+uidsize+1); else if (gidsize == 4 && datasize >= (2 + uidsize + 5)) zip_entry->gid = archive_le32dec( p+offset+2+uidsize+1); } } break; } case 0x9901: /* WinZip AES extra data field. */ if (datasize < 6) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Incomplete AES field"); return ARCHIVE_FAILED; } if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { /* Vendor version. */ zip_entry->aes_extra.vendor = archive_le16dec(p + offset); /* AES encryption strength. */ zip_entry->aes_extra.strength = p[offset + 4]; /* Actual compression method. */ zip_entry->aes_extra.compression = p[offset + 5]; } break; default: break; } offset += datasize; } return ARCHIVE_OK; } /* * Assumes file pointer is at beginning of local file header. */ static int zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, struct zip *zip) { const char *p; const void *h; const wchar_t *wp; const char *cp; size_t len, filename_length, extra_length; struct archive_string_conv *sconv; struct zip_entry *zip_entry = zip->entry; struct zip_entry zip_entry_central_dir; int ret = ARCHIVE_OK; char version; /* Save a copy of the original for consistency checks. */ zip_entry_central_dir = *zip_entry; zip->decompress_init = 0; zip->end_of_entry = 0; zip->entry_uncompressed_bytes_read = 0; zip->entry_compressed_bytes_read = 0; zip->computed_crc32 = zip->crc32func(0, NULL, 0); /* Setup default conversion. */ if (zip->sconv == NULL && !zip->init_default_conversion) { zip->sconv_default = archive_string_default_conversion_for_read(&(a->archive)); zip->init_default_conversion = 1; } if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file header"); return (ARCHIVE_FATAL); } if (memcmp(p, "PK\003\004", 4) != 0) { archive_set_error(&a->archive, -1, "Damaged Zip archive"); return ARCHIVE_FATAL; } version = p[4]; zip_entry->system = p[5]; zip_entry->zip_flags = archive_le16dec(p + 6); if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { zip->has_encrypted_entries = 1; archive_entry_set_is_data_encrypted(entry, 1); if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && zip_entry->zip_flags & ZIP_ENCRYPTED && zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { archive_entry_set_is_metadata_encrypted(entry, 1); return ARCHIVE_FATAL; } } zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); zip_entry->compression = (char)archive_le16dec(p + 8); zip_entry->mtime = dos_to_unix(archive_le32dec(p + 10)); zip_entry->crc32 = archive_le32dec(p + 14); if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) zip_entry->decdat = p[11]; else zip_entry->decdat = p[17]; zip_entry->compressed_size = archive_le32dec(p + 18); zip_entry->uncompressed_size = archive_le32dec(p + 22); filename_length = archive_le16dec(p + 26); extra_length = archive_le16dec(p + 28); __archive_read_consume(a, 30); /* Read the filename. */ if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file header"); return (ARCHIVE_FATAL); } if (zip_entry->zip_flags & ZIP_UTF8_NAME) { /* The filename is stored to be UTF-8. */ if (zip->sconv_utf8 == NULL) { zip->sconv_utf8 = archive_string_conversion_from_charset( &a->archive, "UTF-8", 1); if (zip->sconv_utf8 == NULL) return (ARCHIVE_FATAL); } sconv = zip->sconv_utf8; } else if (zip->sconv != NULL) sconv = zip->sconv; else sconv = zip->sconv_default; if (archive_entry_copy_pathname_l(entry, h, filename_length, sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Pathname cannot be converted " "from %s to current locale.", archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; } __archive_read_consume(a, filename_length); /* Read the extra data. */ if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file header"); return (ARCHIVE_FATAL); } if (ARCHIVE_OK != process_extra(a, entry, h, extra_length, zip_entry)) { return ARCHIVE_FATAL; } __archive_read_consume(a, extra_length); /* Work around a bug in Info-Zip: When reading from a pipe, it * stats the pipe instead of synthesizing a file entry. */ if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { zip_entry->mode &= ~ AE_IFMT; zip_entry->mode |= AE_IFREG; } /* If the mode is totally empty, set some sane default. */ if (zip_entry->mode == 0) { zip_entry->mode |= 0664; } /* Windows archivers sometimes use backslash as the directory * separator. Normalize to slash. */ if (zip_entry->system == 0 && (wp = archive_entry_pathname_w(entry)) != NULL) { if (wcschr(wp, L'/') == NULL && wcschr(wp, L'\\') != NULL) { size_t i; struct archive_wstring s; archive_string_init(&s); archive_wstrcpy(&s, wp); for (i = 0; i < archive_strlen(&s); i++) { if (s.s[i] == '\\') s.s[i] = '/'; } archive_entry_copy_pathname_w(entry, s.s); archive_wstring_free(&s); } } /* Make sure that entries with a trailing '/' are marked as directories * even if the External File Attributes contains bogus values. If this * is not a directory and there is no type, assume a regular file. */ if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) { int has_slash; wp = archive_entry_pathname_w(entry); if (wp != NULL) { len = wcslen(wp); has_slash = len > 0 && wp[len - 1] == L'/'; } else { cp = archive_entry_pathname(entry); len = (cp != NULL)?strlen(cp):0; has_slash = len > 0 && cp[len - 1] == '/'; } /* Correct file type as needed. */ if (has_slash) { zip_entry->mode &= ~AE_IFMT; zip_entry->mode |= AE_IFDIR; zip_entry->mode |= 0111; } else if ((zip_entry->mode & AE_IFMT) == 0) { zip_entry->mode |= AE_IFREG; } } /* Make sure directories end in '/' */ if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { wp = archive_entry_pathname_w(entry); if (wp != NULL) { len = wcslen(wp); if (len > 0 && wp[len - 1] != L'/') { struct archive_wstring s; archive_string_init(&s); archive_wstrcat(&s, wp); archive_wstrappend_wchar(&s, L'/'); archive_entry_copy_pathname_w(entry, s.s); archive_wstring_free(&s); } } else { cp = archive_entry_pathname(entry); len = (cp != NULL)?strlen(cp):0; if (len > 0 && cp[len - 1] != '/') { struct archive_string s; archive_string_init(&s); archive_strcat(&s, cp); archive_strappend_char(&s, '/'); archive_entry_set_pathname(entry, s.s); archive_string_free(&s); } } } if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { /* If this came from the central dir, its size info * is definitive, so ignore the length-at-end flag. */ zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; /* If local header is missing a value, use the one from the central directory. If both have it, warn about mismatches. */ if (zip_entry->crc32 == 0) { zip_entry->crc32 = zip_entry_central_dir.crc32; } else if (!zip->ignore_crc32 && zip_entry->crc32 != zip_entry_central_dir.crc32) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Inconsistent CRC32 values"); ret = ARCHIVE_WARN; } if (zip_entry->compressed_size == 0 || zip_entry->compressed_size == 0xffffffff) { zip_entry->compressed_size = zip_entry_central_dir.compressed_size; } else if (zip_entry->compressed_size != zip_entry_central_dir.compressed_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Inconsistent compressed size: " "%jd in central directory, %jd in local header", (intmax_t)zip_entry_central_dir.compressed_size, (intmax_t)zip_entry->compressed_size); ret = ARCHIVE_WARN; } if (zip_entry->uncompressed_size == 0 || zip_entry->uncompressed_size == 0xffffffff) { zip_entry->uncompressed_size = zip_entry_central_dir.uncompressed_size; } else if (zip_entry->uncompressed_size != zip_entry_central_dir.uncompressed_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Inconsistent uncompressed size: " "%jd in central directory, %jd in local header", (intmax_t)zip_entry_central_dir.uncompressed_size, (intmax_t)zip_entry->uncompressed_size); ret = ARCHIVE_WARN; } } /* Populate some additional entry fields: */ archive_entry_set_mode(entry, zip_entry->mode); archive_entry_set_uid(entry, zip_entry->uid); archive_entry_set_gid(entry, zip_entry->gid); archive_entry_set_mtime(entry, zip_entry->mtime, 0); archive_entry_set_ctime(entry, zip_entry->ctime, 0); archive_entry_set_atime(entry, zip_entry->atime, 0); if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { size_t linkname_length; if (zip_entry->compressed_size > 64 * 1024) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Zip file with oversized link entry"); return ARCHIVE_FATAL; } linkname_length = (size_t)zip_entry->compressed_size; archive_entry_set_size(entry, 0); // take into account link compression if any size_t linkname_full_length = linkname_length; if (zip->entry->compression != 0) { // symlink target string appeared to be compressed int status = ARCHIVE_FATAL; const void *uncompressed_buffer = NULL; switch (zip->entry->compression) { #if HAVE_ZLIB_H case 8: /* Deflate compression. */ zip->entry_bytes_remaining = zip_entry->compressed_size; status = zip_read_data_deflate(a, &uncompressed_buffer, &linkname_full_length, NULL); break; #endif #if HAVE_LZMA_H && HAVE_LIBLZMA case 14: /* ZIPx LZMA compression. */ /*(see zip file format specification, section 4.4.5)*/ zip->entry_bytes_remaining = zip_entry->compressed_size; status = zip_read_data_zipx_lzma_alone(a, &uncompressed_buffer, &linkname_full_length, NULL); break; #endif default: /* Unsupported compression. */ break; } if (status == ARCHIVE_OK) { p = uncompressed_buffer; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported ZIP compression method " "during decompression of link entry (%d: %s)", zip->entry->compression, compression_name(zip->entry->compression)); return ARCHIVE_FAILED; } } else { p = __archive_read_ahead(a, linkname_length, NULL); } if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Truncated Zip file"); return ARCHIVE_FATAL; } sconv = zip->sconv; if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) sconv = zip->sconv_utf8; if (sconv == NULL) sconv = zip->sconv_default; if (archive_entry_copy_symlink_l(entry, p, linkname_full_length, sconv) != 0) { if (errno != ENOMEM && sconv == zip->sconv_utf8 && (zip->entry->zip_flags & ZIP_UTF8_NAME)) archive_entry_copy_symlink_l(entry, p, linkname_full_length, NULL); if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Symlink"); return (ARCHIVE_FATAL); } /* * Since there is no character-set regulation for * symlink name, do not report the conversion error * in an automatic conversion. */ if (sconv != zip->sconv_utf8 || (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Symlink cannot be converted " "from %s to current locale.", archive_string_conversion_charset_name( sconv)); ret = ARCHIVE_WARN; } } zip_entry->uncompressed_size = zip_entry->compressed_size = 0; if (__archive_read_consume(a, linkname_length) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Read error skipping symlink target name"); return ARCHIVE_FATAL; } } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) || (zip_entry->uncompressed_size > 0 && zip_entry->uncompressed_size != 0xffffffff)) { /* Set the size only if it's meaningful. */ archive_entry_set_size(entry, zip_entry->uncompressed_size); } zip->entry_bytes_remaining = zip_entry->compressed_size; /* If there's no body, force read_data() to return EOF immediately. */ if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) && zip->entry_bytes_remaining < 1) zip->end_of_entry = 1; /* Set up a more descriptive format name. */ archive_string_empty(&zip->format_name); archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", version / 10, version % 10, compression_name(zip->entry->compression)); a->archive.archive_format_name = zip->format_name.s; return (ret); } static int check_authentication_code(struct archive_read *a, const void *_p) { struct zip *zip = (struct zip *)(a->format->data); /* Check authentication code. */ if (zip->hctx_valid) { const void *p; uint8_t hmac[20]; size_t hmac_len = 20; int cmp; archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); if (_p == NULL) { /* Read authentication code. */ p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); } } else { p = _p; } cmp = memcmp(hmac, p, AUTH_CODE_SIZE); __archive_read_consume(a, AUTH_CODE_SIZE); if (cmp != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP bad Authentication code"); return (ARCHIVE_WARN); } } return (ARCHIVE_OK); } /* * The Zip end-of-file marker is inherently ambiguous. The specification * in APPNOTE.TXT allows any of four possible formats, and there is no * guaranteed-correct way for a reader to know a priori which one the writer * will have used. The four formats are: * 1. 32-bit format with an initial PK78 marker * 2. 32-bit format without that marker * 3. 64-bit format with the marker * 4. 64-bit format without the marker * * Mark Adler's `sunzip` streaming unzip program solved this ambiguity * by just looking at every possible combination and accepting the * longest one that matches the expected values. His approach always * consumes the longest possible matching EOF marker, based on an * analysis of all the possible failures and how the values could * overlap. * * For example, suppose both of the first two formats listed * above match. In that case, we know the next four * 32-bit words match this pattern: * ``` * [PK\07\08] [CRC32] [compressed size] [uncompressed size] * ``` * but we know they must also match this pattern: * ``` * [CRC32] [compressed size] [uncompressed size] [other PK marker] * ``` * * Since the first word here matches both the PK78 signature in the * first form and the CRC32 in the second, we know those two values * are equal, the CRC32 must be exactly 0x08074b50. Similarly, the * compressed and uncompressed size must also be exactly this value. * So we know these four words are all 0x08074b50. If we were to * accept the shorter pattern, it would be immediately followed by * another PK78 marker, which is not possible in a well-formed ZIP * archive unless there is garbage between entries. This implies we * should not accept the shorter form in such a case; we should accept * the longer form. * * If the second and third possibilities above both match, we * have a slightly different situation. The following words * must match both the 32-bit format * ``` * [CRC32] [compressed size] [uncompressed size] [other PK marker] * ``` * and the 64-bit format * ``` * [CRC32] [compressed low] [compressed high] [uncompressed low] [uncompressed high] [other PK marker] * ``` * Since the 32-bit and 64-bit compressed sizes both match, the * actual size must fit in 32 bits, which implies the high-order * word of the compressed size is zero. So we know the uncompressed * low word is zero, which again implies that if we accept the shorter * format, there will not be a valid PK marker following it. * * Similar considerations rule out the shorter form in every other * possibly-ambiguous pair. So if two of the four possible formats * match, we should accept the longer option. * * If none of the four formats matches, we know the archive must be * corrupted in some fashion. In particular, it's possible that the * length-at-end bit was incorrect and we should not really be looking * for an EOF marker at all. To allow for this possibility, we * evaluate the following words to collect data for a later error * report but do not consume any bytes. We instead rely on the later * search for a new PK marker to re-sync to the next well-formed * entry. */ static void consume_end_of_file_marker(struct archive_read *a, struct zip *zip) { const char *marker; const char *p; uint64_t compressed32, uncompressed32; uint64_t compressed64, uncompressed64; uint64_t compressed_actual, uncompressed_actual; uint32_t crc32_actual; const uint32_t PK78 = 0x08074B50ULL; uint8_t crc32_ignored, crc32_may_be_zero; /* If there shouldn't be a marker, don't consume it. */ if ((zip->entry->zip_flags & ZIP_LENGTH_AT_END) == 0) { return; } /* The longest Zip end-of-file record is 24 bytes. Since an * end-of-file record can never appear at the end of the * archive, we know 24 bytes will be available unless * the archive is severely truncated. */ if (NULL == (marker = __archive_read_ahead(a, 24, NULL))) { return; } p = marker; /* The end-of-file record comprises: * = Optional PK\007\010 marker * = 4-byte CRC32 * = Compressed size * = Uncompressed size * * The last two fields are either both 32 bits or both 64 * bits. We check all possible layouts and accept any one * that gives us a complete match, else we make a best-effort * attempt to parse out the pieces. */ /* CRC32 checking can be tricky: * * Test suites sometimes ignore the CRC32 * * AES AE-2 always writes zero for the CRC32 * * AES AE-1 sometimes writes zero for the CRC32 */ crc32_ignored = zip->ignore_crc32; crc32_may_be_zero = 0; crc32_actual = zip->computed_crc32; if (zip->hctx_valid) { switch (zip->entry->aes_extra.vendor) { case AES_VENDOR_AE_2: crc32_actual = 0; break; case AES_VENDOR_AE_1: default: crc32_may_be_zero = 1; break; } } /* Values computed from the actual data in the archive. */ compressed_actual = (uint64_t)zip->entry_compressed_bytes_read; uncompressed_actual = (uint64_t)zip->entry_uncompressed_bytes_read; /* Longest: PK78 marker, all 64-bit fields (24 bytes total) */ if (archive_le32dec(p) == PK78 && ((archive_le32dec(p + 4) == crc32_actual) || (crc32_may_be_zero && (archive_le32dec(p + 4) == 0)) || crc32_ignored) && (archive_le64dec(p + 8) == compressed_actual) && (archive_le64dec(p + 16) == uncompressed_actual)) { if (!crc32_ignored) { zip->entry->crc32 = crc32_actual; } zip->entry->compressed_size = compressed_actual; zip->entry->uncompressed_size = uncompressed_actual; zip->unconsumed += 24; return; } /* No PK78 marker, 64-bit fields (20 bytes total) */ if (((archive_le32dec(p) == crc32_actual) || (crc32_may_be_zero && (archive_le32dec(p + 4) == 0)) || crc32_ignored) && (archive_le64dec(p + 4) == compressed_actual) && (archive_le64dec(p + 12) == uncompressed_actual)) { if (!crc32_ignored) { zip->entry->crc32 = crc32_actual; } zip->entry->compressed_size = compressed_actual; zip->entry->uncompressed_size = uncompressed_actual; zip->unconsumed += 20; return; } /* PK78 marker and 32-bit fields (16 bytes total) */ if (archive_le32dec(p) == PK78 && ((archive_le32dec(p + 4) == crc32_actual) || (crc32_may_be_zero && (archive_le32dec(p + 4) == 0)) || crc32_ignored) && (archive_le32dec(p + 8) == compressed_actual) && (archive_le32dec(p + 12) == uncompressed_actual)) { if (!crc32_ignored) { zip->entry->crc32 = crc32_actual; } zip->entry->compressed_size = compressed_actual; zip->entry->uncompressed_size = uncompressed_actual; zip->unconsumed += 16; return; } /* Shortest: No PK78 marker, all 32-bit fields (12 bytes total) */ if (((archive_le32dec(p) == crc32_actual) || (crc32_may_be_zero && (archive_le32dec(p + 4) == 0)) || crc32_ignored) && (archive_le32dec(p + 4) == compressed_actual) && (archive_le32dec(p + 8) == uncompressed_actual)) { if (!crc32_ignored) { zip->entry->crc32 = crc32_actual; } zip->entry->compressed_size = compressed_actual; zip->entry->uncompressed_size = uncompressed_actual; zip->unconsumed += 12; return; } /* If none of the above patterns gives us a full exact match, * then there's something definitely amiss. The fallback code * below will parse out some plausible values for error * reporting purposes. Note that this won't actually * consume anything: * * = If there really is a marker here, the logic to resync to * the next entry will suffice to skip it. * * = There might not really be a marker: Corruption or bugs * may have set the length-at-end bit without a marker ever * having actually been written. In this case, we * explicitly should not consume any bytes, since that would * prevent us from correctly reading the next entry. */ if (archive_le32dec(p) == PK78) { p += 4; /* Ignore PK78 if it appears to be present */ } zip->entry->crc32 = archive_le32dec(p); /* Parse CRC32 */ p += 4; /* Consider both 32- and 64-bit interpretations */ compressed32 = archive_le32dec(p); uncompressed32 = archive_le32dec(p + 4); compressed64 = archive_le64dec(p); uncompressed64 = archive_le64dec(p + 8); /* The earlier patterns may have failed because of CRC32 * mismatch, so it's still possible that both sizes match. * Try to match as many as we can... */ if (compressed32 == compressed_actual && uncompressed32 == uncompressed_actual) { /* Both 32-bit fields match */ zip->entry->compressed_size = compressed32; zip->entry->uncompressed_size = uncompressed32; } else if (compressed64 == compressed_actual || uncompressed64 == uncompressed_actual) { /* One or both 64-bit fields match */ zip->entry->compressed_size = compressed64; zip->entry->uncompressed_size = uncompressed64; } else { /* Zero or one 32-bit fields match */ zip->entry->compressed_size = compressed32; zip->entry->uncompressed_size = uncompressed32; } } /* * Read "uncompressed" data. * * This is straightforward if we know the size of the data. This is * always true for the seeking reader (we've examined the Central * Directory already), and will often be true for the streaming reader * (the writer was writing uncompressed so probably knows the size). * * If we don't know the size, then life is more interesting. Note * that a careful reading of the Zip specification says that a writer * must use ZIP_LENGTH_AT_END if it cannot write the CRC into the * local header. And if it uses ZIP_LENGTH_AT_END, then it is * prohibited from storing the sizes in the local header. This * prevents fully-compliant streaming writers from providing any size * clues to a streaming reader. In this case, we have to scan the * data as we read to try to locate the end-of-file marker. * * We assume here that the end-of-file marker always has the * PK\007\010 signature. Although it's technically optional, newer * writers seem to provide it pretty consistently, and it's not clear * how to efficiently recognize an end-of-file marker that lacks it. * * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets * zip->end_of_entry if it consumes all of the data. */ static int zip_read_data_none(struct archive_read *a, const void **_buff, size_t *size, int64_t *offset) { struct zip *zip; const char *buff; ssize_t bytes_avail; ssize_t trailing_extra; int r; (void)offset; /* UNUSED */ zip = (struct zip *)(a->format->data); trailing_extra = zip->hctx_valid ? AUTH_CODE_SIZE : 0; if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { const char *p; ssize_t grabbing_bytes = 24 + trailing_extra; /* Grab at least 24 bytes. */ buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); if (bytes_avail < grabbing_bytes) { /* Zip archives have end-of-archive markers that are longer than this, so a failure to get at least 24 bytes really does indicate a truncated file. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); } /* Check for a complete PK\007\010 signature, followed * by the correct 4-byte CRC. */ p = buff + trailing_extra; if (p[0] == 'P' && p[1] == 'K' && p[2] == '\007' && p[3] == '\010' && (archive_le32dec(p + 4) == zip->computed_crc32 || zip->ignore_crc32 || (zip->hctx_valid && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { zip->end_of_entry = 1; if (zip->hctx_valid) { r = check_authentication_code(a, buff); if (r != ARCHIVE_OK) return (r); } return (ARCHIVE_OK); } /* If not at EOF, ensure we consume at least one byte. */ ++p; /* Scan forward until we see where a PK\007\010 signature * might be. */ /* Return bytes up until that point. On the next call, * the code above will verify the data descriptor. */ while (p < buff + bytes_avail - 4) { if (p[3] == 'P') { p += 3; } else if (p[3] == 'K') { p += 2; } else if (p[3] == '\007') { p += 1; } else if (p[3] == '\010' && p[2] == '\007' && p[1] == 'K' && p[0] == 'P') { break; } else { p += 4; } } p -= trailing_extra; bytes_avail = p - buff; } else { if (zip->entry_bytes_remaining == 0) { zip->end_of_entry = 1; if (zip->hctx_valid) { r = check_authentication_code(a, NULL); if (r != ARCHIVE_OK) return (r); } return (ARCHIVE_OK); } /* Grab a bunch of bytes. */ buff = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); } if (bytes_avail > zip->entry_bytes_remaining) bytes_avail = (ssize_t)zip->entry_bytes_remaining; } if (zip->tctx_valid || zip->cctx_valid) { size_t dec_size = bytes_avail; if (dec_size > zip->decrypted_buffer_size) dec_size = zip->decrypted_buffer_size; if (zip->tctx_valid) { trad_enc_decrypt_update(&zip->tctx, (const uint8_t *)buff, dec_size, zip->decrypted_buffer, dec_size); } else { size_t dsize = dec_size; archive_hmac_sha1_update(&zip->hctx, (const uint8_t *)buff, dec_size); archive_decrypto_aes_ctr_update(&zip->cctx, (const uint8_t *)buff, dec_size, zip->decrypted_buffer, &dsize); } bytes_avail = dec_size; buff = (const char *)zip->decrypted_buffer; } zip->entry_bytes_remaining -= bytes_avail; zip->entry_uncompressed_bytes_read += bytes_avail; zip->entry_compressed_bytes_read += bytes_avail; zip->unconsumed += bytes_avail; *size = bytes_avail; *_buff = buff; return (ARCHIVE_OK); } #if HAVE_LZMA_H && HAVE_LIBLZMA static int zipx_xz_init(struct archive_read *a, struct zip *zip) { lzma_ret r; if(zip->zipx_lzma_valid) { lzma_end(&zip->zipx_lzma_stream); zip->zipx_lzma_valid = 0; } memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream)); r = lzma_stream_decoder(&zip->zipx_lzma_stream, UINT64_MAX, 0); if (r != LZMA_OK) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "xz initialization failed(%d)", r); return (ARCHIVE_FAILED); } zip->zipx_lzma_valid = 1; free(zip->uncompressed_buffer); zip->uncompressed_buffer_size = 256 * 1024; zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); if (zip->uncompressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for xz decompression"); return (ARCHIVE_FATAL); } zip->decompress_init = 1; return (ARCHIVE_OK); } static int zipx_lzma_alone_init(struct archive_read *a, struct zip *zip) { lzma_ret r; const uint8_t* p; #pragma pack(push) #pragma pack(1) struct _alone_header { uint8_t bytes[5]; uint64_t uncompressed_size; } alone_header; #pragma pack(pop) if(zip->zipx_lzma_valid) { lzma_end(&zip->zipx_lzma_stream); zip->zipx_lzma_valid = 0; } /* To unpack ZIPX's "LZMA" (id 14) stream we can use standard liblzma * that is a part of XZ Utils. The stream format stored inside ZIPX * file is a modified "lzma alone" file format, that was used by the * `lzma` utility which was later deprecated in favour of `xz` utility. * Since those formats are nearly the same, we can use a standard * "lzma alone" decoder from XZ Utils. */ memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream)); r = lzma_alone_decoder(&zip->zipx_lzma_stream, UINT64_MAX); if (r != LZMA_OK) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "lzma initialization failed(%d)", r); return (ARCHIVE_FAILED); } /* Flag the cleanup function that we want our lzma-related structures * to be freed later. */ zip->zipx_lzma_valid = 1; /* The "lzma alone" file format and the stream format inside ZIPx are * almost the same. Here's an example of a structure of "lzma alone" * format: * * $ cat /bin/ls | lzma | xxd | head -n 1 * 00000000: 5d00 0080 00ff ffff ffff ffff ff00 2814 * * 5 bytes 8 bytes n bytes * * * lzma_params is a 5-byte blob that has to be decoded to extract * parameters of this LZMA stream. The uncompressed_size field is an * uint64_t value that contains information about the size of the * uncompressed file, or UINT64_MAX if this value is unknown. * The part is the actual lzma-compressed data stream. * * Now here's the structure of the stream inside the ZIPX file: * * $ cat stream_inside_zipx | xxd | head -n 1 * 00000000: 0914 0500 5d00 8000 0000 2814 .... .... * * 2byte 2byte 5 bytes n bytes * * * This means that the ZIPX file contains an additional magic1 and * magic2 headers, the lzma_params field contains the same parameter * set as in the "lzma alone" format, and the field is the * same as in the "lzma alone" format as well. Note that also the zipx * format is missing the uncompressed_size field. * * So, in order to use the "lzma alone" decoder for the zipx lzma * stream, we simply need to shuffle around some fields, prepare a new * lzma alone header, feed it into lzma alone decoder so it will * initialize itself properly, and then we can start feeding normal * zipx lzma stream into the decoder. */ /* Read magic1,magic2,lzma_params from the ZIPX stream. */ if(zip->entry_bytes_remaining < 9 || (p = __archive_read_ahead(a, 9, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated lzma data"); return (ARCHIVE_FATAL); } if(p[2] != 0x05 || p[3] != 0x00) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid lzma data"); return (ARCHIVE_FATAL); } /* Prepare an lzma alone header: copy the lzma_params blob into * a proper place into the lzma alone header. */ memcpy(&alone_header.bytes[0], p + 4, 5); /* Initialize the 'uncompressed size' field to unknown; we'll manually * monitor how many bytes there are still to be uncompressed. */ alone_header.uncompressed_size = UINT64_MAX; if(!zip->uncompressed_buffer) { zip->uncompressed_buffer_size = 256 * 1024; zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); if (zip->uncompressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for lzma decompression"); return (ARCHIVE_FATAL); } } zip->zipx_lzma_stream.next_in = (void*) &alone_header; zip->zipx_lzma_stream.avail_in = sizeof(alone_header); zip->zipx_lzma_stream.total_in = 0; zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size; zip->zipx_lzma_stream.total_out = 0; /* Feed only the header into the lzma alone decoder. This will * effectively initialize the decoder, and will not produce any * output bytes yet. */ r = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); if (r != LZMA_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "lzma stream initialization error"); return ARCHIVE_FATAL; } /* We've already consumed some bytes, so take this into account. */ __archive_read_consume(a, 9); zip->entry_bytes_remaining -= 9; zip->entry_compressed_bytes_read += 9; zip->decompress_init = 1; return (ARCHIVE_OK); } static int zip_read_data_zipx_xz(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct zip* zip = (struct zip *)(a->format->data); int ret; lzma_ret lz_ret; const void* compressed_buf; ssize_t bytes_avail, in_bytes, to_consume = 0; (void) offset; /* UNUSED */ /* Initialize decompressor if not yet initialized. */ if (!zip->decompress_init) { ret = zipx_xz_init(a, zip); if (ret != ARCHIVE_OK) return (ret); } compressed_buf = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated xz file body"); return (ARCHIVE_FATAL); } in_bytes = (ssize_t)zipmin(zip->entry_bytes_remaining, bytes_avail); zip->zipx_lzma_stream.next_in = compressed_buf; zip->zipx_lzma_stream.avail_in = in_bytes; zip->zipx_lzma_stream.total_in = 0; zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size; zip->zipx_lzma_stream.total_out = 0; /* Perform the decompression. */ lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); switch(lz_ret) { case LZMA_DATA_ERROR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xz data error (error %d)", (int) lz_ret); return (ARCHIVE_FATAL); case LZMA_NO_CHECK: case LZMA_OK: break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xz unknown error %d", (int) lz_ret); return (ARCHIVE_FATAL); case LZMA_STREAM_END: lzma_end(&zip->zipx_lzma_stream); zip->zipx_lzma_valid = 0; if((int64_t) zip->zipx_lzma_stream.total_in != zip->entry_bytes_remaining) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xz premature end of stream"); return (ARCHIVE_FATAL); } zip->end_of_entry = 1; break; } to_consume = (ssize_t)zip->zipx_lzma_stream.total_in; __archive_read_consume(a, to_consume); zip->entry_bytes_remaining -= to_consume; zip->entry_compressed_bytes_read += to_consume; zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out; *size = (size_t)zip->zipx_lzma_stream.total_out; *buff = zip->uncompressed_buffer; return (ARCHIVE_OK); } static int zip_read_data_zipx_lzma_alone(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct zip* zip = (struct zip *)(a->format->data); int ret; lzma_ret lz_ret; const void* compressed_buf; ssize_t bytes_avail, in_bytes, to_consume; (void) offset; /* UNUSED */ /* Initialize decompressor if not yet initialized. */ if (!zip->decompress_init) { ret = zipx_lzma_alone_init(a, zip); if (ret != ARCHIVE_OK) return (ret); } /* Fetch more compressed data. The same note as in deflate handler * applies here as well: * * Note: '1' here is a performance optimization. Recall that the * decompression layer returns a count of available bytes; asking for * more than that forces the decompressor to combine reads by copying * data. */ compressed_buf = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated lzma file body"); return (ARCHIVE_FATAL); } /* Set decompressor parameters. */ in_bytes = (ssize_t)zipmin(zip->entry_bytes_remaining, bytes_avail); zip->zipx_lzma_stream.next_in = compressed_buf; zip->zipx_lzma_stream.avail_in = in_bytes; zip->zipx_lzma_stream.total_in = 0; zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; zip->zipx_lzma_stream.avail_out = /* These lzma_alone streams lack end of stream marker, so let's * make sure the unpacker won't try to unpack more than it's * supposed to. */ (size_t)zipmin((int64_t) zip->uncompressed_buffer_size, zip->entry->uncompressed_size - zip->entry_uncompressed_bytes_read); zip->zipx_lzma_stream.total_out = 0; /* Perform the decompression. */ lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); switch(lz_ret) { case LZMA_DATA_ERROR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "lzma data error (error %d)", (int) lz_ret); return (ARCHIVE_FATAL); /* This case is optional in lzma alone format. It can happen, * but most of the files don't have it. (GitHub #1257) */ case LZMA_STREAM_END: if((int64_t) zip->zipx_lzma_stream.total_in != zip->entry_bytes_remaining) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "lzma alone premature end of stream"); return (ARCHIVE_FATAL); } zip->end_of_entry = 1; break; case LZMA_OK: break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "lzma unknown error %d", (int) lz_ret); return (ARCHIVE_FATAL); } to_consume = (ssize_t)zip->zipx_lzma_stream.total_in; /* Update pointers. */ __archive_read_consume(a, to_consume); zip->entry_bytes_remaining -= to_consume; zip->entry_compressed_bytes_read += to_consume; zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out; if(zip->entry_bytes_remaining == 0) { zip->end_of_entry = 1; } /* Free lzma decoder handle because we'll no longer need it. */ /* This cannot be folded into LZMA_STREAM_END handling above * because the stream end marker is not required in this format. */ if(zip->end_of_entry) { lzma_end(&zip->zipx_lzma_stream); zip->zipx_lzma_valid = 0; } /* Return values. */ *size = (size_t)zip->zipx_lzma_stream.total_out; *buff = zip->uncompressed_buffer; /* If we're here, then we're good! */ return (ARCHIVE_OK); } #endif /* HAVE_LZMA_H && HAVE_LIBLZMA */ static int zipx_ppmd8_init(struct archive_read *a, struct zip *zip) { const void* p; uint32_t val; uint32_t order; uint32_t mem; uint32_t restore_method; /* Remove previous decompression context if it exists. */ if(zip->ppmd8_valid) { __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); zip->ppmd8_valid = 0; } /* Create a new decompression context. */ __archive_ppmd8_functions.Ppmd8_Construct(&zip->ppmd8); zip->ppmd8_stream_failed = 0; /* Setup function pointers required by Ppmd8 decompressor. The * 'ppmd_read' function will feed new bytes to the decompressor, * and will increment the 'zip->zipx_ppmd_read_compressed' counter. */ zip->ppmd8.Stream.In = &zip->zipx_ppmd_stream; zip->zipx_ppmd_stream.a = a; zip->zipx_ppmd_stream.Read = &ppmd_read; /* Reset number of read bytes to 0. */ zip->zipx_ppmd_read_compressed = 0; /* Read Ppmd8 header (2 bytes). */ p = __archive_read_ahead(a, 2, NULL); if(!p) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated file data in PPMd8 stream"); return (ARCHIVE_FATAL); } __archive_read_consume(a, 2); /* Decode the stream's compression parameters. */ val = archive_le16dec(p); order = (val & 15) + 1; mem = ((val >> 4) & 0xff) + 1; restore_method = (val >> 12); if(order < 2 || restore_method > 2) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid parameter set in PPMd8 stream (order=%" PRIu32 ", " "restore=%" PRIu32 ")", order, restore_method); return (ARCHIVE_FAILED); } /* Allocate the memory needed to properly decompress the file. */ if(!__archive_ppmd8_functions.Ppmd8_Alloc(&zip->ppmd8, mem << 20)) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for PPMd8 stream: %" PRIu32 " bytes", mem << 20); return (ARCHIVE_FATAL); } /* Signal the cleanup function to release Ppmd8 context in the * cleanup phase. */ zip->ppmd8_valid = 1; /* Perform further Ppmd8 initialization. */ if(!__archive_ppmd8_functions.Ppmd8_RangeDec_Init(&zip->ppmd8)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, "PPMd8 stream range decoder initialization error"); return (ARCHIVE_FATAL); } __archive_ppmd8_functions.Ppmd8_Init(&zip->ppmd8, order, restore_method); /* Allocate the buffer that will hold uncompressed data. */ free(zip->uncompressed_buffer); zip->uncompressed_buffer_size = 256 * 1024; zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); if(zip->uncompressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for PPMd8 decompression"); return ARCHIVE_FATAL; } /* Ppmd8 initialization is done. */ zip->decompress_init = 1; /* We've already read 2 bytes in the output stream. Additionally, * Ppmd8 initialization code could read some data as well. So we * are advancing the stream by 2 bytes plus whatever number of * bytes Ppmd8 init function used. */ zip->entry_compressed_bytes_read += 2 + zip->zipx_ppmd_read_compressed; return ARCHIVE_OK; } static int zip_read_data_zipx_ppmd(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct zip* zip = (struct zip *)(a->format->data); int ret; size_t consumed_bytes = 0; ssize_t bytes_avail = 0; (void) offset; /* UNUSED */ /* If we're here for the first time, initialize Ppmd8 decompression * context first. */ if(!zip->decompress_init) { ret = zipx_ppmd8_init(a, zip); if(ret != ARCHIVE_OK) return ret; } /* Fetch for more data. We're reading 1 byte here, but libarchive * should prefetch more bytes. */ (void) __archive_read_ahead(a, 1, &bytes_avail); if(bytes_avail < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated PPMd8 file body"); return (ARCHIVE_FATAL); } /* This counter will be updated inside ppmd_read(), which at one * point will be called by Ppmd8_DecodeSymbol. */ zip->zipx_ppmd_read_compressed = 0; /* Decompression loop. */ do { int sym = __archive_ppmd8_functions.Ppmd8_DecodeSymbol( &zip->ppmd8); if(sym < 0) { zip->end_of_entry = 1; break; } /* This field is set by ppmd_read() when there was no more data * to be read. */ if(zip->ppmd8_stream_failed) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated PPMd8 file body"); return (ARCHIVE_FATAL); } zip->uncompressed_buffer[consumed_bytes] = (uint8_t) sym; ++consumed_bytes; } while(consumed_bytes < zip->uncompressed_buffer_size); /* Update pointers so we can continue decompression in another call. */ zip->entry_bytes_remaining -= zip->zipx_ppmd_read_compressed; zip->entry_compressed_bytes_read += zip->zipx_ppmd_read_compressed; zip->entry_uncompressed_bytes_read += consumed_bytes; /* If we're at the end of stream, deinitialize Ppmd8 context. */ if(zip->end_of_entry) { __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); zip->ppmd8_valid = 0; } /* Update pointers for libarchive. */ *buff = zip->uncompressed_buffer; *size = consumed_bytes; return ARCHIVE_OK; } #ifdef HAVE_BZLIB_H static int zipx_bzip2_init(struct archive_read *a, struct zip *zip) { int r; /* Deallocate already existing BZ2 decompression context if it * exists. */ if(zip->bzstream_valid) { BZ2_bzDecompressEnd(&zip->bzstream); zip->bzstream_valid = 0; } /* Allocate a new BZ2 decompression context. */ memset(&zip->bzstream, 0, sizeof(bz_stream)); r = BZ2_bzDecompressInit(&zip->bzstream, 0, 1); if(r != BZ_OK) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "bzip2 initialization failed(%d)", r); return ARCHIVE_FAILED; } /* Mark the bzstream field to be released in cleanup phase. */ zip->bzstream_valid = 1; /* (Re)allocate the buffer that will contain decompressed bytes. */ free(zip->uncompressed_buffer); zip->uncompressed_buffer_size = 256 * 1024; zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); if (zip->uncompressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for bzip2 decompression"); return ARCHIVE_FATAL; } /* Initialization done. */ zip->decompress_init = 1; return ARCHIVE_OK; } static int zip_read_data_zipx_bzip2(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct zip *zip = (struct zip *)(a->format->data); ssize_t bytes_avail = 0, in_bytes, to_consume; const void *compressed_buff; int r; uint64_t total_out; (void) offset; /* UNUSED */ /* Initialize decompression context if we're here for the first time. */ if(!zip->decompress_init) { r = zipx_bzip2_init(a, zip); if(r != ARCHIVE_OK) return r; } /* Fetch more compressed bytes. */ compressed_buff = __archive_read_ahead(a, 1, &bytes_avail); if(bytes_avail < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated bzip2 file body"); return (ARCHIVE_FATAL); } in_bytes = (ssize_t)zipmin(zip->entry_bytes_remaining, bytes_avail); if(in_bytes < 1) { /* libbz2 doesn't complain when caller feeds avail_in == 0. * It will actually return success in this case, which is * undesirable. This is why we need to make this check * manually. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated bzip2 file body"); return (ARCHIVE_FATAL); } /* Setup buffer boundaries. */ zip->bzstream.next_in = (char*)(uintptr_t) compressed_buff; zip->bzstream.avail_in = (uint32_t)in_bytes; zip->bzstream.total_in_hi32 = 0; zip->bzstream.total_in_lo32 = 0; zip->bzstream.next_out = (char*) zip->uncompressed_buffer; zip->bzstream.avail_out = (uint32_t)zip->uncompressed_buffer_size; zip->bzstream.total_out_hi32 = 0; zip->bzstream.total_out_lo32 = 0; /* Perform the decompression. */ r = BZ2_bzDecompress(&zip->bzstream); switch(r) { case BZ_STREAM_END: /* If we're at the end of the stream, deinitialize the * decompression context now. */ switch(BZ2_bzDecompressEnd(&zip->bzstream)) { case BZ_OK: break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to clean up bzip2 " "decompressor"); return ARCHIVE_FATAL; } zip->end_of_entry = 1; break; case BZ_OK: /* The decompressor has successfully decoded this * chunk of data, but more data is still in queue. */ break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "bzip2 decompression failed"); return ARCHIVE_FATAL; } /* Update the pointers so decompressor can continue decoding. */ to_consume = zip->bzstream.total_in_lo32; __archive_read_consume(a, to_consume); total_out = ((uint64_t) zip->bzstream.total_out_hi32 << 32) | zip->bzstream.total_out_lo32; zip->entry_bytes_remaining -= to_consume; zip->entry_compressed_bytes_read += to_consume; zip->entry_uncompressed_bytes_read += total_out; /* Give libarchive its due. */ *size = (size_t)total_out; *buff = zip->uncompressed_buffer; return ARCHIVE_OK; } #endif #if HAVE_ZSTD_H && HAVE_LIBZSTD static int zipx_zstd_init(struct archive_read *a, struct zip *zip) { size_t r; /* Deallocate already existing Zstd decompression context if it * exists. */ if(zip->zstdstream_valid) { ZSTD_freeDStream(zip->zstdstream); zip->zstdstream_valid = 0; } /* Allocate a new Zstd decompression context. */ zip->zstdstream = ZSTD_createDStream(); r = ZSTD_initDStream(zip->zstdstream); if (ZSTD_isError(r)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Error initializing zstd decompressor: %s", ZSTD_getErrorName(r)); return ARCHIVE_FAILED; } /* Mark the zstdstream field to be released in cleanup phase. */ zip->zstdstream_valid = 1; /* (Re)allocate the buffer that will contain decompressed bytes. */ free(zip->uncompressed_buffer); zip->uncompressed_buffer_size = ZSTD_DStreamOutSize(); zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); if (zip->uncompressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for Zstd decompression"); return ARCHIVE_FATAL; } /* Initialization done. */ zip->decompress_init = 1; return ARCHIVE_OK; } static int zip_read_data_zipx_zstd(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct zip *zip = (struct zip *)(a->format->data); ssize_t bytes_avail = 0, in_bytes, to_consume; const void *compressed_buff; int r; size_t ret; uint64_t total_out; ZSTD_outBuffer out; ZSTD_inBuffer in; (void) offset; /* UNUSED */ /* Initialize decompression context if we're here for the first time. */ if(!zip->decompress_init) { r = zipx_zstd_init(a, zip); if(r != ARCHIVE_OK) return r; } /* Fetch more compressed bytes */ compressed_buff = __archive_read_ahead(a, 1, &bytes_avail); if(bytes_avail < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated zstd file body"); return (ARCHIVE_FATAL); } in_bytes = (ssize_t)zipmin(zip->entry_bytes_remaining, bytes_avail); if(in_bytes < 1) { /* zstd doesn't complain when caller feeds avail_in == 0. * It will actually return success in this case, which is * undesirable. This is why we need to make this check * manually. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated zstd file body"); return (ARCHIVE_FATAL); } /* Setup buffer boundaries */ in.src = compressed_buff; in.size = in_bytes; in.pos = 0; out = (ZSTD_outBuffer) { zip->uncompressed_buffer, zip->uncompressed_buffer_size, 0 }; /* Perform the decompression. */ ret = ZSTD_decompressStream(zip->zstdstream, &out, &in); if (ZSTD_isError(ret)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Error during zstd decompression: %s", ZSTD_getErrorName(ret)); return (ARCHIVE_FATAL); } /* Check end of the stream. */ if (ret == 0) { if ((in.pos == in.size) && (out.pos < out.size)) { zip->end_of_entry = 1; ZSTD_freeDStream(zip->zstdstream); zip->zstdstream_valid = 0; } } /* Update the pointers so decompressor can continue decoding. */ to_consume = in.pos; __archive_read_consume(a, to_consume); total_out = out.pos; zip->entry_bytes_remaining -= to_consume; zip->entry_compressed_bytes_read += to_consume; zip->entry_uncompressed_bytes_read += total_out; /* Give libarchive its due. */ *size = (size_t)total_out; *buff = zip->uncompressed_buffer; return ARCHIVE_OK; } #endif #ifdef HAVE_ZLIB_H static int zip_deflate_init(struct archive_read *a, struct zip *zip) { int r; /* If we haven't yet read any data, initialize the decompressor. */ if (!zip->decompress_init) { if (zip->stream_valid) r = inflateReset(&zip->stream); else r = inflateInit2(&zip->stream, -15 /* Don't check for zlib header */); if (r != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't initialize ZIP decompression."); return (ARCHIVE_FATAL); } /* Stream structure has been set up. */ zip->stream_valid = 1; /* We've initialized decompression for this stream. */ zip->decompress_init = 1; } return (ARCHIVE_OK); } static int zip_read_data_deflate(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct zip *zip; ssize_t bytes_avail, to_consume = 0; const void *compressed_buff, *sp; int r; (void)offset; /* UNUSED */ zip = (struct zip *)(a->format->data); /* If the buffer hasn't been allocated, allocate it now. */ if (zip->uncompressed_buffer == NULL) { zip->uncompressed_buffer_size = 256 * 1024; zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); if (zip->uncompressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for ZIP decompression"); return (ARCHIVE_FATAL); } } r = zip_deflate_init(a, zip); if (r != ARCHIVE_OK) return (r); /* * Note: '1' here is a performance optimization. * Recall that the decompression layer returns a count of * available bytes; asking for more than that forces the * decompressor to combine reads by copying data. */ compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && bytes_avail > zip->entry_bytes_remaining) { bytes_avail = (ssize_t)zip->entry_bytes_remaining; } if (bytes_avail < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file body"); return (ARCHIVE_FATAL); } if (zip->tctx_valid || zip->cctx_valid) { if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { size_t buff_remaining = (zip->decrypted_buffer + zip->decrypted_buffer_size) - (zip->decrypted_ptr + zip->decrypted_bytes_remaining); if (buff_remaining > (size_t)bytes_avail) buff_remaining = (size_t)bytes_avail; if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && zip->entry_bytes_remaining > 0) { if ((int64_t)(zip->decrypted_bytes_remaining + buff_remaining) > zip->entry_bytes_remaining) { if (zip->entry_bytes_remaining < (int64_t)zip->decrypted_bytes_remaining) buff_remaining = 0; else buff_remaining = (size_t)zip->entry_bytes_remaining - zip->decrypted_bytes_remaining; } } if (buff_remaining > 0) { if (zip->tctx_valid) { trad_enc_decrypt_update(&zip->tctx, compressed_buff, buff_remaining, zip->decrypted_ptr + zip->decrypted_bytes_remaining, buff_remaining); } else { size_t dsize = buff_remaining; archive_decrypto_aes_ctr_update( &zip->cctx, compressed_buff, buff_remaining, zip->decrypted_ptr + zip->decrypted_bytes_remaining, &dsize); } zip->decrypted_bytes_remaining += buff_remaining; } } bytes_avail = zip->decrypted_bytes_remaining; compressed_buff = (const char *)zip->decrypted_ptr; } /* * A bug in zlib.h: stream.next_in should be marked 'const' * but isn't (the library never alters data through the * next_in pointer, only reads it). The result: this ugly * cast to remove 'const'. */ zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; zip->stream.avail_in = (uInt)bytes_avail; zip->stream.total_in = 0; zip->stream.next_out = zip->uncompressed_buffer; zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; zip->stream.total_out = 0; r = inflate(&zip->stream, 0); switch (r) { case Z_OK: break; case Z_STREAM_END: zip->end_of_entry = 1; break; case Z_MEM_ERROR: archive_set_error(&a->archive, ENOMEM, "Out of memory for ZIP decompression"); return (ARCHIVE_FATAL); default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP decompression failed (%d)", r); return (ARCHIVE_FATAL); } /* Consume as much as the compressor actually used. */ to_consume = zip->stream.total_in; __archive_read_consume(a, to_consume); zip->entry_bytes_remaining -= to_consume; zip->entry_compressed_bytes_read += to_consume; zip->entry_uncompressed_bytes_read += zip->stream.total_out; if (zip->tctx_valid || zip->cctx_valid) { zip->decrypted_bytes_remaining -= to_consume; if (zip->decrypted_bytes_remaining == 0) zip->decrypted_ptr = zip->decrypted_buffer; else zip->decrypted_ptr += to_consume; } if (zip->hctx_valid) archive_hmac_sha1_update(&zip->hctx, sp, to_consume); if (zip->end_of_entry) { if (zip->hctx_valid) { r = check_authentication_code(a, NULL); if (r != ARCHIVE_OK) { return (r); } } } *size = zip->stream.total_out; *buff = zip->uncompressed_buffer; return (ARCHIVE_OK); } #endif static int read_decryption_header(struct archive_read *a) { struct zip *zip = (struct zip *)(a->format->data); const char *p; unsigned int remaining_size; unsigned int ts; /* * Read an initialization vector data field. */ p = __archive_read_ahead(a, 2, NULL); if (p == NULL) goto truncated; ts = zip->iv_size; zip->iv_size = archive_le16dec(p); __archive_read_consume(a, 2); if (ts < zip->iv_size) { free(zip->iv); zip->iv = NULL; } p = __archive_read_ahead(a, zip->iv_size, NULL); if (p == NULL) goto truncated; if (zip->iv == NULL) { zip->iv = malloc(zip->iv_size); if (zip->iv == NULL) goto nomem; } memcpy(zip->iv, p, zip->iv_size); __archive_read_consume(a, zip->iv_size); /* * Read a size of remaining decryption header field. */ p = __archive_read_ahead(a, 14, NULL); if (p == NULL) goto truncated; remaining_size = archive_le32dec(p); if (remaining_size < 16 || remaining_size > (1 << 18)) goto corrupted; /* Check if format version is supported. */ if (archive_le16dec(p+4) != 3) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported encryption format version: %u", archive_le16dec(p+4)); return (ARCHIVE_FAILED); } /* * Read an encryption algorithm field. */ zip->alg_id = archive_le16dec(p+6); switch (zip->alg_id) { case 0x6601:/* DES */ case 0x6602:/* RC2 */ case 0x6603:/* 3DES 168 */ case 0x6609:/* 3DES 112 */ case 0x660E:/* AES 128 */ case 0x660F:/* AES 192 */ case 0x6610:/* AES 256 */ case 0x6702:/* RC2 (version >= 5.2) */ case 0x6720:/* Blowfish */ case 0x6721:/* Twofish */ case 0x6801:/* RC4 */ /* Supported encryption algorithm. */ break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unknown encryption algorithm: %u", zip->alg_id); return (ARCHIVE_FAILED); } /* * Read a bit length field. */ zip->bit_len = archive_le16dec(p+8); /* * Read a flags field. */ zip->flags = archive_le16dec(p+10); switch (zip->flags & 0xf000) { case 0x0001: /* Password is required to decrypt. */ case 0x0002: /* Certificates only. */ case 0x0003: /* Password or certificate required to decrypt. */ break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unknown encryption flag: %u", zip->flags); return (ARCHIVE_FAILED); } if ((zip->flags & 0xf000) == 0 || (zip->flags & 0xf000) == 0x4000) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unknown encryption flag: %u", zip->flags); return (ARCHIVE_FAILED); } /* * Read an encrypted random data field. */ ts = zip->erd_size; zip->erd_size = archive_le16dec(p+12); __archive_read_consume(a, 14); if ((zip->erd_size & 0xf) != 0 || (zip->erd_size + 16) > remaining_size || (zip->erd_size + 16) < zip->erd_size) goto corrupted; if (ts < zip->erd_size) { free(zip->erd); zip->erd = NULL; } p = __archive_read_ahead(a, zip->erd_size, NULL); if (p == NULL) goto truncated; if (zip->erd == NULL) { zip->erd = malloc(zip->erd_size); if (zip->erd == NULL) goto nomem; } memcpy(zip->erd, p, zip->erd_size); __archive_read_consume(a, zip->erd_size); /* * Read a reserved data field. */ p = __archive_read_ahead(a, 4, NULL); if (p == NULL) goto truncated; /* Reserved data size should be zero. */ if (archive_le32dec(p) != 0) goto corrupted; __archive_read_consume(a, 4); /* * Read a password validation data field. */ p = __archive_read_ahead(a, 2, NULL); if (p == NULL) goto truncated; ts = zip->v_size; zip->v_size = archive_le16dec(p); __archive_read_consume(a, 2); if ((zip->v_size & 0x0f) != 0 || (zip->erd_size + zip->v_size + 16) > remaining_size || (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) goto corrupted; if (ts < zip->v_size) { free(zip->v_data); zip->v_data = NULL; } p = __archive_read_ahead(a, zip->v_size, NULL); if (p == NULL) goto truncated; if (zip->v_data == NULL) { zip->v_data = malloc(zip->v_size); if (zip->v_data == NULL) goto nomem; } memcpy(zip->v_data, p, zip->v_size); __archive_read_consume(a, zip->v_size); p = __archive_read_ahead(a, 4, NULL); if (p == NULL) goto truncated; zip->v_crc32 = archive_le32dec(p); __archive_read_consume(a, 4); /*return (ARCHIVE_OK); * This is not fully implemented yet.*/ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Encrypted file is unsupported"); return (ARCHIVE_FAILED); truncated: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); corrupted: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Corrupted ZIP file data"); return (ARCHIVE_FATAL); nomem: archive_set_error(&a->archive, ENOMEM, "No memory for ZIP decryption"); return (ARCHIVE_FATAL); } static int zip_alloc_decryption_buffer(struct archive_read *a) { struct zip *zip = (struct zip *)(a->format->data); size_t bs = 256 * 1024; if (zip->decrypted_buffer == NULL) { zip->decrypted_buffer_size = bs; zip->decrypted_buffer = malloc(bs); if (zip->decrypted_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for ZIP decryption"); return (ARCHIVE_FATAL); } } zip->decrypted_ptr = zip->decrypted_buffer; return (ARCHIVE_OK); } static int init_traditional_PKWARE_decryption(struct archive_read *a) { struct zip *zip = (struct zip *)(a->format->data); const void *p; int retry; int r; if (zip->tctx_valid) return (ARCHIVE_OK); /* Read the 12 bytes encryption header stored at the start of the data area. */ #define ENC_HEADER_SIZE 12 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated Zip encrypted body: only %jd bytes available", (intmax_t)zip->entry_bytes_remaining); return (ARCHIVE_FATAL); } p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); } for (retry = 0;; retry++) { const char *passphrase; uint8_t crcchk; passphrase = __archive_read_next_passphrase(a); if (passphrase == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, (retry > 0)? "Incorrect passphrase": "Passphrase required for this entry"); return (ARCHIVE_FAILED); } /* * Initialize ctx for Traditional PKWARE Decryption. */ r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), p, ENC_HEADER_SIZE, &crcchk); if (r == 0 && crcchk == zip->entry->decdat) break;/* The passphrase is OK. */ if (retry > 10000) { /* Avoid infinity loop. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Too many incorrect passphrases"); return (ARCHIVE_FAILED); } } __archive_read_consume(a, ENC_HEADER_SIZE); zip->tctx_valid = 1; if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { zip->entry_bytes_remaining -= ENC_HEADER_SIZE; } /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; zip->decrypted_bytes_remaining = 0; return (zip_alloc_decryption_buffer(a)); #undef ENC_HEADER_SIZE } static int init_WinZip_AES_decryption(struct archive_read *a) { struct zip *zip = (struct zip *)(a->format->data); const void *p; const uint8_t *pv; size_t key_len, salt_len; uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; int retry; int r; if (zip->cctx_valid || zip->hctx_valid) return (ARCHIVE_OK); switch (zip->entry->aes_extra.strength) { case 1: salt_len = 8; key_len = 16; break; case 2: salt_len = 12; key_len = 24; break; case 3: salt_len = 16; key_len = 32; break; default: goto corrupted; } p = __archive_read_ahead(a, salt_len + 2, NULL); if (p == NULL) goto truncated; for (retry = 0;; retry++) { const char *passphrase; passphrase = __archive_read_next_passphrase(a); if (passphrase == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, (retry > 0)? "Incorrect passphrase": "Passphrase required for this entry"); return (ARCHIVE_FAILED); } memset(derived_key, 0, sizeof(derived_key)); r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), p, salt_len, 1000, derived_key, key_len * 2 + 2); if (r != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, - "Decryption is unsupported due to lack of " - "crypto library"); + r == CRYPTOR_STUB_FUNCTION ? "Decryption is unsupported due " + "to lack of crypto library" : "Failed to process passphrase"); return (ARCHIVE_FAILED); } /* Check password verification value. */ pv = ((const uint8_t *)p) + salt_len; if (derived_key[key_len * 2] == pv[0] && derived_key[key_len * 2 + 1] == pv[1]) break;/* The passphrase is OK. */ if (retry > 10000) { /* Avoid infinity loop. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Too many incorrect passphrases"); return (ARCHIVE_FAILED); } } r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); if (r != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Decryption is unsupported due to lack of crypto library"); return (ARCHIVE_FAILED); } r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); if (r != 0) { archive_decrypto_aes_ctr_release(&zip->cctx); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to initialize HMAC-SHA1"); return (ARCHIVE_FAILED); } zip->cctx_valid = zip->hctx_valid = 1; __archive_read_consume(a, salt_len + 2); zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && zip->entry_bytes_remaining < 0) goto corrupted; zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; zip->decrypted_bytes_remaining = 0; zip->entry->compression = zip->entry->aes_extra.compression; return (zip_alloc_decryption_buffer(a)); truncated: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); corrupted: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Corrupted ZIP file data"); return (ARCHIVE_FATAL); } static int archive_read_format_zip_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { int r; struct zip *zip = (struct zip *)(a->format->data); if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { zip->has_encrypted_entries = 0; } *offset = zip->entry_uncompressed_bytes_read; *size = 0; *buff = NULL; /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ if (zip->end_of_entry) return (ARCHIVE_EOF); /* Return EOF immediately if this is a non-regular file. */ if (AE_IFREG != (zip->entry->mode & AE_IFMT)) return (ARCHIVE_EOF); __archive_read_consume(a, zip->unconsumed); zip->unconsumed = 0; if (zip->init_decryption) { zip->has_encrypted_entries = 1; if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) r = read_decryption_header(a); else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) r = init_WinZip_AES_decryption(a); else r = init_traditional_PKWARE_decryption(a); if (r != ARCHIVE_OK) return (r); zip->init_decryption = 0; } switch(zip->entry->compression) { case 0: /* No compression. */ r = zip_read_data_none(a, buff, size, offset); break; #ifdef HAVE_BZLIB_H case 12: /* ZIPx bzip2 compression. */ r = zip_read_data_zipx_bzip2(a, buff, size, offset); break; #endif #if HAVE_LZMA_H && HAVE_LIBLZMA case 14: /* ZIPx LZMA compression. */ r = zip_read_data_zipx_lzma_alone(a, buff, size, offset); break; case 95: /* ZIPx XZ compression. */ r = zip_read_data_zipx_xz(a, buff, size, offset); break; #endif #if HAVE_ZSTD_H && HAVE_LIBZSTD case 93: /* ZIPx Zstd compression. */ r = zip_read_data_zipx_zstd(a, buff, size, offset); break; #endif /* PPMd support is built-in, so we don't need any #if guards. */ case 98: /* ZIPx PPMd compression. */ r = zip_read_data_zipx_ppmd(a, buff, size, offset); break; #ifdef HAVE_ZLIB_H case 8: /* Deflate compression. */ r = zip_read_data_deflate(a, buff, size, offset); break; #endif default: /* Unsupported compression. */ /* Return a warning. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported ZIP compression method (%d: %s)", zip->entry->compression, compression_name(zip->entry->compression)); /* We can't decompress this entry, but we will * be able to skip() it and try the next entry. */ return (ARCHIVE_FAILED); } if (r != ARCHIVE_OK) return (r); if (*size > 0) { zip->computed_crc32 = zip->crc32func(zip->computed_crc32, *buff, (unsigned)*size); } /* If we hit the end, swallow any end-of-data marker and * verify the final check values. */ if (zip->end_of_entry) { consume_end_of_file_marker(a, zip); /* Check computed CRC against header */ if ((!zip->hctx_valid || zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && zip->entry->crc32 != zip->computed_crc32 && !zip->ignore_crc32) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP bad CRC: 0x%lx should be 0x%lx", (unsigned long)zip->computed_crc32, (unsigned long)zip->entry->crc32); return (ARCHIVE_FAILED); } /* Check file size against header. */ if (zip->entry->compressed_size != zip->entry_compressed_bytes_read) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP compressed data is wrong size " "(read %jd, expected %jd)", (intmax_t)zip->entry_compressed_bytes_read, (intmax_t)zip->entry->compressed_size); return (ARCHIVE_FAILED); } /* Size field only stores the lower 32 bits of the actual * size. */ if ((zip->entry->uncompressed_size & UINT32_MAX) != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP uncompressed data is wrong size " "(read %jd, expected %jd)\n", (intmax_t)zip->entry_uncompressed_bytes_read, (intmax_t)zip->entry->uncompressed_size); return (ARCHIVE_FAILED); } } return (ARCHIVE_OK); } static int archive_read_format_zip_cleanup(struct archive_read *a) { struct zip *zip; struct zip_entry *zip_entry, *next_zip_entry; zip = (struct zip *)(a->format->data); #ifdef HAVE_ZLIB_H if (zip->stream_valid) inflateEnd(&zip->stream); #endif #if HAVE_LZMA_H && HAVE_LIBLZMA if (zip->zipx_lzma_valid) { lzma_end(&zip->zipx_lzma_stream); } #endif #ifdef HAVE_BZLIB_H if (zip->bzstream_valid) { BZ2_bzDecompressEnd(&zip->bzstream); } #endif #if HAVE_ZSTD_H && HAVE_LIBZSTD if (zip->zstdstream_valid) { ZSTD_freeDStream(zip->zstdstream); } #endif free(zip->uncompressed_buffer); if (zip->ppmd8_valid) __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); if (zip->zip_entries) { zip_entry = zip->zip_entries; while (zip_entry != NULL) { next_zip_entry = zip_entry->next; archive_string_free(&zip_entry->rsrcname); free(zip_entry); zip_entry = next_zip_entry; } } free(zip->decrypted_buffer); if (zip->cctx_valid) archive_decrypto_aes_ctr_release(&zip->cctx); if (zip->hctx_valid) archive_hmac_sha1_cleanup(&zip->hctx); free(zip->iv); free(zip->erd); free(zip->v_data); archive_string_free(&zip->format_name); free(zip); (a->format->data) = NULL; return (ARCHIVE_OK); } static int archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) { if (_a && _a->format) { struct zip * zip = (struct zip *)_a->format->data; if (zip) { return zip->has_encrypted_entries; } } return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; } static int archive_read_format_zip_options(struct archive_read *a, const char *key, const char *val) { struct zip *zip; int ret = ARCHIVE_FAILED; zip = (struct zip *)(a->format->data); if (strcmp(key, "compat-2x") == 0) { /* Handle filenames as libarchive 2.x */ zip->init_default_conversion = (val != NULL) ? 1 : 0; return (ARCHIVE_OK); } else if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "zip: hdrcharset option needs a character-set name" ); else { zip->sconv = archive_string_conversion_from_charset( &a->archive, val, 0); if (zip->sconv != NULL) { if (strcmp(val, "UTF-8") == 0) zip->sconv_utf8 = zip->sconv; ret = ARCHIVE_OK; } else ret = ARCHIVE_FATAL; } return (ret); } else if (strcmp(key, "ignorecrc32") == 0) { /* Mostly useful for testing. */ if (val == NULL || val[0] == 0) { zip->crc32func = real_crc32; zip->ignore_crc32 = 0; } else { zip->crc32func = fake_crc32; zip->ignore_crc32 = 1; } return (ARCHIVE_OK); } else if (strcmp(key, "mac-ext") == 0) { zip->process_mac_extensions = (val != NULL && val[0] != 0); return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } int archive_read_support_format_zip(struct archive *a) { int r; r = archive_read_support_format_zip_streamable(a); if (r != ARCHIVE_OK) return r; return (archive_read_support_format_zip_seekable(a)); } /* ------------------------------------------------------------------------ */ /* * Streaming-mode support */ static int archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) { (void)a; /* UNUSED */ return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); } static int archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) { const char *p; (void)best_bid; /* UNUSED */ if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) return (-1); /* * Bid of 29 here comes from: * + 16 bits for "PK", * + next 16-bit field has 6 options so contributes * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits * * So we've effectively verified ~29 total bits of check data. */ if (p[0] == 'P' && p[1] == 'K') { if ((p[2] == '\001' && p[3] == '\002') || (p[2] == '\003' && p[3] == '\004') || (p[2] == '\005' && p[3] == '\006') || (p[2] == '\006' && p[3] == '\006') || (p[2] == '\007' && p[3] == '\010') || (p[2] == '0' && p[3] == '0')) return (29); } /* TODO: It's worth looking ahead a little bit for a valid * PK signature. In particular, that would make it possible * to read some UUEncoded SFX files or SFX files coming from * a network socket. */ return (0); } static int archive_read_format_zip_streamable_read_header(struct archive_read *a, struct archive_entry *entry) { struct zip *zip; a->archive.archive_format = ARCHIVE_FORMAT_ZIP; if (a->archive.archive_format_name == NULL) a->archive.archive_format_name = "ZIP"; zip = (struct zip *)(a->format->data); /* * It should be sufficient to call archive_read_next_header() for * a reader to determine if an entry is encrypted or not. If the * encryption of an entry is only detectable when calling * archive_read_data(), so be it. We'll do the same check there * as well. */ if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) zip->has_encrypted_entries = 0; /* Make sure we have a zip_entry structure to use. */ if (zip->zip_entries == NULL) { zip->zip_entries = malloc(sizeof(struct zip_entry)); if (zip->zip_entries == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return ARCHIVE_FATAL; } } zip->entry = zip->zip_entries; memset(zip->entry, 0, sizeof(struct zip_entry)); if (zip->cctx_valid) archive_decrypto_aes_ctr_release(&zip->cctx); if (zip->hctx_valid) archive_hmac_sha1_cleanup(&zip->hctx); zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; __archive_read_reset_passphrase(a); /* Search ahead for the next local file header. */ __archive_read_consume(a, zip->unconsumed); zip->unconsumed = 0; for (;;) { int64_t skipped = 0; const char *p, *end; ssize_t bytes; p = __archive_read_ahead(a, 4, &bytes); if (p == NULL) return (ARCHIVE_FATAL); end = p + bytes; while (p + 4 <= end) { if (p[0] == 'P' && p[1] == 'K') { if (p[2] == '\003' && p[3] == '\004') { /* Regular file entry. */ __archive_read_consume(a, skipped); return zip_read_local_file_header(a, entry, zip); } /* * TODO: We cannot restore permissions * based only on the local file headers. * Consider scanning the central * directory and returning additional * entries for at least directories. * This would allow us to properly set * directory permissions. * * This won't help us fix symlinks * and may not help with regular file * permissions, either. */ if (p[2] == '\001' && p[3] == '\002') { return (ARCHIVE_EOF); } /* End of central directory? Must be an * empty archive. */ if ((p[2] == '\005' && p[3] == '\006') || (p[2] == '\006' && p[3] == '\006')) return (ARCHIVE_EOF); } ++p; ++skipped; } __archive_read_consume(a, skipped); } } static int archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) { struct zip *zip; int64_t bytes_skipped; zip = (struct zip *)(a->format->data); bytes_skipped = __archive_read_consume(a, zip->unconsumed); zip->unconsumed = 0; if (bytes_skipped < 0) return (ARCHIVE_FATAL); /* If we've already read to end of data, we're done. */ if (zip->end_of_entry) return (ARCHIVE_OK); /* So we know we're streaming... */ if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) || zip->entry->compressed_size > 0) { /* We know the compressed length, so we can just skip. */ bytes_skipped = __archive_read_consume(a, zip->entry_bytes_remaining); if (bytes_skipped < 0) return (ARCHIVE_FATAL); return (ARCHIVE_OK); } if (zip->init_decryption) { int r; zip->has_encrypted_entries = 1; if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) r = read_decryption_header(a); else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) r = init_WinZip_AES_decryption(a); else r = init_traditional_PKWARE_decryption(a); if (r != ARCHIVE_OK) return (r); zip->init_decryption = 0; } /* We're streaming and we don't know the length. */ /* If the body is compressed and we know the format, we can * find an exact end-of-entry by decompressing it. */ switch (zip->entry->compression) { #ifdef HAVE_ZLIB_H case 8: /* Deflate compression. */ while (!zip->end_of_entry) { int64_t offset = 0; const void *buff = NULL; size_t size = 0; int r; r = zip_read_data_deflate(a, &buff, &size, &offset); if (r != ARCHIVE_OK) return (r); } return ARCHIVE_OK; #endif default: /* Uncompressed or unknown. */ /* Scan for a PK\007\010 signature. */ for (;;) { const char *p, *buff; ssize_t bytes_avail; buff = __archive_read_ahead(a, 16, &bytes_avail); if (bytes_avail < 16) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); } p = buff; while (p <= buff + bytes_avail - 16) { if (p[3] == 'P') { p += 3; } else if (p[3] == 'K') { p += 2; } else if (p[3] == '\007') { p += 1; } else if (p[3] == '\010' && p[2] == '\007' && p[1] == 'K' && p[0] == 'P') { if (zip->entry->flags & LA_USED_ZIP64) __archive_read_consume(a, p - buff + 24); else __archive_read_consume(a, p - buff + 16); return ARCHIVE_OK; } else { p += 4; } } __archive_read_consume(a, p - buff); } } } int archive_read_support_format_zip_streamable(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct zip *zip; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); zip = calloc(1, sizeof(*zip)); if (zip == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip data"); return (ARCHIVE_FATAL); } /* Streamable reader doesn't support mac extensions. */ zip->process_mac_extensions = 0; /* * Until enough data has been read, we cannot tell about * any encrypted entries yet. */ zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; zip->crc32func = real_crc32; r = __archive_read_register_format(a, zip, "zip", archive_read_format_zip_streamable_bid, archive_read_format_zip_options, archive_read_format_zip_streamable_read_header, archive_read_format_zip_read_data, archive_read_format_zip_read_data_skip_streamable, NULL, archive_read_format_zip_cleanup, archive_read_support_format_zip_capabilities_streamable, archive_read_format_zip_has_encrypted_entries); if (r != ARCHIVE_OK) free(zip); return (ARCHIVE_OK); } /* ------------------------------------------------------------------------ */ /* * Seeking-mode support */ static int archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) { (void)a; /* UNUSED */ return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); } /* * TODO: This is a performance sink because it forces the read core to * drop buffered data from the start of file, which will then have to * be re-read again if this bidder loses. * * We workaround this a little by passing in the best bid so far so * that later bidders can do nothing if they know they'll never * outbid. But we can certainly do better... */ static int read_eocd(struct zip *zip, const char *p, int64_t current_offset) { uint16_t disk_num; uint32_t cd_size, cd_offset; disk_num = archive_le16dec(p + 4); cd_size = archive_le32dec(p + 12); cd_offset = archive_le32dec(p + 16); /* Sanity-check the EOCD we've found. */ /* This must be the first volume. */ if (disk_num != 0) return 0; /* Central directory must be on this volume. */ if (disk_num != archive_le16dec(p + 6)) return 0; /* All central directory entries must be on this volume. */ if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) return 0; /* Central directory can't extend beyond start of EOCD record. */ if ((int64_t)cd_offset + cd_size > current_offset) return 0; /* Save the central directory location for later use. */ zip->central_directory_offset = cd_offset; zip->central_directory_offset_adjusted = current_offset - cd_size; /* This is just a tiny bit higher than the maximum returned by the streaming Zip bidder. This ensures that the more accurate seeking Zip parser wins whenever seek is available. */ return 32; } /* * Examine Zip64 EOCD locator: If it's valid, store the information * from it. */ static int read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) { int64_t eocd64_offset; int64_t eocd64_size; /* Sanity-check the locator record. */ /* Central dir must be on first volume. */ if (archive_le32dec(p + 4) != 0) return 0; /* Must be only a single volume. */ if (archive_le32dec(p + 16) != 1) return 0; /* Find the Zip64 EOCD record. */ eocd64_offset = archive_le64dec(p + 8); if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) return 0; if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) return 0; /* Make sure we can read all of it. */ eocd64_size = archive_le64dec(p + 4) + 12; if (eocd64_size < 56 || eocd64_size > 16384) return 0; if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) return 0; /* Sanity-check the EOCD64 */ if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ return 0; if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ return 0; /* CD can't be split. */ if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) return 0; /* Save the central directory offset for later use. */ zip->central_directory_offset = archive_le64dec(p + 48); /* TODO: Needs scanning backwards to find the eocd64 instead of assuming */ zip->central_directory_offset_adjusted = zip->central_directory_offset; return 32; } static int archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) { struct zip *zip = (struct zip *)a->format->data; int64_t file_size, current_offset; const char *p; int i, tail; /* If someone has already bid more than 32, then avoid trashing the look-ahead buffers with a seek. */ if (best_bid > 32) return (-1); file_size = __archive_read_seek(a, 0, SEEK_END); if (file_size <= 0) return 0; /* Search last 16k of file for end-of-central-directory * record (which starts with PK\005\006) */ tail = (int)zipmin(1024 * 16, file_size); current_offset = __archive_read_seek(a, -tail, SEEK_END); if (current_offset < 0) return 0; if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) return 0; /* Boyer-Moore search backwards from the end, since we want * to match the last EOCD in the file (there can be more than * one if there is an uncompressed Zip archive as a member * within this Zip archive). */ for (i = tail - 22; i > 0;) { switch (p[i]) { case 'P': if (memcmp(p + i, "PK\005\006", 4) == 0) { int ret = read_eocd(zip, p + i, current_offset + i); /* Zip64 EOCD locator precedes * regular EOCD if present. */ if (i >= 20 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { int ret_zip64 = read_zip64_eocd(a, zip, p + i - 20); if (ret_zip64 > ret) ret = ret_zip64; } return (ret); } i -= 4; break; case 'K': i -= 1; break; case 005: i -= 2; break; case 006: i -= 3; break; default: i -= 4; break; } } return 0; } /* The red-black trees are only used in seeking mode to manage * the in-memory copy of the central directory. */ static int cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct zip_entry *e1 = (const struct zip_entry *)n1; const struct zip_entry *e2 = (const struct zip_entry *)n2; if (e1->local_header_offset > e2->local_header_offset) return -1; if (e1->local_header_offset < e2->local_header_offset) return 1; return 0; } static int cmp_key(const struct archive_rb_node *n, const void *key) { /* This function won't be called */ (void)n; /* UNUSED */ (void)key; /* UNUSED */ return 1; } static const struct archive_rb_tree_ops rb_ops = { &cmp_node, &cmp_key }; static int rsrc_cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct zip_entry *e1 = (const struct zip_entry *)n1; const struct zip_entry *e2 = (const struct zip_entry *)n2; return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); } static int rsrc_cmp_key(const struct archive_rb_node *n, const void *key) { const struct zip_entry *e = (const struct zip_entry *)n; return (strcmp((const char *)key, e->rsrcname.s)); } static const struct archive_rb_tree_ops rb_rsrc_ops = { &rsrc_cmp_node, &rsrc_cmp_key }; static const char * rsrc_basename(const char *name, size_t name_length) { const char *s, *r; r = s = name; for (;;) { s = memchr(s, '/', name_length - (s - name)); if (s == NULL) break; r = ++s; } return (r); } static void expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) { struct archive_string str; struct zip_entry *dir; char *s; archive_string_init(&str); archive_strncpy(&str, name, name_length); for (;;) { s = strrchr(str.s, '/'); if (s == NULL) break; *s = '\0'; /* Transfer the parent directory from zip->tree_rsrc RB * tree to zip->tree RB tree to expose. */ dir = (struct zip_entry *) __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); if (dir == NULL) break; __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); archive_string_free(&dir->rsrcname); __archive_rb_tree_insert_node(&zip->tree, &dir->node); } archive_string_free(&str); } static int slurp_central_directory(struct archive_read *a, struct archive_entry* entry, struct zip *zip) { ssize_t i; unsigned found; int64_t correction; ssize_t bytes_avail; const char *p; /* * Find the start of the central directory. The end-of-CD * record has our starting point, but there are lots of * Zip archives which have had other data prepended to the * file, which makes the recorded offsets all too small. * So we search forward from the specified offset until we * find the real start of the central directory. Then we * know the correction we need to apply to account for leading * padding. */ if (__archive_read_seek(a, zip->central_directory_offset_adjusted, SEEK_SET) < 0) return ARCHIVE_FATAL; found = 0; while (!found) { if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) return ARCHIVE_FATAL; for (found = 0, i = 0; !found && i < bytes_avail - 4;) { switch (p[i + 3]) { case 'P': i += 3; break; case 'K': i += 2; break; case 001: i += 1; break; case 002: if (memcmp(p + i, "PK\001\002", 4) == 0) { p += i; found = 1; } else i += 4; break; case 005: i += 1; break; case 006: if (memcmp(p + i, "PK\005\006", 4) == 0) { p += i; found = 1; } else if (memcmp(p + i, "PK\006\006", 4) == 0) { p += i; found = 1; } else i += 1; break; default: i += 4; break; } } __archive_read_consume(a, i); } correction = archive_filter_bytes(&a->archive, 0) - zip->central_directory_offset; __archive_rb_tree_init(&zip->tree, &rb_ops); __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); zip->central_directory_entries_total = 0; while (1) { struct zip_entry *zip_entry; size_t filename_length, extra_length, comment_length; uint32_t external_attributes; const char *name, *r; if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) return ARCHIVE_FATAL; if (memcmp(p, "PK\006\006", 4) == 0 || memcmp(p, "PK\005\006", 4) == 0) { break; } else if (memcmp(p, "PK\001\002", 4) != 0) { archive_set_error(&a->archive, -1, "Invalid central directory signature"); return ARCHIVE_FATAL; } if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) return ARCHIVE_FATAL; zip_entry = calloc(1, sizeof(struct zip_entry)); if (zip_entry == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip entry"); return ARCHIVE_FATAL; } zip_entry->next = zip->zip_entries; zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; zip->zip_entries = zip_entry; zip->central_directory_entries_total++; /* version = p[4]; */ zip_entry->system = p[5]; /* version_required = archive_le16dec(p + 6); */ zip_entry->zip_flags = archive_le16dec(p + 8); if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ zip->has_encrypted_entries = 1; } zip_entry->compression = (char)archive_le16dec(p + 10); zip_entry->mtime = dos_to_unix(archive_le32dec(p + 12)); zip_entry->crc32 = archive_le32dec(p + 16); if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) zip_entry->decdat = p[13]; else zip_entry->decdat = p[19]; zip_entry->compressed_size = archive_le32dec(p + 20); zip_entry->uncompressed_size = archive_le32dec(p + 24); filename_length = archive_le16dec(p + 28); extra_length = archive_le16dec(p + 30); comment_length = archive_le16dec(p + 32); /* disk_start = archive_le16dec(p + 34); * Better be zero. * internal_attributes = archive_le16dec(p + 36); * text bit */ external_attributes = archive_le32dec(p + 38); zip_entry->local_header_offset = archive_le32dec(p + 42) + correction; /* If we can't guess the mode, leave it zero here; when we read the local file header we might get more information. */ if (zip_entry->system == 3) { zip_entry->mode = external_attributes >> 16; } else if (zip_entry->system == 0) { // Interpret MSDOS directory bit if (0x10 == (external_attributes & 0x10)) { zip_entry->mode = AE_IFDIR | 0775; } else { zip_entry->mode = AE_IFREG | 0664; } if (0x01 == (external_attributes & 0x01)) { // Read-only bit; strip write permissions zip_entry->mode &= 0555; } } else { zip_entry->mode = 0; } /* We're done with the regular data; get the filename and * extra data. */ __archive_read_consume(a, 46); p = __archive_read_ahead(a, filename_length + extra_length, NULL); if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file header"); return ARCHIVE_FATAL; } if (ARCHIVE_OK != process_extra(a, entry, p + filename_length, extra_length, zip_entry)) { return ARCHIVE_FATAL; } /* * Mac resource fork files are stored under the * "__MACOSX/" directory, so we should check if * it is. */ if (!zip->process_mac_extensions) { /* Treat every entry as a regular entry. */ __archive_rb_tree_insert_node(&zip->tree, &zip_entry->node); } else { name = p; r = rsrc_basename(name, filename_length); if (filename_length >= 9 && strncmp("__MACOSX/", name, 9) == 0) { /* If this file is not a resource fork nor * a directory. We should treat it as a non * resource fork file to expose it. */ if (name[filename_length-1] != '/' && (r - name < 3 || r[0] != '.' || r[1] != '_')) { __archive_rb_tree_insert_node( &zip->tree, &zip_entry->node); /* Expose its parent directories. */ expose_parent_dirs(zip, name, filename_length); } else { /* This file is a resource fork file or * a directory. */ archive_strncpy(&(zip_entry->rsrcname), name, filename_length); __archive_rb_tree_insert_node( &zip->tree_rsrc, &zip_entry->node); } } else { /* Generate resource fork name to find its * resource file at zip->tree_rsrc. */ /* If this is an entry ending with slash, * make the resource for name slash-less * as the actual resource fork doesn't end with '/'. */ size_t tmp_length = filename_length; if (tmp_length > 0 && name[tmp_length - 1] == '/') { tmp_length--; r = rsrc_basename(name, tmp_length); } archive_strcpy(&(zip_entry->rsrcname), "__MACOSX/"); archive_strncat(&(zip_entry->rsrcname), name, r - name); archive_strcat(&(zip_entry->rsrcname), "._"); archive_strncat(&(zip_entry->rsrcname), name + (r - name), tmp_length - (r - name)); /* Register an entry to RB tree to sort it by * file offset. */ __archive_rb_tree_insert_node(&zip->tree, &zip_entry->node); } } /* Skip the comment too ... */ __archive_read_consume(a, filename_length + extra_length + comment_length); } return ARCHIVE_OK; } static ssize_t zip_get_local_file_header_size(struct archive_read *a, size_t extra) { const char *p; ssize_t filename_length, extra_length; if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file header"); return (ARCHIVE_WARN); } p += extra; if (memcmp(p, "PK\003\004", 4) != 0) { archive_set_error(&a->archive, -1, "Damaged Zip archive"); return ARCHIVE_WARN; } filename_length = archive_le16dec(p + 26); extra_length = archive_le16dec(p + 28); return (30 + filename_length + extra_length); } static int zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, struct zip_entry *rsrc) { struct zip *zip = (struct zip *)a->format->data; unsigned char *metadata, *mp; int64_t offset = archive_filter_bytes(&a->archive, 0); size_t remaining_bytes, metadata_bytes; ssize_t hsize; int ret = ARCHIVE_OK, eof; switch(rsrc->compression) { case 0: /* No compression. */ if (rsrc->uncompressed_size != rsrc->compressed_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Malformed OS X metadata entry: " "inconsistent size"); return (ARCHIVE_FATAL); } #ifdef HAVE_ZLIB_H case 8: /* Deflate compression. */ #endif break; default: /* Unsupported compression. */ /* Return a warning. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported ZIP compression method (%s)", compression_name(rsrc->compression)); /* We can't decompress this entry, but we will * be able to skip() it and try the next entry. */ return (ARCHIVE_WARN); } if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Mac metadata is too large: %jd > 4M bytes", (intmax_t)rsrc->uncompressed_size); return (ARCHIVE_WARN); } if (rsrc->compressed_size > (4 * 1024 * 1024)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Mac metadata is too large: %jd > 4M bytes", (intmax_t)rsrc->compressed_size); return (ARCHIVE_WARN); } metadata = malloc((size_t)rsrc->uncompressed_size); if (metadata == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Mac metadata"); return (ARCHIVE_FATAL); } if (offset < rsrc->local_header_offset) __archive_read_consume(a, rsrc->local_header_offset - offset); else if (offset != rsrc->local_header_offset) { __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); } hsize = zip_get_local_file_header_size(a, 0); __archive_read_consume(a, hsize); remaining_bytes = (size_t)rsrc->compressed_size; metadata_bytes = (size_t)rsrc->uncompressed_size; mp = metadata; eof = 0; while (!eof && remaining_bytes) { const unsigned char *p; ssize_t bytes_avail; size_t bytes_used; p = __archive_read_ahead(a, 1, &bytes_avail); if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file header"); ret = ARCHIVE_WARN; goto exit_mac_metadata; } if ((size_t)bytes_avail > remaining_bytes) bytes_avail = remaining_bytes; switch(rsrc->compression) { case 0: /* No compression. */ if ((size_t)bytes_avail > metadata_bytes) bytes_avail = metadata_bytes; memcpy(mp, p, bytes_avail); bytes_used = (size_t)bytes_avail; metadata_bytes -= bytes_used; mp += bytes_used; if (metadata_bytes == 0) eof = 1; break; #ifdef HAVE_ZLIB_H case 8: /* Deflate compression. */ { int r; ret = zip_deflate_init(a, zip); if (ret != ARCHIVE_OK) goto exit_mac_metadata; zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)p; zip->stream.avail_in = (uInt)bytes_avail; zip->stream.total_in = 0; zip->stream.next_out = mp; zip->stream.avail_out = (uInt)metadata_bytes; zip->stream.total_out = 0; r = inflate(&zip->stream, 0); switch (r) { case Z_OK: break; case Z_STREAM_END: eof = 1; break; case Z_MEM_ERROR: archive_set_error(&a->archive, ENOMEM, "Out of memory for ZIP decompression"); ret = ARCHIVE_FATAL; goto exit_mac_metadata; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP decompression failed (%d)", r); ret = ARCHIVE_FATAL; goto exit_mac_metadata; } bytes_used = zip->stream.total_in; metadata_bytes -= zip->stream.total_out; mp += zip->stream.total_out; break; } #endif default: bytes_used = 0; break; } __archive_read_consume(a, bytes_used); remaining_bytes -= bytes_used; } archive_entry_copy_mac_metadata(entry, metadata, (size_t)rsrc->uncompressed_size - metadata_bytes); exit_mac_metadata: __archive_read_seek(a, offset, SEEK_SET); zip->decompress_init = 0; free(metadata); return (ret); } static int archive_read_format_zip_seekable_read_header(struct archive_read *a, struct archive_entry *entry) { struct zip *zip = (struct zip *)a->format->data; struct zip_entry *rsrc; int64_t offset; int r, ret = ARCHIVE_OK; /* * It should be sufficient to call archive_read_next_header() for * a reader to determine if an entry is encrypted or not. If the * encryption of an entry is only detectable when calling * archive_read_data(), so be it. We'll do the same check there * as well. */ if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) zip->has_encrypted_entries = 0; a->archive.archive_format = ARCHIVE_FORMAT_ZIP; if (a->archive.archive_format_name == NULL) a->archive.archive_format_name = "ZIP"; if (zip->zip_entries == NULL) { r = slurp_central_directory(a, entry, zip); if (r != ARCHIVE_OK) return r; /* Get first entry whose local header offset is lower than * other entries in the archive file. */ zip->entry = (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); } else if (zip->entry != NULL) { /* Get next entry in local header offset order. */ zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); } if (zip->entry == NULL) return ARCHIVE_EOF; if (zip->entry->rsrcname.s) rsrc = (struct zip_entry *)__archive_rb_tree_find_node( &zip->tree_rsrc, zip->entry->rsrcname.s); else rsrc = NULL; if (zip->cctx_valid) archive_decrypto_aes_ctr_release(&zip->cctx); if (zip->hctx_valid) archive_hmac_sha1_cleanup(&zip->hctx); zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; __archive_read_reset_passphrase(a); /* File entries are sorted by the header offset, we should mostly * use __archive_read_consume to advance a read point to avoid * redundant data reading. */ offset = archive_filter_bytes(&a->archive, 0); if (offset < zip->entry->local_header_offset) __archive_read_consume(a, zip->entry->local_header_offset - offset); else if (offset != zip->entry->local_header_offset) { __archive_read_seek(a, zip->entry->local_header_offset, SEEK_SET); } zip->unconsumed = 0; r = zip_read_local_file_header(a, entry, zip); if (r != ARCHIVE_OK) return r; if (rsrc) { int ret2 = zip_read_mac_metadata(a, entry, rsrc); if (ret2 < ret) ret = ret2; } return (ret); } /* * We're going to seek for the next header anyway, so we don't * need to bother doing anything here. */ static int archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) { struct zip *zip; zip = (struct zip *)(a->format->data); zip->unconsumed = 0; return (ARCHIVE_OK); } int archive_read_support_format_zip_seekable(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct zip *zip; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); zip = calloc(1, sizeof(*zip)); if (zip == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip data"); return (ARCHIVE_FATAL); } #ifdef HAVE_COPYFILE_H /* Set this by default on Mac OS. */ zip->process_mac_extensions = 1; #endif /* * Until enough data has been read, we cannot tell about * any encrypted entries yet. */ zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; zip->crc32func = real_crc32; r = __archive_read_register_format(a, zip, "zip", archive_read_format_zip_seekable_bid, archive_read_format_zip_options, archive_read_format_zip_seekable_read_header, archive_read_format_zip_read_data, archive_read_format_zip_read_data_skip_seekable, NULL, archive_read_format_zip_cleanup, archive_read_support_format_zip_capabilities_seekable, archive_read_format_zip_has_encrypted_entries); if (r != ARCHIVE_OK) free(zip); return (ARCHIVE_OK); } /*# vim:set noet:*/ diff --git a/contrib/libarchive/libarchive/archive_string.c b/contrib/libarchive/libarchive/archive_string.c index 7437715f9122..3bb978335eb8 100644 --- a/contrib/libarchive/libarchive/archive_string.c +++ b/contrib/libarchive/libarchive/archive_string.c @@ -1,4347 +1,4348 @@ /*- * Copyright (c) 2003-2011 Tim Kientzle * Copyright (c) 2011-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" /* * Basic resizable string support, to simplify manipulating arbitrary-sized * strings while minimizing heap activity. * * In particular, the buffer used by a string object is only grown, it * never shrinks, so you can clear and reuse the same string object * without incurring additional memory allocations. */ #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_ICONV_H #include #endif #ifdef HAVE_LANGINFO_H #include #endif #ifdef HAVE_LOCALCHARSET_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_WCHAR_H #include #endif #if defined(_WIN32) && !defined(__CYGWIN__) #include #include #endif #include "archive_endian.h" #include "archive_private.h" #include "archive_string.h" #include "archive_string_composition.h" #if !defined(HAVE_WMEMCPY) && !defined(wmemcpy) #define wmemcpy(a,b,i) (wchar_t *)memcpy((a), (b), (i) * sizeof(wchar_t)) #endif #if !defined(HAVE_WMEMMOVE) && !defined(wmemmove) #define wmemmove(a,b,i) (wchar_t *)memmove((a), (b), (i) * sizeof(wchar_t)) #endif #undef max #define max(a, b) ((a)>(b)?(a):(b)) struct archive_string_conv { struct archive_string_conv *next; char *from_charset; char *to_charset; unsigned from_cp; unsigned to_cp; /* Set 1 if from_charset and to_charset are the same. */ int same; int flag; #define SCONV_TO_CHARSET 1 /* MBS is being converted to specified * charset. */ #define SCONV_FROM_CHARSET (1<<1) /* MBS is being converted from * specified charset. */ #define SCONV_BEST_EFFORT (1<<2) /* Copy at least ASCII code. */ #define SCONV_WIN_CP (1<<3) /* Use Windows API for converting * MBS. */ #define SCONV_UTF8_LIBARCHIVE_2 (1<<4) /* Incorrect UTF-8 made by libarchive * 2.x in the wrong assumption. */ #define SCONV_NORMALIZATION_C (1<<6) /* Need normalization to be Form C. * Before UTF-8 characters are actually * processed. */ #define SCONV_NORMALIZATION_D (1<<7) /* Need normalization to be Form D. * Before UTF-8 characters are actually * processed. * Currently this only for MAC OS X. */ #define SCONV_TO_UTF8 (1<<8) /* "to charset" side is UTF-8. */ #define SCONV_FROM_UTF8 (1<<9) /* "from charset" side is UTF-8. */ #define SCONV_TO_UTF16BE (1<<10) /* "to charset" side is UTF-16BE. */ #define SCONV_FROM_UTF16BE (1<<11) /* "from charset" side is UTF-16BE. */ #define SCONV_TO_UTF16LE (1<<12) /* "to charset" side is UTF-16LE. */ #define SCONV_FROM_UTF16LE (1<<13) /* "from charset" side is UTF-16LE. */ #define SCONV_TO_UTF16 (SCONV_TO_UTF16BE | SCONV_TO_UTF16LE) #define SCONV_FROM_UTF16 (SCONV_FROM_UTF16BE | SCONV_FROM_UTF16LE) #if HAVE_ICONV iconv_t cd; iconv_t cd_w;/* Use at archive_mstring on * Windows. */ #endif /* A temporary buffer for normalization. */ struct archive_string utftmp; int (*converter[2])(struct archive_string *, const void *, size_t, struct archive_string_conv *); int nconverter; }; #define CP_C_LOCALE 0 /* "C" locale only for this file. */ #define CP_UTF16LE 1200 #define CP_UTF16BE 1201 #define IS_HIGH_SURROGATE_LA(uc) ((uc) >= 0xD800 && (uc) <= 0xDBFF) #define IS_LOW_SURROGATE_LA(uc) ((uc) >= 0xDC00 && (uc) <= 0xDFFF) #define IS_SURROGATE_PAIR_LA(uc) ((uc) >= 0xD800 && (uc) <= 0xDFFF) #define UNICODE_MAX 0x10FFFF #define UNICODE_R_CHAR 0xFFFD /* Replacement character. */ /* Set U+FFFD(Replacement character) in UTF-8. */ static const char utf8_replacement_char[] = {0xef, 0xbf, 0xbd}; static struct archive_string_conv *find_sconv_object(struct archive *, const char *, const char *); static void add_sconv_object(struct archive *, struct archive_string_conv *); static struct archive_string_conv *create_sconv_object(const char *, const char *, unsigned, int); static void free_sconv_object(struct archive_string_conv *); static struct archive_string_conv *get_sconv_object(struct archive *, const char *, const char *, int); static unsigned make_codepage_from_charset(const char *); static unsigned get_current_codepage(void); static unsigned get_current_oemcp(void); static size_t mbsnbytes(const void *, size_t); static size_t utf16nbytes(const void *, size_t); #if defined(_WIN32) && !defined(__CYGWIN__) static int archive_wstring_append_from_mbs_in_codepage( struct archive_wstring *, const char *, size_t, struct archive_string_conv *); static int archive_string_append_from_wcs_in_codepage(struct archive_string *, const wchar_t *, size_t, struct archive_string_conv *); static int strncat_in_codepage(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int win_strncat_from_utf16be(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int win_strncat_from_utf16le(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int win_strncat_to_utf16be(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int win_strncat_to_utf16le(struct archive_string *, const void *, size_t, struct archive_string_conv *); #endif static int best_effort_strncat_from_utf16be(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int best_effort_strncat_from_utf16le(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int best_effort_strncat_to_utf16be(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int best_effort_strncat_to_utf16le(struct archive_string *, const void *, size_t, struct archive_string_conv *); #if defined(HAVE_ICONV) static int iconv_strncat_in_locale(struct archive_string *, const void *, size_t, struct archive_string_conv *); #endif static int best_effort_strncat_in_locale(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int _utf8_to_unicode(uint32_t *, const char *, size_t); static int utf8_to_unicode(uint32_t *, const char *, size_t); static inline uint32_t combine_surrogate_pair(uint32_t, uint32_t); static int cesu8_to_unicode(uint32_t *, const char *, size_t); static size_t unicode_to_utf8(char *, size_t, uint32_t); static int utf16_to_unicode(uint32_t *, const char *, size_t, int); static size_t unicode_to_utf16be(char *, size_t, uint32_t); static size_t unicode_to_utf16le(char *, size_t, uint32_t); static int strncat_from_utf8_libarchive2(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int strncat_from_utf8_to_utf8(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int archive_string_normalize_C(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int archive_string_normalize_D(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int archive_string_append_unicode(struct archive_string *, const void *, size_t, struct archive_string_conv *); #if defined __LITTLE_ENDIAN__ #define IS_BIG_ENDIAN 0 #elif defined __BIG_ENDIAN__ #define IS_BIG_ENDIAN 1 #elif defined(__BYTE_ORDER__) && defined(__ORDER_LITTLE_ENDIAN__) && (__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__) #define IS_BIG_ENDIAN 0 #elif defined(__BYTE_ORDER__) && defined(__ORDER_BIG_ENDIAN__) && (__BYTE_ORDER__ == __ORDER_BIG_ENDIAN__) #define IS_BIG_ENDIAN 1 #elif defined(_MSC_VER) && (defined(_M_IX86) || defined(_M_AMD64) || defined(_M_X64) || defined(_M_ARM64)) #define IS_BIG_ENDIAN 0 #else // Detect endianness at runtime. static int is_big_endian(void) { uint16_t d = 1; return (archive_be16dec(&d) == 1); } #define IS_BIG_ENDIAN is_big_endian() #endif static struct archive_string * archive_string_append(struct archive_string *as, const char *p, size_t s) { if (archive_string_ensure(as, as->length + s + 1) == NULL) return (NULL); if (s) memmove(as->s + as->length, p, s); as->length += s; as->s[as->length] = 0; return (as); } static struct archive_wstring * archive_wstring_append(struct archive_wstring *as, const wchar_t *p, size_t s) { if (archive_wstring_ensure(as, as->length + s + 1) == NULL) return (NULL); if (s) wmemmove(as->s + as->length, p, s); as->length += s; as->s[as->length] = 0; return (as); } struct archive_string * archive_array_append(struct archive_string *as, const char *p, size_t s) { return archive_string_append(as, p, s); } void archive_string_concat(struct archive_string *dest, struct archive_string *src) { if (archive_string_append(dest, src->s, src->length) == NULL) __archive_errx(1, "Out of memory"); } void archive_wstring_concat(struct archive_wstring *dest, struct archive_wstring *src) { if (archive_wstring_append(dest, src->s, src->length) == NULL) __archive_errx(1, "Out of memory"); } void archive_string_free(struct archive_string *as) { as->length = 0; as->buffer_length = 0; free(as->s); as->s = NULL; } void archive_wstring_free(struct archive_wstring *as) { as->length = 0; as->buffer_length = 0; free(as->s); as->s = NULL; } struct archive_wstring * archive_wstring_ensure(struct archive_wstring *as, size_t s) { return (struct archive_wstring *) archive_string_ensure((struct archive_string *)as, s * sizeof(wchar_t)); } /* Returns NULL on any allocation failure. */ struct archive_string * archive_string_ensure(struct archive_string *as, size_t s) { char *p; size_t new_length; /* If buffer is already big enough, don't reallocate. */ if (as->s && (s <= as->buffer_length)) return (as); /* * Growing the buffer at least exponentially ensures that * append operations are always linear in the number of * characters appended. Using a smaller growth rate for * larger buffers reduces memory waste somewhat at the cost of * a larger constant factor. */ if (as->buffer_length < 32) /* Start with a minimum 32-character buffer. */ new_length = 32; else if (as->buffer_length < 8192) /* Buffers under 8k are doubled for speed. */ new_length = as->buffer_length + as->buffer_length; else { /* Buffers 8k and over grow by at least 25% each time. */ new_length = as->buffer_length + as->buffer_length / 4; /* Be safe: If size wraps, fail. */ if (new_length < as->buffer_length) { /* On failure, wipe the string and return NULL. */ archive_string_free(as); errno = ENOMEM;/* Make sure errno has ENOMEM. */ return (NULL); } } /* * The computation above is a lower limit to how much we'll * grow the buffer. In any case, we have to grow it enough to * hold the request. */ if (new_length < s) new_length = s; /* Now we can reallocate the buffer. */ p = realloc(as->s, new_length); if (p == NULL) { /* On failure, wipe the string and return NULL. */ archive_string_free(as); errno = ENOMEM;/* Make sure errno has ENOMEM. */ return (NULL); } as->s = p; as->buffer_length = new_length; return (as); } /* * TODO: See if there's a way to avoid scanning * the source string twice. Then test to see * if it actually helps (remember that we're almost * always called with pretty short arguments, so * such an optimization might not help). */ struct archive_string * archive_strncat(struct archive_string *as, const void *_p, size_t n) { size_t s; const char *p, *pp; p = (const char *)_p; /* Like strlen(p), except won't examine positions beyond p[n]. */ s = 0; pp = p; while (s < n && *pp) { pp++; s++; } if ((as = archive_string_append(as, p, s)) == NULL) __archive_errx(1, "Out of memory"); return (as); } struct archive_wstring * archive_wstrncat(struct archive_wstring *as, const wchar_t *p, size_t n) { size_t s; const wchar_t *pp; /* Like strlen(p), except won't examine positions beyond p[n]. */ s = 0; pp = p; while (s < n && *pp) { pp++; s++; } if ((as = archive_wstring_append(as, p, s)) == NULL) __archive_errx(1, "Out of memory"); return (as); } struct archive_string * archive_strcat(struct archive_string *as, const void *p) { /* strcat is just strncat without an effective limit. * Assert that we'll never get called with a source * string over 16MB. * TODO: Review all uses of strcat in the source * and try to replace them with strncat(). */ return archive_strncat(as, p, 0x1000000); } struct archive_wstring * archive_wstrcat(struct archive_wstring *as, const wchar_t *p) { /* Ditto. */ return archive_wstrncat(as, p, 0x1000000); } struct archive_string * archive_strappend_char(struct archive_string *as, char c) { if ((as = archive_string_append(as, &c, 1)) == NULL) __archive_errx(1, "Out of memory"); return (as); } struct archive_wstring * archive_wstrappend_wchar(struct archive_wstring *as, wchar_t c) { if ((as = archive_wstring_append(as, &c, 1)) == NULL) __archive_errx(1, "Out of memory"); return (as); } /* * Get the "current character set" name to use with iconv. * On FreeBSD, the empty character set name "" chooses * the correct character encoding for the current locale, * so this isn't necessary. * But iconv on Mac OS 10.6 doesn't seem to handle this correctly; * on that system, we have to explicitly call nl_langinfo() * to get the right name. Not sure about other platforms. * * NOTE: GNU libiconv does not recognize the character-set name * which some platform nl_langinfo(CODESET) returns, so we should * use locale_charset() instead of nl_langinfo(CODESET) for GNU libiconv. */ static const char * default_iconv_charset(const char *charset) { if (charset != NULL && charset[0] != '\0') return charset; #if HAVE_LOCALE_CHARSET && !defined(__APPLE__) /* locale_charset() is broken on Mac OS */ return locale_charset(); #elif HAVE_NL_LANGINFO return nl_langinfo(CODESET); #else return ""; #endif } #if defined(_WIN32) && !defined(__CYGWIN__) /* * Convert MBS to WCS. * Note: returns -1 if conversion fails. */ int archive_wstring_append_from_mbs(struct archive_wstring *dest, const char *p, size_t len) { return archive_wstring_append_from_mbs_in_codepage(dest, p, len, NULL); } static int archive_wstring_append_from_mbs_in_codepage(struct archive_wstring *dest, const char *s, size_t length, struct archive_string_conv *sc) { int ret = 0; size_t count; UINT from_cp; if (sc != NULL) from_cp = sc->from_cp; else from_cp = get_current_codepage(); if (from_cp == CP_C_LOCALE) { /* * "C" locale special processing. */ wchar_t *ws; const unsigned char *mp; if (NULL == archive_wstring_ensure(dest, dest->length + length + 1)) return (-1); ws = dest->s + dest->length; mp = (const unsigned char *)s; count = 0; while (count < length && *mp) { *ws++ = (wchar_t)*mp++; count++; } } else if (sc != NULL && (sc->flag & (SCONV_NORMALIZATION_C | SCONV_NORMALIZATION_D))) { /* * Normalize UTF-8 and UTF-16BE and convert it directly * to UTF-16 as wchar_t. */ struct archive_string u16; int saved_flag = sc->flag;/* save current flag. */ if (IS_BIG_ENDIAN) sc->flag |= SCONV_TO_UTF16BE; else sc->flag |= SCONV_TO_UTF16LE; if (sc->flag & SCONV_FROM_UTF16) { /* * UTF-16BE/LE NFD ===> UTF-16 NFC * UTF-16BE/LE NFC ===> UTF-16 NFD */ count = utf16nbytes(s, length); } else { /* * UTF-8 NFD ===> UTF-16 NFC * UTF-8 NFC ===> UTF-16 NFD */ count = mbsnbytes(s, length); } u16.s = (char *)dest->s; u16.length = dest->length << 1; u16.buffer_length = dest->buffer_length; if (sc->flag & SCONV_NORMALIZATION_C) ret = archive_string_normalize_C(&u16, s, count, sc); else ret = archive_string_normalize_D(&u16, s, count, sc); dest->s = (wchar_t *)u16.s; dest->length = u16.length >> 1; dest->buffer_length = u16.buffer_length; sc->flag = saved_flag;/* restore the saved flag. */ return (ret); } else if (sc != NULL && (sc->flag & SCONV_FROM_UTF16)) { count = utf16nbytes(s, length); count >>= 1; /* to be WCS length */ /* Allocate memory for WCS. */ if (NULL == archive_wstring_ensure(dest, dest->length + count + 1)) return (-1); wmemcpy(dest->s + dest->length, (const wchar_t *)s, count); if ((sc->flag & SCONV_FROM_UTF16BE) && !IS_BIG_ENDIAN) { uint16_t *u16 = (uint16_t *)(dest->s + dest->length); size_t b; for (b = 0; b < count; b++) { uint16_t val = archive_le16dec(u16+b); archive_be16enc(u16+b, val); } } else if ((sc->flag & SCONV_FROM_UTF16LE) && IS_BIG_ENDIAN) { uint16_t *u16 = (uint16_t *)(dest->s + dest->length); size_t b; for (b = 0; b < count; b++) { uint16_t val = archive_be16dec(u16+b); archive_le16enc(u16+b, val); } } } else { DWORD mbflag; size_t buffsize; if (sc == NULL) mbflag = 0; else if (sc->flag & SCONV_FROM_CHARSET) { /* Do not trust the length which comes from * an archive file. */ length = mbsnbytes(s, length); mbflag = 0; } else mbflag = MB_PRECOMPOSED; mbflag |= MB_ERR_INVALID_CHARS; buffsize = dest->length + length + 1; do { int r; /* MultiByteToWideChar is limited to int. */ if (length > (size_t)INT_MAX || (dest->buffer_length >> 1) > (size_t)INT_MAX) return (-1); /* Allocate memory for WCS. */ if (NULL == archive_wstring_ensure(dest, buffsize)) return (-1); /* Convert MBS to WCS. */ r = MultiByteToWideChar(from_cp, mbflag, s, (int)length, dest->s + dest->length, (int)(dest->buffer_length >> 1) -1); if (r == 0 && GetLastError() == ERROR_INSUFFICIENT_BUFFER) { /* Expand the WCS buffer. */ buffsize = dest->buffer_length << 1; continue; } if (r == 0 && length != 0) ret = -1; count = (size_t)r; break; } while (1); } dest->length += count; dest->s[dest->length] = L'\0'; return (ret); } #else /* * Convert MBS to WCS. * Note: returns -1 if conversion fails. */ int archive_wstring_append_from_mbs(struct archive_wstring *dest, const char *p, size_t len) { size_t r; int ret_val = 0; /* * No single byte will be more than one wide character, * so this length estimate will always be big enough. */ // size_t wcs_length = len; size_t mbs_length = len; const char *mbs = p; wchar_t *wcs; #if HAVE_MBRTOWC mbstate_t shift_state; memset(&shift_state, 0, sizeof(shift_state)); #endif /* * As we decided to have wcs_length == mbs_length == len * we can use len here instead of wcs_length */ if (NULL == archive_wstring_ensure(dest, dest->length + len + 1)) return (-1); wcs = dest->s + dest->length; /* * We cannot use mbsrtowcs/mbstowcs here because those may convert * extra MBS when strlen(p) > len and one wide character consists of * multi bytes. */ while (*mbs && mbs_length > 0) { /* * The buffer we allocated is always big enough. * Keep this code path in a comment if we decide to choose * smaller wcs_length in the future */ /* if (wcs_length == 0) { dest->length = wcs - dest->s; dest->s[dest->length] = L'\0'; wcs_length = mbs_length; if (NULL == archive_wstring_ensure(dest, dest->length + wcs_length + 1)) return (-1); wcs = dest->s + dest->length; } */ #if HAVE_MBRTOWC r = mbrtowc(wcs, mbs, mbs_length, &shift_state); #else r = mbtowc(wcs, mbs, mbs_length); #endif if (r == (size_t)-1 || r == (size_t)-2) { ret_val = -1; break; } if (r == 0 || r > mbs_length) break; wcs++; // wcs_length--; mbs += r; mbs_length -= r; } dest->length = wcs - dest->s; dest->s[dest->length] = L'\0'; return (ret_val); } #endif #if defined(_WIN32) && !defined(__CYGWIN__) /* * WCS ==> MBS. * Note: returns -1 if conversion fails. * * Win32 builds use WideCharToMultiByte from the Windows API. * (Maybe Cygwin should too? WideCharToMultiByte will know a * lot more about local character encodings than the wcrtomb() * wrapper is going to know.) */ int archive_string_append_from_wcs(struct archive_string *as, const wchar_t *w, size_t len) { return archive_string_append_from_wcs_in_codepage(as, w, len, NULL); } static int archive_string_append_from_wcs_in_codepage(struct archive_string *as, const wchar_t *ws, size_t len, struct archive_string_conv *sc) { BOOL defchar_used, *dp; int ret = 0; UINT to_cp; size_t count, wslen = len; if (sc != NULL) to_cp = sc->to_cp; else to_cp = get_current_codepage(); if (to_cp == CP_C_LOCALE) { /* * "C" locale special processing. */ const wchar_t *wp = ws; char *p; if (NULL == archive_string_ensure(as, as->length + wslen +1)) return (-1); p = as->s + as->length; count = 0; defchar_used = 0; while (count < wslen && *wp) { if (*wp > 255) { *p++ = '?'; wp++; defchar_used = 1; } else *p++ = (char)*wp++; count++; } } else if (sc != NULL && (sc->flag & SCONV_TO_UTF16)) { uint16_t *u16; if (NULL == archive_string_ensure(as, as->length + len * 2 + 2)) return (-1); u16 = (uint16_t *)(as->s + as->length); count = 0; defchar_used = 0; if (sc->flag & SCONV_TO_UTF16BE) { while (count < len && *ws) { archive_be16enc(u16+count, *ws); ws++; count++; } } else { while (count < len && *ws) { archive_le16enc(u16+count, *ws); ws++; count++; } } count <<= 1; /* to be byte size */ } else { /* Make sure the MBS buffer has plenty to set. */ if (NULL == archive_string_ensure(as, as->length + len * 2 + 1)) return (-1); do { int r; defchar_used = 0; if (to_cp == CP_UTF8 || sc == NULL) dp = NULL; else dp = &defchar_used; /* WideCharToMultiByte is limited to int. */ if (as->buffer_length - as->length - 1 > (size_t)INT_MAX || wslen > (size_t)INT_MAX) return (-1); r = WideCharToMultiByte(to_cp, 0, ws, (int)wslen, as->s + as->length, (int)(as->buffer_length - as->length - 1), NULL, dp); if (r == 0 && GetLastError() == ERROR_INSUFFICIENT_BUFFER) { /* Expand the MBS buffer and retry. */ if (NULL == archive_string_ensure(as, as->buffer_length + len)) return (-1); continue; } if (r == 0) ret = -1; count = (size_t)r; break; } while (1); } as->length += count; as->s[as->length] = '\0'; return (defchar_used?-1:ret); } #elif defined(HAVE_WCTOMB) || defined(HAVE_WCRTOMB) /* * Translates a wide character string into current locale character set * and appends to the archive_string. Note: returns -1 if conversion * fails. */ int archive_string_append_from_wcs(struct archive_string *as, const wchar_t *w, size_t len) { /* We cannot use the standard wcstombs() here because it * cannot tell us how big the output buffer should be. So * I've built a loop around wcrtomb() or wctomb() that * converts a character at a time and resizes the string as * needed. We prefer wcrtomb() when it's available because * it's thread-safe. */ int n, ret_val = 0; char *p; char *end; #if HAVE_WCRTOMB mbstate_t shift_state; memset(&shift_state, 0, sizeof(shift_state)); #else /* Clear the shift state before starting. */ wctomb(NULL, L'\0'); #endif /* * Allocate buffer for MBS. * We need this allocation here since it is possible that * as->s is still NULL. */ if (archive_string_ensure(as, as->length + len + 1) == NULL) return (-1); p = as->s + as->length; end = as->s + as->buffer_length - MB_CUR_MAX -1; while (*w != L'\0' && len > 0) { if (p >= end) { as->length = p - as->s; as->s[as->length] = '\0'; /* Re-allocate buffer for MBS. */ if (archive_string_ensure(as, as->length + max(len * 2, (size_t)MB_CUR_MAX) + 1) == NULL) return (-1); p = as->s + as->length; end = as->s + as->buffer_length - MB_CUR_MAX -1; } #if HAVE_WCRTOMB n = wcrtomb(p, *w++, &shift_state); #else n = wctomb(p, *w++); #endif if (n == -1) { if (errno == EILSEQ) { /* Skip an illegal wide char. */ *p++ = '?'; ret_val = -1; } else { ret_val = -1; break; } } else p += n; len--; } as->length = p - as->s; as->s[as->length] = '\0'; return (ret_val); } #else /* HAVE_WCTOMB || HAVE_WCRTOMB */ /* * TODO: Test if __STDC_ISO_10646__ is defined. * Non-Windows uses ISO C wcrtomb() or wctomb() to perform the conversion * one character at a time. If a non-Windows platform doesn't have * either of these, fall back to the built-in UTF8 conversion. */ int archive_string_append_from_wcs(struct archive_string *as, const wchar_t *w, size_t len) { (void)as;/* UNUSED */ (void)w;/* UNUSED */ (void)len;/* UNUSED */ errno = ENOSYS; return (-1); } #endif /* HAVE_WCTOMB || HAVE_WCRTOMB */ /* * Find a string conversion object by a pair of 'from' charset name * and 'to' charset name from an archive object. * Return NULL if not found. */ static struct archive_string_conv * find_sconv_object(struct archive *a, const char *fc, const char *tc) { struct archive_string_conv *sc; if (a == NULL) return (NULL); for (sc = a->sconv; sc != NULL; sc = sc->next) { if (strcmp(sc->from_charset, fc) == 0 && strcmp(sc->to_charset, tc) == 0) break; } return (sc); } /* * Register a string object to an archive object. */ static void add_sconv_object(struct archive *a, struct archive_string_conv *sc) { struct archive_string_conv **psc; /* Add a new sconv to sconv list. */ psc = &(a->sconv); while (*psc != NULL) psc = &((*psc)->next); *psc = sc; } static void add_converter(struct archive_string_conv *sc, int (*converter) (struct archive_string *, const void *, size_t, struct archive_string_conv *)) { if (sc == NULL || sc->nconverter >= 2) __archive_errx(1, "Programming error"); sc->converter[sc->nconverter++] = converter; } static void setup_converter(struct archive_string_conv *sc) { /* Reset. */ sc->nconverter = 0; /* * Perform special sequence for the incorrect UTF-8 filenames * made by libarchive2.x. */ if (sc->flag & SCONV_UTF8_LIBARCHIVE_2) { add_converter(sc, strncat_from_utf8_libarchive2); return; } /* * Convert a string to UTF-16BE/LE. */ if (sc->flag & SCONV_TO_UTF16) { /* * If the current locale is UTF-8, we can translate * a UTF-8 string into a UTF-16BE string. */ if (sc->flag & SCONV_FROM_UTF8) { add_converter(sc, archive_string_append_unicode); return; } #if defined(_WIN32) && !defined(__CYGWIN__) if (sc->flag & SCONV_WIN_CP) { if (sc->flag & SCONV_TO_UTF16BE) add_converter(sc, win_strncat_to_utf16be); else add_converter(sc, win_strncat_to_utf16le); return; } #endif #if defined(HAVE_ICONV) if (sc->cd != (iconv_t)-1) { add_converter(sc, iconv_strncat_in_locale); return; } #endif if (sc->flag & SCONV_BEST_EFFORT) { if (sc->flag & SCONV_TO_UTF16BE) add_converter(sc, best_effort_strncat_to_utf16be); else add_converter(sc, best_effort_strncat_to_utf16le); } else /* Make sure we have no converter. */ sc->nconverter = 0; return; } /* * Convert a string from UTF-16BE/LE. */ if (sc->flag & SCONV_FROM_UTF16) { /* * At least we should normalize a UTF-16BE string. */ if (sc->flag & SCONV_NORMALIZATION_D) add_converter(sc,archive_string_normalize_D); else if (sc->flag & SCONV_NORMALIZATION_C) add_converter(sc, archive_string_normalize_C); if (sc->flag & SCONV_TO_UTF8) { /* * If the current locale is UTF-8, we can translate * a UTF-16BE/LE string into a UTF-8 string directly. */ if (!(sc->flag & (SCONV_NORMALIZATION_D |SCONV_NORMALIZATION_C))) add_converter(sc, archive_string_append_unicode); return; } #if defined(_WIN32) && !defined(__CYGWIN__) if (sc->flag & SCONV_WIN_CP) { if (sc->flag & SCONV_FROM_UTF16BE) add_converter(sc, win_strncat_from_utf16be); else add_converter(sc, win_strncat_from_utf16le); return; } #endif #if defined(HAVE_ICONV) if (sc->cd != (iconv_t)-1) { add_converter(sc, iconv_strncat_in_locale); return; } #endif if ((sc->flag & (SCONV_BEST_EFFORT | SCONV_FROM_UTF16BE)) == (SCONV_BEST_EFFORT | SCONV_FROM_UTF16BE)) add_converter(sc, best_effort_strncat_from_utf16be); else if ((sc->flag & (SCONV_BEST_EFFORT | SCONV_FROM_UTF16LE)) == (SCONV_BEST_EFFORT | SCONV_FROM_UTF16LE)) add_converter(sc, best_effort_strncat_from_utf16le); else /* Make sure we have no converter. */ sc->nconverter = 0; return; } if (sc->flag & SCONV_FROM_UTF8) { /* * At least we should normalize a UTF-8 string. */ if (sc->flag & SCONV_NORMALIZATION_D) add_converter(sc,archive_string_normalize_D); else if (sc->flag & SCONV_NORMALIZATION_C) add_converter(sc, archive_string_normalize_C); /* * Copy UTF-8 string with a check of CESU-8. * Apparently, iconv does not check surrogate pairs in UTF-8 * when both from-charset and to-charset are UTF-8, and then * we use our UTF-8 copy code. */ if (sc->flag & SCONV_TO_UTF8) { /* * If the current locale is UTF-8, we can translate * a UTF-16BE string into a UTF-8 string directly. */ if (!(sc->flag & (SCONV_NORMALIZATION_D |SCONV_NORMALIZATION_C))) add_converter(sc, strncat_from_utf8_to_utf8); return; } } #if defined(_WIN32) && !defined(__CYGWIN__) /* * On Windows we can use Windows API for a string conversion. */ if (sc->flag & SCONV_WIN_CP) { add_converter(sc, strncat_in_codepage); return; } #endif #if HAVE_ICONV if (sc->cd != (iconv_t)-1) { add_converter(sc, iconv_strncat_in_locale); /* * iconv generally does not support UTF-8-MAC and so * we have to the output of iconv from NFC to NFD if * need. */ if ((sc->flag & SCONV_FROM_CHARSET) && (sc->flag & SCONV_TO_UTF8)) { if (sc->flag & SCONV_NORMALIZATION_D) add_converter(sc, archive_string_normalize_D); } return; } #endif /* * Try conversion in the best effort or no conversion. */ if ((sc->flag & SCONV_BEST_EFFORT) || sc->same) add_converter(sc, best_effort_strncat_in_locale); else /* Make sure we have no converter. */ sc->nconverter = 0; } /* * Return canonicalized charset-name but this supports just UTF-8, UTF-16BE * and CP932 which are referenced in create_sconv_object(). */ static const char * canonical_charset_name(const char *charset) { char cs[16]; char *p; const char *s; if (charset == NULL || charset[0] == '\0' || strlen(charset) > 15) return (charset); /* Copy name to uppercase. */ p = cs; s = charset; while (*s) { char c = *s++; if (c >= 'a' && c <= 'z') c -= 'a' - 'A'; *p++ = c; } *p++ = '\0'; if (strcmp(cs, "UTF-8") == 0 || strcmp(cs, "UTF8") == 0) return ("UTF-8"); if (strcmp(cs, "UTF-16BE") == 0 || strcmp(cs, "UTF16BE") == 0) return ("UTF-16BE"); if (strcmp(cs, "UTF-16LE") == 0 || strcmp(cs, "UTF16LE") == 0) return ("UTF-16LE"); if (strcmp(cs, "CP932") == 0) return ("CP932"); return (charset); } /* * Create a string conversion object. */ static struct archive_string_conv * create_sconv_object(const char *fc, const char *tc, unsigned current_codepage, int flag) { struct archive_string_conv *sc; sc = calloc(1, sizeof(*sc)); if (sc == NULL) return (NULL); sc->next = NULL; sc->from_charset = strdup(fc); if (sc->from_charset == NULL) { free(sc); return (NULL); } sc->to_charset = strdup(tc); if (sc->to_charset == NULL) { free(sc->from_charset); free(sc); return (NULL); } archive_string_init(&sc->utftmp); if (flag & SCONV_TO_CHARSET) { /* * Convert characters from the current locale charset to * a specified charset. */ sc->from_cp = current_codepage; sc->to_cp = make_codepage_from_charset(tc); #if defined(_WIN32) && !defined(__CYGWIN__) if (IsValidCodePage(sc->to_cp)) flag |= SCONV_WIN_CP; #endif } else if (flag & SCONV_FROM_CHARSET) { /* * Convert characters from a specified charset to * the current locale charset. */ sc->to_cp = current_codepage; sc->from_cp = make_codepage_from_charset(fc); #if defined(_WIN32) && !defined(__CYGWIN__) if (IsValidCodePage(sc->from_cp)) flag |= SCONV_WIN_CP; #endif } /* * Check if "from charset" and "to charset" are the same. */ if (strcmp(fc, tc) == 0 || (sc->from_cp != (unsigned)-1 && sc->from_cp == sc->to_cp)) sc->same = 1; else sc->same = 0; /* * Mark if "from charset" or "to charset" are UTF-8 or UTF-16BE/LE. */ if (strcmp(tc, "UTF-8") == 0) flag |= SCONV_TO_UTF8; else if (strcmp(tc, "UTF-16BE") == 0) flag |= SCONV_TO_UTF16BE; else if (strcmp(tc, "UTF-16LE") == 0) flag |= SCONV_TO_UTF16LE; if (strcmp(fc, "UTF-8") == 0) flag |= SCONV_FROM_UTF8; else if (strcmp(fc, "UTF-16BE") == 0) flag |= SCONV_FROM_UTF16BE; else if (strcmp(fc, "UTF-16LE") == 0) flag |= SCONV_FROM_UTF16LE; #if defined(_WIN32) && !defined(__CYGWIN__) if (sc->to_cp == CP_UTF8) flag |= SCONV_TO_UTF8; else if (sc->to_cp == CP_UTF16BE) flag |= SCONV_TO_UTF16BE | SCONV_WIN_CP; else if (sc->to_cp == CP_UTF16LE) flag |= SCONV_TO_UTF16LE | SCONV_WIN_CP; if (sc->from_cp == CP_UTF8) flag |= SCONV_FROM_UTF8; else if (sc->from_cp == CP_UTF16BE) flag |= SCONV_FROM_UTF16BE | SCONV_WIN_CP; else if (sc->from_cp == CP_UTF16LE) flag |= SCONV_FROM_UTF16LE | SCONV_WIN_CP; #endif /* * Set a flag for Unicode NFD. Usually iconv cannot correctly * handle it. So we have to translate NFD characters to NFC ones * ourselves before iconv handles. Another reason is to prevent * that the same sight of two filenames, one is NFC and other * is NFD, would be in its directory. * On Mac OS X, although its filesystem layer automatically * convert filenames to NFD, it would be useful for filename * comparing to find out the same filenames that we normalize * that to be NFD ourselves. */ if ((flag & SCONV_FROM_CHARSET) && (flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8))) { #if defined(__APPLE__) if (flag & SCONV_TO_UTF8) flag |= SCONV_NORMALIZATION_D; else #endif flag |= SCONV_NORMALIZATION_C; } #if defined(__APPLE__) /* * In case writing an archive file, make sure that a filename * going to be passed to iconv is a Unicode NFC string since * a filename in HFS Plus filesystem is a Unicode NFD one and * iconv cannot handle it with "UTF-8" charset. It is simpler * than a use of "UTF-8-MAC" charset. */ if ((flag & SCONV_TO_CHARSET) && (flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8)) && !(flag & (SCONV_TO_UTF16 | SCONV_TO_UTF8))) flag |= SCONV_NORMALIZATION_C; /* * In case reading an archive file. make sure that a filename * will be passed to users is a Unicode NFD string in order to * correctly compare the filename with other one which comes * from HFS Plus filesystem. */ if ((flag & SCONV_FROM_CHARSET) && !(flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8)) && (flag & SCONV_TO_UTF8)) flag |= SCONV_NORMALIZATION_D; #endif #if defined(HAVE_ICONV) sc->cd_w = (iconv_t)-1; /* * Create an iconv object. */ if (((flag & (SCONV_TO_UTF8 | SCONV_TO_UTF16)) && (flag & (SCONV_FROM_UTF8 | SCONV_FROM_UTF16))) || (flag & SCONV_WIN_CP)) { /* This case we won't use iconv. */ sc->cd = (iconv_t)-1; } else { sc->cd = iconv_open(tc, fc); if (sc->cd == (iconv_t)-1 && (sc->flag & SCONV_BEST_EFFORT)) { /* * Unfortunately, all of iconv implements do support * "CP932" character-set, so we should use "SJIS" * instead if iconv_open failed. */ if (strcmp(tc, "CP932") == 0) sc->cd = iconv_open("SJIS", fc); else if (strcmp(fc, "CP932") == 0) sc->cd = iconv_open(tc, "SJIS"); } #if defined(_WIN32) && !defined(__CYGWIN__) /* * archive_mstring on Windows directly convert multi-bytes * into archive_wstring in order not to depend on locale * so that you can do a I18N programming. This will be * used only in archive_mstring_copy_mbs_len_l so far. */ if (flag & SCONV_FROM_CHARSET) { sc->cd_w = iconv_open("UTF-8", fc); if (sc->cd_w == (iconv_t)-1 && (sc->flag & SCONV_BEST_EFFORT)) { if (strcmp(fc, "CP932") == 0) sc->cd_w = iconv_open("UTF-8", "SJIS"); } } #endif /* _WIN32 && !__CYGWIN__ */ } #endif /* HAVE_ICONV */ sc->flag = flag; /* * Set up converters. */ setup_converter(sc); return (sc); } /* * Free a string conversion object. */ static void free_sconv_object(struct archive_string_conv *sc) { free(sc->from_charset); free(sc->to_charset); archive_string_free(&sc->utftmp); #if HAVE_ICONV if (sc->cd != (iconv_t)-1) iconv_close(sc->cd); if (sc->cd_w != (iconv_t)-1) iconv_close(sc->cd_w); #endif free(sc); } #if defined(_WIN32) && !defined(__CYGWIN__) # if defined(WINAPI_FAMILY_PARTITION) && !WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP) # define GetOEMCP() CP_OEMCP # endif static unsigned my_atoi(const char *p) { unsigned cp; cp = 0; while (*p) { if (*p >= '0' && *p <= '9') cp = cp * 10 + (*p - '0'); else return (-1); p++; } return (cp); } /* * Translate Charset name (as used by iconv) into CodePage (as used by Windows) * Return -1 if failed. * * Note: This translation code may be insufficient. */ static struct charset { const char *name; unsigned cp; } charsets[] = { /* MUST BE SORTED! */ {"ASCII", 1252}, {"ASMO-708", 708}, {"BIG5", 950}, {"CHINESE", 936}, {"CP367", 1252}, {"CP819", 1252}, {"CP1025", 21025}, {"DOS-720", 720}, {"DOS-862", 862}, {"EUC-CN", 51936}, {"EUC-JP", 51932}, {"EUC-KR", 949}, {"EUCCN", 51936}, {"EUCJP", 51932}, {"EUCKR", 949}, {"GB18030", 54936}, {"GB2312", 936}, {"HEBREW", 1255}, {"HZ-GB-2312", 52936}, {"IBM273", 20273}, {"IBM277", 20277}, {"IBM278", 20278}, {"IBM280", 20280}, {"IBM284", 20284}, {"IBM285", 20285}, {"IBM290", 20290}, {"IBM297", 20297}, {"IBM367", 1252}, {"IBM420", 20420}, {"IBM423", 20423}, {"IBM424", 20424}, {"IBM819", 1252}, {"IBM871", 20871}, {"IBM880", 20880}, {"IBM905", 20905}, {"IBM924", 20924}, {"ISO-8859-1", 28591}, {"ISO-8859-13", 28603}, {"ISO-8859-15", 28605}, {"ISO-8859-2", 28592}, {"ISO-8859-3", 28593}, {"ISO-8859-4", 28594}, {"ISO-8859-5", 28595}, {"ISO-8859-6", 28596}, {"ISO-8859-7", 28597}, {"ISO-8859-8", 28598}, {"ISO-8859-9", 28599}, {"ISO8859-1", 28591}, {"ISO8859-13", 28603}, {"ISO8859-15", 28605}, {"ISO8859-2", 28592}, {"ISO8859-3", 28593}, {"ISO8859-4", 28594}, {"ISO8859-5", 28595}, {"ISO8859-6", 28596}, {"ISO8859-7", 28597}, {"ISO8859-8", 28598}, {"ISO8859-9", 28599}, {"JOHAB", 1361}, {"KOI8-R", 20866}, {"KOI8-U", 21866}, {"KS_C_5601-1987", 949}, {"LATIN1", 1252}, {"LATIN2", 28592}, {"MACINTOSH", 10000}, {"SHIFT-JIS", 932}, {"SHIFT_JIS", 932}, {"SJIS", 932}, {"US", 1252}, {"US-ASCII", 1252}, {"UTF-16", 1200}, {"UTF-16BE", 1201}, {"UTF-16LE", 1200}, {"UTF-8", CP_UTF8}, {"X-EUROPA", 29001}, {"X-MAC-ARABIC", 10004}, {"X-MAC-CE", 10029}, {"X-MAC-CHINESEIMP", 10008}, {"X-MAC-CHINESETRAD", 10002}, {"X-MAC-CROATIAN", 10082}, {"X-MAC-CYRILLIC", 10007}, {"X-MAC-GREEK", 10006}, {"X-MAC-HEBREW", 10005}, {"X-MAC-ICELANDIC", 10079}, {"X-MAC-JAPANESE", 10001}, {"X-MAC-KOREAN", 10003}, {"X-MAC-ROMANIAN", 10010}, {"X-MAC-THAI", 10021}, {"X-MAC-TURKISH", 10081}, {"X-MAC-UKRAINIAN", 10017}, }; static unsigned make_codepage_from_charset(const char *charset) { char cs[16]; char *p; unsigned cp; int a, b; if (charset == NULL || strlen(charset) > 15) return -1; /* Copy name to uppercase. */ p = cs; while (*charset) { char c = *charset++; if (c >= 'a' && c <= 'z') c -= 'a' - 'A'; *p++ = c; } *p++ = '\0'; cp = -1; /* Look it up in the table first, so that we can easily * override CP367, which we map to 1252 instead of 367. */ a = 0; b = sizeof(charsets)/sizeof(charsets[0]); while (b > a) { int c = (b + a) / 2; int r = strcmp(charsets[c].name, cs); if (r < 0) a = c + 1; else if (r > 0) b = c; else return charsets[c].cp; } /* If it's not in the table, try to parse it. */ switch (*cs) { case 'C': if (cs[1] == 'P' && cs[2] >= '0' && cs[2] <= '9') { cp = my_atoi(cs + 2); } else if (strcmp(cs, "CP_ACP") == 0) cp = get_current_codepage(); else if (strcmp(cs, "CP_OEMCP") == 0) cp = get_current_oemcp(); break; case 'I': if (cs[1] == 'B' && cs[2] == 'M' && cs[3] >= '0' && cs[3] <= '9') { cp = my_atoi(cs + 3); } break; case 'W': if (strncmp(cs, "WINDOWS-", 8) == 0) { cp = my_atoi(cs + 8); if (cp != 874 && (cp < 1250 || cp > 1258)) cp = -1;/* This may invalid code. */ } break; } return (cp); } /* * Return ANSI Code Page of current locale set by setlocale(). */ static unsigned get_current_codepage(void) { char *locale, *p; unsigned cp; locale = setlocale(LC_CTYPE, NULL); if (locale == NULL) return (GetACP()); if (locale[0] == 'C' && locale[1] == '\0') return (CP_C_LOCALE); p = strrchr(locale, '.'); if (p == NULL) return (GetACP()); if ((strcmp(p+1, "utf8") == 0) || (strcmp(p+1, "UTF-8") == 0)) return CP_UTF8; cp = my_atoi(p+1); if ((int)cp <= 0) return (GetACP()); return (cp); } /* * Translation table between Locale Name and ACP/OEMCP. */ static struct { unsigned acp; unsigned ocp; const char *locale; } acp_ocp_map[] = { { 950, 950, "Chinese_Taiwan" }, { 936, 936, "Chinese_People's Republic of China" }, { 950, 950, "Chinese_Taiwan" }, { 1250, 852, "Czech_Czech Republic" }, { 1252, 850, "Danish_Denmark" }, { 1252, 850, "Dutch_Netherlands" }, { 1252, 850, "Dutch_Belgium" }, { 1252, 437, "English_United States" }, { 1252, 850, "English_Australia" }, { 1252, 850, "English_Canada" }, { 1252, 850, "English_New Zealand" }, { 1252, 850, "English_United Kingdom" }, { 1252, 437, "English_United States" }, { 1252, 850, "Finnish_Finland" }, { 1252, 850, "French_France" }, { 1252, 850, "French_Belgium" }, { 1252, 850, "French_Canada" }, { 1252, 850, "French_Switzerland" }, { 1252, 850, "German_Germany" }, { 1252, 850, "German_Austria" }, { 1252, 850, "German_Switzerland" }, { 1253, 737, "Greek_Greece" }, { 1250, 852, "Hungarian_Hungary" }, { 1252, 850, "Icelandic_Iceland" }, { 1252, 850, "Italian_Italy" }, { 1252, 850, "Italian_Switzerland" }, { 932, 932, "Japanese_Japan" }, { 949, 949, "Korean_Korea" }, { 1252, 850, "Norwegian (BokmOl)_Norway" }, { 1252, 850, "Norwegian (BokmOl)_Norway" }, { 1252, 850, "Norwegian-Nynorsk_Norway" }, { 1250, 852, "Polish_Poland" }, { 1252, 850, "Portuguese_Portugal" }, { 1252, 850, "Portuguese_Brazil" }, { 1251, 866, "Russian_Russia" }, { 1250, 852, "Slovak_Slovakia" }, { 1252, 850, "Spanish_Spain" }, { 1252, 850, "Spanish_Mexico" }, { 1252, 850, "Spanish_Spain" }, { 1252, 850, "Swedish_Sweden" }, { 1254, 857, "Turkish_Turkey" }, { 0, 0, NULL} }; /* * Return OEM Code Page of current locale set by setlocale(). */ static unsigned get_current_oemcp(void) { int i; char *locale, *p; size_t len; locale = setlocale(LC_CTYPE, NULL); if (locale == NULL) return (GetOEMCP()); if (locale[0] == 'C' && locale[1] == '\0') return (CP_C_LOCALE); p = strrchr(locale, '.'); if (p == NULL) return (GetOEMCP()); len = p - locale; for (i = 0; acp_ocp_map[i].acp; i++) { if (strncmp(acp_ocp_map[i].locale, locale, len) == 0) return (acp_ocp_map[i].ocp); } return (GetOEMCP()); } #else /* * POSIX platform does not use CodePage. */ static unsigned get_current_codepage(void) { return (-1);/* Unknown */ } static unsigned make_codepage_from_charset(const char *charset) { (void)charset; /* UNUSED */ return (-1);/* Unknown */ } static unsigned get_current_oemcp(void) { return (-1);/* Unknown */ } #endif /* defined(_WIN32) && !defined(__CYGWIN__) */ /* * Return a string conversion object. */ static struct archive_string_conv * get_sconv_object(struct archive *a, const char *fc, const char *tc, int flag) { struct archive_string_conv *sc; unsigned current_codepage; /* Check if we have made the sconv object. */ sc = find_sconv_object(a, fc, tc); if (sc != NULL) return (sc); if (a == NULL) current_codepage = get_current_codepage(); else current_codepage = a->current_codepage; sc = create_sconv_object(canonical_charset_name(fc), canonical_charset_name(tc), current_codepage, flag); if (sc == NULL) { if (a != NULL) archive_set_error(a, ENOMEM, "Could not allocate memory for " "a string conversion object"); return (NULL); } /* * If there is no converter for current string conversion object, * we cannot handle this conversion. */ if (sc->nconverter == 0) { if (a != NULL) { #if HAVE_ICONV archive_set_error(a, ARCHIVE_ERRNO_MISC, "iconv_open failed : Cannot handle ``%s''", (flag & SCONV_TO_CHARSET)?tc:fc); #else archive_set_error(a, ARCHIVE_ERRNO_MISC, "A character-set conversion not fully supported " "on this platform"); #endif } /* Failed; free a sconv object. */ free_sconv_object(sc); return (NULL); } /* * Success! */ if (a != NULL) add_sconv_object(a, sc); return (sc); } static const char * get_current_charset(struct archive *a) { const char *cur_charset; if (a == NULL) cur_charset = default_iconv_charset(""); else { cur_charset = default_iconv_charset(a->current_code); if (a->current_code == NULL) { a->current_code = strdup(cur_charset); a->current_codepage = get_current_codepage(); a->current_oemcp = get_current_oemcp(); } } return (cur_charset); } /* * Make and Return a string conversion object. * Return NULL if the platform does not support the specified conversion * and best_effort is 0. * If best_effort is set, A string conversion object must be returned * unless memory allocation for the object fails, but the conversion * might fail when non-ASCII code is found. */ struct archive_string_conv * archive_string_conversion_to_charset(struct archive *a, const char *charset, int best_effort) { int flag = SCONV_TO_CHARSET; if (best_effort) flag |= SCONV_BEST_EFFORT; return (get_sconv_object(a, get_current_charset(a), charset, flag)); } struct archive_string_conv * archive_string_conversion_from_charset(struct archive *a, const char *charset, int best_effort) { int flag = SCONV_FROM_CHARSET; if (best_effort) flag |= SCONV_BEST_EFFORT; return (get_sconv_object(a, charset, get_current_charset(a), flag)); } /* * archive_string_default_conversion_*_archive() are provided for Windows * platform because other archiver application use CP_OEMCP for * MultiByteToWideChar() and WideCharToMultiByte() for the filenames * in tar or zip files. But mbstowcs/wcstombs(CRT) usually use CP_ACP * unless you use setlocale(LC_ALL, ".OCP")(specify CP_OEMCP). * So we should make a string conversion between CP_ACP and CP_OEMCP * for compatibility. */ #if defined(_WIN32) && !defined(__CYGWIN__) struct archive_string_conv * archive_string_default_conversion_for_read(struct archive *a) { const char *cur_charset = get_current_charset(a); char oemcp[16]; /* NOTE: a check of cur_charset is unneeded but we need * that get_current_charset() has been surely called at * this time whatever C compiler optimized. */ if (cur_charset != NULL && (a->current_codepage == CP_C_LOCALE || a->current_codepage == a->current_oemcp)) return (NULL);/* no conversion. */ _snprintf(oemcp, sizeof(oemcp)-1, "CP%d", a->current_oemcp); /* Make sure a null termination must be set. */ oemcp[sizeof(oemcp)-1] = '\0'; return (get_sconv_object(a, oemcp, cur_charset, SCONV_FROM_CHARSET)); } struct archive_string_conv * archive_string_default_conversion_for_write(struct archive *a) { const char *cur_charset = get_current_charset(a); char oemcp[16]; /* NOTE: a check of cur_charset is unneeded but we need * that get_current_charset() has been surely called at * this time whatever C compiler optimized. */ if (cur_charset != NULL && (a->current_codepage == CP_C_LOCALE || a->current_codepage == a->current_oemcp)) return (NULL);/* no conversion. */ _snprintf(oemcp, sizeof(oemcp)-1, "CP%d", a->current_oemcp); /* Make sure a null termination must be set. */ oemcp[sizeof(oemcp)-1] = '\0'; return (get_sconv_object(a, cur_charset, oemcp, SCONV_TO_CHARSET)); } #else struct archive_string_conv * archive_string_default_conversion_for_read(struct archive *a) { (void)a; /* UNUSED */ return (NULL); } struct archive_string_conv * archive_string_default_conversion_for_write(struct archive *a) { (void)a; /* UNUSED */ return (NULL); } #endif /* * Dispose of all character conversion objects in the archive object. */ void archive_string_conversion_free(struct archive *a) { struct archive_string_conv *sc; struct archive_string_conv *sc_next; for (sc = a->sconv; sc != NULL; sc = sc_next) { sc_next = sc->next; free_sconv_object(sc); } a->sconv = NULL; free(a->current_code); a->current_code = NULL; } /* * Return a conversion charset name. */ const char * archive_string_conversion_charset_name(struct archive_string_conv *sc) { if (sc->flag & SCONV_TO_CHARSET) return (sc->to_charset); else return (sc->from_charset); } /* * Change the behavior of a string conversion. */ void archive_string_conversion_set_opt(struct archive_string_conv *sc, int opt) { switch (opt) { /* * A filename in UTF-8 was made with libarchive 2.x in a wrong * assumption that wchar_t was Unicode. * This option enables simulating the assumption in order to read * that filename correctly. */ case SCONV_SET_OPT_UTF8_LIBARCHIVE2X: #if (defined(_WIN32) && !defined(__CYGWIN__)) \ || defined(__STDC_ISO_10646__) || defined(__APPLE__) /* * Nothing to do for it since wchar_t on these platforms * is really Unicode. */ (void)sc; /* UNUSED */ #else if ((sc->flag & SCONV_UTF8_LIBARCHIVE_2) == 0) { sc->flag |= SCONV_UTF8_LIBARCHIVE_2; /* Set up string converters. */ setup_converter(sc); } #endif break; case SCONV_SET_OPT_NORMALIZATION_C: if ((sc->flag & SCONV_NORMALIZATION_C) == 0) { sc->flag |= SCONV_NORMALIZATION_C; sc->flag &= ~SCONV_NORMALIZATION_D; /* Set up string converters. */ setup_converter(sc); } break; case SCONV_SET_OPT_NORMALIZATION_D: #if defined(HAVE_ICONV) /* * If iconv will take the string, do not change the * setting of the normalization. */ if (!(sc->flag & SCONV_WIN_CP) && (sc->flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8)) && !(sc->flag & (SCONV_TO_UTF16 | SCONV_TO_UTF8))) break; #endif if ((sc->flag & SCONV_NORMALIZATION_D) == 0) { sc->flag |= SCONV_NORMALIZATION_D; sc->flag &= ~SCONV_NORMALIZATION_C; /* Set up string converters. */ setup_converter(sc); } break; default: break; } } /* * * Copy one archive_string to another in locale conversion. * * archive_strncat_l(); * archive_strncpy_l(); * */ static size_t mbsnbytes(const void *_p, size_t n) { size_t s; const char *p, *pp; if (_p == NULL) return (0); p = (const char *)_p; /* Like strlen(p), except won't examine positions beyond p[n]. */ s = 0; pp = p; while (s < n && *pp) { pp++; s++; } return (s); } static size_t utf16nbytes(const void *_p, size_t n) { size_t s; const char *p, *pp; if (_p == NULL) return (0); p = (const char *)_p; /* Like strlen(p), except won't examine positions beyond p[n]. */ s = 0; pp = p; n >>= 1; while (s < n && (pp[0] || pp[1])) { pp += 2; s++; } return (s<<1); } int archive_strncpy_l(struct archive_string *as, const void *_p, size_t n, struct archive_string_conv *sc) { as->length = 0; return (archive_strncat_l(as, _p, n, sc)); } int archive_strncat_l(struct archive_string *as, const void *_p, size_t n, struct archive_string_conv *sc) { const void *s; size_t length = 0; int i, r = 0, r2; if (_p != NULL && n > 0) { if (sc != NULL && (sc->flag & SCONV_FROM_UTF16)) length = utf16nbytes(_p, n); else length = mbsnbytes(_p, n); } /* We must allocate memory even if there is no data for conversion * or copy. This simulates archive_string_append behavior. */ if (length == 0) { - int tn = 1; + size_t tn = 1; if (sc != NULL && (sc->flag & SCONV_TO_UTF16)) tn = 2; if (archive_string_ensure(as, as->length + tn) == NULL) return (-1); as->s[as->length] = 0; if (tn == 2) as->s[as->length+1] = 0; return (0); } /* * If sc is NULL, we just make a copy. */ if (sc == NULL) { if (archive_string_append(as, _p, length) == NULL) return (-1);/* No memory */ return (0); } s = _p; i = 0; if (sc->nconverter > 1) { sc->utftmp.length = 0; r2 = sc->converter[0](&(sc->utftmp), s, length, sc); if (r2 != 0 && errno == ENOMEM) return (r2); if (r > r2) r = r2; s = sc->utftmp.s; length = sc->utftmp.length; ++i; } r2 = sc->converter[i](as, s, length, sc); if (r > r2) r = r2; return (r); } #if HAVE_ICONV /* * Return -1 if conversion fails. */ static int iconv_strncat_in_locale(struct archive_string *as, const void *_p, size_t length, struct archive_string_conv *sc) { ICONV_CONST char *itp; size_t remaining; iconv_t cd; char *outp; size_t avail, bs; int return_value = 0; /* success */ size_t to_size, from_size; if (sc->flag & SCONV_TO_UTF16) to_size = 2; else to_size = 1; if (sc->flag & SCONV_FROM_UTF16) from_size = 2; else from_size = 1; if (archive_string_ensure(as, as->length + length*2+to_size) == NULL) return (-1); cd = sc->cd; itp = (char *)(uintptr_t)_p; remaining = length; outp = as->s + as->length; avail = as->buffer_length - as->length - to_size; while (remaining >= from_size) { size_t result = iconv(cd, &itp, &remaining, &outp, &avail); if (result != (size_t)-1) break; /* Conversion completed. */ if (errno == EILSEQ || errno == EINVAL) { /* * If an output charset is UTF-8 or UTF-16BE/LE, * unknown character should be U+FFFD * (replacement character). */ if (sc->flag & (SCONV_TO_UTF8 | SCONV_TO_UTF16)) { size_t rbytes; if (sc->flag & SCONV_TO_UTF8) rbytes = sizeof(utf8_replacement_char); else rbytes = 2; if (avail < rbytes) { as->length = outp - as->s; bs = as->buffer_length + (remaining * to_size) + rbytes; if (NULL == archive_string_ensure(as, bs)) return (-1); outp = as->s + as->length; avail = as->buffer_length - as->length - to_size; } if (sc->flag & SCONV_TO_UTF8) memcpy(outp, utf8_replacement_char, sizeof(utf8_replacement_char)); else if (sc->flag & SCONV_TO_UTF16BE) archive_be16enc(outp, UNICODE_R_CHAR); else archive_le16enc(outp, UNICODE_R_CHAR); outp += rbytes; avail -= rbytes; } else { /* Skip the illegal input bytes. */ *outp++ = '?'; avail--; } itp += from_size; remaining -= from_size; return_value = -1; /* failure */ } else { /* E2BIG no output buffer, * Increase an output buffer. */ as->length = outp - as->s; bs = as->buffer_length + remaining * 2; if (NULL == archive_string_ensure(as, bs)) return (-1); outp = as->s + as->length; avail = as->buffer_length - as->length - to_size; } } as->length = outp - as->s; as->s[as->length] = 0; if (to_size == 2) as->s[as->length+1] = 0; return (return_value); } #endif /* HAVE_ICONV */ #if defined(_WIN32) && !defined(__CYGWIN__) /* * Translate a string from a some CodePage to an another CodePage by * Windows APIs, and copy the result. Return -1 if conversion fails. */ static int strncat_in_codepage(struct archive_string *as, const void *_p, size_t length, struct archive_string_conv *sc) { const char *s = (const char *)_p; struct archive_wstring aws; size_t l; int r, saved_flag; archive_string_init(&aws); saved_flag = sc->flag; sc->flag &= ~(SCONV_NORMALIZATION_D | SCONV_NORMALIZATION_C); r = archive_wstring_append_from_mbs_in_codepage(&aws, s, length, sc); sc->flag = saved_flag; if (r != 0) { archive_wstring_free(&aws); if (errno != ENOMEM) archive_string_append(as, s, length); return (-1); } l = as->length; r = archive_string_append_from_wcs_in_codepage( as, aws.s, aws.length, sc); if (r != 0 && errno != ENOMEM && l == as->length) archive_string_append(as, s, length); archive_wstring_free(&aws); return (r); } /* * Test whether MBS ==> WCS is okay. */ static int invalid_mbs(const void *_p, size_t n, struct archive_string_conv *sc) { const char *p = (const char *)_p; unsigned codepage; DWORD mbflag = MB_ERR_INVALID_CHARS; if (sc->flag & SCONV_FROM_CHARSET) codepage = sc->to_cp; else codepage = sc->from_cp; if (codepage == CP_C_LOCALE) return (0); if (codepage != CP_UTF8) mbflag |= MB_PRECOMPOSED; if (n > (size_t)INT_MAX) return (-1); /* Invalid */ if (MultiByteToWideChar(codepage, mbflag, p, (int)n, NULL, 0) == 0) return (-1); /* Invalid */ return (0); /* Okay */ } #else /* * Test whether MBS ==> WCS is okay. */ static int invalid_mbs(const void *_p, size_t n, struct archive_string_conv *sc) { const char *p = (const char *)_p; size_t r; #if HAVE_MBRTOWC mbstate_t shift_state; memset(&shift_state, 0, sizeof(shift_state)); #else /* Clear the shift state before starting. */ mbtowc(NULL, NULL, 0); #endif while (n) { wchar_t wc; #if HAVE_MBRTOWC r = mbrtowc(&wc, p, n, &shift_state); #else r = mbtowc(&wc, p, n); #endif if (r == (size_t)-1 || r == (size_t)-2) return (-1);/* Invalid. */ if (r == 0) break; p += r; n -= r; } (void)sc; /* UNUSED */ return (0); /* All Okey. */ } #endif /* defined(_WIN32) && !defined(__CYGWIN__) */ /* * Basically returns -1 because we cannot make a conversion of charset * without iconv but in some cases this would return 0. * Returns 0 if all copied characters are ASCII. * Returns 0 if both from-locale and to-locale are the same and those * can be WCS with no error. */ static int best_effort_strncat_in_locale(struct archive_string *as, const void *_p, size_t length, struct archive_string_conv *sc) { size_t remaining; const uint8_t *itp; int return_value = 0; /* success */ /* * If both from-locale and to-locale is the same, this makes a copy. * And then this checks all copied MBS can be WCS if so returns 0. */ if (sc->same) { if (archive_string_append(as, _p, length) == NULL) return (-1);/* No memory */ return (invalid_mbs(_p, length, sc)); } /* * If a character is ASCII, this just copies it. If not, this * assigns '?' character instead but in UTF-8 locale this assigns * byte sequence 0xEF 0xBD 0xBD, which are code point U+FFFD, * a Replacement Character in Unicode. */ remaining = length; itp = (const uint8_t *)_p; while (*itp && remaining > 0) { if (*itp > 127) { // Non-ASCII: Substitute with suitable replacement if (sc->flag & SCONV_TO_UTF8) { if (archive_string_append(as, utf8_replacement_char, sizeof(utf8_replacement_char)) == NULL) { __archive_errx(1, "Out of memory"); } } else { archive_strappend_char(as, '?'); } return_value = -1; } else { archive_strappend_char(as, *itp); } ++itp; } return (return_value); } /* * Unicode conversion functions. * - UTF-8 <===> UTF-8 in removing surrogate pairs. * - UTF-8 NFD ===> UTF-8 NFC in removing surrogate pairs. * - UTF-8 made by libarchive 2.x ===> UTF-8. * - UTF-16BE <===> UTF-8. * */ /* * Utility to convert a single UTF-8 sequence. * * Usually return used bytes, return used byte in negative value when * a unicode character is replaced with U+FFFD. * See also http://unicode.org/review/pr-121.html Public Review Issue #121 * Recommended Practice for Replacement Characters. */ static int _utf8_to_unicode(uint32_t *pwc, const char *s, size_t n) { static const char utf8_count[256] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 00 - 0F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 10 - 1F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 20 - 2F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 30 - 3F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 40 - 4F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 50 - 5F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 60 - 6F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 70 - 7F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 80 - 8F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 90 - 9F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* A0 - AF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* B0 - BF */ 0, 0, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* C0 - CF */ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* D0 - DF */ 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,/* E0 - EF */ 4, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 /* F0 - FF */ }; int ch, i; int cnt; uint32_t wc; /* Sanity check. */ if (n == 0) return (0); /* * Decode 1-4 bytes depending on the value of the first byte. */ ch = (unsigned char)*s; if (ch == 0) return (0); /* Standard: return 0 for end-of-string. */ cnt = utf8_count[ch]; /* Invalid sequence or there are not plenty bytes. */ if (n < (size_t)cnt) { cnt = (int)n; for (i = 1; i < cnt; i++) { if ((s[i] & 0xc0) != 0x80) { cnt = i; break; } } goto invalid_sequence; } /* Make a Unicode code point from a single UTF-8 sequence. */ switch (cnt) { case 1: /* 1 byte sequence. */ *pwc = ch & 0x7f; return (cnt); case 2: /* 2 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) { cnt = 1; goto invalid_sequence; } *pwc = ((ch & 0x1f) << 6) | (s[1] & 0x3f); return (cnt); case 3: /* 3 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) { cnt = 1; goto invalid_sequence; } if ((s[2] & 0xc0) != 0x80) { cnt = 2; goto invalid_sequence; } wc = ((ch & 0x0f) << 12) | ((s[1] & 0x3f) << 6) | (s[2] & 0x3f); if (wc < 0x800) goto invalid_sequence;/* Overlong sequence. */ break; case 4: /* 4 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) { cnt = 1; goto invalid_sequence; } if ((s[2] & 0xc0) != 0x80) { cnt = 2; goto invalid_sequence; } if ((s[3] & 0xc0) != 0x80) { cnt = 3; goto invalid_sequence; } wc = ((ch & 0x07) << 18) | ((s[1] & 0x3f) << 12) | ((s[2] & 0x3f) << 6) | (s[3] & 0x3f); if (wc < 0x10000) goto invalid_sequence;/* Overlong sequence. */ break; default: /* Others are all invalid sequence. */ if (ch == 0xc0 || ch == 0xc1) cnt = 2; else if (ch >= 0xf5 && ch <= 0xf7) cnt = 4; else if (ch >= 0xf8 && ch <= 0xfb) cnt = 5; else if (ch == 0xfc || ch == 0xfd) cnt = 6; else cnt = 1; if (n < (size_t)cnt) cnt = (int)n; for (i = 1; i < cnt; i++) { if ((s[i] & 0xc0) != 0x80) { cnt = i; break; } } goto invalid_sequence; } /* The code point larger than 0x10FFFF is not legal * Unicode values. */ if (wc > UNICODE_MAX) goto invalid_sequence; /* Correctly gets a Unicode, returns used bytes. */ *pwc = wc; return (cnt); invalid_sequence: *pwc = UNICODE_R_CHAR;/* set the Replacement Character instead. */ return (cnt * -1); } static int utf8_to_unicode(uint32_t *pwc, const char *s, size_t n) { int cnt; cnt = _utf8_to_unicode(pwc, s, n); /* Any of Surrogate pair is not legal Unicode values. */ if (cnt == 3 && IS_SURROGATE_PAIR_LA(*pwc)) return (-3); return (cnt); } static inline uint32_t combine_surrogate_pair(uint32_t uc, uint32_t uc2) { uc -= 0xD800; uc *= 0x400; uc += uc2 - 0xDC00; uc += 0x10000; return (uc); } /* * Convert a single UTF-8/CESU-8 sequence to a Unicode code point in * removing surrogate pairs. * * CESU-8: The Compatibility Encoding Scheme for UTF-16. * * Usually return used bytes, return used byte in negative value when * a unicode character is replaced with U+FFFD. */ static int cesu8_to_unicode(uint32_t *pwc, const char *s, size_t n) { uint32_t wc = 0; int cnt; cnt = _utf8_to_unicode(&wc, s, n); if (cnt == 3 && IS_HIGH_SURROGATE_LA(wc)) { uint32_t wc2 = 0; if (n - 3 < 3) { /* Invalid byte sequence. */ goto invalid_sequence; } cnt = _utf8_to_unicode(&wc2, s+3, n-3); if (cnt != 3 || !IS_LOW_SURROGATE_LA(wc2)) { /* Invalid byte sequence. */ goto invalid_sequence; } wc = combine_surrogate_pair(wc, wc2); cnt = 6; } else if (cnt == 3 && IS_LOW_SURROGATE_LA(wc)) { /* Invalid byte sequence. */ goto invalid_sequence; } *pwc = wc; return (cnt); invalid_sequence: *pwc = UNICODE_R_CHAR;/* set the Replacement Character instead. */ if (cnt > 0) cnt *= -1; return (cnt); } /* * Convert a Unicode code point to a single UTF-8 sequence. * * NOTE:This function does not check if the Unicode is legal or not. * Please you definitely check it before calling this. */ static size_t unicode_to_utf8(char *p, size_t remaining, uint32_t uc) { char *_p = p; /* Invalid Unicode char maps to Replacement character */ if (uc > UNICODE_MAX) uc = UNICODE_R_CHAR; /* Translate code point to UTF8 */ if (uc <= 0x7f) { if (remaining == 0) return (0); *p++ = (char)uc; } else if (uc <= 0x7ff) { if (remaining < 2) return (0); *p++ = 0xc0 | ((uc >> 6) & 0x1f); *p++ = 0x80 | (uc & 0x3f); } else if (uc <= 0xffff) { if (remaining < 3) return (0); *p++ = 0xe0 | ((uc >> 12) & 0x0f); *p++ = 0x80 | ((uc >> 6) & 0x3f); *p++ = 0x80 | (uc & 0x3f); } else { if (remaining < 4) return (0); *p++ = 0xf0 | ((uc >> 18) & 0x07); *p++ = 0x80 | ((uc >> 12) & 0x3f); *p++ = 0x80 | ((uc >> 6) & 0x3f); *p++ = 0x80 | (uc & 0x3f); } return (p - _p); } static int utf16be_to_unicode(uint32_t *pwc, const char *s, size_t n) { return (utf16_to_unicode(pwc, s, n, 1)); } static int utf16le_to_unicode(uint32_t *pwc, const char *s, size_t n) { return (utf16_to_unicode(pwc, s, n, 0)); } static int utf16_to_unicode(uint32_t *pwc, const char *s, size_t n, int be) { const char *utf16 = s; unsigned uc; if (n == 0) return (0); if (n == 1) { /* set the Replacement Character instead. */ *pwc = UNICODE_R_CHAR; return (-1); } if (be) uc = archive_be16dec(utf16); else uc = archive_le16dec(utf16); utf16 += 2; /* If this is a surrogate pair, assemble the full code point.*/ if (IS_HIGH_SURROGATE_LA(uc)) { unsigned uc2; if (n >= 4) { if (be) uc2 = archive_be16dec(utf16); else uc2 = archive_le16dec(utf16); } else uc2 = 0; if (IS_LOW_SURROGATE_LA(uc2)) { uc = combine_surrogate_pair(uc, uc2); utf16 += 2; } else { /* Undescribed code point should be U+FFFD * (replacement character). */ *pwc = UNICODE_R_CHAR; return (-2); } } /* * Surrogate pair values(0xd800 through 0xdfff) are only * used by UTF-16, so, after above calculation, the code * must not be surrogate values, and Unicode has no codes * larger than 0x10ffff. Thus, those are not legal Unicode * values. */ if (IS_SURROGATE_PAIR_LA(uc) || uc > UNICODE_MAX) { /* Undescribed code point should be U+FFFD * (replacement character). */ *pwc = UNICODE_R_CHAR; return (((int)(utf16 - s)) * -1); } *pwc = uc; return ((int)(utf16 - s)); } static size_t unicode_to_utf16be(char *p, size_t remaining, uint32_t uc) { char *utf16 = p; if (uc > 0xffff) { /* We have a code point that won't fit into a * wchar_t; convert it to a surrogate pair. */ if (remaining < 4) return (0); uc -= 0x10000; archive_be16enc(utf16, ((uc >> 10) & 0x3ff) + 0xD800); archive_be16enc(utf16+2, (uc & 0x3ff) + 0xDC00); return (4); } else { if (remaining < 2) return (0); archive_be16enc(utf16, (uint16_t)uc); return (2); } } static size_t unicode_to_utf16le(char *p, size_t remaining, uint32_t uc) { char *utf16 = p; if (uc > 0xffff) { /* We have a code point that won't fit into a * wchar_t; convert it to a surrogate pair. */ if (remaining < 4) return (0); uc -= 0x10000; archive_le16enc(utf16, ((uc >> 10) & 0x3ff) + 0xD800); archive_le16enc(utf16+2, (uc & 0x3ff) + 0xDC00); return (4); } else { if (remaining < 2) return (0); archive_le16enc(utf16, (uint16_t)uc); return (2); } } /* * Append new UTF-8 string to existing UTF-8 string. * Existing string is assumed to already be in proper form; * the new string will have invalid sequences replaced and * surrogate pairs canonicalized. */ static int strncat_from_utf8_to_utf8(struct archive_string *as, const void *_src, size_t len, struct archive_string_conv *sc) { int ret = 0; const char *src = _src; (void)sc; /* UNUSED */ /* Pre-extend the destination */ if (archive_string_ensure(as, as->length + len + 1) == NULL) return (-1); /* Invariant: src points to the first UTF8 byte that hasn't * been copied to the destination `as`. */ for (;;) { int n; uint32_t uc; const char *e = src; /* Skip UTF-8 sequences until we reach end-of-string or * a code point that needs conversion. */ while ((n = utf8_to_unicode(&uc, e, len)) > 0) { e += n; len -= n; } /* Copy the part that doesn't need conversion */ if (e > src) { if (archive_string_append(as, src, e - src) == NULL) return (-1); src = e; } if (n == 0) { /* We reached end-of-string */ return (ret); } else { /* Next code point needs conversion */ char t[4]; size_t w; /* Try decoding a surrogate pair */ if (n == -3 && IS_SURROGATE_PAIR_LA(uc)) { n = cesu8_to_unicode(&uc, src, len); } /* Not a (valid) surrogate, so use a replacement char */ if (n < 0) { ret = -1; /* Return -1 if we used any replacement */ n *= -1; } /* Consume converted code point */ src += n; len -= n; /* Convert and append new UTF-8 sequence. */ w = unicode_to_utf8(t, sizeof(t), uc); if (archive_string_append(as, t, w) == NULL) return (-1); } } } static int archive_string_append_unicode(struct archive_string *as, const void *_p, size_t len, struct archive_string_conv *sc) { const char *s; char *p, *endp; uint32_t uc; size_t w; - int n, ret = 0, ts, tm; + size_t ts, tm; + int n, ret = 0; int (*parse)(uint32_t *, const char *, size_t); size_t (*unparse)(char *, size_t, uint32_t); if (sc->flag & SCONV_TO_UTF16BE) { unparse = unicode_to_utf16be; ts = 2; } else if (sc->flag & SCONV_TO_UTF16LE) { unparse = unicode_to_utf16le; ts = 2; } else if (sc->flag & SCONV_TO_UTF8) { unparse = unicode_to_utf8; ts = 1; } else { /* * This case is going to be converted to another * character-set through iconv. */ if (sc->flag & SCONV_FROM_UTF16BE) { unparse = unicode_to_utf16be; ts = 2; } else if (sc->flag & SCONV_FROM_UTF16LE) { unparse = unicode_to_utf16le; ts = 2; } else { unparse = unicode_to_utf8; ts = 1; } } if (sc->flag & SCONV_FROM_UTF16BE) { parse = utf16be_to_unicode; tm = 1; } else if (sc->flag & SCONV_FROM_UTF16LE) { parse = utf16le_to_unicode; tm = 1; } else { parse = cesu8_to_unicode; tm = ts; } if (archive_string_ensure(as, as->length + len * tm + ts) == NULL) return (-1); s = (const char *)_p; p = as->s + as->length; endp = as->s + as->buffer_length - ts; while ((n = parse(&uc, s, len)) != 0) { if (n < 0) { /* Use a replaced unicode character. */ n *= -1; ret = -1; } s += n; len -= n; while ((w = unparse(p, endp - p, uc)) == 0) { /* There is not enough output buffer so * we have to expand it. */ as->length = p - as->s; if (archive_string_ensure(as, as->buffer_length + len * tm + ts) == NULL) return (-1); p = as->s + as->length; endp = as->s + as->buffer_length - ts; } p += w; } as->length = p - as->s; as->s[as->length] = '\0'; if (ts == 2) as->s[as->length+1] = '\0'; return (ret); } /* * Following Constants for Hangul compositions this information comes from * Unicode Standard Annex #15 http://unicode.org/reports/tr15/ */ #define HC_SBASE 0xAC00 #define HC_LBASE 0x1100 #define HC_VBASE 0x1161 #define HC_TBASE 0x11A7 #define HC_LCOUNT 19 #define HC_VCOUNT 21 #define HC_TCOUNT 28 #define HC_NCOUNT (HC_VCOUNT * HC_TCOUNT) #define HC_SCOUNT (HC_LCOUNT * HC_NCOUNT) static uint32_t get_nfc(uint32_t uc, uint32_t uc2) { int t, b; t = 0; b = sizeof(u_composition_table)/sizeof(u_composition_table[0]) -1; while (b >= t) { int m = (t + b) / 2; if (u_composition_table[m].cp1 < uc) t = m + 1; else if (u_composition_table[m].cp1 > uc) b = m - 1; else if (u_composition_table[m].cp2 < uc2) t = m + 1; else if (u_composition_table[m].cp2 > uc2) b = m - 1; else return (u_composition_table[m].nfc); } return (0); } #define FDC_MAX 10 /* The maximum number of Following Decomposable * Characters. */ /* * Update first code point. */ #define UPDATE_UC(new_uc) do { \ uc = new_uc; \ ucptr = NULL; \ } while (0) /* * Replace first code point with second code point. */ #define REPLACE_UC_WITH_UC2() do { \ uc = uc2; \ ucptr = uc2ptr; \ n = n2; \ } while (0) #define EXPAND_BUFFER() do { \ as->length = p - as->s; \ if (archive_string_ensure(as, \ as->buffer_length + len * tm + ts) == NULL)\ return (-1); \ p = as->s + as->length; \ endp = as->s + as->buffer_length - ts; \ } while (0) #define UNPARSE(p, endp, uc) do { \ while ((w = unparse(p, (endp) - (p), uc)) == 0) {\ EXPAND_BUFFER(); \ } \ p += w; \ } while (0) /* * Write first code point. * If the code point has not be changed from its original code, * this just copies it from its original buffer pointer. * If not, this converts it to UTF-8 byte sequence and copies it. */ #define WRITE_UC() do { \ if (ucptr) { \ if (p + n > endp) \ EXPAND_BUFFER(); \ switch (n) { \ case 4: \ *p++ = *ucptr++; \ /* FALL THROUGH */ \ case 3: \ *p++ = *ucptr++; \ /* FALL THROUGH */ \ case 2: \ *p++ = *ucptr++; \ /* FALL THROUGH */ \ case 1: \ *p++ = *ucptr; \ break; \ } \ ucptr = NULL; \ } else { \ UNPARSE(p, endp, uc); \ } \ } while (0) /* * Collect following decomposable code points. */ #define COLLECT_CPS(start) do { \ int _i; \ for (_i = start; _i < FDC_MAX ; _i++) { \ nx = parse(&ucx[_i], s, len); \ if (nx <= 0) \ break; \ cx = CCC(ucx[_i]); \ if (cl >= cx && cl != 228 && cx != 228)\ break; \ s += nx; \ len -= nx; \ cl = cx; \ ccx[_i] = cx; \ } \ if (_i >= FDC_MAX) { \ ret = -1; \ ucx_size = FDC_MAX; \ } else \ ucx_size = _i; \ } while (0) /* * Normalize UTF-8/UTF-16BE characters to Form C and copy the result. * * TODO: Convert composition exclusions, which are never converted * from NFC,NFD,NFKC and NFKD, to Form C. */ static int archive_string_normalize_C(struct archive_string *as, const void *_p, size_t len, struct archive_string_conv *sc) { const char *s = (const char *)_p; char *p, *endp; uint32_t uc, uc2; size_t w; int always_replace, n, n2, ret = 0, spair, ts, tm; int (*parse)(uint32_t *, const char *, size_t); size_t (*unparse)(char *, size_t, uint32_t); always_replace = 1; ts = 1;/* text size. */ if (sc->flag & SCONV_TO_UTF16BE) { unparse = unicode_to_utf16be; ts = 2; if (sc->flag & SCONV_FROM_UTF16BE) always_replace = 0; } else if (sc->flag & SCONV_TO_UTF16LE) { unparse = unicode_to_utf16le; ts = 2; if (sc->flag & SCONV_FROM_UTF16LE) always_replace = 0; } else if (sc->flag & SCONV_TO_UTF8) { unparse = unicode_to_utf8; if (sc->flag & SCONV_FROM_UTF8) always_replace = 0; } else { /* * This case is going to be converted to another * character-set through iconv. */ always_replace = 0; if (sc->flag & SCONV_FROM_UTF16BE) { unparse = unicode_to_utf16be; ts = 2; } else if (sc->flag & SCONV_FROM_UTF16LE) { unparse = unicode_to_utf16le; ts = 2; } else { unparse = unicode_to_utf8; } } if (sc->flag & SCONV_FROM_UTF16BE) { parse = utf16be_to_unicode; tm = 1; spair = 4;/* surrogate pair size in UTF-16. */ } else if (sc->flag & SCONV_FROM_UTF16LE) { parse = utf16le_to_unicode; tm = 1; spair = 4;/* surrogate pair size in UTF-16. */ } else { parse = cesu8_to_unicode; tm = ts; spair = 6;/* surrogate pair size in UTF-8. */ } if (archive_string_ensure(as, as->length + len * tm + ts) == NULL) return (-1); p = as->s + as->length; endp = as->s + as->buffer_length - ts; while ((n = parse(&uc, s, len)) != 0) { const char *ucptr, *uc2ptr; if (n < 0) { /* Use a replaced unicode character. */ UNPARSE(p, endp, uc); s += n*-1; len -= n*-1; ret = -1; continue; } else if (n == spair || always_replace) /* uc is converted from a surrogate pair. * this should be treated as a changed code. */ ucptr = NULL; else ucptr = s; s += n; len -= n; /* Read second code point. */ while ((n2 = parse(&uc2, s, len)) > 0) { uint32_t ucx[FDC_MAX]; int ccx[FDC_MAX]; int cl, cx, i, nx, ucx_size; int LIndex,SIndex; uint32_t nfc; if (n2 == spair || always_replace) /* uc2 is converted from a surrogate pair. * this should be treated as a changed code. */ uc2ptr = NULL; else uc2ptr = s; s += n2; len -= n2; /* * If current second code point is out of decomposable * code points, finding compositions is unneeded. */ if (!IS_DECOMPOSABLE_BLOCK(uc2)) { WRITE_UC(); REPLACE_UC_WITH_UC2(); continue; } /* * Try to combine current code points. */ /* * We have to combine Hangul characters according to * http://uniicode.org/reports/tr15/#Hangul */ if (0 <= (LIndex = uc - HC_LBASE) && LIndex < HC_LCOUNT) { /* * Hangul Composition. * 1. Two current code points are L and V. */ int VIndex = uc2 - HC_VBASE; if (0 <= VIndex && VIndex < HC_VCOUNT) { /* Make syllable of form LV. */ UPDATE_UC(HC_SBASE + (LIndex * HC_VCOUNT + VIndex) * HC_TCOUNT); } else { WRITE_UC(); REPLACE_UC_WITH_UC2(); } continue; } else if (0 <= (SIndex = uc - HC_SBASE) && SIndex < HC_SCOUNT && (SIndex % HC_TCOUNT) == 0) { /* * Hangul Composition. * 2. Two current code points are LV and T. */ int TIndex = uc2 - HC_TBASE; if (0 < TIndex && TIndex < HC_TCOUNT) { /* Make syllable of form LVT. */ UPDATE_UC(uc + TIndex); } else { WRITE_UC(); REPLACE_UC_WITH_UC2(); } continue; } else if ((nfc = get_nfc(uc, uc2)) != 0) { /* A composition to current code points * is found. */ UPDATE_UC(nfc); continue; } else if ((cl = CCC(uc2)) == 0) { /* Clearly 'uc2' the second code point is not * a decomposable code. */ WRITE_UC(); REPLACE_UC_WITH_UC2(); continue; } /* * Collect following decomposable code points. */ cx = 0; ucx[0] = uc2; ccx[0] = cl; COLLECT_CPS(1); /* * Find a composed code in the collected code points. */ i = 1; while (i < ucx_size) { int j; if ((nfc = get_nfc(uc, ucx[i])) == 0) { i++; continue; } /* * nfc is composed of uc and ucx[i]. */ UPDATE_UC(nfc); /* * Remove ucx[i] by shifting * following code points. */ for (j = i; j+1 < ucx_size; j++) { ucx[j] = ucx[j+1]; ccx[j] = ccx[j+1]; } ucx_size --; /* * Collect following code points blocked * by ucx[i] the removed code point. */ if (ucx_size > 0 && i == ucx_size && nx > 0 && cx == cl) { cl = ccx[ucx_size-1]; COLLECT_CPS(ucx_size); } /* * Restart finding a composed code with * the updated uc from the top of the * collected code points. */ i = 0; } /* * Apparently the current code points are not * decomposed characters or already composed. */ WRITE_UC(); for (i = 0; i < ucx_size; i++) UNPARSE(p, endp, ucx[i]); /* * Flush out remaining canonical combining characters. */ if (nx > 0 && cx == cl && len > 0) { while ((nx = parse(&ucx[0], s, len)) > 0) { cx = CCC(ucx[0]); if (cl > cx) break; s += nx; len -= nx; cl = cx; UNPARSE(p, endp, ucx[0]); } } break; } if (n2 < 0) { WRITE_UC(); /* Use a replaced unicode character. */ UNPARSE(p, endp, uc2); s += n2*-1; len -= n2*-1; ret = -1; continue; } else if (n2 == 0) { WRITE_UC(); break; } } as->length = p - as->s; as->s[as->length] = '\0'; if (ts == 2) as->s[as->length+1] = '\0'; return (ret); } static int get_nfd(uint32_t *cp1, uint32_t *cp2, uint32_t uc) { int t, b; /* * These are not converted to NFD on Mac OS. */ if ((uc >= 0x2000 && uc <= 0x2FFF) || (uc >= 0xF900 && uc <= 0xFAFF) || (uc >= 0x2F800 && uc <= 0x2FAFF)) return (0); /* * Those code points are not converted to NFD on Mac OS. * I do not know the reason because it is undocumented. * NFC NFD * 1109A ==> 11099 110BA * 1109C ==> 1109B 110BA * 110AB ==> 110A5 110BA */ if (uc == 0x1109A || uc == 0x1109C || uc == 0x110AB) return (0); t = 0; b = sizeof(u_decomposition_table)/sizeof(u_decomposition_table[0]) -1; while (b >= t) { int m = (t + b) / 2; if (u_decomposition_table[m].nfc < uc) t = m + 1; else if (u_decomposition_table[m].nfc > uc) b = m - 1; else { *cp1 = u_decomposition_table[m].cp1; *cp2 = u_decomposition_table[m].cp2; return (1); } } return (0); } #define REPLACE_UC_WITH(cp) do { \ uc = cp; \ ucptr = NULL; \ } while (0) /* * Normalize UTF-8 characters to Form D and copy the result. */ static int archive_string_normalize_D(struct archive_string *as, const void *_p, size_t len, struct archive_string_conv *sc) { const char *s = (const char *)_p; char *p, *endp; uint32_t uc, uc2; size_t w; int always_replace, n, n2, ret = 0, spair, ts, tm; int (*parse)(uint32_t *, const char *, size_t); size_t (*unparse)(char *, size_t, uint32_t); always_replace = 1; ts = 1;/* text size. */ if (sc->flag & SCONV_TO_UTF16BE) { unparse = unicode_to_utf16be; ts = 2; if (sc->flag & SCONV_FROM_UTF16BE) always_replace = 0; } else if (sc->flag & SCONV_TO_UTF16LE) { unparse = unicode_to_utf16le; ts = 2; if (sc->flag & SCONV_FROM_UTF16LE) always_replace = 0; } else if (sc->flag & SCONV_TO_UTF8) { unparse = unicode_to_utf8; if (sc->flag & SCONV_FROM_UTF8) always_replace = 0; } else { /* * This case is going to be converted to another * character-set through iconv. */ always_replace = 0; if (sc->flag & SCONV_FROM_UTF16BE) { unparse = unicode_to_utf16be; ts = 2; } else if (sc->flag & SCONV_FROM_UTF16LE) { unparse = unicode_to_utf16le; ts = 2; } else { unparse = unicode_to_utf8; } } if (sc->flag & SCONV_FROM_UTF16BE) { parse = utf16be_to_unicode; tm = 1; spair = 4;/* surrogate pair size in UTF-16. */ } else if (sc->flag & SCONV_FROM_UTF16LE) { parse = utf16le_to_unicode; tm = 1; spair = 4;/* surrogate pair size in UTF-16. */ } else { parse = cesu8_to_unicode; tm = ts; spair = 6;/* surrogate pair size in UTF-8. */ } if (archive_string_ensure(as, as->length + len * tm + ts) == NULL) return (-1); p = as->s + as->length; endp = as->s + as->buffer_length - ts; while ((n = parse(&uc, s, len)) != 0) { const char *ucptr; uint32_t cp1, cp2; int SIndex; struct { uint32_t uc; int ccc; } fdc[FDC_MAX]; int fdi, fdj; int ccc; check_first_code: if (n < 0) { /* Use a replaced unicode character. */ UNPARSE(p, endp, uc); s += n*-1; len -= n*-1; ret = -1; continue; } else if (n == spair || always_replace) /* uc is converted from a surrogate pair. * this should be treated as a changed code. */ ucptr = NULL; else ucptr = s; s += n; len -= n; /* Hangul Decomposition. */ if ((SIndex = uc - HC_SBASE) >= 0 && SIndex < HC_SCOUNT) { int L = HC_LBASE + SIndex / HC_NCOUNT; int V = HC_VBASE + (SIndex % HC_NCOUNT) / HC_TCOUNT; int T = HC_TBASE + SIndex % HC_TCOUNT; REPLACE_UC_WITH(L); WRITE_UC(); REPLACE_UC_WITH(V); WRITE_UC(); if (T != HC_TBASE) { REPLACE_UC_WITH(T); WRITE_UC(); } continue; } if (IS_DECOMPOSABLE_BLOCK(uc) && CCC(uc) != 0) { WRITE_UC(); continue; } fdi = 0; while (get_nfd(&cp1, &cp2, uc) && fdi < FDC_MAX) { int k; for (k = fdi; k > 0; k--) fdc[k] = fdc[k-1]; fdc[0].ccc = CCC(cp2); fdc[0].uc = cp2; fdi++; REPLACE_UC_WITH(cp1); } /* Read following code points. */ while ((n2 = parse(&uc2, s, len)) > 0 && (ccc = CCC(uc2)) != 0 && fdi < FDC_MAX) { int j, k; s += n2; len -= n2; for (j = 0; j < fdi; j++) { if (fdc[j].ccc > ccc) break; } if (j < fdi) { for (k = fdi; k > j; k--) fdc[k] = fdc[k-1]; fdc[j].ccc = ccc; fdc[j].uc = uc2; } else { fdc[fdi].ccc = ccc; fdc[fdi].uc = uc2; } fdi++; } WRITE_UC(); for (fdj = 0; fdj < fdi; fdj++) { REPLACE_UC_WITH(fdc[fdj].uc); WRITE_UC(); } if (n2 == 0) break; REPLACE_UC_WITH(uc2); n = n2; goto check_first_code; } as->length = p - as->s; as->s[as->length] = '\0'; if (ts == 2) as->s[as->length+1] = '\0'; return (ret); } /* * libarchive 2.x made incorrect UTF-8 strings in the wrong assumption * that WCS is Unicode. It is true for several platforms but some are false. * And then people who did not use UTF-8 locale on the non Unicode WCS * platform and made a tar file with libarchive(mostly bsdtar) 2.x. Those * now cannot get right filename from libarchive 3.x and later since we * fixed the wrong assumption and it is incompatible to older its versions. * So we provide special option, "compat-2x.x", for resolving it. * That option enable the string conversion of libarchive 2.x. * * Translates the wrong UTF-8 string made by libarchive 2.x into current * locale character set and appends to the archive_string. * Note: returns -1 if conversion fails. */ static int strncat_from_utf8_libarchive2(struct archive_string *as, const void *_p, size_t len, struct archive_string_conv *sc) { const char *s; int n; char *p; char *end; uint32_t unicode; #if HAVE_WCRTOMB mbstate_t shift_state; memset(&shift_state, 0, sizeof(shift_state)); #else /* Clear the shift state before starting. */ wctomb(NULL, L'\0'); #endif (void)sc; /* UNUSED */ /* * Allocate buffer for MBS. * We need this allocation here since it is possible that * as->s is still NULL. */ if (archive_string_ensure(as, as->length + len + 1) == NULL) return (-1); s = (const char *)_p; p = as->s + as->length; end = as->s + as->buffer_length - MB_CUR_MAX -1; while ((n = _utf8_to_unicode(&unicode, s, len)) != 0) { wchar_t wc; if (p >= end) { as->length = p - as->s; /* Re-allocate buffer for MBS. */ if (archive_string_ensure(as, as->length + max(len * 2, (size_t)MB_CUR_MAX) + 1) == NULL) return (-1); p = as->s + as->length; end = as->s + as->buffer_length - MB_CUR_MAX -1; } /* * As libarchive 2.x, translates the UTF-8 characters into * wide-characters in the assumption that WCS is Unicode. */ if (n < 0) { n *= -1; wc = L'?'; } else wc = (wchar_t)unicode; s += n; len -= n; /* * Translates the wide-character into the current locale MBS. */ #if HAVE_WCRTOMB n = (int)wcrtomb(p, wc, &shift_state); #else n = (int)wctomb(p, wc); #endif if (n == -1) return (-1); p += n; } as->length = p - as->s; as->s[as->length] = '\0'; return (0); } /* * Conversion functions between current locale dependent MBS and UTF-16BE. * strncat_from_utf16be() : UTF-16BE --> MBS * strncat_to_utf16be() : MBS --> UTF16BE */ #if defined(_WIN32) && !defined(__CYGWIN__) /* * Convert a UTF-16BE/LE string to current locale and copy the result. * Return -1 if conversion fails. */ static int win_strncat_from_utf16(struct archive_string *as, const void *_p, size_t bytes, struct archive_string_conv *sc, int be) { struct archive_string tmp; const char *u16; BOOL defchar; char *mbs; size_t mbs_size, b, ll; int ret = 0; bytes &= ~1; if (archive_string_ensure(as, as->length + bytes +1) == NULL) return (-1); mbs = as->s + as->length; mbs_size = as->buffer_length - as->length -1; if (sc->to_cp == CP_C_LOCALE) { /* * "C" locale special process. */ u16 = _p; ll = 0; for (b = 0; b < bytes; b += 2) { uint16_t val; if (be) val = archive_be16dec(u16+b); else val = archive_le16dec(u16+b); if (val > 255) { *mbs++ = '?'; ret = -1; } else *mbs++ = (char)(val&0xff); ll++; } as->length += ll; as->s[as->length] = '\0'; return (ret); } archive_string_init(&tmp); if (be) { if (IS_BIG_ENDIAN) { u16 = _p; } else { if (archive_string_ensure(&tmp, bytes+2) == NULL) return (-1); memcpy(tmp.s, _p, bytes); for (b = 0; b < bytes; b += 2) { uint16_t val = archive_be16dec(tmp.s+b); archive_le16enc(tmp.s+b, val); } u16 = tmp.s; } } else { if (!IS_BIG_ENDIAN) { u16 = _p; } else { if (archive_string_ensure(&tmp, bytes+2) == NULL) return (-1); memcpy(tmp.s, _p, bytes); for (b = 0; b < bytes; b += 2) { uint16_t val = archive_le16dec(tmp.s+b); archive_be16enc(tmp.s+b, val); } u16 = tmp.s; } } do { int r; defchar = 0; /* WideCharToMultiByte is limited to int. */ if (bytes > (size_t)INT_MAX || mbs_size > (size_t)INT_MAX) return (-1); r = WideCharToMultiByte(sc->to_cp, 0, (LPCWSTR)u16, (int)bytes>>1, mbs, (int)mbs_size, NULL, &defchar); /* Exit loop if we succeeded */ if (r != 0 || GetLastError() != ERROR_INSUFFICIENT_BUFFER) { ll = (size_t)r; break; } /* Else expand buffer and loop to try again. */ r = WideCharToMultiByte(sc->to_cp, 0, (LPCWSTR)u16, (int)bytes, NULL, 0, NULL, NULL); ll = (size_t)r; if (archive_string_ensure(as, ll +1) == NULL) return (-1); mbs = as->s + as->length; mbs_size = as->buffer_length - as->length -1; } while (1); archive_string_free(&tmp); as->length += ll; as->s[as->length] = '\0'; if (ll == 0 || defchar) ret = -1; return (ret); } static int win_strncat_from_utf16be(struct archive_string *as, const void *_p, size_t bytes, struct archive_string_conv *sc) { return (win_strncat_from_utf16(as, _p, bytes, sc, 1)); } static int win_strncat_from_utf16le(struct archive_string *as, const void *_p, size_t bytes, struct archive_string_conv *sc) { return (win_strncat_from_utf16(as, _p, bytes, sc, 0)); } /* * Convert a current locale string to UTF-16BE/LE and copy the result. * Return -1 if conversion fails. */ static int win_strncat_to_utf16(struct archive_string *as16, const void *_p, size_t length, struct archive_string_conv *sc, int bigendian) { const char *s = (const char *)_p; char *u16; size_t count, avail; if (archive_string_ensure(as16, as16->length + (length + 1) * 2) == NULL) return (-1); u16 = as16->s + as16->length; avail = as16->buffer_length - 2; if (sc->from_cp == CP_C_LOCALE) { /* * "C" locale special process. */ count = 0; while (count < length && *s) { if (bigendian) archive_be16enc(u16, *s); else archive_le16enc(u16, *s); u16 += 2; s++; count++; } as16->length += count << 1; as16->s[as16->length] = 0; as16->s[as16->length+1] = 0; return (0); } do { int r; if (length > (size_t)INT_MAX || (avail >> 1) > (size_t)INT_MAX) return (-1); r = MultiByteToWideChar(sc->from_cp, MB_PRECOMPOSED, s, (int)length, (LPWSTR)u16, (int)avail>>1); /* Exit loop if we succeeded */ if (r != 0 || GetLastError() != ERROR_INSUFFICIENT_BUFFER) { count = (size_t)r; break; } /* Expand buffer and try again */ r = MultiByteToWideChar(sc->from_cp, MB_PRECOMPOSED, s, (int)length, NULL, 0); count = (size_t)r; if (archive_string_ensure(as16, (count +1) * 2) == NULL) return (-1); u16 = as16->s + as16->length; avail = as16->buffer_length - 2; } while (1); as16->length += count * 2; as16->s[as16->length] = 0; as16->s[as16->length+1] = 0; if (count == 0) return (-1); if (IS_BIG_ENDIAN) { if (!bigendian) { while (count > 0) { uint16_t v = archive_be16dec(u16); archive_le16enc(u16, v); u16 += 2; count--; } } } else { if (bigendian) { while (count > 0) { uint16_t v = archive_le16dec(u16); archive_be16enc(u16, v); u16 += 2; count--; } } } return (0); } static int win_strncat_to_utf16be(struct archive_string *as16, const void *_p, size_t length, struct archive_string_conv *sc) { return (win_strncat_to_utf16(as16, _p, length, sc, 1)); } static int win_strncat_to_utf16le(struct archive_string *as16, const void *_p, size_t length, struct archive_string_conv *sc) { return (win_strncat_to_utf16(as16, _p, length, sc, 0)); } #endif /* _WIN32 && !__CYGWIN__ */ /* * Do the best effort for conversions. * We cannot handle UTF-16BE character-set without such iconv, * but there is a chance if a string consists just ASCII code or * a current locale is UTF-8. */ /* * Convert a UTF-16BE string to current locale and copy the result. * Return -1 if conversion fails. */ static int best_effort_strncat_from_utf16(struct archive_string *as, const void *_p, size_t bytes, struct archive_string_conv *sc, int be) { const char *utf16 = (const char *)_p; char *mbs; uint32_t uc; int n, ret; (void)sc; /* UNUSED */ /* * Other case, we should do the best effort. * If all character are ASCII(<0x7f), we can convert it. * if not , we set a alternative character and return -1. */ ret = 0; if (archive_string_ensure(as, as->length + bytes +1) == NULL) return (-1); mbs = as->s + as->length; while ((n = utf16_to_unicode(&uc, utf16, bytes, be)) != 0) { if (n < 0) { n *= -1; ret = -1; } bytes -= n; utf16 += n; if (uc > 127) { /* We cannot handle it. */ *mbs++ = '?'; ret = -1; } else *mbs++ = (char)uc; } as->length = mbs - as->s; as->s[as->length] = '\0'; return (ret); } static int best_effort_strncat_from_utf16be(struct archive_string *as, const void *_p, size_t bytes, struct archive_string_conv *sc) { return (best_effort_strncat_from_utf16(as, _p, bytes, sc, 1)); } static int best_effort_strncat_from_utf16le(struct archive_string *as, const void *_p, size_t bytes, struct archive_string_conv *sc) { return (best_effort_strncat_from_utf16(as, _p, bytes, sc, 0)); } /* * Convert a current locale string to UTF-16BE/LE and copy the result. * Return -1 if conversion fails. */ static int best_effort_strncat_to_utf16(struct archive_string *as16, const void *_p, size_t length, struct archive_string_conv *sc, int bigendian) { const char *s = (const char *)_p; char *utf16; size_t remaining; int ret; (void)sc; /* UNUSED */ /* * Other case, we should do the best effort. * If all character are ASCII(<0x7f), we can convert it. * if not , we set a alternative character and return -1. */ ret = 0; remaining = length; if (archive_string_ensure(as16, as16->length + (length + 1) * 2) == NULL) return (-1); utf16 = as16->s + as16->length; while (remaining--) { unsigned c = *s++; if (c > 127) { /* We cannot handle it. */ c = UNICODE_R_CHAR; ret = -1; } if (bigendian) archive_be16enc(utf16, (uint16_t)c); else archive_le16enc(utf16, (uint16_t)c); utf16 += 2; } as16->length = utf16 - as16->s; as16->s[as16->length] = 0; as16->s[as16->length+1] = 0; return (ret); } static int best_effort_strncat_to_utf16be(struct archive_string *as16, const void *_p, size_t length, struct archive_string_conv *sc) { return (best_effort_strncat_to_utf16(as16, _p, length, sc, 1)); } static int best_effort_strncat_to_utf16le(struct archive_string *as16, const void *_p, size_t length, struct archive_string_conv *sc) { return (best_effort_strncat_to_utf16(as16, _p, length, sc, 0)); } /* * Multistring operations. */ void archive_mstring_clean(struct archive_mstring *aes) { archive_wstring_free(&(aes->aes_wcs)); archive_string_free(&(aes->aes_mbs)); archive_string_free(&(aes->aes_utf8)); archive_string_free(&(aes->aes_mbs_in_locale)); aes->aes_set = 0; } void archive_mstring_copy(struct archive_mstring *dest, struct archive_mstring *src) { dest->aes_set = src->aes_set; archive_string_copy(&(dest->aes_mbs), &(src->aes_mbs)); archive_string_copy(&(dest->aes_utf8), &(src->aes_utf8)); archive_wstring_copy(&(dest->aes_wcs), &(src->aes_wcs)); } int archive_mstring_get_utf8(struct archive *a, struct archive_mstring *aes, const char **p) { struct archive_string_conv *sc; int r; /* If we already have a UTF8 form, return that immediately. */ if (aes->aes_set & AES_SET_UTF8) { *p = aes->aes_utf8.s; return (0); } *p = NULL; #if defined(_WIN32) && !defined(__CYGWIN__) /* * On Windows, first try converting from WCS because (1) there's no * guarantee that the conversion to MBS will succeed, e.g. when using * CP_ACP, and (2) that's more efficient than converting to MBS, just to * convert back to WCS again before finally converting to UTF-8 */ if ((aes->aes_set & AES_SET_WCS) != 0) { sc = archive_string_conversion_to_charset(a, "UTF-8", 1); if (sc == NULL) return (-1);/* Couldn't allocate memory for sc. */ archive_string_empty(&(aes->aes_utf8)); r = archive_string_append_from_wcs_in_codepage(&(aes->aes_utf8), aes->aes_wcs.s, aes->aes_wcs.length, sc); if (a == NULL) free_sconv_object(sc); if (r == 0) { aes->aes_set |= AES_SET_UTF8; *p = aes->aes_utf8.s; return (0);/* success. */ } else return (-1);/* failure. */ } #endif /* Try converting WCS to MBS first if MBS does not exist yet. */ if ((aes->aes_set & AES_SET_MBS) == 0) { const char *pm; /* unused */ archive_mstring_get_mbs(a, aes, &pm); /* ignore errors, we'll handle it later */ } if (aes->aes_set & AES_SET_MBS) { sc = archive_string_conversion_to_charset(a, "UTF-8", 1); if (sc == NULL) return (-1);/* Couldn't allocate memory for sc. */ r = archive_strncpy_l(&(aes->aes_utf8), aes->aes_mbs.s, aes->aes_mbs.length, sc); if (a == NULL) free_sconv_object(sc); if (r == 0) { aes->aes_set |= AES_SET_UTF8; *p = aes->aes_utf8.s; return (0);/* success. */ } else return (-1);/* failure. */ } return (0);/* success. */ } int archive_mstring_get_mbs(struct archive *a, struct archive_mstring *aes, const char **p) { struct archive_string_conv *sc; int r, ret = 0; /* If we already have an MBS form, return that immediately. */ if (aes->aes_set & AES_SET_MBS) { *p = aes->aes_mbs.s; return (ret); } *p = NULL; /* If there's a WCS form, try converting with the native locale. */ if (aes->aes_set & AES_SET_WCS) { archive_string_empty(&(aes->aes_mbs)); r = archive_string_append_from_wcs(&(aes->aes_mbs), aes->aes_wcs.s, aes->aes_wcs.length); *p = aes->aes_mbs.s; if (r == 0) { aes->aes_set |= AES_SET_MBS; return (ret); } else ret = -1; } /* If there's a UTF-8 form, try converting with the native locale. */ if (aes->aes_set & AES_SET_UTF8) { archive_string_empty(&(aes->aes_mbs)); sc = archive_string_conversion_from_charset(a, "UTF-8", 1); if (sc == NULL) return (-1);/* Couldn't allocate memory for sc. */ r = archive_strncpy_l(&(aes->aes_mbs), aes->aes_utf8.s, aes->aes_utf8.length, sc); if (a == NULL) free_sconv_object(sc); *p = aes->aes_mbs.s; if (r == 0) { aes->aes_set |= AES_SET_MBS; ret = 0;/* success; overwrite previous error. */ } else ret = -1;/* failure. */ } return (ret); } int archive_mstring_get_wcs(struct archive *a, struct archive_mstring *aes, const wchar_t **wp) { int r, ret = 0; (void)a;/* UNUSED */ /* Return WCS form if we already have it. */ if (aes->aes_set & AES_SET_WCS) { *wp = aes->aes_wcs.s; return (ret); } *wp = NULL; #if defined(_WIN32) && !defined(__CYGWIN__) /* * On Windows, prefer converting from UTF-8 directly to WCS because: * (1) there's no guarantee that the string can be represented in MBS (e.g. * with CP_ACP), and (2) in order to convert from UTF-8 to MBS, we're going * to need to convert from UTF-8 to WCS anyway and its wasteful to throw * away that intermediate result */ if (aes->aes_set & AES_SET_UTF8) { struct archive_string_conv *sc; sc = archive_string_conversion_from_charset(a, "UTF-8", 1); if (sc != NULL) { archive_wstring_empty((&aes->aes_wcs)); r = archive_wstring_append_from_mbs_in_codepage(&(aes->aes_wcs), aes->aes_utf8.s, aes->aes_utf8.length, sc); if (a == NULL) free_sconv_object(sc); if (r == 0) { aes->aes_set |= AES_SET_WCS; *wp = aes->aes_wcs.s; return (0); } } } #endif /* Try converting UTF8 to MBS first if MBS does not exist yet. */ if ((aes->aes_set & AES_SET_MBS) == 0) { const char *p; /* unused */ archive_mstring_get_mbs(a, aes, &p); /* ignore errors, we'll handle it later */ } /* Try converting MBS to WCS using native locale. */ if (aes->aes_set & AES_SET_MBS) { archive_wstring_empty(&(aes->aes_wcs)); r = archive_wstring_append_from_mbs(&(aes->aes_wcs), aes->aes_mbs.s, aes->aes_mbs.length); if (r == 0) { aes->aes_set |= AES_SET_WCS; *wp = aes->aes_wcs.s; } else ret = -1;/* failure. */ } return (ret); } int archive_mstring_get_mbs_l(struct archive *a, struct archive_mstring *aes, const char **p, size_t *length, struct archive_string_conv *sc) { int ret = 0; #if defined(_WIN32) && !defined(__CYGWIN__) int r; /* * Internationalization programming on Windows must use Wide * characters because Windows platform cannot make locale UTF-8. */ if (sc != NULL && (aes->aes_set & AES_SET_WCS) != 0) { archive_string_empty(&(aes->aes_mbs_in_locale)); r = archive_string_append_from_wcs_in_codepage( &(aes->aes_mbs_in_locale), aes->aes_wcs.s, aes->aes_wcs.length, sc); if (r == 0) { *p = aes->aes_mbs_in_locale.s; if (length != NULL) *length = aes->aes_mbs_in_locale.length; return (0); } else if (errno == ENOMEM) return (-1); else ret = -1; } #endif /* If there is not an MBS form but there is a WCS or UTF8 form, try converting * with the native locale to be used for translating it to specified * character-set. */ if ((aes->aes_set & AES_SET_MBS) == 0) { const char *pm; /* unused */ archive_mstring_get_mbs(a, aes, &pm); /* ignore errors, we'll handle it later */ } /* If we already have an MBS form, use it to be translated to * specified character-set. */ if (aes->aes_set & AES_SET_MBS) { if (sc == NULL) { /* Conversion is unneeded. */ *p = aes->aes_mbs.s; if (length != NULL) *length = aes->aes_mbs.length; return (0); } ret = archive_strncpy_l(&(aes->aes_mbs_in_locale), aes->aes_mbs.s, aes->aes_mbs.length, sc); *p = aes->aes_mbs_in_locale.s; if (length != NULL) *length = aes->aes_mbs_in_locale.length; } else { *p = NULL; if (length != NULL) *length = 0; } return (ret); } int archive_mstring_copy_mbs(struct archive_mstring *aes, const char *mbs) { if (mbs == NULL) { aes->aes_set = 0; return (0); } return (archive_mstring_copy_mbs_len(aes, mbs, strlen(mbs))); } int archive_mstring_copy_mbs_len(struct archive_mstring *aes, const char *mbs, size_t len) { if (mbs == NULL) { aes->aes_set = 0; return (0); } aes->aes_set = AES_SET_MBS; /* Only MBS form is set now. */ archive_strncpy(&(aes->aes_mbs), mbs, len); archive_string_empty(&(aes->aes_utf8)); archive_wstring_empty(&(aes->aes_wcs)); return (0); } int archive_mstring_copy_wcs(struct archive_mstring *aes, const wchar_t *wcs) { return archive_mstring_copy_wcs_len(aes, wcs, wcs == NULL ? 0 : wcslen(wcs)); } int archive_mstring_copy_utf8(struct archive_mstring *aes, const char *utf8) { if (utf8 == NULL) { aes->aes_set = 0; return (0); } aes->aes_set = AES_SET_UTF8; archive_string_empty(&(aes->aes_mbs)); archive_string_empty(&(aes->aes_wcs)); archive_strncpy(&(aes->aes_utf8), utf8, strlen(utf8)); return (int)strlen(utf8); } int archive_mstring_copy_wcs_len(struct archive_mstring *aes, const wchar_t *wcs, size_t len) { if (wcs == NULL) { aes->aes_set = 0; return (0); } aes->aes_set = AES_SET_WCS; /* Only WCS form set. */ archive_string_empty(&(aes->aes_mbs)); archive_string_empty(&(aes->aes_utf8)); archive_wstrncpy(&(aes->aes_wcs), wcs, len); return (0); } int archive_mstring_copy_mbs_len_l(struct archive_mstring *aes, const char *mbs, size_t len, struct archive_string_conv *sc) { int r; if (mbs == NULL) { aes->aes_set = 0; return (0); } archive_string_empty(&(aes->aes_mbs)); archive_wstring_empty(&(aes->aes_wcs)); archive_string_empty(&(aes->aes_utf8)); #if defined(_WIN32) && !defined(__CYGWIN__) /* * Internationalization programming on Windows must use Wide * characters because Windows platform cannot make locale UTF-8. */ if (sc == NULL) { if (archive_string_append(&(aes->aes_mbs), mbs, mbsnbytes(mbs, len)) == NULL) { aes->aes_set = 0; r = -1; } else { aes->aes_set = AES_SET_MBS; r = 0; } #if defined(HAVE_ICONV) } else if (sc != NULL && sc->cd_w != (iconv_t)-1) { /* * This case happens only when MultiByteToWideChar() cannot * handle sc->from_cp, and we have to iconv in order to * translate character-set to wchar_t,UTF-16. */ iconv_t cd = sc->cd; unsigned from_cp; int flag; /* * Translate multi-bytes from some character-set to UTF-8. */ sc->cd = sc->cd_w; r = archive_strncpy_l(&(aes->aes_utf8), mbs, len, sc); sc->cd = cd; if (r != 0) { aes->aes_set = 0; return (r); } aes->aes_set = AES_SET_UTF8; /* * Append the UTF-8 string into wstring. */ flag = sc->flag; sc->flag &= ~(SCONV_NORMALIZATION_C | SCONV_TO_UTF16| SCONV_FROM_UTF16); from_cp = sc->from_cp; sc->from_cp = CP_UTF8; r = archive_wstring_append_from_mbs_in_codepage(&(aes->aes_wcs), aes->aes_utf8.s, aes->aes_utf8.length, sc); sc->flag = flag; sc->from_cp = from_cp; if (r == 0) aes->aes_set |= AES_SET_WCS; #endif } else { r = archive_wstring_append_from_mbs_in_codepage( &(aes->aes_wcs), mbs, len, sc); if (r == 0) aes->aes_set = AES_SET_WCS; else aes->aes_set = 0; } #else r = archive_strncpy_l(&(aes->aes_mbs), mbs, len, sc); if (r == 0) aes->aes_set = AES_SET_MBS; /* Only MBS form is set now. */ else aes->aes_set = 0; #endif return (r); } /* * The 'update' form tries to proactively update all forms of * this string (WCS and MBS) and returns an error if any of * them fail. This is used by the 'pax' handler, for instance, * to detect and report character-conversion failures early while * still allowing clients to get potentially useful values from * the more tolerant lazy conversions. (get_mbs and get_wcs will * strive to give the user something useful, so you can get hopefully * usable values even if some of the character conversions are failing.) */ int archive_mstring_update_utf8(struct archive *a, struct archive_mstring *aes, const char *utf8) { struct archive_string_conv *sc; int r; if (utf8 == NULL) { aes->aes_set = 0; return (0); /* Succeeded in clearing everything. */ } /* Save the UTF8 string. */ archive_strcpy(&(aes->aes_utf8), utf8); /* Empty the mbs and wcs strings. */ archive_string_empty(&(aes->aes_mbs)); archive_wstring_empty(&(aes->aes_wcs)); aes->aes_set = AES_SET_UTF8; /* Only UTF8 is set now. */ sc = archive_string_conversion_from_charset(a, "UTF-8", 1); if (sc == NULL) return (-1);/* Couldn't allocate memory for sc. */ #if defined(_WIN32) && !defined(__CYGWIN__) /* On Windows, there's no good way to convert from UTF8 -> MBS directly, so * prefer to first convert to WCS as (1) it's wasteful to throw away the * intermediate result, and (2) WCS will still be set even if we fail to * convert to MBS (e.g. with ACP that can't represent the characters) */ r = archive_wstring_append_from_mbs_in_codepage(&(aes->aes_wcs), aes->aes_utf8.s, aes->aes_utf8.length, sc); if (a == NULL) free_sconv_object(sc); if (r != 0) return (-1); /* This will guarantee we can't convert to MBS */ aes->aes_set = AES_SET_UTF8 | AES_SET_WCS; /* Both UTF8 and WCS set. */ /* Try converting WCS to MBS, return false on failure. */ if (archive_string_append_from_wcs(&(aes->aes_mbs), aes->aes_wcs.s, aes->aes_wcs.length)) return (-1); #else /* Try converting UTF-8 to MBS, return false on failure. */ r = archive_strcpy_l(&(aes->aes_mbs), utf8, sc); if (a == NULL) free_sconv_object(sc); if (r != 0) return (-1); aes->aes_set = AES_SET_UTF8 | AES_SET_MBS; /* Both UTF8 and MBS set. */ /* Try converting MBS to WCS, return false on failure. */ if (archive_wstring_append_from_mbs(&(aes->aes_wcs), aes->aes_mbs.s, aes->aes_mbs.length)) return (-1); #endif /* All conversions succeeded. */ aes->aes_set = AES_SET_UTF8 | AES_SET_WCS | AES_SET_MBS; return (0); } diff --git a/contrib/libarchive/libarchive/archive_string_sprintf.c b/contrib/libarchive/libarchive/archive_string_sprintf.c index 1c5910e0b2d3..69b0cdcd83d5 100644 --- a/contrib/libarchive/libarchive/archive_string_sprintf.c +++ b/contrib/libarchive/libarchive/archive_string_sprintf.c @@ -1,200 +1,200 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" /* * The use of printf()-family functions can be troublesome * for space-constrained applications. In addition, correctly * implementing this function in terms of vsnprintf() requires * two calls (one to determine the size, another to format the * result), which in turn requires duplicating the argument list * using va_copy, which isn't yet universally available. * * So, I've implemented a bare minimum of printf()-like capability * here. This is only used to format error messages, so doesn't * require any floating-point support or field-width handling. */ #ifdef HAVE_ERRNO_H #include #endif #include #include "archive_string.h" #include "archive_private.h" /* * Utility functions to format signed/unsigned integers and append * them to an archive_string. */ static void append_uint(struct archive_string *as, uintmax_t d, unsigned base) { static const char digits[] = "0123456789abcdef"; if (d >= base) append_uint(as, d/base, base); archive_strappend_char(as, digits[d % base]); } static void append_int(struct archive_string *as, intmax_t d, unsigned base) { uintmax_t ud; if (d < 0) { archive_strappend_char(as, '-'); ud = (d == INTMAX_MIN) ? (uintmax_t)(INTMAX_MAX) + 1 : (uintmax_t)(-d); } else ud = d; append_uint(as, ud, base); } void archive_string_sprintf(struct archive_string *as, const char *fmt, ...) { va_list ap; va_start(ap, fmt); archive_string_vsprintf(as, fmt, ap); va_end(ap); } /* * Like 'vsprintf', but ensures the target is big enough, resizing if * necessary. */ void archive_string_vsprintf(struct archive_string *as, const char *fmt, va_list ap) { char long_flag; intmax_t s; /* Signed integer temp. */ uintmax_t u; /* Unsigned integer temp. */ const char *p, *p2; const wchar_t *pw; if (archive_string_ensure(as, 64) == NULL) __archive_errx(1, "Out of memory"); if (fmt == NULL) { as->s[0] = 0; return; } for (p = fmt; *p != '\0'; p++) { const char *saved_p = p; if (*p != '%') { archive_strappend_char(as, *p); continue; } p++; long_flag = '\0'; switch(*p) { case 'l': if (p[1] == 'l') { long_flag = 'L'; p += 2; break; } __LA_FALLTHROUGH; case 'j': case 'z': long_flag = *p; p++; break; } switch (*p) { case '%': archive_strappend_char(as, '%'); break; case 'c': s = va_arg(ap, int); archive_strappend_char(as, (char)s); break; case 'd': switch(long_flag) { case 'j': s = va_arg(ap, intmax_t); break; case 'l': s = va_arg(ap, long); break; case 'L': s = va_arg(ap, long long); break; case 'z': s = va_arg(ap, ssize_t); break; default: s = va_arg(ap, int); break; } - append_int(as, s, 10); + append_int(as, s, 10); break; case 's': switch(long_flag) { case 'l': case 'L': pw = va_arg(ap, wchar_t *); if (pw == NULL) pw = L"(null)"; if (archive_string_append_from_wcs(as, pw, wcslen(pw)) != 0 && errno == ENOMEM) __archive_errx(1, "Out of memory"); break; default: p2 = va_arg(ap, char *); if (p2 == NULL) p2 = "(null)"; archive_strcat(as, p2); break; } break; case 'S': pw = va_arg(ap, wchar_t *); if (pw == NULL) pw = L"(null)"; if (archive_string_append_from_wcs(as, pw, wcslen(pw)) != 0 && errno == ENOMEM) __archive_errx(1, "Out of memory"); break; case 'o': case 'u': case 'x': case 'X': /* Common handling for unsigned integer formats. */ switch(long_flag) { case 'j': u = va_arg(ap, uintmax_t); break; case 'l': u = va_arg(ap, unsigned long); break; case 'L': u = va_arg(ap, unsigned long long); break; case 'z': u = va_arg(ap, size_t); break; default: u = va_arg(ap, unsigned int); break; } /* Format it in the correct base. */ switch (*p) { case 'o': append_uint(as, u, 8); break; case 'u': append_uint(as, u, 10); break; default: append_uint(as, u, 16); break; } break; default: /* Rewind and print the initial '%' literally. */ p = saved_p; archive_strappend_char(as, *p); } } } diff --git a/contrib/libarchive/libarchive/archive_util.c b/contrib/libarchive/libarchive/archive_util.c index 900abd0c3c62..d048bbc94650 100644 --- a/contrib/libarchive/libarchive/archive_util.c +++ b/contrib/libarchive/libarchive/archive_util.c @@ -1,658 +1,686 @@ /*- * Copyright (c) 2009-2012,2014 Michihiro NAKAJIMA * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_TYPES_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #if defined(_WIN32) && !defined(__CYGWIN__) #if defined(HAVE_BCRYPT_H) && _WIN32_WINNT >= _WIN32_WINNT_VISTA /* don't use bcrypt when XP needs to be supported */ #include /* Common in other bcrypt implementations, but missing from VS2008. */ #ifndef BCRYPT_SUCCESS #define BCRYPT_SUCCESS(r) ((NTSTATUS)(r) == STATUS_SUCCESS) #endif #elif defined(HAVE_WINCRYPT_H) #include #endif #endif #ifdef HAVE_ZLIB_H #include #endif #ifdef HAVE_LZMA_H #include #endif #ifdef HAVE_BZLIB_H #include #endif #ifdef HAVE_LZ4_H #include #endif #include "archive.h" #include "archive_private.h" #include "archive_random_private.h" #include "archive_string.h" #ifndef O_CLOEXEC #define O_CLOEXEC 0 #endif #if ARCHIVE_VERSION_NUMBER < 4000000 static int __LA_LIBC_CC archive_utility_string_sort_helper(const void *, const void *); #endif /* Generic initialization of 'struct archive' objects. */ int __archive_clean(struct archive *a) { archive_string_conversion_free(a); return (ARCHIVE_OK); } int archive_version_number(void) { return (ARCHIVE_VERSION_NUMBER); } const char * archive_version_string(void) { return (ARCHIVE_VERSION_STRING); } int archive_errno(struct archive *a) { return (a->archive_error_number); } const char * archive_error_string(struct archive *a) { if (a->error != NULL && *a->error != '\0') return (a->error); else return (NULL); } int archive_file_count(struct archive *a) { return (a->file_count); } int archive_format(struct archive *a) { return (a->archive_format); } const char * archive_format_name(struct archive *a) { return (a->archive_format_name); } int archive_compression(struct archive *a) { return archive_filter_code(a, 0); } const char * archive_compression_name(struct archive *a) { return archive_filter_name(a, 0); } /* * Return a count of the number of compressed bytes processed. */ la_int64_t archive_position_compressed(struct archive *a) { return archive_filter_bytes(a, -1); } /* * Return a count of the number of uncompressed bytes processed. */ la_int64_t archive_position_uncompressed(struct archive *a) { return archive_filter_bytes(a, 0); } void archive_clear_error(struct archive *a) { archive_string_empty(&a->error_string); a->error = NULL; a->archive_error_number = 0; } void archive_set_error(struct archive *a, int error_number, const char *fmt, ...) { va_list ap; a->archive_error_number = error_number; if (fmt == NULL) { a->error = NULL; return; } archive_string_empty(&(a->error_string)); va_start(ap, fmt); archive_string_vsprintf(&(a->error_string), fmt, ap); va_end(ap); a->error = a->error_string.s; } void archive_copy_error(struct archive *dest, struct archive *src) { dest->archive_error_number = src->archive_error_number; archive_string_copy(&dest->error_string, &src->error_string); dest->error = dest->error_string.s; } void __archive_errx(int retvalue, const char *msg) { static const char msg1[] = "Fatal Internal Error in libarchive: "; size_t s; s = write(2, msg1, strlen(msg1)); (void)s; /* UNUSED */ s = write(2, msg, strlen(msg)); (void)s; /* UNUSED */ s = write(2, "\n", 1); (void)s; /* UNUSED */ exit(retvalue); } /* * Create a temporary file */ #if defined(_WIN32) && !defined(__CYGWIN__) /* * Do not use Windows tmpfile() function. * It will make a temporary file under the root directory * and it'll cause permission error if a user who is * non-Administrator creates temporary files. * Also Windows version of mktemp family including _mktemp_s * are not secure. */ static int __archive_mktempx(const char *tmpdir, wchar_t *template) { static const wchar_t prefix[] = L"libarchive_"; static const wchar_t suffix[] = L"XXXXXXXXXX"; static const wchar_t num[] = { L'0', L'1', L'2', L'3', L'4', L'5', L'6', L'7', L'8', L'9', L'A', L'B', L'C', L'D', L'E', L'F', L'G', L'H', L'I', L'J', L'K', L'L', L'M', L'N', L'O', L'P', L'Q', L'R', L'S', L'T', L'U', L'V', L'W', L'X', L'Y', L'Z', L'a', L'b', L'c', L'd', L'e', L'f', L'g', L'h', L'i', L'j', L'k', L'l', L'm', L'n', L'o', L'p', L'q', L'r', L's', L't', L'u', L'v', L'w', L'x', L'y', L'z' }; struct archive_wstring temp_name; wchar_t *ws; DWORD attr; wchar_t *xp, *ep; int fd; #if defined(HAVE_BCRYPT_H) && _WIN32_WINNT >= _WIN32_WINNT_VISTA BCRYPT_ALG_HANDLE hAlg = NULL; #else HCRYPTPROV hProv = (HCRYPTPROV)NULL; #endif fd = -1; ws = NULL; archive_string_init(&temp_name); if (template == NULL) { /* Get a temporary directory. */ if (tmpdir == NULL) { size_t l; wchar_t *tmp; l = GetTempPathW(0, NULL); if (l == 0) { la_dosmaperr(GetLastError()); goto exit_tmpfile; } tmp = malloc(l*sizeof(wchar_t)); if (tmp == NULL) { errno = ENOMEM; goto exit_tmpfile; } GetTempPathW((DWORD)l, tmp); archive_wstrcpy(&temp_name, tmp); free(tmp); } else { if (archive_wstring_append_from_mbs(&temp_name, tmpdir, strlen(tmpdir)) < 0) goto exit_tmpfile; if (temp_name.length == 0 || temp_name.s[temp_name.length-1] != L'/') archive_wstrappend_wchar(&temp_name, L'/'); } /* Check if temp_name is a directory. */ attr = GetFileAttributesW(temp_name.s); if (attr == (DWORD)-1) { if (GetLastError() != ERROR_FILE_NOT_FOUND) { la_dosmaperr(GetLastError()); goto exit_tmpfile; } ws = __la_win_permissive_name_w(temp_name.s); if (ws == NULL) { errno = EINVAL; goto exit_tmpfile; } attr = GetFileAttributesW(ws); if (attr == (DWORD)-1) { la_dosmaperr(GetLastError()); goto exit_tmpfile; } } if (!(attr & FILE_ATTRIBUTE_DIRECTORY)) { errno = ENOTDIR; goto exit_tmpfile; } /* * Create a temporary file. */ archive_wstrcat(&temp_name, prefix); archive_wstrcat(&temp_name, suffix); ep = temp_name.s + archive_strlen(&temp_name); xp = ep - wcslen(suffix); template = temp_name.s; } else { xp = wcschr(template, L'X'); if (xp == NULL) /* No X, programming error */ abort(); for (ep = xp; *ep == L'X'; ep++) continue; if (*ep) /* X followed by non X, programming error */ abort(); } #if defined(HAVE_BCRYPT_H) && _WIN32_WINNT >= _WIN32_WINNT_VISTA if (!BCRYPT_SUCCESS(BCryptOpenAlgorithmProvider(&hAlg, BCRYPT_RNG_ALGORITHM, NULL, 0))) { la_dosmaperr(GetLastError()); goto exit_tmpfile; } #else if (!CryptAcquireContext(&hProv, NULL, NULL, PROV_RSA_FULL, CRYPT_VERIFYCONTEXT)) { la_dosmaperr(GetLastError()); goto exit_tmpfile; } #endif for (;;) { wchar_t *p; HANDLE h; # if _WIN32_WINNT >= 0x0602 /* _WIN32_WINNT_WIN8 */ CREATEFILE2_EXTENDED_PARAMETERS createExParams; #endif /* Generate a random file name through CryptGenRandom(). */ p = xp; #if defined(HAVE_BCRYPT_H) && _WIN32_WINNT >= _WIN32_WINNT_VISTA if (!BCRYPT_SUCCESS(BCryptGenRandom(hAlg, (PUCHAR)p, (DWORD)(ep - p)*sizeof(wchar_t), 0))) { la_dosmaperr(GetLastError()); goto exit_tmpfile; } #else if (!CryptGenRandom(hProv, (DWORD)(ep - p)*sizeof(wchar_t), (BYTE*)p)) { la_dosmaperr(GetLastError()); goto exit_tmpfile; } #endif for (; p < ep; p++) *p = num[((DWORD)*p) % (sizeof(num)/sizeof(num[0]))]; free(ws); ws = __la_win_permissive_name_w(template); if (ws == NULL) { errno = EINVAL; goto exit_tmpfile; } if (template == temp_name.s) { attr = FILE_ATTRIBUTE_TEMPORARY | FILE_FLAG_DELETE_ON_CLOSE; } else { /* mkstemp */ attr = FILE_ATTRIBUTE_NORMAL; } # if _WIN32_WINNT >= 0x0602 /* _WIN32_WINNT_WIN8 */ ZeroMemory(&createExParams, sizeof(createExParams)); createExParams.dwSize = sizeof(createExParams); createExParams.dwFileAttributes = attr & 0xFFFF; createExParams.dwFileFlags = attr & 0xFFF00000; h = CreateFile2(ws, GENERIC_READ | GENERIC_WRITE | DELETE, 0,/* Not share */ CREATE_NEW, &createExParams); #else h = CreateFileW(ws, GENERIC_READ | GENERIC_WRITE | DELETE, 0,/* Not share */ NULL, CREATE_NEW,/* Create a new file only */ attr, NULL); #endif if (h == INVALID_HANDLE_VALUE) { /* The same file already exists. retry with * a new filename. */ if (GetLastError() == ERROR_FILE_EXISTS) continue; /* Otherwise, fail creation temporary file. */ la_dosmaperr(GetLastError()); goto exit_tmpfile; } fd = _open_osfhandle((intptr_t)h, _O_BINARY | _O_RDWR); if (fd == -1) { la_dosmaperr(GetLastError()); CloseHandle(h); goto exit_tmpfile; } else break;/* success! */ } exit_tmpfile: #if defined(HAVE_BCRYPT_H) && _WIN32_WINNT >= _WIN32_WINNT_VISTA if (hAlg != NULL) BCryptCloseAlgorithmProvider(hAlg, 0); #else if (hProv != (HCRYPTPROV)NULL) CryptReleaseContext(hProv, 0); #endif free(ws); if (template == temp_name.s) archive_wstring_free(&temp_name); return (fd); } int __archive_mktemp(const char *tmpdir) { return __archive_mktempx(tmpdir, NULL); } int __archive_mkstemp(wchar_t *template) { return __archive_mktempx(NULL, template); } #else static int -get_tempdir(struct archive_string *temppath) +__archive_issetugid(void) { - const char *tmp; +#ifdef HAVE_ISSETUGID + return (issetugid()); +#elif HAVE_GETRESUID + uid_t ruid, euid, suid; + gid_t rgid, egid, sgid; + if (getresuid(&ruid, &euid, &suid) != 0) + return (-1); + if (ruid != euid || ruid != suid) + return (1); + if (getresgid(&ruid, &egid, &sgid) != 0) + return (-1); + if (rgid != egid || rgid != sgid) + return (1); +#elif HAVE_GETEUID + if (geteuid() != getuid()) + return (1); +#if HAVE_GETEGID + if (getegid() != getgid()) + return (1); +#endif +#endif + return (0); +} + +int +__archive_get_tempdir(struct archive_string *temppath) +{ + const char *tmp = NULL; - tmp = getenv("TMPDIR"); + if (__archive_issetugid() == 0) + tmp = getenv("TMPDIR"); if (tmp == NULL) #ifdef _PATH_TMP tmp = _PATH_TMP; #else tmp = "/tmp"; #endif archive_strcpy(temppath, tmp); if (temppath->length == 0 || temppath->s[temppath->length-1] != '/') archive_strappend_char(temppath, '/'); return (ARCHIVE_OK); } #if defined(HAVE_MKSTEMP) /* * We can use mkstemp(). */ int __archive_mktemp(const char *tmpdir) { struct archive_string temp_name; int fd = -1; archive_string_init(&temp_name); if (tmpdir == NULL) { - if (get_tempdir(&temp_name) != ARCHIVE_OK) + if (__archive_get_tempdir(&temp_name) != ARCHIVE_OK) goto exit_tmpfile; } else { archive_strcpy(&temp_name, tmpdir); if (temp_name.length == 0 || temp_name.s[temp_name.length-1] != '/') archive_strappend_char(&temp_name, '/'); } #ifdef O_TMPFILE fd = open(temp_name.s, O_RDWR|O_CLOEXEC|O_TMPFILE|O_EXCL, 0600); if(fd >= 0) goto exit_tmpfile; #endif archive_strcat(&temp_name, "libarchive_XXXXXX"); fd = mkstemp(temp_name.s); if (fd < 0) goto exit_tmpfile; __archive_ensure_cloexec_flag(fd); unlink(temp_name.s); exit_tmpfile: archive_string_free(&temp_name); return (fd); } int __archive_mkstemp(char *template) { int fd = -1; fd = mkstemp(template); if (fd >= 0) __archive_ensure_cloexec_flag(fd); return (fd); } #else /* !HAVE_MKSTEMP */ /* * We use a private routine. */ static int __archive_mktempx(const char *tmpdir, char *template) { static const char num[] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z' }; struct archive_string temp_name; struct stat st; int fd; char *tp, *ep; fd = -1; if (template == NULL) { archive_string_init(&temp_name); if (tmpdir == NULL) { - if (get_tempdir(&temp_name) != ARCHIVE_OK) + if (__archive_get_tempdir(&temp_name) != ARCHIVE_OK) goto exit_tmpfile; } else archive_strcpy(&temp_name, tmpdir); if (temp_name.length > 0 && temp_name.s[temp_name.length-1] == '/') { temp_name.s[temp_name.length-1] = '\0'; temp_name.length --; } if (la_stat(temp_name.s, &st) < 0) goto exit_tmpfile; if (!S_ISDIR(st.st_mode)) { errno = ENOTDIR; goto exit_tmpfile; } archive_strcat(&temp_name, "/libarchive_"); tp = temp_name.s + archive_strlen(&temp_name); archive_strcat(&temp_name, "XXXXXXXXXX"); ep = temp_name.s + archive_strlen(&temp_name); template = temp_name.s; } else { tp = strchr(template, 'X'); if (tp == NULL) /* No X, programming error */ abort(); for (ep = tp; *ep == 'X'; ep++) continue; if (*ep) /* X followed by non X, programming error */ abort(); } do { char *p; p = tp; archive_random(p, ep - p); while (p < ep) { int d = *((unsigned char *)p) % sizeof(num); *p++ = num[d]; } fd = open(template, O_CREAT | O_EXCL | O_RDWR | O_CLOEXEC, 0600); } while (fd < 0 && errno == EEXIST); if (fd < 0) goto exit_tmpfile; __archive_ensure_cloexec_flag(fd); if (template == temp_name.s) unlink(temp_name.s); exit_tmpfile: if (template == temp_name.s) archive_string_free(&temp_name); return (fd); } int __archive_mktemp(const char *tmpdir) { return __archive_mktempx(tmpdir, NULL); } int __archive_mkstemp(char *template) { return __archive_mktempx(NULL, template); } #endif /* !HAVE_MKSTEMP */ #endif /* !_WIN32 || __CYGWIN__ */ /* * Set FD_CLOEXEC flag to a file descriptor if it is not set. * We have to set the flag if the platform does not provide O_CLOEXEC * or F_DUPFD_CLOEXEC flags. * * Note: This function is absolutely called after creating a new file * descriptor even if the platform seemingly provides O_CLOEXEC or * F_DUPFD_CLOEXEC macros because it is possible that the platform * merely declares those macros, especially Linux 2.6.18 - 2.6.24 do it. */ void __archive_ensure_cloexec_flag(int fd) { #if defined(_WIN32) && !defined(__CYGWIN__) (void)fd; /* UNUSED */ #else int flags; if (fd >= 0) { flags = fcntl(fd, F_GETFD); if (flags != -1 && (flags & FD_CLOEXEC) == 0) fcntl(fd, F_SETFD, flags | FD_CLOEXEC); } #endif } #if ARCHIVE_VERSION_NUMBER < 4000000 /* * Utility functions to sort a group of strings using quicksort. */ static int __LA_LIBC_CC archive_utility_string_sort_helper(const void *p1, const void *p2) { const char * const * const s1 = p1; const char * const * const s2 = p2; return strcmp(*s1, *s2); } int archive_utility_string_sort(char **strings) { size_t size = 0; while (strings[size] != NULL) size++; qsort(strings, size, sizeof(char *), archive_utility_string_sort_helper); return (ARCHIVE_OK); } #endif diff --git a/contrib/libarchive/libarchive/archive_write.c b/contrib/libarchive/libarchive/archive_write.c index a8e7b63b5bfe..9b9cb196f0f9 100644 --- a/contrib/libarchive/libarchive/archive_write.c +++ b/contrib/libarchive/libarchive/archive_write.c @@ -1,867 +1,873 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" /* * This file contains the "essential" portions of the write API, that * is, stuff that will essentially always be used by any client that * actually needs to write an archive. Optional pieces have been, as * far as possible, separated out into separate files to reduce * needlessly bloating statically-linked clients. */ #ifdef HAVE_SYS_WAIT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include #ifdef HAVE_UNISTD_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_write_private.h" static int _archive_filter_code(struct archive *, int); static const char *_archive_filter_name(struct archive *, int); static int64_t _archive_filter_bytes(struct archive *, int); static int _archive_write_filter_count(struct archive *); static int _archive_write_close(struct archive *); static int _archive_write_free(struct archive *); static int _archive_write_header(struct archive *, struct archive_entry *); static int _archive_write_finish_entry(struct archive *); static ssize_t _archive_write_data(struct archive *, const void *, size_t); struct archive_none { size_t buffer_size; size_t avail; char *buffer; char *next; }; static const struct archive_vtable archive_write_vtable = { .archive_close = _archive_write_close, .archive_filter_bytes = _archive_filter_bytes, .archive_filter_code = _archive_filter_code, .archive_filter_name = _archive_filter_name, .archive_filter_count = _archive_write_filter_count, .archive_free = _archive_write_free, .archive_write_header = _archive_write_header, .archive_write_finish_entry = _archive_write_finish_entry, .archive_write_data = _archive_write_data, }; /* * Allocate, initialize and return an archive object. */ struct archive * archive_write_new(void) { struct archive_write *a; unsigned char *nulls; a = calloc(1, sizeof(*a)); if (a == NULL) return (NULL); a->archive.magic = ARCHIVE_WRITE_MAGIC; a->archive.state = ARCHIVE_STATE_NEW; a->archive.vtable = &archive_write_vtable; /* * The value 10240 here matches the traditional tar default, * but is otherwise arbitrary. * TODO: Set the default block size from the format selected. */ a->bytes_per_block = 10240; a->bytes_in_last_block = -1; /* Default */ /* Initialize a block of nulls for padding purposes. */ a->null_length = 1024; nulls = calloc(a->null_length, sizeof(unsigned char)); if (nulls == NULL) { free(a); return (NULL); } a->nulls = nulls; return (&a->archive); } /* * Set the block size. Returns 0 if successful. */ int archive_write_set_bytes_per_block(struct archive *_a, int bytes_per_block) { struct archive_write *a = (struct archive_write *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_bytes_per_block"); if (bytes_per_block < 0) { // Do nothing if the bytes_per_block is negative return 0; } a->bytes_per_block = bytes_per_block; return (ARCHIVE_OK); } /* * Get the current block size. */ int archive_write_get_bytes_per_block(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_ANY, "archive_write_get_bytes_per_block"); if (a->bytes_per_block < 0) { // Don't return a negative value return 1; } return (a->bytes_per_block); } /* * Set the size for the last block. * Returns 0 if successful. */ int archive_write_set_bytes_in_last_block(struct archive *_a, int bytes) { struct archive_write *a = (struct archive_write *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_ANY, "archive_write_set_bytes_in_last_block"); a->bytes_in_last_block = bytes; return (ARCHIVE_OK); } /* * Return the value set above. -1 indicates it has not been set. */ int archive_write_get_bytes_in_last_block(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_ANY, "archive_write_get_bytes_in_last_block"); return (a->bytes_in_last_block); } /* * dev/ino of a file to be rejected. Used to prevent adding * an archive to itself recursively. */ int archive_write_set_skip_file(struct archive *_a, la_int64_t d, la_int64_t i) { struct archive_write *a = (struct archive_write *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_ANY, "archive_write_set_skip_file"); a->skip_file_set = 1; a->skip_file_dev = d; a->skip_file_ino = i; return (ARCHIVE_OK); } /* * Allocate and return the next filter structure. */ struct archive_write_filter * __archive_write_allocate_filter(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct archive_write_filter *f; f = calloc(1, sizeof(*f)); if (f == NULL) return (NULL); f->archive = _a; f->state = ARCHIVE_WRITE_FILTER_STATE_NEW; if (a->filter_first == NULL) a->filter_first = f; else a->filter_last->next_filter = f; a->filter_last = f; return f; } /* * Write data to a particular filter. */ int __archive_write_filter(struct archive_write_filter *f, const void *buff, size_t length) { int r; /* Never write to non-open filters */ if (f->state != ARCHIVE_WRITE_FILTER_STATE_OPEN) return(ARCHIVE_FATAL); if (length == 0) return(ARCHIVE_OK); if (f->write == NULL) /* If unset, a fatal error has already occurred, so this filter * didn't open. We cannot write anything. */ return(ARCHIVE_FATAL); r = (f->write)(f, buff, length); f->bytes_written += length; return (r); } /* * Recursive function for opening the filter chain * Last filter is opened first */ static int __archive_write_open_filter(struct archive_write_filter *f) { int ret; ret = ARCHIVE_OK; if (f->next_filter != NULL) ret = __archive_write_open_filter(f->next_filter); if (ret != ARCHIVE_OK) return (ret); if (f->state != ARCHIVE_WRITE_FILTER_STATE_NEW) return (ARCHIVE_FATAL); if (f->open == NULL) { f->state = ARCHIVE_WRITE_FILTER_STATE_OPEN; return (ARCHIVE_OK); } ret = (f->open)(f); if (ret == ARCHIVE_OK) f->state = ARCHIVE_WRITE_FILTER_STATE_OPEN; else f->state = ARCHIVE_WRITE_FILTER_STATE_FATAL; return (ret); } /* * Open all filters */ static int __archive_write_filters_open(struct archive_write *a) { return (__archive_write_open_filter(a->filter_first)); } /* * Close all filters */ static int __archive_write_filters_close(struct archive_write *a) { struct archive_write_filter *f; int ret, ret1; ret = ARCHIVE_OK; for (f = a->filter_first; f != NULL; f = f->next_filter) { /* Do not close filters that are not open */ if (f->state == ARCHIVE_WRITE_FILTER_STATE_OPEN) { if (f->close != NULL) { ret1 = (f->close)(f); if (ret1 < ret) ret = ret1; if (ret1 == ARCHIVE_OK) { f->state = ARCHIVE_WRITE_FILTER_STATE_CLOSED; } else { f->state = ARCHIVE_WRITE_FILTER_STATE_FATAL; } } else f->state = ARCHIVE_WRITE_FILTER_STATE_CLOSED; } } return (ret); } int __archive_write_output(struct archive_write *a, const void *buff, size_t length) { return (__archive_write_filter(a->filter_first, buff, length)); } static int __archive_write_filters_flush(struct archive_write *a) { struct archive_write_filter *f; int ret, ret1; ret = ARCHIVE_OK; for (f = a->filter_first; f != NULL; f = f->next_filter) { if (f->flush != NULL && f->bytes_written > 0) { ret1 = (f->flush)(f); if (ret1 < ret) ret = ret1; if (ret1 < ARCHIVE_WARN) f->state = ARCHIVE_WRITE_FILTER_STATE_FATAL; } } return (ret); } int __archive_write_nulls(struct archive_write *a, size_t length) { if (length == 0) return (ARCHIVE_OK); while (length > 0) { size_t to_write = length < a->null_length ? length : a->null_length; int r = __archive_write_output(a, a->nulls, to_write); if (r < ARCHIVE_OK) return (r); length -= to_write; } return (ARCHIVE_OK); } static int archive_write_client_open(struct archive_write_filter *f) { struct archive_write *a = (struct archive_write *)f->archive; struct archive_none *state; void *buffer; size_t buffer_size; - int ret; f->bytes_per_block = archive_write_get_bytes_per_block(f->archive); f->bytes_in_last_block = archive_write_get_bytes_in_last_block(f->archive); buffer_size = f->bytes_per_block; state = calloc(1, sizeof(*state)); buffer = malloc(buffer_size); if (state == NULL || buffer == NULL) { free(state); free(buffer); archive_set_error(f->archive, ENOMEM, "Can't allocate data for output buffering"); return (ARCHIVE_FATAL); } state->buffer_size = buffer_size; state->buffer = buffer; state->next = state->buffer; state->avail = state->buffer_size; f->data = state; if (a->client_opener == NULL) return (ARCHIVE_OK); - ret = a->client_opener(f->archive, a->client_data); - if (ret != ARCHIVE_OK) { - free(state->buffer); - free(state); - f->data = NULL; - } - return (ret); + return (a->client_opener(f->archive, a->client_data)); } static int archive_write_client_write(struct archive_write_filter *f, const void *_buff, size_t length) { struct archive_write *a = (struct archive_write *)f->archive; struct archive_none *state = (struct archive_none *)f->data; const char *buff = (const char *)_buff; ssize_t remaining, to_copy; ssize_t bytes_written; remaining = length; /* * If there is no buffer for blocking, just pass the data * straight through to the client write callback. In * particular, this supports "no write delay" operation for * special applications. Just set the block size to zero. */ if (state->buffer_size == 0) { while (remaining > 0) { bytes_written = (a->client_writer)(&a->archive, a->client_data, buff, remaining); if (bytes_written <= 0) return (ARCHIVE_FATAL); remaining -= bytes_written; buff += bytes_written; } return (ARCHIVE_OK); } /* If the copy buffer isn't empty, try to fill it. */ if (state->avail < state->buffer_size) { /* If buffer is not empty... */ /* ... copy data into buffer ... */ to_copy = ((size_t)remaining > state->avail) ? state->avail : (size_t)remaining; memcpy(state->next, buff, to_copy); state->next += to_copy; state->avail -= to_copy; buff += to_copy; remaining -= to_copy; /* ... if it's full, write it out. */ if (state->avail == 0) { char *p = state->buffer; size_t to_write = state->buffer_size; while (to_write > 0) { bytes_written = (a->client_writer)(&a->archive, a->client_data, p, to_write); if (bytes_written <= 0) return (ARCHIVE_FATAL); if ((size_t)bytes_written > to_write) { archive_set_error(&(a->archive), -1, "write overrun"); return (ARCHIVE_FATAL); } p += bytes_written; to_write -= bytes_written; } state->next = state->buffer; state->avail = state->buffer_size; } } while ((size_t)remaining >= state->buffer_size) { /* Write out full blocks directly to client. */ bytes_written = (a->client_writer)(&a->archive, a->client_data, buff, state->buffer_size); if (bytes_written <= 0) return (ARCHIVE_FATAL); buff += bytes_written; remaining -= bytes_written; } if (remaining > 0) { /* Copy last bit into copy buffer. */ memcpy(state->next, buff, remaining); state->next += remaining; state->avail -= remaining; } return (ARCHIVE_OK); } static int archive_write_client_free(struct archive_write_filter *f) { struct archive_write *a = (struct archive_write *)f->archive; + struct archive_none *state = (struct archive_none *)f->data; if (a->client_freer) (*a->client_freer)(&a->archive, a->client_data); a->client_data = NULL; /* Clear passphrase. */ if (a->passphrase != NULL) { memset(a->passphrase, 0, strlen(a->passphrase)); free(a->passphrase); a->passphrase = NULL; } + /* Free state. */ + if (state != NULL) { + free(state->buffer); + free(state); + f->data = NULL; + } + return (ARCHIVE_OK); } static int archive_write_client_close(struct archive_write_filter *f) { struct archive_write *a = (struct archive_write *)f->archive; struct archive_none *state = (struct archive_none *)f->data; ssize_t block_length; ssize_t target_block_length; ssize_t bytes_written; size_t to_write; char *p; int ret = ARCHIVE_OK; /* If there's pending data, pad and write the last block */ if (state->next != state->buffer) { block_length = state->buffer_size - state->avail; /* Tricky calculation to determine size of last block */ if (a->bytes_in_last_block <= 0) /* Default or Zero: pad to full block */ target_block_length = a->bytes_per_block; else /* Round to next multiple of bytes_in_last_block. */ target_block_length = a->bytes_in_last_block * ( (block_length + a->bytes_in_last_block - 1) / a->bytes_in_last_block); if (target_block_length > a->bytes_per_block) target_block_length = a->bytes_per_block; if (block_length < target_block_length) { memset(state->next, 0, target_block_length - block_length); block_length = target_block_length; } p = state->buffer; to_write = block_length; while (to_write > 0) { bytes_written = (a->client_writer)(&a->archive, a->client_data, p, to_write); if (bytes_written <= 0) { ret = ARCHIVE_FATAL; break; } if ((size_t)bytes_written > to_write) { archive_set_error(&(a->archive), -1, "write overrun"); ret = ARCHIVE_FATAL; break; } p += bytes_written; to_write -= bytes_written; } } if (a->client_closer) (*a->client_closer)(&a->archive, a->client_data); - free(state->buffer); - free(state); /* Clear the close handler myself not to be called again. */ f->state = ARCHIVE_WRITE_FILTER_STATE_CLOSED; return (ret); } /* * Open the archive using the current settings. */ int archive_write_open2(struct archive *_a, void *client_data, archive_open_callback *opener, archive_write_callback *writer, archive_close_callback *closer, archive_free_callback *freer) { struct archive_write *a = (struct archive_write *)_a; struct archive_write_filter *client_filter; int ret, r1; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_open"); archive_clear_error(&a->archive); a->client_writer = writer; a->client_opener = opener; a->client_closer = closer; a->client_freer = freer; a->client_data = client_data; client_filter = __archive_write_allocate_filter(_a); if (client_filter == NULL) return (ARCHIVE_FATAL); client_filter->open = archive_write_client_open; client_filter->write = archive_write_client_write; client_filter->close = archive_write_client_close; client_filter->free = archive_write_client_free; ret = __archive_write_filters_open(a); if (ret < ARCHIVE_WARN) { r1 = __archive_write_filters_close(a); __archive_write_filters_free(_a); return (r1 < ret ? r1 : ret); } a->archive.state = ARCHIVE_STATE_HEADER; if (a->format_init) ret = (a->format_init)(a); return (ret); } int archive_write_open(struct archive *_a, void *client_data, archive_open_callback *opener, archive_write_callback *writer, archive_close_callback *closer) { return archive_write_open2(_a, client_data, opener, writer, closer, NULL); } /* * Close out the archive. */ static int _archive_write_close(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int r = ARCHIVE_OK, r1 = ARCHIVE_OK; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_write_close"); if (a->archive.state == ARCHIVE_STATE_NEW || a->archive.state == ARCHIVE_STATE_CLOSED) return (ARCHIVE_OK); /* Okay to close() when not open. */ archive_clear_error(&a->archive); /* Finish the last entry if a finish callback is specified */ if (a->archive.state == ARCHIVE_STATE_DATA && a->format_finish_entry != NULL) r = ((a->format_finish_entry)(a)); /* Finish off the archive. */ /* TODO: have format closers invoke compression close. */ if (a->format_close != NULL) { r1 = (a->format_close)(a); if (r1 < r) r = r1; } /* Finish the compression and close the stream. */ r1 = __archive_write_filters_close(a); if (r1 < r) r = r1; if (a->archive.state != ARCHIVE_STATE_FATAL) a->archive.state = ARCHIVE_STATE_CLOSED; return (r); } static int _archive_write_filter_count(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct archive_write_filter *p = a->filter_first; int count = 0; while(p) { count++; p = p->next_filter; } return count; } void __archive_write_filters_free(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int r = ARCHIVE_OK, r1; while (a->filter_first != NULL) { struct archive_write_filter *next = a->filter_first->next_filter; if (a->filter_first->free != NULL) { r1 = (*a->filter_first->free)(a->filter_first); if (r > r1) r = r1; } free(a->filter_first); a->filter_first = next; } a->filter_last = NULL; } /* * Destroy the archive structure. * * Be careful: user might just call write_new and then write_free. * Don't assume we actually wrote anything or performed any non-trivial * initialization. */ static int _archive_write_free(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int r = ARCHIVE_OK, r1; if (_a == NULL) return (ARCHIVE_OK); /* It is okay to call free() in state FATAL. */ archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_write_free"); if (a->archive.state != ARCHIVE_STATE_FATAL) r = archive_write_close(&a->archive); /* Release format resources. */ if (a->format_free != NULL) { r1 = (a->format_free)(a); if (r1 < r) r = r1; } __archive_write_filters_free(_a); /* Release various dynamic buffers. */ free((void *)(uintptr_t)(const void *)a->nulls); archive_string_free(&a->archive.error_string); if (a->passphrase != NULL) { /* A passphrase should be cleaned. */ memset(a->passphrase, 0, strlen(a->passphrase)); free(a->passphrase); } a->archive.magic = 0; __archive_clean(&a->archive); free(a); return (r); } /* * Write the appropriate header. */ static int _archive_write_header(struct archive *_a, struct archive_entry *entry) { struct archive_write *a = (struct archive_write *)_a; int ret, r2; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_DATA | ARCHIVE_STATE_HEADER, "archive_write_header"); archive_clear_error(&a->archive); if (a->format_write_header == NULL) { archive_set_error(&(a->archive), -1, "Format must be set before you can write to an archive."); a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } /* In particular, "retry" and "fatal" get returned immediately. */ ret = archive_write_finish_entry(&a->archive); if (ret == ARCHIVE_FATAL) { a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } if (ret < ARCHIVE_OK && ret != ARCHIVE_WARN) return (ret); if (a->skip_file_set && archive_entry_dev_is_set(entry) && archive_entry_ino_is_set(entry) && archive_entry_dev(entry) == (dev_t)a->skip_file_dev && archive_entry_ino64(entry) == a->skip_file_ino) { archive_set_error(&a->archive, 0, "Can't add archive to itself"); return (ARCHIVE_FAILED); } /* Flush filters at boundary. */ r2 = __archive_write_filters_flush(a); if (r2 == ARCHIVE_FAILED) { return (ARCHIVE_FAILED); } if (r2 == ARCHIVE_FATAL) { a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } if (r2 < ret) ret = r2; /* Format and write header. */ r2 = ((a->format_write_header)(a, entry)); if (r2 == ARCHIVE_FAILED) { return (ARCHIVE_FAILED); } if (r2 == ARCHIVE_FATAL) { a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } if (r2 < ret) ret = r2; a->archive.state = ARCHIVE_STATE_DATA; return (ret); } static int _archive_write_finish_entry(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int ret = ARCHIVE_OK; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_finish_entry"); if (a->archive.state & ARCHIVE_STATE_DATA && a->format_finish_entry != NULL) ret = (a->format_finish_entry)(a); - a->archive.state = ARCHIVE_STATE_HEADER; + if (ret == ARCHIVE_FATAL) + a->archive.state = ARCHIVE_STATE_FATAL; + else + a->archive.state = ARCHIVE_STATE_HEADER; return (ret); } /* * Note that the compressor is responsible for blocking. */ static ssize_t _archive_write_data(struct archive *_a, const void *buff, size_t s) { struct archive_write *a = (struct archive_write *)_a; const size_t max_write = INT_MAX; + int ret; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_DATA, "archive_write_data"); /* In particular, this catches attempts to pass negative values. */ if (s > max_write) s = max_write; archive_clear_error(&a->archive); - return ((a->format_write_data)(a, buff, s)); + ret = (a->format_write_data)(a, buff, s); + if (ret == ARCHIVE_FATAL) + a->archive.state = ARCHIVE_STATE_FATAL; + return (ret); } static struct archive_write_filter * filter_lookup(struct archive *_a, int n) { struct archive_write *a = (struct archive_write *)_a; struct archive_write_filter *f = a->filter_first; if (n == -1) return a->filter_last; if (n < 0) return NULL; while (n > 0 && f != NULL) { f = f->next_filter; --n; } return f; } static int _archive_filter_code(struct archive *_a, int n) { struct archive_write_filter *f = filter_lookup(_a, n); return f == NULL ? -1 : f->code; } static const char * _archive_filter_name(struct archive *_a, int n) { struct archive_write_filter *f = filter_lookup(_a, n); return f != NULL ? f->name : NULL; } static int64_t _archive_filter_bytes(struct archive *_a, int n) { struct archive_write_filter *f = filter_lookup(_a, n); return f == NULL ? -1 : f->bytes_written; } diff --git a/contrib/libarchive/libarchive/archive_write_add_filter_bzip2.c b/contrib/libarchive/libarchive/archive_write_add_filter_bzip2.c index 0726f08936ec..2434528d5133 100644 --- a/contrib/libarchive/libarchive/archive_write_add_filter_bzip2.c +++ b/contrib/libarchive/libarchive/archive_write_add_filter_bzip2.c @@ -1,398 +1,402 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * Copyright (c) 2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_BZLIB_H #include #endif #include "archive.h" #include "archive_private.h" #include "archive_write_private.h" #if ARCHIVE_VERSION_NUMBER < 4000000 int archive_write_set_compression_bzip2(struct archive *a) { __archive_write_filters_free(a); return (archive_write_add_filter_bzip2(a)); } #endif struct private_data { int compression_level; #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) bz_stream stream; int64_t total_in; char *compressed; size_t compressed_buffer_size; #else struct archive_write_program_data *pdata; #endif }; static int archive_compressor_bzip2_close(struct archive_write_filter *); static int archive_compressor_bzip2_free(struct archive_write_filter *); static int archive_compressor_bzip2_open(struct archive_write_filter *); static int archive_compressor_bzip2_options(struct archive_write_filter *, const char *, const char *); static int archive_compressor_bzip2_write(struct archive_write_filter *, const void *, size_t); /* * Add a bzip2 compression filter to this write handle. */ int archive_write_add_filter_bzip2(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct archive_write_filter *f = __archive_write_allocate_filter(_a); struct private_data *data; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_add_filter_bzip2"); data = calloc(1, sizeof(*data)); if (data == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } data->compression_level = 9; /* default */ f->data = data; f->options = &archive_compressor_bzip2_options; f->close = &archive_compressor_bzip2_close; f->free = &archive_compressor_bzip2_free; f->open = &archive_compressor_bzip2_open; f->code = ARCHIVE_FILTER_BZIP2; f->name = "bzip2"; #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) return (ARCHIVE_OK); #else data->pdata = __archive_write_program_allocate("bzip2"); if (data->pdata == NULL) { free(data); archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } data->compression_level = 0; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Using external bzip2 program"); return (ARCHIVE_WARN); #endif } /* * Set write options. */ static int archive_compressor_bzip2_options(struct archive_write_filter *f, const char *key, const char *value) { struct private_data *data = (struct private_data *)f->data; if (strcmp(key, "compression-level") == 0) { if (value == NULL || !(value[0] >= '0' && value[0] <= '9') || value[1] != '\0') return (ARCHIVE_WARN); data->compression_level = value[0] - '0'; /* Make '0' be a synonym for '1'. */ /* This way, bzip2 compressor supports the same 0..9 * range of levels as gzip. */ if (data->compression_level < 1) data->compression_level = 1; return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) /* Don't compile this if we don't have bzlib. */ /* * Yuck. bzlib.h is not const-correct, so I need this one bit * of ugly hackery to convert a const * pointer to a non-const pointer. */ #define SET_NEXT_IN(st,src) \ (st)->stream.next_in = (char *)(uintptr_t)(const void *)(src) static int drive_compressor(struct archive_write_filter *, struct private_data *, int finishing); /* * Setup callback. */ static int archive_compressor_bzip2_open(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; int ret; if (data->compressed == NULL) { size_t bs = 65536, bpb; if (f->archive->magic == ARCHIVE_WRITE_MAGIC) { /* Buffer size should be a multiple number of the bytes * per block for performance. */ bpb = archive_write_get_bytes_per_block(f->archive); if (bpb > bs) bs = bpb; else if (bpb != 0) bs -= bs % bpb; } data->compressed_buffer_size = bs; data->compressed = malloc(data->compressed_buffer_size); if (data->compressed == NULL) { archive_set_error(f->archive, ENOMEM, "Can't allocate data for compression buffer"); return (ARCHIVE_FATAL); } } memset(&data->stream, 0, sizeof(data->stream)); data->stream.next_out = data->compressed; data->stream.avail_out = (uint32_t)data->compressed_buffer_size; f->write = archive_compressor_bzip2_write; /* Initialize compression library */ ret = BZ2_bzCompressInit(&(data->stream), data->compression_level, 0, 30); if (ret == BZ_OK) { f->data = data; return (ARCHIVE_OK); } /* Library setup failed: clean up. */ archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library"); /* Override the error message if we know what really went wrong. */ switch (ret) { case BZ_PARAM_ERROR: archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library: " "invalid setup parameter"); break; case BZ_MEM_ERROR: archive_set_error(f->archive, ENOMEM, "Internal error initializing compression library: " "out of memory"); break; case BZ_CONFIG_ERROR: archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library: " "mis-compiled library"); break; } return (ARCHIVE_FATAL); } /* * Write data to the compressed stream. * * Returns ARCHIVE_OK if all data written, error otherwise. */ static int archive_compressor_bzip2_write(struct archive_write_filter *f, const void *buff, size_t length) { struct private_data *data = (struct private_data *)f->data; /* Update statistics */ data->total_in += length; /* Compress input data to output buffer */ SET_NEXT_IN(data, buff); data->stream.avail_in = (uint32_t)length; if (drive_compressor(f, data, 0)) return (ARCHIVE_FATAL); return (ARCHIVE_OK); } /* * Finish the compression. */ static int archive_compressor_bzip2_close(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; int ret; /* Finish compression cycle. */ ret = drive_compressor(f, data, 1); if (ret == ARCHIVE_OK) { /* Write the last block */ ret = __archive_write_filter(f->next_filter, data->compressed, data->compressed_buffer_size - data->stream.avail_out); } switch (BZ2_bzCompressEnd(&(data->stream))) { case BZ_OK: break; default: archive_set_error(f->archive, ARCHIVE_ERRNO_PROGRAMMER, "Failed to clean up compressor"); ret = ARCHIVE_FATAL; } return ret; } static int archive_compressor_bzip2_free(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; + + /* May already have been called, but not necessarily. */ + (void)BZ2_bzCompressEnd(&(data->stream)); + free(data->compressed); free(data); f->data = NULL; return (ARCHIVE_OK); } /* * Utility function to push input data through compressor, writing * full output blocks as necessary. * * Note that this handles both the regular write case (finishing == * false) and the end-of-archive case (finishing == true). */ static int drive_compressor(struct archive_write_filter *f, struct private_data *data, int finishing) { int ret; for (;;) { if (data->stream.avail_out == 0) { ret = __archive_write_filter(f->next_filter, data->compressed, data->compressed_buffer_size); if (ret != ARCHIVE_OK) { /* TODO: Handle this write failure */ return (ARCHIVE_FATAL); } data->stream.next_out = data->compressed; data->stream.avail_out = (uint32_t)data->compressed_buffer_size; } /* If there's nothing to do, we're done. */ if (!finishing && data->stream.avail_in == 0) return (ARCHIVE_OK); ret = BZ2_bzCompress(&(data->stream), finishing ? BZ_FINISH : BZ_RUN); switch (ret) { case BZ_RUN_OK: /* In non-finishing case, did compressor * consume everything? */ if (!finishing && data->stream.avail_in == 0) return (ARCHIVE_OK); break; case BZ_FINISH_OK: /* Finishing: There's more work to do */ break; case BZ_STREAM_END: /* Finishing: all done */ /* Only occurs in finishing case */ return (ARCHIVE_OK); default: /* Any other return value indicates an error */ archive_set_error(f->archive, ARCHIVE_ERRNO_PROGRAMMER, "Bzip2 compression failed;" " BZ2_bzCompress() returned %d", ret); return (ARCHIVE_FATAL); } } } #else /* HAVE_BZLIB_H && BZ_CONFIG_ERROR */ static int archive_compressor_bzip2_open(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; struct archive_string as; int r; archive_string_init(&as); archive_strcpy(&as, "bzip2"); /* Specify compression level. */ if (data->compression_level > 0) { archive_strcat(&as, " -"); archive_strappend_char(&as, '0' + data->compression_level); } f->write = archive_compressor_bzip2_write; r = __archive_write_program_open(f, data->pdata, as.s); archive_string_free(&as); return (r); } static int archive_compressor_bzip2_write(struct archive_write_filter *f, const void *buff, size_t length) { struct private_data *data = (struct private_data *)f->data; return __archive_write_program_write(f, data->pdata, buff, length); } static int archive_compressor_bzip2_close(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; return __archive_write_program_close(f, data->pdata); } static int archive_compressor_bzip2_free(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; __archive_write_program_free(data->pdata); free(data); return (ARCHIVE_OK); } #endif /* HAVE_BZLIB_H && BZ_CONFIG_ERROR */ diff --git a/contrib/libarchive/libarchive/archive_write_add_filter_gzip.c b/contrib/libarchive/libarchive/archive_write_add_filter_gzip.c index 5ef43c1936ed..b09e669b753d 100644 --- a/contrib/libarchive/libarchive/archive_write_add_filter_gzip.c +++ b/contrib/libarchive/libarchive/archive_write_add_filter_gzip.c @@ -1,456 +1,479 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include #ifdef HAVE_ZLIB_H #include #endif #include "archive.h" #include "archive_private.h" #include "archive_string.h" #include "archive_write_private.h" #if ARCHIVE_VERSION_NUMBER < 4000000 int archive_write_set_compression_gzip(struct archive *a) { __archive_write_filters_free(a); return (archive_write_add_filter_gzip(a)); } #endif /* Don't compile this if we don't have zlib. */ struct private_data { int compression_level; int timestamp; char *original_filename; #ifdef HAVE_ZLIB_H z_stream stream; int64_t total_in; unsigned char *compressed; size_t compressed_buffer_size; unsigned long crc; #else struct archive_write_program_data *pdata; #endif }; /* * Yuck. zlib.h is not const-correct, so I need this one bit * of ugly hackery to convert a const * pointer to a non-const pointer. */ #define SET_NEXT_IN(st,src) \ (st)->stream.next_in = (Bytef *)(uintptr_t)(const void *)(src) static int archive_compressor_gzip_options(struct archive_write_filter *, const char *, const char *); static int archive_compressor_gzip_open(struct archive_write_filter *); static int archive_compressor_gzip_write(struct archive_write_filter *, const void *, size_t); static int archive_compressor_gzip_close(struct archive_write_filter *); static int archive_compressor_gzip_free(struct archive_write_filter *); #ifdef HAVE_ZLIB_H static int drive_compressor(struct archive_write_filter *, struct private_data *, int finishing); #endif /* * Add a gzip compression filter to this write handle. */ int archive_write_add_filter_gzip(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct archive_write_filter *f = __archive_write_allocate_filter(_a); struct private_data *data; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_add_filter_gzip"); data = calloc(1, sizeof(*data)); if (data == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } f->data = data; f->open = &archive_compressor_gzip_open; f->options = &archive_compressor_gzip_options; f->close = &archive_compressor_gzip_close; f->free = &archive_compressor_gzip_free; f->code = ARCHIVE_FILTER_GZIP; f->name = "gzip"; data->original_filename = NULL; #ifdef HAVE_ZLIB_H data->compression_level = Z_DEFAULT_COMPRESSION; return (ARCHIVE_OK); #else data->pdata = __archive_write_program_allocate("gzip"); if (data->pdata == NULL) { free(data); archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } data->compression_level = 0; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Using external gzip program"); return (ARCHIVE_WARN); #endif } static int archive_compressor_gzip_free(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; #ifdef HAVE_ZLIB_H free(data->compressed); #else __archive_write_program_free(data->pdata); #endif free((void*)data->original_filename); free(data); f->data = NULL; return (ARCHIVE_OK); } /* * Set write options. */ static int archive_compressor_gzip_options(struct archive_write_filter *f, const char *key, const char *value) { struct private_data *data = (struct private_data *)f->data; if (strcmp(key, "compression-level") == 0) { if (value == NULL || !(value[0] >= '0' && value[0] <= '9') || value[1] != '\0') return (ARCHIVE_WARN); data->compression_level = value[0] - '0'; return (ARCHIVE_OK); } if (strcmp(key, "timestamp") == 0) { data->timestamp = (value == NULL)?-1:1; return (ARCHIVE_OK); } if (strcmp(key, "original-filename") == 0) { free((void*)data->original_filename); data->original_filename = NULL; if (value) data->original_filename = strdup(value); return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } #ifdef HAVE_ZLIB_H /* * Setup callback. */ static int archive_compressor_gzip_open(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; - int ret; + int ret = ARCHIVE_OK; + int init_success; if (data->compressed == NULL) { size_t bs = 65536, bpb; if (f->archive->magic == ARCHIVE_WRITE_MAGIC) { /* Buffer size should be a multiple number of * the of bytes per block for performance. */ bpb = archive_write_get_bytes_per_block(f->archive); if (bpb > bs) bs = bpb; else if (bpb != 0) bs -= bs % bpb; } data->compressed_buffer_size = bs; data->compressed = malloc(data->compressed_buffer_size); if (data->compressed == NULL) { archive_set_error(f->archive, ENOMEM, "Can't allocate data for compression buffer"); return (ARCHIVE_FATAL); } } data->crc = crc32(0L, NULL, 0); data->stream.next_out = data->compressed; data->stream.avail_out = (uInt)data->compressed_buffer_size; /* Prime output buffer with a gzip header. */ data->compressed[0] = 0x1f; /* GZip signature bytes */ data->compressed[1] = 0x8b; data->compressed[2] = 0x08; /* "Deflate" compression */ - data->compressed[3] = data->original_filename == NULL ? 0 : 0x8; + data->compressed[3] = 0x00; /* Flags */ if (data->timestamp >= 0) { time_t t = time(NULL); data->compressed[4] = (uint8_t)(t)&0xff; /* Timestamp */ data->compressed[5] = (uint8_t)(t>>8)&0xff; data->compressed[6] = (uint8_t)(t>>16)&0xff; data->compressed[7] = (uint8_t)(t>>24)&0xff; - } else + } else { memset(&data->compressed[4], 0, 4); - if (data->compression_level == 9) - data->compressed[8] = 2; - else if(data->compression_level == 1) - data->compressed[8] = 4; - else - data->compressed[8] = 0; + } + if (data->compression_level == 9) { + data->compressed[8] = 2; + } else if(data->compression_level == 1) { + data->compressed[8] = 4; + } else { + data->compressed[8] = 0; + } data->compressed[9] = 3; /* OS=Unix */ data->stream.next_out += 10; data->stream.avail_out -= 10; if (data->original_filename != NULL) { - strcpy((char*)data->compressed + 10, data->original_filename); - data->stream.next_out += strlen(data->original_filename) + 1; - data->stream.avail_out -= strlen(data->original_filename) + 1; + /* Limit "original filename" to 32k or the + * remaining space in the buffer, whichever is smaller. + */ + int ofn_length = strlen(data->original_filename); + int ofn_max_length = 32768; + int ofn_space_available = data->compressed + + data->compressed_buffer_size + - data->stream.next_out + - 1; + if (ofn_max_length > ofn_space_available) { + ofn_max_length = ofn_space_available; + } + if (ofn_length < ofn_max_length) { + data->compressed[3] |= 0x8; + strcpy((char*)data->compressed + 10, + data->original_filename); + data->stream.next_out += ofn_length + 1; + data->stream.avail_out -= ofn_length + 1; + } else { + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "Gzip 'Original Filename' ignored because it is too long"); + ret = ARCHIVE_WARN; + } } f->write = archive_compressor_gzip_write; /* Initialize compression library. */ - ret = deflateInit2(&(data->stream), + init_success = deflateInit2(&(data->stream), data->compression_level, Z_DEFLATED, -15 /* < 0 to suppress zlib header */, 8, Z_DEFAULT_STRATEGY); - if (ret == Z_OK) { + if (init_success == Z_OK) { f->data = data; - return (ARCHIVE_OK); + return (ret); } /* Library setup failed: clean up. */ archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Internal error " "initializing compression library"); /* Override the error message if we know what really went wrong. */ - switch (ret) { + switch (init_success) { case Z_STREAM_ERROR: archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Internal error initializing " "compression library: invalid setup parameter"); break; case Z_MEM_ERROR: archive_set_error(f->archive, ENOMEM, "Internal error initializing compression library"); break; case Z_VERSION_ERROR: archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Internal error initializing " "compression library: invalid library version"); break; } return (ARCHIVE_FATAL); } /* * Write data to the compressed stream. */ static int archive_compressor_gzip_write(struct archive_write_filter *f, const void *buff, size_t length) { struct private_data *data = (struct private_data *)f->data; int ret; /* Update statistics */ data->crc = crc32(data->crc, (const Bytef *)buff, (uInt)length); data->total_in += length; /* Compress input data to output buffer */ SET_NEXT_IN(data, buff); data->stream.avail_in = (uInt)length; if ((ret = drive_compressor(f, data, 0)) != ARCHIVE_OK) return (ret); return (ARCHIVE_OK); } /* * Finish the compression... */ static int archive_compressor_gzip_close(struct archive_write_filter *f) { unsigned char trailer[8]; struct private_data *data = (struct private_data *)f->data; int ret; /* Finish compression cycle */ ret = drive_compressor(f, data, 1); if (ret == ARCHIVE_OK) { /* Write the last compressed data. */ ret = __archive_write_filter(f->next_filter, data->compressed, data->compressed_buffer_size - data->stream.avail_out); } if (ret == ARCHIVE_OK) { /* Build and write out 8-byte trailer. */ trailer[0] = (uint8_t)(data->crc)&0xff; trailer[1] = (uint8_t)(data->crc >> 8)&0xff; trailer[2] = (uint8_t)(data->crc >> 16)&0xff; trailer[3] = (uint8_t)(data->crc >> 24)&0xff; trailer[4] = (uint8_t)(data->total_in)&0xff; trailer[5] = (uint8_t)(data->total_in >> 8)&0xff; trailer[6] = (uint8_t)(data->total_in >> 16)&0xff; trailer[7] = (uint8_t)(data->total_in >> 24)&0xff; ret = __archive_write_filter(f->next_filter, trailer, 8); } switch (deflateEnd(&(data->stream))) { case Z_OK: break; default: archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Failed to clean up compressor"); ret = ARCHIVE_FATAL; } return ret; } /* * Utility function to push input data through compressor, * writing full output blocks as necessary. * * Note that this handles both the regular write case (finishing == * false) and the end-of-archive case (finishing == true). */ static int drive_compressor(struct archive_write_filter *f, struct private_data *data, int finishing) { int ret; for (;;) { if (data->stream.avail_out == 0) { ret = __archive_write_filter(f->next_filter, data->compressed, data->compressed_buffer_size); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); data->stream.next_out = data->compressed; data->stream.avail_out = (uInt)data->compressed_buffer_size; } /* If there's nothing to do, we're done. */ if (!finishing && data->stream.avail_in == 0) return (ARCHIVE_OK); ret = deflate(&(data->stream), finishing ? Z_FINISH : Z_NO_FLUSH ); switch (ret) { case Z_OK: /* In non-finishing case, check if compressor * consumed everything */ if (!finishing && data->stream.avail_in == 0) return (ARCHIVE_OK); /* In finishing case, this return always means * there's more work */ break; case Z_STREAM_END: /* This return can only occur in finishing case. */ return (ARCHIVE_OK); default: /* Any other return value indicates an error. */ archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "GZip compression failed:" " deflate() call returned status %d", ret); return (ARCHIVE_FATAL); } } } #else /* HAVE_ZLIB_H */ static int archive_compressor_gzip_open(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; struct archive_string as; int r; archive_string_init(&as); archive_strcpy(&as, "gzip"); /* Specify compression level. */ if (data->compression_level > 0) { archive_strcat(&as, " -"); archive_strappend_char(&as, '0' + data->compression_level); } if (data->timestamp < 0) /* Do not save timestamp. */ archive_strcat(&as, " -n"); else if (data->timestamp > 0) /* Save timestamp. */ archive_strcat(&as, " -N"); f->write = archive_compressor_gzip_write; r = __archive_write_program_open(f, data->pdata, as.s); archive_string_free(&as); return (r); } static int archive_compressor_gzip_write(struct archive_write_filter *f, const void *buff, size_t length) { struct private_data *data = (struct private_data *)f->data; return __archive_write_program_write(f, data->pdata, buff, length); } static int archive_compressor_gzip_close(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; return __archive_write_program_close(f, data->pdata); } #endif /* HAVE_ZLIB_H */ diff --git a/contrib/libarchive/libarchive/archive_write_add_filter_program.c b/contrib/libarchive/libarchive/archive_write_add_filter_program.c index c661cc7f412f..f12db3373883 100644 --- a/contrib/libarchive/libarchive/archive_write_add_filter_program.c +++ b/contrib/libarchive/libarchive/archive_write_add_filter_program.c @@ -1,390 +1,389 @@ /*- * Copyright (c) 2007 Joerg Sonnenberger * Copyright (c) 2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_WAIT_H # include #endif #ifdef HAVE_ERRNO_H # include #endif #ifdef HAVE_FCNTL_H # include #endif #ifdef HAVE_STDLIB_H # include #endif #ifdef HAVE_STRING_H # include #endif #include "archive.h" #include "archive_private.h" #include "archive_string.h" #include "archive_write_private.h" #include "filter_fork.h" #if ARCHIVE_VERSION_NUMBER < 4000000 int archive_write_set_compression_program(struct archive *a, const char *cmd) { __archive_write_filters_free(a); return (archive_write_add_filter_program(a, cmd)); } #endif struct archive_write_program_data { #if defined(_WIN32) && !defined(__CYGWIN__) HANDLE child; #else pid_t child; #endif int child_stdin, child_stdout; char *child_buf; size_t child_buf_len, child_buf_avail; char *program_name; }; struct private_data { struct archive_write_program_data *pdata; struct archive_string description; char *cmd; }; static int archive_compressor_program_open(struct archive_write_filter *); static int archive_compressor_program_write(struct archive_write_filter *, const void *, size_t); static int archive_compressor_program_close(struct archive_write_filter *); static int archive_compressor_program_free(struct archive_write_filter *); /* * Add a filter to this write handle that passes all data through an * external program. */ int archive_write_add_filter_program(struct archive *_a, const char *cmd) { struct archive_write_filter *f = __archive_write_allocate_filter(_a); struct private_data *data; static const char prefix[] = "Program: "; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_add_filter_program"); f->data = calloc(1, sizeof(*data)); if (f->data == NULL) goto memerr; data = (struct private_data *)f->data; data->cmd = strdup(cmd); if (data->cmd == NULL) goto memerr; data->pdata = __archive_write_program_allocate(cmd); if (data->pdata == NULL) goto memerr; /* Make up a description string. */ if (archive_string_ensure(&data->description, strlen(prefix) + strlen(cmd) + 1) == NULL) goto memerr; archive_strcpy(&data->description, prefix); archive_strcat(&data->description, cmd); f->name = data->description.s; f->code = ARCHIVE_FILTER_PROGRAM; f->open = archive_compressor_program_open; f->write = archive_compressor_program_write; f->close = archive_compressor_program_close; f->free = archive_compressor_program_free; return (ARCHIVE_OK); memerr: archive_compressor_program_free(f); archive_set_error(_a, ENOMEM, "Can't allocate memory for filter program"); return (ARCHIVE_FATAL); } static int archive_compressor_program_open(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; return __archive_write_program_open(f, data->pdata, data->cmd); } static int archive_compressor_program_write(struct archive_write_filter *f, const void *buff, size_t length) { struct private_data *data = (struct private_data *)f->data; return __archive_write_program_write(f, data->pdata, buff, length); } static int archive_compressor_program_close(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; return __archive_write_program_close(f, data->pdata); } static int archive_compressor_program_free(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; if (data) { free(data->cmd); archive_string_free(&data->description); __archive_write_program_free(data->pdata); free(data); f->data = NULL; } return (ARCHIVE_OK); } /* * Allocate resources for executing an external program. */ struct archive_write_program_data * __archive_write_program_allocate(const char *program) { struct archive_write_program_data *data; data = calloc(1, sizeof(struct archive_write_program_data)); if (data == NULL) return (data); data->child_stdin = -1; data->child_stdout = -1; data->program_name = strdup(program); return (data); } /* * Release the resources. */ int __archive_write_program_free(struct archive_write_program_data *data) { if (data) { free(data->program_name); free(data->child_buf); free(data); } return (ARCHIVE_OK); } int __archive_write_program_open(struct archive_write_filter *f, struct archive_write_program_data *data, const char *cmd) { int ret; if (data->child_buf == NULL) { data->child_buf_len = 65536; data->child_buf_avail = 0; data->child_buf = malloc(data->child_buf_len); if (data->child_buf == NULL) { archive_set_error(f->archive, ENOMEM, "Can't allocate compression buffer"); return (ARCHIVE_FATAL); } } ret = __archive_create_child(cmd, &data->child_stdin, &data->child_stdout, &data->child); if (ret != ARCHIVE_OK) { archive_set_error(f->archive, EINVAL, "Can't launch external program: %s", cmd); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } static ssize_t child_write(struct archive_write_filter *f, struct archive_write_program_data *data, const char *buf, size_t buf_len) { ssize_t ret; if (data->child_stdin == -1) return (-1); if (buf_len == 0) return (-1); for (;;) { do { ret = write(data->child_stdin, buf, buf_len); } while (ret == -1 && errno == EINTR); if (ret > 0) return (ret); if (ret == 0) { close(data->child_stdin); data->child_stdin = -1; fcntl(data->child_stdout, F_SETFL, 0); return (0); } if (ret == -1 && errno != EAGAIN) return (-1); if (data->child_stdout == -1) { fcntl(data->child_stdin, F_SETFL, 0); __archive_check_child(data->child_stdin, data->child_stdout); continue; } do { ret = read(data->child_stdout, data->child_buf + data->child_buf_avail, data->child_buf_len - data->child_buf_avail); } while (ret == -1 && errno == EINTR); if (ret == 0 || (ret == -1 && errno == EPIPE)) { close(data->child_stdout); data->child_stdout = -1; fcntl(data->child_stdin, F_SETFL, 0); continue; } if (ret == -1 && errno == EAGAIN) { __archive_check_child(data->child_stdin, data->child_stdout); continue; } if (ret == -1) return (-1); data->child_buf_avail += ret; ret = __archive_write_filter(f->next_filter, data->child_buf, data->child_buf_avail); if (ret != ARCHIVE_OK) return (-1); data->child_buf_avail = 0; } } /* * Write data to the filter stream. */ int __archive_write_program_write(struct archive_write_filter *f, struct archive_write_program_data *data, const void *buff, size_t length) { ssize_t ret; const char *buf; if (data->child == 0) return (ARCHIVE_OK); buf = buff; while (length > 0) { ret = child_write(f, data, buf, length); if (ret == -1 || ret == 0) { archive_set_error(f->archive, EIO, "Can't write to program: %s", data->program_name); return (ARCHIVE_FATAL); } length -= ret; buf += ret; } return (ARCHIVE_OK); } /* * Finish the filtering... */ int __archive_write_program_close(struct archive_write_filter *f, struct archive_write_program_data *data) { int ret, status; + pid_t pid; ssize_t bytes_read; if (data->child == 0) return ARCHIVE_OK; ret = 0; close(data->child_stdin); data->child_stdin = -1; fcntl(data->child_stdout, F_SETFL, 0); for (;;) { do { bytes_read = read(data->child_stdout, data->child_buf + data->child_buf_avail, data->child_buf_len - data->child_buf_avail); } while (bytes_read == -1 && errno == EINTR); if (bytes_read == 0 || (bytes_read == -1 && errno == EPIPE)) break; if (bytes_read == -1) { archive_set_error(f->archive, errno, "Error reading from program: %s", data->program_name); ret = ARCHIVE_FATAL; goto cleanup; } data->child_buf_avail += bytes_read; ret = __archive_write_filter(f->next_filter, data->child_buf, data->child_buf_avail); if (ret != ARCHIVE_OK) { ret = ARCHIVE_FATAL; goto cleanup; } data->child_buf_avail = 0; } cleanup: /* Shut down the child. */ if (data->child_stdin != -1) close(data->child_stdin); if (data->child_stdout != -1) close(data->child_stdout); - while (waitpid(data->child, &status, 0) == -1 && errno == EINTR) - continue; -#if defined(_WIN32) && !defined(__CYGWIN__) - CloseHandle(data->child); -#endif + do { + pid = waitpid(data->child, &status, 0); + } while (pid == -1 && errno == EINTR); data->child = 0; - if (status != 0) { + if (pid < 0 || status != 0) { archive_set_error(f->archive, EIO, "Error closing program: %s", data->program_name); ret = ARCHIVE_FATAL; } return ret; } diff --git a/contrib/libarchive/libarchive/archive_write_add_filter_zstd.c b/contrib/libarchive/libarchive/archive_write_add_filter_zstd.c index c0a6e5a37a66..d4752c247157 100644 --- a/contrib/libarchive/libarchive/archive_write_add_filter_zstd.c +++ b/contrib/libarchive/libarchive/archive_write_add_filter_zstd.c @@ -1,570 +1,572 @@ /*- * Copyright (c) 2017 Sean Purcell * Copyright (c) 2023-2024 Klara, Inc. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_STDINT_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_ZSTD_H #include #endif #include "archive.h" #include "archive_private.h" #include "archive_string.h" #include "archive_write_private.h" /* Don't compile this if we don't have zstd.h */ struct private_data { int compression_level; int threads; int long_distance; #if HAVE_ZSTD_H && HAVE_ZSTD_compressStream enum { running, finishing, resetting, } state; int frame_per_file; size_t min_frame_in; size_t max_frame_in; size_t min_frame_out; size_t max_frame_out; size_t cur_frame; size_t cur_frame_in; size_t cur_frame_out; size_t total_in; ZSTD_CStream *cstream; ZSTD_outBuffer out; #else struct archive_write_program_data *pdata; #endif }; /* If we don't have the library use default range values (zstdcli.c v1.4.0) */ #define CLEVEL_MIN -99 #define CLEVEL_STD_MIN 0 /* prior to 1.3.4 and more recent without using --fast */ #define CLEVEL_DEFAULT 3 #define CLEVEL_STD_MAX 19 /* without using --ultra */ #define CLEVEL_MAX 22 #define LONG_STD 27 #define MINVER_NEGCLEVEL 10304 #define MINVER_MINCLEVEL 10306 #define MINVER_LONG 10302 static int archive_compressor_zstd_options(struct archive_write_filter *, const char *, const char *); static int archive_compressor_zstd_open(struct archive_write_filter *); static int archive_compressor_zstd_write(struct archive_write_filter *, const void *, size_t); static int archive_compressor_zstd_flush(struct archive_write_filter *); static int archive_compressor_zstd_close(struct archive_write_filter *); static int archive_compressor_zstd_free(struct archive_write_filter *); #if HAVE_ZSTD_H && HAVE_ZSTD_compressStream static int drive_compressor(struct archive_write_filter *, struct private_data *, int, const void *, size_t); #endif /* * Add a zstd compression filter to this write handle. */ int archive_write_add_filter_zstd(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct archive_write_filter *f = __archive_write_allocate_filter(_a); struct private_data *data; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_add_filter_zstd"); data = calloc(1, sizeof(*data)); if (data == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } f->data = data; f->open = &archive_compressor_zstd_open; f->options = &archive_compressor_zstd_options; f->flush = &archive_compressor_zstd_flush; f->close = &archive_compressor_zstd_close; f->free = &archive_compressor_zstd_free; f->code = ARCHIVE_FILTER_ZSTD; f->name = "zstd"; data->compression_level = CLEVEL_DEFAULT; data->threads = 0; data->long_distance = 0; #if HAVE_ZSTD_H && HAVE_ZSTD_compressStream data->frame_per_file = 0; data->min_frame_in = 0; data->max_frame_in = SIZE_MAX; data->min_frame_out = 0; data->max_frame_out = SIZE_MAX; data->cur_frame_in = 0; data->cur_frame_out = 0; data->cstream = ZSTD_createCStream(); if (data->cstream == NULL) { free(data); archive_set_error(&a->archive, ENOMEM, "Failed to allocate zstd compressor object"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); #else data->pdata = __archive_write_program_allocate("zstd"); if (data->pdata == NULL) { free(data); archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Using external zstd program"); return (ARCHIVE_WARN); #endif } static int archive_compressor_zstd_free(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; #if HAVE_ZSTD_H && HAVE_ZSTD_compressStream ZSTD_freeCStream(data->cstream); free(data->out.dst); #else __archive_write_program_free(data->pdata); #endif free(data); f->data = NULL; return (ARCHIVE_OK); } static int string_to_number(const char *string, intmax_t *numberp) { char *end; if (string == NULL || *string == '\0') return (ARCHIVE_WARN); *numberp = strtoimax(string, &end, 10); if (end == string || *end != '\0' || errno == EOVERFLOW) { *numberp = 0; return (ARCHIVE_WARN); } return (ARCHIVE_OK); } #if HAVE_ZSTD_H && HAVE_ZSTD_compressStream static int string_to_size(const char *string, size_t *numberp) { uintmax_t number; char *end; unsigned int shift = 0; if (string == NULL || *string == '\0' || *string == '-') return (ARCHIVE_WARN); number = strtoumax(string, &end, 10); if (end > string) { if (*end == 'K' || *end == 'k') { shift = 10; end++; } else if (*end == 'M' || *end == 'm') { shift = 20; end++; } else if (*end == 'G' || *end == 'g') { shift = 30; end++; } if (*end == 'B' || *end == 'b') { end++; } } if (end == string || *end != '\0' || errno == EOVERFLOW) { return (ARCHIVE_WARN); } if (number > (uintmax_t)SIZE_MAX >> shift) { return (ARCHIVE_WARN); } *numberp = (size_t)(number << shift); return (ARCHIVE_OK); } #endif /* * Set write options. */ static int archive_compressor_zstd_options(struct archive_write_filter *f, const char *key, const char *value) { struct private_data *data = (struct private_data *)f->data; if (strcmp(key, "compression-level") == 0) { intmax_t level; if (string_to_number(value, &level) != ARCHIVE_OK) { return (ARCHIVE_WARN); } /* If we don't have the library, hard-code the max level */ int minimum = CLEVEL_MIN; int maximum = CLEVEL_MAX; #if HAVE_ZSTD_H && HAVE_ZSTD_compressStream maximum = ZSTD_maxCLevel(); #if ZSTD_VERSION_NUMBER >= MINVER_MINCLEVEL if (ZSTD_versionNumber() >= MINVER_MINCLEVEL) { minimum = ZSTD_minCLevel(); } else #endif if (ZSTD_versionNumber() < MINVER_NEGCLEVEL) { minimum = CLEVEL_STD_MIN; } #endif if (level < minimum || level > maximum) { return (ARCHIVE_WARN); } data->compression_level = (int)level; return (ARCHIVE_OK); } else if (strcmp(key, "threads") == 0) { intmax_t threads; if (string_to_number(value, &threads) != ARCHIVE_OK) { return (ARCHIVE_WARN); } #if defined(HAVE_SYSCONF) && defined(_SC_NPROCESSORS_ONLN) if (threads == 0) { threads = sysconf(_SC_NPROCESSORS_ONLN); } #elif !defined(__CYGWIN__) && defined(_WIN32_WINNT) && \ _WIN32_WINNT >= 0x0601 /* _WIN32_WINNT_WIN7 */ if (threads == 0) { DWORD winCores = GetActiveProcessorCount( ALL_PROCESSOR_GROUPS); threads = (intmax_t)winCores; } #endif if (threads < 0 || threads > INT_MAX) { return (ARCHIVE_WARN); } data->threads = (int)threads; return (ARCHIVE_OK); #if HAVE_ZSTD_H && HAVE_ZSTD_compressStream } else if (strcmp(key, "frame-per-file") == 0) { data->frame_per_file = 1; return (ARCHIVE_OK); } else if (strcmp(key, "min-frame-in") == 0) { if (string_to_size(value, &data->min_frame_in) != ARCHIVE_OK) { return (ARCHIVE_WARN); } return (ARCHIVE_OK); } else if (strcmp(key, "min-frame-out") == 0 || strcmp(key, "min-frame-size") == 0) { if (string_to_size(value, &data->min_frame_out) != ARCHIVE_OK) { return (ARCHIVE_WARN); } return (ARCHIVE_OK); } else if (strcmp(key, "max-frame-in") == 0 || strcmp(key, "max-frame-size") == 0) { if (string_to_size(value, &data->max_frame_in) != ARCHIVE_OK || data->max_frame_in < 1024) { return (ARCHIVE_WARN); } return (ARCHIVE_OK); } else if (strcmp(key, "max-frame-out") == 0) { if (string_to_size(value, &data->max_frame_out) != ARCHIVE_OK || data->max_frame_out < 1024) { return (ARCHIVE_WARN); } return (ARCHIVE_OK); #endif } else if (strcmp(key, "long") == 0) { intmax_t long_distance; if (string_to_number(value, &long_distance) != ARCHIVE_OK) { return (ARCHIVE_WARN); } #if HAVE_ZSTD_H && HAVE_ZSTD_compressStream && ZSTD_VERSION_NUMBER >= MINVER_LONG ZSTD_bounds bounds = ZSTD_cParam_getBounds(ZSTD_c_windowLog); if (ZSTD_isError(bounds.error)) { int max_distance = ((int)(sizeof(size_t) == 4 ? 30 : 31)); if (((int)long_distance) < 10 || (int)long_distance > max_distance) return (ARCHIVE_WARN); } else { if ((int)long_distance < bounds.lowerBound || (int)long_distance > bounds.upperBound) return (ARCHIVE_WARN); } #else int max_distance = ((int)(sizeof(size_t) == 4 ? 30 : 31)); if (((int)long_distance) < 10 || (int)long_distance > max_distance) return (ARCHIVE_WARN); #endif data->long_distance = (int)long_distance; return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } #if HAVE_ZSTD_H && HAVE_ZSTD_compressStream /* * Setup callback. */ static int archive_compressor_zstd_open(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; if (data->out.dst == NULL) { size_t bs = ZSTD_CStreamOutSize(), bpb; if (f->archive->magic == ARCHIVE_WRITE_MAGIC) { /* Buffer size should be a multiple number of * the of bytes per block for performance. */ bpb = archive_write_get_bytes_per_block(f->archive); if (bpb > bs) bs = bpb; else if (bpb != 0) bs -= bs % bpb; } data->out.size = bs; data->out.pos = 0; data->out.dst = malloc(data->out.size); if (data->out.dst == NULL) { archive_set_error(f->archive, ENOMEM, "Can't allocate data for compression buffer"); return (ARCHIVE_FATAL); } } f->write = archive_compressor_zstd_write; if (ZSTD_isError(ZSTD_initCStream(data->cstream, data->compression_level))) { archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Internal error initializing zstd compressor object"); return (ARCHIVE_FATAL); } ZSTD_CCtx_setParameter(data->cstream, ZSTD_c_nbWorkers, data->threads); + ZSTD_CCtx_setParameter(data->cstream, ZSTD_c_checksumFlag, 1); + #if ZSTD_VERSION_NUMBER >= MINVER_LONG ZSTD_CCtx_setParameter(data->cstream, ZSTD_c_windowLog, data->long_distance); #endif return (ARCHIVE_OK); } /* * Write data to the compressed stream. */ static int archive_compressor_zstd_write(struct archive_write_filter *f, const void *buff, size_t length) { struct private_data *data = (struct private_data *)f->data; return (drive_compressor(f, data, 0, buff, length)); } /* * Flush the compressed stream. */ static int archive_compressor_zstd_flush(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; if (data->frame_per_file && data->state == running) { if (data->cur_frame_in > data->min_frame_in && data->cur_frame_out > data->min_frame_out) { data->state = finishing; } } return (drive_compressor(f, data, 1, NULL, 0)); } /* * Finish the compression... */ static int archive_compressor_zstd_close(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; if (data->state == running) data->state = finishing; return (drive_compressor(f, data, 1, NULL, 0)); } /* * Utility function to push input data through compressor, * writing full output blocks as necessary. */ static int drive_compressor(struct archive_write_filter *f, struct private_data *data, int flush, const void *src, size_t length) { ZSTD_inBuffer in = { .src = src, .size = length, .pos = 0 }; size_t ipos, opos, zstdret = 0; int ret; for (;;) { ipos = in.pos; opos = data->out.pos; switch (data->state) { case running: if (in.pos == in.size) return (ARCHIVE_OK); zstdret = ZSTD_compressStream(data->cstream, &data->out, &in); if (ZSTD_isError(zstdret)) goto zstd_fatal; break; case finishing: zstdret = ZSTD_endStream(data->cstream, &data->out); if (ZSTD_isError(zstdret)) goto zstd_fatal; if (zstdret == 0) data->state = resetting; break; case resetting: ZSTD_CCtx_reset(data->cstream, ZSTD_reset_session_only); data->cur_frame++; data->cur_frame_in = 0; data->cur_frame_out = 0; data->state = running; break; } data->total_in += in.pos - ipos; data->cur_frame_in += in.pos - ipos; data->cur_frame_out += data->out.pos - opos; if (data->state == running) { if (data->cur_frame_in >= data->max_frame_in || data->cur_frame_out >= data->max_frame_out) { data->state = finishing; } } if (data->out.pos == data->out.size || (flush && data->out.pos > 0)) { ret = __archive_write_filter(f->next_filter, data->out.dst, data->out.pos); if (ret != ARCHIVE_OK) goto fatal; data->out.pos = 0; } } zstd_fatal: archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Zstd compression failed: %s", ZSTD_getErrorName(zstdret)); fatal: return (ARCHIVE_FATAL); } #else /* HAVE_ZSTD_H && HAVE_ZSTD_compressStream */ static int archive_compressor_zstd_open(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; struct archive_string as; int r; archive_string_init(&as); /* --no-check matches library default */ archive_strcpy(&as, "zstd --no-check"); if (data->compression_level < CLEVEL_STD_MIN) { archive_string_sprintf(&as, " --fast=%d", -data->compression_level); } else { archive_string_sprintf(&as, " -%d", data->compression_level); } if (data->compression_level > CLEVEL_STD_MAX) { archive_strcat(&as, " --ultra"); } if (data->threads != 0) { archive_string_sprintf(&as, " --threads=%d", data->threads); } if (data->long_distance != 0) { archive_string_sprintf(&as, " --long=%d", data->long_distance); } f->write = archive_compressor_zstd_write; r = __archive_write_program_open(f, data->pdata, as.s); archive_string_free(&as); return (r); } static int archive_compressor_zstd_write(struct archive_write_filter *f, const void *buff, size_t length) { struct private_data *data = (struct private_data *)f->data; return __archive_write_program_write(f, data->pdata, buff, length); } static int archive_compressor_zstd_flush(struct archive_write_filter *f) { (void)f; /* UNUSED */ return (ARCHIVE_OK); } static int archive_compressor_zstd_close(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; return __archive_write_program_close(f, data->pdata); } #endif /* HAVE_ZSTD_H && HAVE_ZSTD_compressStream */ diff --git a/contrib/libarchive/libarchive/archive_write_disk_posix.c b/contrib/libarchive/libarchive/archive_write_disk_posix.c index 1bbfd7a3d779..aeb27e1270ad 100644 --- a/contrib/libarchive/libarchive/archive_write_disk_posix.c +++ b/contrib/libarchive/libarchive/archive_write_disk_posix.c @@ -1,4760 +1,4764 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * Copyright (c) 2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #if !defined(_WIN32) || defined(__CYGWIN__) #ifdef HAVE_SYS_TYPES_H #include #endif #ifdef HAVE_SYS_ACL_H #include #endif #ifdef HAVE_SYS_EXTATTR_H #include #endif #if HAVE_SYS_XATTR_H #include #elif HAVE_ATTR_XATTR_H #include #endif #ifdef HAVE_SYS_EA_H #include #endif #ifdef HAVE_SYS_IOCTL_H #include #endif #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_SYS_TIME_H #include #endif #ifdef HAVE_SYS_UTIME_H #include #endif #ifdef HAVE_COPYFILE_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_GRP_H #include #endif #ifdef HAVE_LANGINFO_H #include #endif #ifdef HAVE_LINUX_FS_H #include /* for Linux file flags */ #endif /* * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. * As the include guards don't agree, the order of include is important. */ #ifdef HAVE_LINUX_EXT2_FS_H #include /* for Linux file flags */ #endif #if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) #include /* Linux file flags, broken on Cygwin */ #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_PWD_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_UTIME_H #include #endif #ifdef F_GETTIMES /* Tru64 specific */ #include #endif /* * Macro to cast st_mtime and time_t to an int64 so that 2 numbers can reliably be compared. * * It assumes that the input is an integer type of no more than 64 bits. * If the number is less than zero, t must be a signed type, so it fits in * int64_t. Otherwise, it's a nonnegative value so we can cast it to uint64_t * without loss. But it could be a large unsigned value, so we have to clip it * to INT64_MAX.* */ #define to_int64_time(t) \ ((t) < 0 ? (int64_t)(t) : (uint64_t)(t) > (uint64_t)INT64_MAX ? INT64_MAX : (int64_t)(t)) #if __APPLE__ #include #if TARGET_OS_MAC && !TARGET_OS_EMBEDDED && HAVE_QUARANTINE_H #include #define HAVE_QUARANTINE 1 #endif #endif #ifdef HAVE_ZLIB_H #include #endif /* TODO: Support Mac OS 'quarantine' feature. This is really just a * standard tag to mark files that have been downloaded as "tainted". * On Mac OS, we should mark the extracted files as tainted if the * archive being read was tainted. Windows has a similar feature; we * should investigate ways to support this generically. */ #include "archive.h" #include "archive_acl_private.h" #include "archive_string.h" #include "archive_endian.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_write_disk_private.h" #ifndef O_BINARY #define O_BINARY 0 #endif #ifndef O_CLOEXEC #define O_CLOEXEC 0 #endif /* Ignore non-int O_NOFOLLOW constant. */ /* gnulib's fcntl.h does this on AIX, but it seems practical everywhere */ #if defined O_NOFOLLOW && !(INT_MIN <= O_NOFOLLOW && O_NOFOLLOW <= INT_MAX) #undef O_NOFOLLOW #endif #ifndef O_NOFOLLOW #define O_NOFOLLOW 0 #endif #ifndef AT_FDCWD #define AT_FDCWD -100 #endif struct fixup_entry { struct fixup_entry *next; struct archive_acl acl; mode_t mode; __LA_MODE_T filetype; int64_t atime; int64_t birthtime; int64_t mtime; int64_t ctime; unsigned long atime_nanos; unsigned long birthtime_nanos; unsigned long mtime_nanos; unsigned long ctime_nanos; unsigned long fflags_set; size_t mac_metadata_size; void *mac_metadata; int fixup; /* bitmask of what needs fixing */ char *name; }; /* * We use a bitmask to track which operations remain to be done for * this file. In particular, this helps us avoid unnecessary * operations when it's possible to take care of one step as a * side-effect of another. For example, mkdir() can specify the mode * for the newly-created object but symlink() cannot. This means we * can skip chmod() if mkdir() succeeded, but we must explicitly * chmod() if we're trying to create a directory that already exists * (mkdir() failed) or if we're restoring a symlink. Similarly, we * need to verify UID/GID before trying to restore SUID/SGID bits; * that verification can occur explicitly through a stat() call or * implicitly because of a successful chown() call. */ #define TODO_MODE_FORCE 0x40000000 #define TODO_MODE_BASE 0x20000000 #define TODO_SUID 0x10000000 #define TODO_SUID_CHECK 0x08000000 #define TODO_SGID 0x04000000 #define TODO_SGID_CHECK 0x02000000 #define TODO_APPLEDOUBLE 0x01000000 #define TODO_MODE (TODO_MODE_BASE|TODO_SUID|TODO_SGID) #define TODO_TIMES ARCHIVE_EXTRACT_TIME #define TODO_OWNER ARCHIVE_EXTRACT_OWNER #define TODO_FFLAGS ARCHIVE_EXTRACT_FFLAGS #define TODO_ACLS ARCHIVE_EXTRACT_ACL #define TODO_XATTR ARCHIVE_EXTRACT_XATTR #define TODO_MAC_METADATA ARCHIVE_EXTRACT_MAC_METADATA #define TODO_HFS_COMPRESSION ARCHIVE_EXTRACT_HFS_COMPRESSION_FORCED struct archive_write_disk { struct archive archive; mode_t user_umask; struct fixup_entry *fixup_list; struct fixup_entry *current_fixup; int64_t user_uid; int skip_file_set; int64_t skip_file_dev; int64_t skip_file_ino; time_t start_time; int64_t (*lookup_gid)(void *private, const char *gname, int64_t gid); void (*cleanup_gid)(void *private); void *lookup_gid_data; int64_t (*lookup_uid)(void *private, const char *uname, int64_t uid); void (*cleanup_uid)(void *private); void *lookup_uid_data; /* * Full path of last file to satisfy symlink checks. */ struct archive_string path_safe; /* * Cached stat data from disk for the current entry. * If this is valid, pst points to st. Otherwise, * pst is null. */ struct stat st; struct stat *pst; /* Information about the object being restored right now. */ struct archive_entry *entry; /* Entry being extracted. */ char *name; /* Name of entry, possibly edited. */ struct archive_string _name_data; /* backing store for 'name' */ char *tmpname; /* Temporary name * */ struct archive_string _tmpname_data; /* backing store for 'tmpname' */ /* Tasks remaining for this object. */ int todo; /* Tasks deferred until end-of-archive. */ int deferred; /* Options requested by the client. */ int flags; /* Handle for the file we're restoring. */ int fd; /* Current offset for writing data to the file. */ int64_t offset; /* Last offset actually written to disk. */ int64_t fd_offset; /* Total bytes actually written to files. */ int64_t total_bytes_written; /* Maximum size of file, -1 if unknown. */ int64_t filesize; /* Dir we were in before this restore; only for deep paths. */ int restore_pwd; /* Mode we should use for this entry; affected by _PERM and umask. */ mode_t mode; /* UID/GID to use in restoring this entry. */ int64_t uid; int64_t gid; /* * HFS+ Compression. */ /* Xattr "com.apple.decmpfs". */ uint32_t decmpfs_attr_size; unsigned char *decmpfs_header_p; /* ResourceFork set options used for fsetxattr. */ int rsrc_xattr_options; /* Xattr "com.apple.ResourceFork". */ unsigned char *resource_fork; size_t resource_fork_allocated_size; unsigned int decmpfs_block_count; uint32_t *decmpfs_block_info; /* Buffer for compressed data. */ unsigned char *compressed_buffer; size_t compressed_buffer_size; size_t compressed_buffer_remaining; /* The offset of the ResourceFork where compressed data will * be placed. */ uint32_t compressed_rsrc_position; uint32_t compressed_rsrc_position_v; /* Buffer for uncompressed data. */ char *uncompressed_buffer; size_t block_remaining_bytes; size_t file_remaining_bytes; #ifdef HAVE_ZLIB_H z_stream stream; int stream_valid; int decmpfs_compression_level; #endif }; /* * Default mode for dirs created automatically (will be modified by umask). * Note that POSIX specifies 0777 for implicitly-created dirs, "modified * by the process' file creation mask." */ #define DEFAULT_DIR_MODE 0777 /* * Dir modes are restored in two steps: During the extraction, the permissions * in the archive are modified to match the following limits. During * the post-extract fixup pass, the permissions from the archive are * applied. */ #define MINIMUM_DIR_MODE 0700 #define MAXIMUM_DIR_MODE 0775 /* * Maximum uncompressed size of a decmpfs block. */ #define MAX_DECMPFS_BLOCK_SIZE (64 * 1024) /* * HFS+ compression type. */ #define CMP_XATTR 3/* Compressed data in xattr. */ #define CMP_RESOURCE_FORK 4/* Compressed data in resource fork. */ /* * HFS+ compression resource fork. */ #define RSRC_H_SIZE 260 /* Base size of Resource fork header. */ #define RSRC_F_SIZE 50 /* Size of Resource fork footer. */ /* Size to write compressed data to resource fork. */ #define COMPRESSED_W_SIZE (64 * 1024) /* decmpfs definitions. */ #define MAX_DECMPFS_XATTR_SIZE 3802 #ifndef DECMPFS_XATTR_NAME #define DECMPFS_XATTR_NAME "com.apple.decmpfs" #endif #define DECMPFS_MAGIC 0x636d7066 #define DECMPFS_COMPRESSION_MAGIC 0 #define DECMPFS_COMPRESSION_TYPE 4 #define DECMPFS_UNCOMPRESSED_SIZE 8 #define DECMPFS_HEADER_SIZE 16 #define HFS_BLOCKS(s) ((s) >> 12) static int la_opendirat(int, const char *); static int la_mktemp(struct archive_write_disk *); static int la_verify_filetype(mode_t, __LA_MODE_T); static void fsobj_error(int *, struct archive_string *, int, const char *, const char *); static int check_symlinks_fsobj(char *, int *, struct archive_string *, int, int); static int check_symlinks(struct archive_write_disk *); static int create_filesystem_object(struct archive_write_disk *); static struct fixup_entry *current_fixup(struct archive_write_disk *, const char *pathname); #if defined(HAVE_FCHDIR) && defined(PATH_MAX) static void edit_deep_directories(struct archive_write_disk *ad); #endif static int cleanup_pathname_fsobj(char *, int *, struct archive_string *, int); static int cleanup_pathname(struct archive_write_disk *); static int create_dir(struct archive_write_disk *, char *); static int create_parent_dir(struct archive_write_disk *, char *); static ssize_t hfs_write_data_block(struct archive_write_disk *, const char *, size_t); static int fixup_appledouble(struct archive_write_disk *, const char *); static int older(struct stat *, struct archive_entry *); static int restore_entry(struct archive_write_disk *); static int set_mac_metadata(struct archive_write_disk *, const char *, const void *, size_t); static int set_xattrs(struct archive_write_disk *); static int clear_nochange_fflags(struct archive_write_disk *); static int set_fflags(struct archive_write_disk *); static int set_fflags_platform(struct archive_write_disk *, int fd, const char *name, mode_t mode, unsigned long fflags_set, unsigned long fflags_clear); static int set_ownership(struct archive_write_disk *); static int set_mode(struct archive_write_disk *, int mode); static int set_time(int, int, const char *, time_t, long, time_t, long); static int set_times(struct archive_write_disk *, int, int, const char *, time_t, long, time_t, long, time_t, long, time_t, long); static int set_times_from_entry(struct archive_write_disk *); static struct fixup_entry *sort_dir_list(struct fixup_entry *p); static ssize_t write_data_block(struct archive_write_disk *, const char *, size_t); static void close_file_descriptor(struct archive_write_disk *); static int _archive_write_disk_close(struct archive *); static int _archive_write_disk_free(struct archive *); static int _archive_write_disk_header(struct archive *, struct archive_entry *); static int64_t _archive_write_disk_filter_bytes(struct archive *, int); static int _archive_write_disk_finish_entry(struct archive *); static ssize_t _archive_write_disk_data(struct archive *, const void *, size_t); static ssize_t _archive_write_disk_data_block(struct archive *, const void *, size_t, int64_t); static int la_mktemp(struct archive_write_disk *a) { int oerrno, fd; mode_t mode; archive_string_empty(&a->_tmpname_data); archive_string_sprintf(&a->_tmpname_data, "%s.XXXXXX", a->name); a->tmpname = a->_tmpname_data.s; fd = __archive_mkstemp(a->tmpname); if (fd == -1) return -1; mode = a->mode & 0777 & ~a->user_umask; if (fchmod(fd, mode) == -1) { oerrno = errno; close(fd); errno = oerrno; return -1; } return fd; } static int la_opendirat(int fd, const char *path) { const int flags = O_CLOEXEC #if defined(O_BINARY) | O_BINARY #endif #if defined(O_DIRECTORY) | O_DIRECTORY #endif #if defined(O_PATH) | O_PATH #elif defined(O_SEARCH) | O_SEARCH #elif defined(__FreeBSD__) && defined(O_EXEC) | O_EXEC #else | O_RDONLY #endif ; #if !defined(HAVE_OPENAT) if (fd != AT_FDCWD) { errno = ENOTSUP; return (-1); } else return (open(path, flags)); #else return (openat(fd, path, flags)); #endif } static int la_verify_filetype(mode_t mode, __LA_MODE_T filetype) { int ret = 0; switch (filetype) { case AE_IFREG: ret = (S_ISREG(mode)); break; case AE_IFDIR: ret = (S_ISDIR(mode)); break; case AE_IFLNK: ret = (S_ISLNK(mode)); break; #ifdef S_ISSOCK case AE_IFSOCK: ret = (S_ISSOCK(mode)); break; #endif case AE_IFCHR: ret = (S_ISCHR(mode)); break; case AE_IFBLK: ret = (S_ISBLK(mode)); break; case AE_IFIFO: ret = (S_ISFIFO(mode)); break; default: break; } return (ret); } static int lazy_stat(struct archive_write_disk *a) { if (a->pst != NULL) { /* Already have stat() data available. */ return (ARCHIVE_OK); } #ifdef HAVE_FSTAT if (a->fd >= 0 && fstat(a->fd, &a->st) == 0) { a->pst = &a->st; return (ARCHIVE_OK); } #endif /* * XXX At this point, symlinks should not be hit, otherwise * XXX a race occurred. Do we want to check explicitly for that? */ #ifdef HAVE_LSTAT if (lstat(a->name, &a->st) == 0) #else if (la_stat(a->name, &a->st) == 0) #endif { a->pst = &a->st; return (ARCHIVE_OK); } archive_set_error(&a->archive, errno, "Couldn't stat file"); return (ARCHIVE_WARN); } static const struct archive_vtable archive_write_disk_vtable = { .archive_close = _archive_write_disk_close, .archive_filter_bytes = _archive_write_disk_filter_bytes, .archive_free = _archive_write_disk_free, .archive_write_header = _archive_write_disk_header, .archive_write_finish_entry = _archive_write_disk_finish_entry, .archive_write_data = _archive_write_disk_data, .archive_write_data_block = _archive_write_disk_data_block, }; static int64_t _archive_write_disk_filter_bytes(struct archive *_a, int n) { struct archive_write_disk *a = (struct archive_write_disk *)_a; (void)n; /* UNUSED */ if (n == -1 || n == 0) return (a->total_bytes_written); return (-1); } int archive_write_disk_set_options(struct archive *_a, int flags) { struct archive_write_disk *a = (struct archive_write_disk *)_a; a->flags = flags; return (ARCHIVE_OK); } /* * Extract this entry to disk. * * TODO: Validate hardlinks. According to the standards, we're * supposed to check each extracted hardlink and squawk if it refers * to a file that we didn't restore. I'm not entirely convinced this * is a good idea, but more importantly: Is there any way to validate * hardlinks without keeping a complete list of filenames from the * entire archive?? Ugh. * */ static int _archive_write_disk_header(struct archive *_a, struct archive_entry *entry) { struct archive_write_disk *a = (struct archive_write_disk *)_a; struct fixup_entry *fe; const char *linkname; int ret, r; archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_disk_header"); archive_clear_error(&a->archive); if (a->archive.state & ARCHIVE_STATE_DATA) { r = _archive_write_disk_finish_entry(&a->archive); if (r == ARCHIVE_FATAL) return (r); } /* Set up for this particular entry. */ a->pst = NULL; a->current_fixup = NULL; a->deferred = 0; if (a->entry) { archive_entry_free(a->entry); a->entry = NULL; } a->entry = archive_entry_clone(entry); a->fd = -1; a->fd_offset = 0; a->offset = 0; a->restore_pwd = -1; a->uid = a->user_uid; a->mode = archive_entry_mode(a->entry); if (archive_entry_size_is_set(a->entry)) a->filesize = archive_entry_size(a->entry); else a->filesize = -1; archive_strcpy(&(a->_name_data), archive_entry_pathname(a->entry)); a->name = a->_name_data.s; archive_clear_error(&a->archive); /* * Clean up the requested path. This is necessary for correct * dir restores; the dir restore logic otherwise gets messed * up by nonsense like "dir/.". */ ret = cleanup_pathname(a); if (ret != ARCHIVE_OK) return (ret); /* * Check if we have a hardlink that points to itself. */ linkname = archive_entry_hardlink(a->entry); if (linkname != NULL && strcmp(a->name, linkname) == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Skipping hardlink pointing to itself: %s", a->name); return (ARCHIVE_WARN); } /* * Query the umask so we get predictable mode settings. * This gets done on every call to _write_header in case the * user edits their umask during the extraction for some * reason. */ umask(a->user_umask = umask(0)); /* Figure out what we need to do for this entry. */ a->todo = TODO_MODE_BASE; if (a->flags & ARCHIVE_EXTRACT_PERM) { a->todo |= TODO_MODE_FORCE; /* Be pushy about permissions. */ /* * SGID requires an extra "check" step because we * cannot easily predict the GID that the system will * assign. (Different systems assign GIDs to files * based on a variety of criteria, including process * credentials and the gid of the enclosing * directory.) We can only restore the SGID bit if * the file has the right GID, and we only know the * GID if we either set it (see set_ownership) or if * we've actually called stat() on the file after it * was restored. Since there are several places at * which we might verify the GID, we need a TODO bit * to keep track. */ if (a->mode & S_ISGID) a->todo |= TODO_SGID | TODO_SGID_CHECK; /* * Verifying the SUID is simpler, but can still be * done in multiple ways, hence the separate "check" bit. */ if (a->mode & S_ISUID) a->todo |= TODO_SUID | TODO_SUID_CHECK; } else { /* * User didn't request full permissions, so don't * restore SUID, SGID bits and obey umask. */ a->mode &= ~S_ISUID; a->mode &= ~S_ISGID; a->mode &= ~S_ISVTX; a->mode &= ~a->user_umask; } if (a->flags & ARCHIVE_EXTRACT_OWNER) a->todo |= TODO_OWNER; if (a->flags & ARCHIVE_EXTRACT_TIME) a->todo |= TODO_TIMES; if (a->flags & ARCHIVE_EXTRACT_ACL) { #if ARCHIVE_ACL_DARWIN /* * On MacOS, platform ACLs get stored in mac_metadata, too. * If we intend to extract mac_metadata and it is present * we skip extracting libarchive NFSv4 ACLs. */ size_t metadata_size; if ((a->flags & ARCHIVE_EXTRACT_MAC_METADATA) == 0 || archive_entry_mac_metadata(a->entry, &metadata_size) == NULL || metadata_size == 0) #endif #if ARCHIVE_ACL_LIBRICHACL /* * RichACLs are stored in an extended attribute. * If we intend to extract extended attributes and have this * attribute we skip extracting libarchive NFSv4 ACLs. */ short extract_acls = 1; if (a->flags & ARCHIVE_EXTRACT_XATTR && ( archive_entry_acl_types(a->entry) & ARCHIVE_ENTRY_ACL_TYPE_NFS4)) { const char *attr_name; const void *attr_value; size_t attr_size; int i = archive_entry_xattr_reset(a->entry); while (i--) { archive_entry_xattr_next(a->entry, &attr_name, &attr_value, &attr_size); if (attr_name != NULL && attr_value != NULL && attr_size > 0 && strcmp(attr_name, "trusted.richacl") == 0) { extract_acls = 0; break; } } } if (extract_acls) #endif #if ARCHIVE_ACL_DARWIN || ARCHIVE_ACL_LIBRICHACL { #endif if (archive_entry_filetype(a->entry) == AE_IFDIR) a->deferred |= TODO_ACLS; else a->todo |= TODO_ACLS; #if ARCHIVE_ACL_DARWIN || ARCHIVE_ACL_LIBRICHACL } #endif } if (a->flags & ARCHIVE_EXTRACT_MAC_METADATA) { if (archive_entry_filetype(a->entry) == AE_IFDIR) a->deferred |= TODO_MAC_METADATA; else a->todo |= TODO_MAC_METADATA; } #if defined(__APPLE__) && defined(UF_COMPRESSED) && defined(HAVE_ZLIB_H) if ((a->flags & ARCHIVE_EXTRACT_NO_HFS_COMPRESSION) == 0) { unsigned long set, clear; archive_entry_fflags(a->entry, &set, &clear); if ((set & ~clear) & UF_COMPRESSED) { a->todo |= TODO_HFS_COMPRESSION; a->decmpfs_block_count = (unsigned)-1; } } if ((a->flags & ARCHIVE_EXTRACT_HFS_COMPRESSION_FORCED) != 0 && (a->mode & AE_IFMT) == AE_IFREG && a->filesize > 0) { a->todo |= TODO_HFS_COMPRESSION; a->decmpfs_block_count = (unsigned)-1; } { const char *p; /* Check if the current file name is a type of the * resource fork file. */ p = strrchr(a->name, '/'); if (p == NULL) p = a->name; else p++; if (p[0] == '.' && p[1] == '_') { /* Do not compress "._XXX" files. */ a->todo &= ~TODO_HFS_COMPRESSION; if (a->filesize > 0) a->todo |= TODO_APPLEDOUBLE; } } #endif if (a->flags & ARCHIVE_EXTRACT_XATTR) { #if ARCHIVE_XATTR_DARWIN /* * On MacOS, extended attributes get stored in mac_metadata, * too. If we intend to extract mac_metadata and it is present * we skip extracting extended attributes. */ size_t metadata_size; if ((a->flags & ARCHIVE_EXTRACT_MAC_METADATA) == 0 || archive_entry_mac_metadata(a->entry, &metadata_size) == NULL || metadata_size == 0) #endif a->todo |= TODO_XATTR; } if (a->flags & ARCHIVE_EXTRACT_FFLAGS) a->todo |= TODO_FFLAGS; if (a->flags & ARCHIVE_EXTRACT_SECURE_SYMLINKS) { ret = check_symlinks(a); if (ret != ARCHIVE_OK) return (ret); } #if defined(HAVE_FCHDIR) && defined(PATH_MAX) /* If path exceeds PATH_MAX, shorten the path. */ edit_deep_directories(a); #endif ret = restore_entry(a); #if defined(__APPLE__) && defined(UF_COMPRESSED) && defined(HAVE_ZLIB_H) /* * Check if the filesystem the file is restoring on supports * HFS+ Compression. If not, cancel HFS+ Compression. */ if (a->todo | TODO_HFS_COMPRESSION) { /* * NOTE: UF_COMPRESSED is ignored even if the filesystem * supports HFS+ Compression because the file should * have at least an extended attribute "com.apple.decmpfs" * before the flag is set to indicate that the file have * been compressed. If the filesystem does not support * HFS+ Compression the system call will fail. */ if (a->fd < 0 || fchflags(a->fd, UF_COMPRESSED) != 0) a->todo &= ~TODO_HFS_COMPRESSION; } #endif /* * TODO: There are rumours that some extended attributes must * be restored before file data is written. If this is true, * then we either need to write all extended attributes both * before and after restoring the data, or find some rule for * determining which must go first and which last. Due to the * many ways people are using xattrs, this may prove to be an * intractable problem. */ #ifdef HAVE_FCHDIR /* If we changed directory above, restore it here. */ if (a->restore_pwd >= 0) { r = fchdir(a->restore_pwd); if (r != 0) { archive_set_error(&a->archive, errno, "chdir() failure"); ret = ARCHIVE_FATAL; } close(a->restore_pwd); a->restore_pwd = -1; } #endif /* * Fixup uses the unedited pathname from archive_entry_pathname(), * because it is relative to the base dir and the edited path * might be relative to some intermediate dir as a result of the * deep restore logic. */ if (a->deferred & TODO_MODE) { fe = current_fixup(a, archive_entry_pathname(entry)); if (fe == NULL) return (ARCHIVE_FATAL); fe->filetype = archive_entry_filetype(entry); fe->fixup |= TODO_MODE_BASE; fe->mode = a->mode; } if ((a->deferred & TODO_TIMES) && (archive_entry_mtime_is_set(entry) || archive_entry_atime_is_set(entry))) { fe = current_fixup(a, archive_entry_pathname(entry)); if (fe == NULL) return (ARCHIVE_FATAL); fe->filetype = archive_entry_filetype(entry); fe->mode = a->mode; fe->fixup |= TODO_TIMES; if (archive_entry_atime_is_set(entry)) { fe->atime = archive_entry_atime(entry); fe->atime_nanos = archive_entry_atime_nsec(entry); } else { /* If atime is unset, use start time. */ fe->atime = a->start_time; fe->atime_nanos = 0; } if (archive_entry_mtime_is_set(entry)) { fe->mtime = archive_entry_mtime(entry); fe->mtime_nanos = archive_entry_mtime_nsec(entry); } else { /* If mtime is unset, use start time. */ fe->mtime = a->start_time; fe->mtime_nanos = 0; } if (archive_entry_birthtime_is_set(entry)) { fe->birthtime = archive_entry_birthtime(entry); fe->birthtime_nanos = archive_entry_birthtime_nsec( entry); } else { /* If birthtime is unset, use mtime. */ fe->birthtime = fe->mtime; fe->birthtime_nanos = fe->mtime_nanos; } } if (a->deferred & TODO_ACLS) { fe = current_fixup(a, archive_entry_pathname(entry)); if (fe == NULL) return (ARCHIVE_FATAL); fe->filetype = archive_entry_filetype(entry); fe->fixup |= TODO_ACLS; archive_acl_copy(&fe->acl, archive_entry_acl(entry)); } if (a->deferred & TODO_MAC_METADATA) { const void *metadata; size_t metadata_size; metadata = archive_entry_mac_metadata(a->entry, &metadata_size); if (metadata != NULL && metadata_size > 0) { fe = current_fixup(a, archive_entry_pathname(entry)); if (fe == NULL) return (ARCHIVE_FATAL); fe->filetype = archive_entry_filetype(entry); fe->mac_metadata = malloc(metadata_size); if (fe->mac_metadata != NULL) { memcpy(fe->mac_metadata, metadata, metadata_size); fe->mac_metadata_size = metadata_size; fe->fixup |= TODO_MAC_METADATA; } } } if (a->deferred & TODO_FFLAGS) { fe = current_fixup(a, archive_entry_pathname(entry)); if (fe == NULL) return (ARCHIVE_FATAL); fe->filetype = archive_entry_filetype(entry); fe->fixup |= TODO_FFLAGS; /* TODO: Complete this.. defer fflags from below. */ } /* We've created the object and are ready to pour data into it. */ if (ret >= ARCHIVE_WARN) a->archive.state = ARCHIVE_STATE_DATA; /* * If it's not open, tell our client not to try writing. * In particular, dirs, links, etc, don't get written to. */ if (a->fd < 0) { archive_entry_set_size(entry, 0); a->filesize = 0; } return (ret); } int archive_write_disk_set_skip_file(struct archive *_a, la_int64_t d, la_int64_t i) { struct archive_write_disk *a = (struct archive_write_disk *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_write_disk_set_skip_file"); a->skip_file_set = 1; a->skip_file_dev = d; a->skip_file_ino = i; return (ARCHIVE_OK); } static ssize_t write_data_block(struct archive_write_disk *a, const char *buff, size_t size) { uint64_t start_size = size; ssize_t bytes_written = 0; ssize_t block_size = 0, bytes_to_write; if (size == 0) return (ARCHIVE_OK); if (a->filesize == 0 || a->fd < 0) { archive_set_error(&a->archive, 0, "Attempt to write to an empty file"); return (ARCHIVE_WARN); } if (a->flags & ARCHIVE_EXTRACT_SPARSE) { #if HAVE_STRUCT_STAT_ST_BLKSIZE int r; if ((r = lazy_stat(a)) != ARCHIVE_OK) return (r); block_size = a->pst->st_blksize; #else /* XXX TODO XXX Is there a more appropriate choice here ? */ /* This needn't match the filesystem allocation size. */ block_size = 16*1024; #endif } /* If this write would run beyond the file size, truncate it. */ if (a->filesize >= 0 && (int64_t)(a->offset + size) > a->filesize) start_size = size = (size_t)(a->filesize - a->offset); /* Write the data. */ while (size > 0) { if (block_size == 0) { bytes_to_write = size; } else { /* We're sparsifying the file. */ const char *p, *end; int64_t block_end; /* Skip leading zero bytes. */ for (p = buff, end = buff + size; p < end; ++p) { if (*p != '\0') break; } a->offset += p - buff; size -= p - buff; buff = p; if (size == 0) break; /* Calculate next block boundary after offset. */ block_end = (a->offset / block_size + 1) * block_size; /* If the adjusted write would cross block boundary, * truncate it to the block boundary. */ bytes_to_write = size; if (a->offset + bytes_to_write > block_end) bytes_to_write = block_end - a->offset; } /* Seek if necessary to the specified offset. */ if (a->offset != a->fd_offset) { if (lseek(a->fd, a->offset, SEEK_SET) < 0) { archive_set_error(&a->archive, errno, "Seek failed"); return (ARCHIVE_FATAL); } a->fd_offset = a->offset; } bytes_written = write(a->fd, buff, bytes_to_write); if (bytes_written < 0) { archive_set_error(&a->archive, errno, "Write failed"); return (ARCHIVE_WARN); } buff += bytes_written; size -= bytes_written; a->total_bytes_written += bytes_written; a->offset += bytes_written; a->fd_offset = a->offset; } return (start_size - size); } #if defined(__APPLE__) && defined(UF_COMPRESSED) && defined(HAVE_SYS_XATTR_H)\ && defined(HAVE_ZLIB_H) /* * Set UF_COMPRESSED file flag. * This have to be called after hfs_write_decmpfs() because if the * file does not have "com.apple.decmpfs" xattr the flag is ignored. */ static int hfs_set_compressed_fflag(struct archive_write_disk *a) { int r; if ((r = lazy_stat(a)) != ARCHIVE_OK) return (r); a->st.st_flags |= UF_COMPRESSED; if (fchflags(a->fd, a->st.st_flags) != 0) { archive_set_error(&a->archive, errno, "Failed to set UF_COMPRESSED file flag"); return (ARCHIVE_WARN); } return (ARCHIVE_OK); } /* * HFS+ Compression decmpfs * * +------------------------------+ +0 * | Magic(LE 4 bytes) | * +------------------------------+ * | Type(LE 4 bytes) | * +------------------------------+ * | Uncompressed size(LE 8 bytes)| * +------------------------------+ +16 * | | * | Compressed data | * | (Placed only if Type == 3) | * | | * +------------------------------+ +3802 = MAX_DECMPFS_XATTR_SIZE * * Type is 3: decmpfs has compressed data. * Type is 4: Resource Fork has compressed data. */ /* * Write "com.apple.decmpfs" */ static int hfs_write_decmpfs(struct archive_write_disk *a) { int r; uint32_t compression_type; r = fsetxattr(a->fd, DECMPFS_XATTR_NAME, a->decmpfs_header_p, a->decmpfs_attr_size, 0, 0); if (r < 0) { archive_set_error(&a->archive, errno, "Cannot restore xattr:%s", DECMPFS_XATTR_NAME); compression_type = archive_le32dec( &a->decmpfs_header_p[DECMPFS_COMPRESSION_TYPE]); if (compression_type == CMP_RESOURCE_FORK) fremovexattr(a->fd, XATTR_RESOURCEFORK_NAME, XATTR_SHOWCOMPRESSION); return (ARCHIVE_WARN); } return (ARCHIVE_OK); } /* * HFS+ Compression Resource Fork * * +-----------------------------+ * | Header(260 bytes) | * +-----------------------------+ * | Block count(LE 4 bytes) | * +-----------------------------+ --+ * +-- | Offset (LE 4 bytes) | | * | | [distance from Block count] | | Block 0 * | +-----------------------------+ | * | | Compressed size(LE 4 bytes) | | * | +-----------------------------+ --+ * | | | * | | .................. | * | | | * | +-----------------------------+ --+ * | | Offset (LE 4 bytes) | | * | +-----------------------------+ | Block (Block count -1) * | | Compressed size(LE 4 bytes) | | * +-> +-----------------------------+ --+ * | Compressed data(n bytes) | Block 0 * +-----------------------------+ * | | * | .................. | * | | * +-----------------------------+ * | Compressed data(n bytes) | Block (Block count -1) * +-----------------------------+ * | Footer(50 bytes) | * +-----------------------------+ * */ /* * Write the header of "com.apple.ResourceFork" */ static int hfs_write_resource_fork(struct archive_write_disk *a, unsigned char *buff, size_t bytes, uint32_t position) { int ret; ret = fsetxattr(a->fd, XATTR_RESOURCEFORK_NAME, buff, bytes, position, a->rsrc_xattr_options); if (ret < 0) { archive_set_error(&a->archive, errno, "Cannot restore xattr: %s at %u pos %u bytes", XATTR_RESOURCEFORK_NAME, (unsigned)position, (unsigned)bytes); return (ARCHIVE_WARN); } a->rsrc_xattr_options &= ~XATTR_CREATE; return (ARCHIVE_OK); } static int hfs_write_compressed_data(struct archive_write_disk *a, size_t bytes_compressed) { int ret; ret = hfs_write_resource_fork(a, a->compressed_buffer, bytes_compressed, a->compressed_rsrc_position); if (ret == ARCHIVE_OK) a->compressed_rsrc_position += bytes_compressed; return (ret); } static int hfs_write_resource_fork_header(struct archive_write_disk *a) { unsigned char *buff; uint32_t rsrc_bytes; uint32_t rsrc_header_bytes; /* * Write resource fork header + block info. */ buff = a->resource_fork; rsrc_bytes = a->compressed_rsrc_position - RSRC_F_SIZE; rsrc_header_bytes = RSRC_H_SIZE + /* Header base size. */ 4 + /* Block count. */ (a->decmpfs_block_count * 8);/* Block info */ archive_be32enc(buff, 0x100); archive_be32enc(buff + 4, rsrc_bytes); archive_be32enc(buff + 8, rsrc_bytes - 256); archive_be32enc(buff + 12, 0x32); memset(buff + 16, 0, 240); archive_be32enc(buff + 256, rsrc_bytes - 260); return hfs_write_resource_fork(a, buff, rsrc_header_bytes, 0); } static size_t hfs_set_resource_fork_footer(unsigned char *buff, size_t buff_size) { static const char rsrc_footer[RSRC_F_SIZE] = { 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1c, 0x00, 0x32, 0x00, 0x00, 'c', 'm', 'p', 'f', 0x00, 0x00, 0x00, 0x0a, 0x00, 0x01, 0xff, 0xff, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 }; if (buff_size < sizeof(rsrc_footer)) return (0); memcpy(buff, rsrc_footer, sizeof(rsrc_footer)); return (sizeof(rsrc_footer)); } static int hfs_reset_compressor(struct archive_write_disk *a) { int ret; if (a->stream_valid) ret = deflateReset(&a->stream); else ret = deflateInit(&a->stream, a->decmpfs_compression_level); if (ret != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to initialize compressor"); return (ARCHIVE_FATAL); } else a->stream_valid = 1; return (ARCHIVE_OK); } static int hfs_decompress(struct archive_write_disk *a) { uint32_t *block_info; unsigned int block_count; uint32_t data_pos, data_size; ssize_t r; ssize_t bytes_written, bytes_to_write; unsigned char *b; block_info = (uint32_t *)(a->resource_fork + RSRC_H_SIZE); block_count = archive_le32dec(block_info++); while (block_count--) { data_pos = RSRC_H_SIZE + archive_le32dec(block_info++); data_size = archive_le32dec(block_info++); r = fgetxattr(a->fd, XATTR_RESOURCEFORK_NAME, a->compressed_buffer, data_size, data_pos, 0); if (r != data_size) { archive_set_error(&a->archive, (r < 0)?errno:ARCHIVE_ERRNO_MISC, "Failed to read resource fork"); return (ARCHIVE_WARN); } if (a->compressed_buffer[0] == 0xff) { bytes_to_write = data_size -1; b = a->compressed_buffer + 1; } else { uLong dest_len = MAX_DECMPFS_BLOCK_SIZE; int zr; zr = uncompress((Bytef *)a->uncompressed_buffer, &dest_len, a->compressed_buffer, data_size); if (zr != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to decompress resource fork"); return (ARCHIVE_WARN); } bytes_to_write = dest_len; b = (unsigned char *)a->uncompressed_buffer; } do { bytes_written = write(a->fd, b, bytes_to_write); if (bytes_written < 0) { archive_set_error(&a->archive, errno, "Write failed"); return (ARCHIVE_WARN); } bytes_to_write -= bytes_written; b += bytes_written; } while (bytes_to_write > 0); } r = fremovexattr(a->fd, XATTR_RESOURCEFORK_NAME, 0); if (r == -1) { archive_set_error(&a->archive, errno, "Failed to remove resource fork"); return (ARCHIVE_WARN); } return (ARCHIVE_OK); } static int hfs_drive_compressor(struct archive_write_disk *a, const char *buff, size_t size) { unsigned char *buffer_compressed; size_t bytes_compressed; size_t bytes_used; int ret; ret = hfs_reset_compressor(a); if (ret != ARCHIVE_OK) return (ret); if (a->compressed_buffer == NULL) { size_t block_size; block_size = COMPRESSED_W_SIZE + RSRC_F_SIZE + + compressBound(MAX_DECMPFS_BLOCK_SIZE); a->compressed_buffer = malloc(block_size); if (a->compressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Resource Fork"); return (ARCHIVE_FATAL); } a->compressed_buffer_size = block_size; a->compressed_buffer_remaining = block_size; } buffer_compressed = a->compressed_buffer + a->compressed_buffer_size - a->compressed_buffer_remaining; a->stream.next_in = (Bytef *)(uintptr_t)(const void *)buff; a->stream.avail_in = size; a->stream.next_out = buffer_compressed; a->stream.avail_out = a->compressed_buffer_remaining; do { ret = deflate(&a->stream, Z_FINISH); switch (ret) { case Z_OK: case Z_STREAM_END: break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to compress data"); return (ARCHIVE_FAILED); } } while (ret == Z_OK); bytes_compressed = a->compressed_buffer_remaining - a->stream.avail_out; /* * If the compressed size is larger than the original size, * throw away compressed data, use uncompressed data instead. */ if (bytes_compressed > size) { buffer_compressed[0] = 0xFF;/* uncompressed marker. */ memcpy(buffer_compressed + 1, buff, size); bytes_compressed = size + 1; } a->compressed_buffer_remaining -= bytes_compressed; /* * If the compressed size is smaller than MAX_DECMPFS_XATTR_SIZE * and the block count in the file is only one, store compressed * data to decmpfs xattr instead of the resource fork. */ if (a->decmpfs_block_count == 1 && (a->decmpfs_attr_size + bytes_compressed) <= MAX_DECMPFS_XATTR_SIZE) { archive_le32enc(&a->decmpfs_header_p[DECMPFS_COMPRESSION_TYPE], CMP_XATTR); memcpy(a->decmpfs_header_p + DECMPFS_HEADER_SIZE, buffer_compressed, bytes_compressed); a->decmpfs_attr_size += bytes_compressed; a->compressed_buffer_remaining = a->compressed_buffer_size; /* * Finish HFS+ Compression. * - Write the decmpfs xattr. * - Set the UF_COMPRESSED file flag. */ ret = hfs_write_decmpfs(a); if (ret == ARCHIVE_OK) ret = hfs_set_compressed_fflag(a); return (ret); } /* Update block info. */ archive_le32enc(a->decmpfs_block_info++, a->compressed_rsrc_position_v - RSRC_H_SIZE); archive_le32enc(a->decmpfs_block_info++, bytes_compressed); a->compressed_rsrc_position_v += bytes_compressed; /* * Write the compressed data to the resource fork. */ bytes_used = a->compressed_buffer_size - a->compressed_buffer_remaining; while (bytes_used >= COMPRESSED_W_SIZE) { ret = hfs_write_compressed_data(a, COMPRESSED_W_SIZE); if (ret != ARCHIVE_OK) return (ret); bytes_used -= COMPRESSED_W_SIZE; if (bytes_used > COMPRESSED_W_SIZE) memmove(a->compressed_buffer, a->compressed_buffer + COMPRESSED_W_SIZE, bytes_used); else memcpy(a->compressed_buffer, a->compressed_buffer + COMPRESSED_W_SIZE, bytes_used); } a->compressed_buffer_remaining = a->compressed_buffer_size - bytes_used; /* * If the current block is the last block, write the remaining * compressed data and the resource fork footer. */ if (a->file_remaining_bytes == 0) { size_t rsrc_size; int64_t bk; /* Append the resource footer. */ rsrc_size = hfs_set_resource_fork_footer( a->compressed_buffer + bytes_used, a->compressed_buffer_remaining); ret = hfs_write_compressed_data(a, bytes_used + rsrc_size); a->compressed_buffer_remaining = a->compressed_buffer_size; /* If the compressed size is not enough smaller than * the uncompressed size. cancel HFS+ compression. * TODO: study a behavior of ditto utility and improve * the condition to fall back into no HFS+ compression. */ bk = HFS_BLOCKS(a->compressed_rsrc_position); bk += bk >> 7; if (bk > HFS_BLOCKS(a->filesize)) return hfs_decompress(a); /* * Write the resourcefork header. */ if (ret == ARCHIVE_OK) ret = hfs_write_resource_fork_header(a); /* * Finish HFS+ Compression. * - Write the decmpfs xattr. * - Set the UF_COMPRESSED file flag. */ if (ret == ARCHIVE_OK) ret = hfs_write_decmpfs(a); if (ret == ARCHIVE_OK) ret = hfs_set_compressed_fflag(a); } return (ret); } static ssize_t hfs_write_decmpfs_block(struct archive_write_disk *a, const char *buff, size_t size) { const char *buffer_to_write; size_t bytes_to_write; int ret; if (a->decmpfs_block_count == (unsigned)-1) { void *new_block; size_t new_size; unsigned int block_count; if (a->decmpfs_header_p == NULL) { new_block = malloc(MAX_DECMPFS_XATTR_SIZE + sizeof(uint32_t)); if (new_block == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for decmpfs"); return (ARCHIVE_FATAL); } a->decmpfs_header_p = new_block; } a->decmpfs_attr_size = DECMPFS_HEADER_SIZE; archive_le32enc(&a->decmpfs_header_p[DECMPFS_COMPRESSION_MAGIC], DECMPFS_MAGIC); archive_le32enc(&a->decmpfs_header_p[DECMPFS_COMPRESSION_TYPE], CMP_RESOURCE_FORK); archive_le64enc(&a->decmpfs_header_p[DECMPFS_UNCOMPRESSED_SIZE], a->filesize); /* Calculate a block count of the file. */ block_count = (a->filesize + MAX_DECMPFS_BLOCK_SIZE -1) / MAX_DECMPFS_BLOCK_SIZE; /* * Allocate buffer for resource fork. * Set up related pointers; */ new_size = RSRC_H_SIZE + /* header */ 4 + /* Block count */ (block_count * sizeof(uint32_t) * 2) + RSRC_F_SIZE; /* footer */ if (new_size > a->resource_fork_allocated_size) { new_block = realloc(a->resource_fork, new_size); if (new_block == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for ResourceFork"); return (ARCHIVE_FATAL); } a->resource_fork_allocated_size = new_size; a->resource_fork = new_block; } /* Allocate uncompressed buffer */ if (a->uncompressed_buffer == NULL) { new_block = malloc(MAX_DECMPFS_BLOCK_SIZE); if (new_block == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for decmpfs"); return (ARCHIVE_FATAL); } a->uncompressed_buffer = new_block; } a->block_remaining_bytes = MAX_DECMPFS_BLOCK_SIZE; a->file_remaining_bytes = a->filesize; a->compressed_buffer_remaining = a->compressed_buffer_size; /* * Set up a resource fork. */ a->rsrc_xattr_options = XATTR_CREATE; /* Get the position where we are going to set a bunch * of block info. */ a->decmpfs_block_info = (uint32_t *)(a->resource_fork + RSRC_H_SIZE); /* Set the block count to the resource fork. */ archive_le32enc(a->decmpfs_block_info++, block_count); /* Get the position where we are going to set compressed * data. */ a->compressed_rsrc_position = RSRC_H_SIZE + 4 + (block_count * 8); a->compressed_rsrc_position_v = a->compressed_rsrc_position; a->decmpfs_block_count = block_count; } /* Ignore redundant bytes. */ if (a->file_remaining_bytes == 0) return ((ssize_t)size); /* Do not overrun a block size. */ if (size > a->block_remaining_bytes) bytes_to_write = a->block_remaining_bytes; else bytes_to_write = size; /* Do not overrun the file size. */ if (bytes_to_write > a->file_remaining_bytes) bytes_to_write = a->file_remaining_bytes; /* For efficiency, if a copy length is full of the uncompressed * buffer size, do not copy writing data to it. */ if (bytes_to_write == MAX_DECMPFS_BLOCK_SIZE) buffer_to_write = buff; else { memcpy(a->uncompressed_buffer + MAX_DECMPFS_BLOCK_SIZE - a->block_remaining_bytes, buff, bytes_to_write); buffer_to_write = a->uncompressed_buffer; } a->block_remaining_bytes -= bytes_to_write; a->file_remaining_bytes -= bytes_to_write; if (a->block_remaining_bytes == 0 || a->file_remaining_bytes == 0) { ret = hfs_drive_compressor(a, buffer_to_write, MAX_DECMPFS_BLOCK_SIZE - a->block_remaining_bytes); if (ret < 0) return (ret); a->block_remaining_bytes = MAX_DECMPFS_BLOCK_SIZE; } /* Ignore redundant bytes. */ if (a->file_remaining_bytes == 0) return ((ssize_t)size); return (bytes_to_write); } static ssize_t hfs_write_data_block(struct archive_write_disk *a, const char *buff, size_t size) { uint64_t start_size = size; ssize_t bytes_written = 0; ssize_t bytes_to_write; if (size == 0) return (ARCHIVE_OK); if (a->filesize == 0 || a->fd < 0) { archive_set_error(&a->archive, 0, "Attempt to write to an empty file"); return (ARCHIVE_WARN); } /* If this write would run beyond the file size, truncate it. */ if (a->filesize >= 0 && (int64_t)(a->offset + size) > a->filesize) start_size = size = (size_t)(a->filesize - a->offset); /* Write the data. */ while (size > 0) { bytes_to_write = size; /* Seek if necessary to the specified offset. */ if (a->offset < a->fd_offset) { /* Can't support backward move. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Seek failed"); return (ARCHIVE_FATAL); } else if (a->offset > a->fd_offset) { uint64_t skip = a->offset - a->fd_offset; char nullblock[1024]; memset(nullblock, 0, sizeof(nullblock)); while (skip > 0) { if (skip > sizeof(nullblock)) bytes_written = hfs_write_decmpfs_block( a, nullblock, sizeof(nullblock)); else bytes_written = hfs_write_decmpfs_block( a, nullblock, skip); if (bytes_written < 0) { archive_set_error(&a->archive, errno, "Write failed"); return (ARCHIVE_WARN); } skip -= bytes_written; } a->fd_offset = a->offset; } bytes_written = hfs_write_decmpfs_block(a, buff, bytes_to_write); if (bytes_written < 0) return (bytes_written); buff += bytes_written; size -= bytes_written; a->total_bytes_written += bytes_written; a->offset += bytes_written; a->fd_offset = a->offset; } return (start_size - size); } #else static ssize_t hfs_write_data_block(struct archive_write_disk *a, const char *buff, size_t size) { return (write_data_block(a, buff, size)); } #endif static ssize_t _archive_write_disk_data_block(struct archive *_a, const void *buff, size_t size, int64_t offset) { struct archive_write_disk *a = (struct archive_write_disk *)_a; ssize_t r; archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, ARCHIVE_STATE_DATA, "archive_write_data_block"); a->offset = offset; if (a->todo & TODO_HFS_COMPRESSION) r = hfs_write_data_block(a, buff, size); else r = write_data_block(a, buff, size); if (r < ARCHIVE_OK) return (r); if ((size_t)r < size) { archive_set_error(&a->archive, 0, "Too much data: Truncating file at %ju bytes", (uintmax_t)a->filesize); return (ARCHIVE_WARN); } #if ARCHIVE_VERSION_NUMBER < 3999000 return (ARCHIVE_OK); #else return (size); #endif } static ssize_t _archive_write_disk_data(struct archive *_a, const void *buff, size_t size) { struct archive_write_disk *a = (struct archive_write_disk *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, ARCHIVE_STATE_DATA, "archive_write_data"); if (a->todo & TODO_HFS_COMPRESSION) return (hfs_write_data_block(a, buff, size)); return (write_data_block(a, buff, size)); } static int _archive_write_disk_finish_entry(struct archive *_a) { struct archive_write_disk *a = (struct archive_write_disk *)_a; int ret = ARCHIVE_OK; archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_finish_entry"); if (a->archive.state & ARCHIVE_STATE_HEADER) return (ARCHIVE_OK); archive_clear_error(&a->archive); /* Pad or truncate file to the right size. */ if (a->fd < 0) { /* There's no file. */ } else if (a->filesize < 0) { /* File size is unknown, so we can't set the size. */ } else if (a->fd_offset == a->filesize) { /* Last write ended at exactly the filesize; we're done. */ /* Hopefully, this is the common case. */ #if defined(__APPLE__) && defined(UF_COMPRESSED) && defined(HAVE_ZLIB_H) } else if (a->todo & TODO_HFS_COMPRESSION) { char null_d[1024]; ssize_t r; if (a->file_remaining_bytes) memset(null_d, 0, sizeof(null_d)); while (a->file_remaining_bytes) { if (a->file_remaining_bytes > sizeof(null_d)) r = hfs_write_data_block( a, null_d, sizeof(null_d)); else r = hfs_write_data_block( a, null_d, a->file_remaining_bytes); if (r < 0) { close_file_descriptor(a); return ((int)r); } } #endif } else { #if HAVE_FTRUNCATE if (ftruncate(a->fd, a->filesize) == -1 && a->filesize == 0) { archive_set_error(&a->archive, errno, "File size could not be restored"); close_file_descriptor(a); return (ARCHIVE_FAILED); } #endif /* * Not all platforms implement the XSI option to * extend files via ftruncate. Stat() the file again * to see what happened. */ a->pst = NULL; if ((ret = lazy_stat(a)) != ARCHIVE_OK) { close_file_descriptor(a); return (ret); } /* We can use lseek()/write() to extend the file if * ftruncate didn't work or isn't available. */ if (a->st.st_size < a->filesize) { const char nul = '\0'; if (lseek(a->fd, a->filesize - 1, SEEK_SET) < 0) { archive_set_error(&a->archive, errno, "Seek failed"); close_file_descriptor(a); return (ARCHIVE_FATAL); } if (write(a->fd, &nul, 1) < 0) { archive_set_error(&a->archive, errno, "Write to restore size failed"); close_file_descriptor(a); return (ARCHIVE_FATAL); } a->pst = NULL; } } /* Restore metadata. */ /* * This is specific to Mac OS X. * If the current file is an AppleDouble file, it should be * linked with the data fork file and remove it. */ if (a->todo & TODO_APPLEDOUBLE) { int r2 = fixup_appledouble(a, a->name); if (r2 == ARCHIVE_EOF) { /* The current file has been successfully linked * with the data fork file and removed. So there * is nothing to do on the current file. */ goto finish_metadata; } if (r2 < ret) ret = r2; } /* * Look up the "real" UID only if we're going to need it. * TODO: the TODO_SGID condition can be dropped here, can't it? */ if (a->todo & (TODO_OWNER | TODO_SUID | TODO_SGID)) { a->uid = archive_write_disk_uid(&a->archive, archive_entry_uname(a->entry), archive_entry_uid(a->entry)); } /* Look up the "real" GID only if we're going to need it. */ /* TODO: the TODO_SUID condition can be dropped here, can't it? */ if (a->todo & (TODO_OWNER | TODO_SGID | TODO_SUID)) { a->gid = archive_write_disk_gid(&a->archive, archive_entry_gname(a->entry), archive_entry_gid(a->entry)); } /* * Restore ownership before set_mode tries to restore suid/sgid * bits. If we set the owner, we know what it is and can skip * a stat() call to examine the ownership of the file on disk. */ if (a->todo & TODO_OWNER) { int r2 = set_ownership(a); if (r2 < ret) ret = r2; } /* * HYPOTHESIS: * If we're not root, we won't be setting any security * attributes that may be wiped by the set_mode() routine * below. We also can't set xattr on non-owner-writable files, * which may be the state after set_mode(). Perform * set_xattrs() first based on these constraints. */ if (a->user_uid != 0 && (a->todo & TODO_XATTR)) { int r2 = set_xattrs(a); if (r2 < ret) ret = r2; } /* * set_mode must precede ACLs on systems such as Solaris and * FreeBSD where setting the mode implicitly clears extended ACLs */ if (a->todo & TODO_MODE) { int r2 = set_mode(a, a->mode); if (r2 < ret) ret = r2; } /* * Security-related extended attributes (such as * security.capability on Linux) have to be restored last, * since they're implicitly removed by other file changes. * We do this last only when root. */ if (a->user_uid == 0 && (a->todo & TODO_XATTR)) { int r2 = set_xattrs(a); if (r2 < ret) ret = r2; } /* * Some flags prevent file modification; they must be restored after * file contents are written. */ if (a->todo & TODO_FFLAGS) { int r2 = set_fflags(a); if (r2 < ret) ret = r2; } /* * Time must follow most other metadata; * otherwise atime will get changed. */ if (a->todo & TODO_TIMES) { int r2 = set_times_from_entry(a); if (r2 < ret) ret = r2; } /* * Mac extended metadata includes ACLs. */ if (a->todo & TODO_MAC_METADATA) { const void *metadata; size_t metadata_size; metadata = archive_entry_mac_metadata(a->entry, &metadata_size); if (metadata != NULL && metadata_size > 0) { int r2 = set_mac_metadata(a, archive_entry_pathname( a->entry), metadata, metadata_size); if (r2 < ret) ret = r2; } } /* * ACLs must be restored after timestamps because there are * ACLs that prevent attribute changes (including time). */ if (a->todo & TODO_ACLS) { int r2; r2 = archive_write_disk_set_acls(&a->archive, a->fd, archive_entry_pathname(a->entry), archive_entry_acl(a->entry), archive_entry_mode(a->entry)); if (r2 < ret) ret = r2; } finish_metadata: /* If there's an fd, we can close it now. */ if (a->fd >= 0) { close(a->fd); a->fd = -1; if (a->tmpname) { if (rename(a->tmpname, a->name) == -1) { archive_set_error(&a->archive, errno, "Failed to rename temporary file"); ret = ARCHIVE_FAILED; unlink(a->tmpname); } a->tmpname = NULL; } } /* If there's an entry, we can release it now. */ archive_entry_free(a->entry); a->entry = NULL; a->archive.state = ARCHIVE_STATE_HEADER; return (ret); } int archive_write_disk_set_group_lookup(struct archive *_a, void *private_data, la_int64_t (*lookup_gid)(void *private, const char *gname, la_int64_t gid), void (*cleanup_gid)(void *private)) { struct archive_write_disk *a = (struct archive_write_disk *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_write_disk_set_group_lookup"); if (a->cleanup_gid != NULL && a->lookup_gid_data != NULL) (a->cleanup_gid)(a->lookup_gid_data); a->lookup_gid = lookup_gid; a->cleanup_gid = cleanup_gid; a->lookup_gid_data = private_data; return (ARCHIVE_OK); } int archive_write_disk_set_user_lookup(struct archive *_a, void *private_data, int64_t (*lookup_uid)(void *private, const char *uname, int64_t uid), void (*cleanup_uid)(void *private)) { struct archive_write_disk *a = (struct archive_write_disk *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_write_disk_set_user_lookup"); if (a->cleanup_uid != NULL && a->lookup_uid_data != NULL) (a->cleanup_uid)(a->lookup_uid_data); a->lookup_uid = lookup_uid; a->cleanup_uid = cleanup_uid; a->lookup_uid_data = private_data; return (ARCHIVE_OK); } int64_t archive_write_disk_gid(struct archive *_a, const char *name, la_int64_t id) { struct archive_write_disk *a = (struct archive_write_disk *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_write_disk_gid"); if (a->lookup_gid) return (a->lookup_gid)(a->lookup_gid_data, name, id); return (id); } int64_t archive_write_disk_uid(struct archive *_a, const char *name, la_int64_t id) { struct archive_write_disk *a = (struct archive_write_disk *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_write_disk_uid"); if (a->lookup_uid) return (a->lookup_uid)(a->lookup_uid_data, name, id); return (id); } /* * Create a new archive_write_disk object and initialize it with global state. */ struct archive * archive_write_disk_new(void) { struct archive_write_disk *a; a = calloc(1, sizeof(*a)); if (a == NULL) return (NULL); a->archive.magic = ARCHIVE_WRITE_DISK_MAGIC; /* We're ready to write a header immediately. */ a->archive.state = ARCHIVE_STATE_HEADER; a->archive.vtable = &archive_write_disk_vtable; a->start_time = time(NULL); /* Query and restore the umask. */ umask(a->user_umask = umask(0)); #ifdef HAVE_GETEUID a->user_uid = geteuid(); #endif /* HAVE_GETEUID */ if (archive_string_ensure(&a->path_safe, 512) == NULL) { free(a); return (NULL); } a->path_safe.s[0] = 0; #ifdef HAVE_ZLIB_H a->decmpfs_compression_level = 5; #endif return (&a->archive); } /* * If pathname is longer than PATH_MAX, chdir to a suitable * intermediate dir and edit the path down to a shorter suffix. Note * that this routine never returns an error; if the chdir() attempt * fails for any reason, we just go ahead with the long pathname. The * object creation is likely to fail, but any error will get handled * at that time. */ #if defined(HAVE_FCHDIR) && defined(PATH_MAX) static void edit_deep_directories(struct archive_write_disk *a) { int ret; char *tail = a->name; /* If path is short, avoid the open() below. */ if (strlen(tail) < PATH_MAX) return; /* Try to record our starting dir. */ a->restore_pwd = la_opendirat(AT_FDCWD, "."); __archive_ensure_cloexec_flag(a->restore_pwd); if (a->restore_pwd < 0) return; /* As long as the path is too long... */ while (strlen(tail) >= PATH_MAX) { /* Locate a dir prefix shorter than PATH_MAX. */ tail += PATH_MAX - 8; while (tail > a->name && *tail != '/') tail--; /* Exit if we find a too-long path component. */ if (tail <= a->name) return; /* Create the intermediate dir and chdir to it. */ *tail = '\0'; /* Terminate dir portion */ ret = create_dir(a, a->name); if (ret == ARCHIVE_OK && chdir(a->name) != 0) ret = ARCHIVE_FAILED; *tail = '/'; /* Restore the / we removed. */ if (ret != ARCHIVE_OK) return; tail++; /* The chdir() succeeded; we've now shortened the path. */ a->name = tail; } return; } #endif /* * The main restore function. */ static int restore_entry(struct archive_write_disk *a) { int ret = ARCHIVE_OK, en; if (a->flags & ARCHIVE_EXTRACT_UNLINK && !S_ISDIR(a->mode)) { /* * TODO: Fix this. Apparently, there are platforms * that still allow root to hose the entire filesystem * by unlinking a dir. The S_ISDIR() test above * prevents us from using unlink() here if the new * object is a dir, but that doesn't mean the old * object isn't a dir. */ if (a->flags & ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS) (void)clear_nochange_fflags(a); if (unlink(a->name) == 0) { /* We removed it, reset cached stat. */ a->pst = NULL; } else if (errno == ENOENT) { /* File didn't exist, that's just as good. */ } else if (rmdir(a->name) == 0) { /* It was a dir, but now it's gone. */ a->pst = NULL; } else { /* We tried, but couldn't get rid of it. */ archive_set_error(&a->archive, errno, "Could not unlink"); return(ARCHIVE_FAILED); } } /* Try creating it first; if this fails, we'll try to recover. */ en = create_filesystem_object(a); if ((en == ENOTDIR || en == ENOENT) && !(a->flags & ARCHIVE_EXTRACT_NO_AUTODIR)) { /* If the parent dir doesn't exist, try creating it. */ create_parent_dir(a, a->name); /* Now try to create the object again. */ en = create_filesystem_object(a); } if ((en == ENOENT) && (archive_entry_hardlink(a->entry) != NULL)) { archive_set_error(&a->archive, en, "Hard-link target '%s' does not exist.", archive_entry_hardlink(a->entry)); return (ARCHIVE_FAILED); } if ((en == EISDIR || en == EEXIST) && (a->flags & ARCHIVE_EXTRACT_NO_OVERWRITE)) { /* If we're not overwriting, we're done. */ if (S_ISDIR(a->mode)) { /* Don't overwrite any settings on existing directories. */ a->todo = 0; } archive_entry_unset_size(a->entry); return (ARCHIVE_OK); } /* * Some platforms return EISDIR if you call * open(O_WRONLY | O_EXCL | O_CREAT) on a directory, some * return EEXIST. POSIX is ambiguous, requiring EISDIR * for open(O_WRONLY) on a dir and EEXIST for open(O_EXCL | O_CREAT) * on an existing item. */ if (en == EISDIR) { /* A dir is in the way of a non-dir, rmdir it. */ if (rmdir(a->name) != 0) { archive_set_error(&a->archive, errno, "Can't remove already-existing dir"); return (ARCHIVE_FAILED); } a->pst = NULL; /* Try again. */ en = create_filesystem_object(a); } else if (en == EEXIST) { /* * We know something is in the way, but we don't know what; * we need to find out before we go any further. */ int r = 0; /* * The SECURE_SYMLINKS logic has already removed a * symlink to a dir if the client wants that. So * follow the symlink if we're creating a dir. */ if (S_ISDIR(a->mode)) r = la_stat(a->name, &a->st); /* * If it's not a dir (or it's a broken symlink), * then don't follow it. */ if (r != 0 || !S_ISDIR(a->mode)) #ifdef HAVE_LSTAT r = lstat(a->name, &a->st); #else r = la_stat(a->name, &a->st); #endif if (r != 0) { archive_set_error(&a->archive, errno, "Can't stat existing object"); return (ARCHIVE_FAILED); } /* * NO_OVERWRITE_NEWER doesn't apply to directories. */ if ((a->flags & ARCHIVE_EXTRACT_NO_OVERWRITE_NEWER) && !S_ISDIR(a->st.st_mode)) { if (!older(&(a->st), a->entry)) { archive_entry_unset_size(a->entry); return (ARCHIVE_OK); } } /* If it's our archive, we're done. */ if (a->skip_file_set && a->st.st_dev == (dev_t)a->skip_file_dev && a->st.st_ino == (ino_t)a->skip_file_ino) { archive_set_error(&a->archive, 0, "Refusing to overwrite archive"); return (ARCHIVE_FAILED); } if (!S_ISDIR(a->st.st_mode)) { if (a->flags & ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS) (void)clear_nochange_fflags(a); if ((a->flags & ARCHIVE_EXTRACT_SAFE_WRITES) && - S_ISREG(a->st.st_mode)) { + S_ISREG(a->mode)) { /* Use a temporary file to extract */ if ((a->fd = la_mktemp(a)) == -1) { archive_set_error(&a->archive, errno, "Can't create temporary file"); return ARCHIVE_FAILED; } a->pst = NULL; en = 0; } else { /* A non-dir is in the way, unlink it. */ if (unlink(a->name) != 0) { archive_set_error(&a->archive, errno, "Can't unlink already-existing " "object"); return (ARCHIVE_FAILED); } a->pst = NULL; /* Try again. */ en = create_filesystem_object(a); } } else if (!S_ISDIR(a->mode)) { /* A dir is in the way of a non-dir, rmdir it. */ if (a->flags & ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS) (void)clear_nochange_fflags(a); if (rmdir(a->name) != 0) { archive_set_error(&a->archive, errno, "Can't replace existing directory with non-directory"); return (ARCHIVE_FAILED); } /* Try again. */ en = create_filesystem_object(a); } else { /* * There's a dir in the way of a dir. Don't * waste time with rmdir()/mkdir(), just fix * up the permissions on the existing dir. * Note that we don't change perms on existing * dirs unless _EXTRACT_PERM is specified. */ if ((a->mode != a->st.st_mode) && (a->todo & TODO_MODE_FORCE)) a->deferred |= (a->todo & TODO_MODE); /* Ownership doesn't need deferred fixup. */ en = 0; /* Forget the EEXIST. */ } } if (en) { /* Everything failed; give up here. */ if ((&a->archive)->error == NULL) archive_set_error(&a->archive, en, "Can't create '%s'", a->name); return (ARCHIVE_FAILED); } a->pst = NULL; /* Cached stat data no longer valid. */ return (ret); } /* * Returns 0 if creation succeeds, or else returns errno value from * the failed system call. Note: This function should only ever perform * a single system call. */ static int create_filesystem_object(struct archive_write_disk *a) { /* Create the entry. */ const char *linkname; mode_t final_mode, mode; int r; /* these for check_symlinks_fsobj */ char *linkname_copy; /* non-const copy of linkname */ struct stat st; struct archive_string error_string; int error_number; /* We identify hard/symlinks according to the link names. */ /* Since link(2) and symlink(2) don't handle modes, we're done here. */ linkname = archive_entry_hardlink(a->entry); if (linkname != NULL) { #if !HAVE_LINK return (EPERM); #else archive_string_init(&error_string); linkname_copy = strdup(linkname); if (linkname_copy == NULL) { return (EPERM); } /* * TODO: consider using the cleaned-up path as the link * target? */ r = cleanup_pathname_fsobj(linkname_copy, &error_number, &error_string, a->flags); if (r != ARCHIVE_OK) { archive_set_error(&a->archive, error_number, "%s", error_string.s); free(linkname_copy); archive_string_free(&error_string); /* * EPERM is more appropriate than error_number for our * callers */ return (EPERM); } r = check_symlinks_fsobj(linkname_copy, &error_number, &error_string, a->flags, 1); if (r != ARCHIVE_OK) { archive_set_error(&a->archive, error_number, "%s", error_string.s); free(linkname_copy); archive_string_free(&error_string); /* * EPERM is more appropriate than error_number for our * callers */ return (EPERM); } free(linkname_copy); archive_string_free(&error_string); /* * Unlinking and linking here is really not atomic, * but doing it right, would require us to construct * an mktemplink() function, and then use rename(2). */ if (a->flags & ARCHIVE_EXTRACT_SAFE_WRITES) unlink(a->name); #ifdef HAVE_LINKAT r = linkat(AT_FDCWD, linkname, AT_FDCWD, a->name, 0) ? errno : 0; #else r = link(linkname, a->name) ? errno : 0; #endif /* * New cpio and pax formats allow hardlink entries * to carry data, so we may have to open the file * for hardlink entries. * * If the hardlink was successfully created and * the archive doesn't have carry data for it, * consider it to be non-authoritative for meta data. * This is consistent with GNU tar and BSD pax. * If the hardlink does carry data, let the last * archive entry decide ownership. */ if (r == 0 && a->filesize <= 0) { a->todo = 0; a->deferred = 0; } else if (r == 0 && a->filesize > 0) { #ifdef HAVE_LSTAT r = lstat(a->name, &st); #else r = la_stat(a->name, &st); #endif if (r != 0) r = errno; else if ((st.st_mode & AE_IFMT) == AE_IFREG) { a->fd = open(a->name, O_WRONLY | O_TRUNC | O_BINARY | O_CLOEXEC | O_NOFOLLOW); __archive_ensure_cloexec_flag(a->fd); if (a->fd < 0) r = errno; } } return (r); #endif } linkname = archive_entry_symlink(a->entry); if (linkname != NULL) { #if HAVE_SYMLINK /* * Unlinking and linking here is really not atomic, * but doing it right, would require us to construct * an mktempsymlink() function, and then use rename(2). */ if (a->flags & ARCHIVE_EXTRACT_SAFE_WRITES) unlink(a->name); return symlink(linkname, a->name) ? errno : 0; #else return (EPERM); #endif } /* * The remaining system calls all set permissions, so let's * try to take advantage of that to avoid an extra chmod() * call. (Recall that umask is set to zero right now!) */ /* Mode we want for the final restored object (w/o file type bits). */ final_mode = a->mode & 07777; /* * The mode that will actually be restored in this step. Note * that SUID, SGID, etc, require additional work to ensure * security, so we never restore them at this point. */ mode = final_mode & 0777 & ~a->user_umask; /* * Always create writable such that [f]setxattr() works if we're not * root. */ if (a->user_uid != 0 && a->todo & (TODO_HFS_COMPRESSION | TODO_XATTR)) { mode |= 0200; } switch (a->mode & AE_IFMT) { default: /* POSIX requires that we fall through here. */ /* FALLTHROUGH */ case AE_IFREG: a->tmpname = NULL; a->fd = open(a->name, O_WRONLY | O_CREAT | O_EXCL | O_BINARY | O_CLOEXEC, mode); __archive_ensure_cloexec_flag(a->fd); r = (a->fd < 0); break; case AE_IFCHR: #ifdef HAVE_MKNOD /* Note: we use AE_IFCHR for the case label, and * S_IFCHR for the mknod() call. This is correct. */ r = mknod(a->name, mode | S_IFCHR, archive_entry_rdev(a->entry)); break; #else /* TODO: Find a better way to warn about our inability * to restore a char device node. */ return (EINVAL); #endif /* HAVE_MKNOD */ case AE_IFBLK: #ifdef HAVE_MKNOD r = mknod(a->name, mode | S_IFBLK, archive_entry_rdev(a->entry)); break; #else /* TODO: Find a better way to warn about our inability * to restore a block device node. */ return (EINVAL); #endif /* HAVE_MKNOD */ case AE_IFDIR: mode = (mode | MINIMUM_DIR_MODE) & MAXIMUM_DIR_MODE; r = mkdir(a->name, mode); if (r == 0) { /* Defer setting dir times. */ a->deferred |= (a->todo & TODO_TIMES); a->todo &= ~TODO_TIMES; /* Never use an immediate chmod(). */ /* We can't avoid the chmod() entirely if EXTRACT_PERM * because of SysV SGID inheritance. */ if ((mode != final_mode) || (a->flags & ARCHIVE_EXTRACT_PERM)) a->deferred |= (a->todo & TODO_MODE); a->todo &= ~TODO_MODE; } break; case AE_IFIFO: #ifdef HAVE_MKFIFO r = mkfifo(a->name, mode); break; #else /* TODO: Find a better way to warn about our inability * to restore a fifo. */ return (EINVAL); #endif /* HAVE_MKFIFO */ } /* All the system calls above set errno on failure. */ if (r) return (errno); /* If we managed to set the final mode, we've avoided a chmod(). */ if (mode == final_mode) a->todo &= ~TODO_MODE; return (0); } /* * Cleanup function for archive_extract. Mostly, this involves processing * the fixup list, which is used to address a number of problems: * * Dir permissions might prevent us from restoring a file in that * dir, so we restore the dir with minimum 0700 permissions first, * then correct the mode at the end. * * Similarly, the act of restoring a file touches the directory * and changes the timestamp on the dir, so we have to touch-up dir * timestamps at the end as well. * * Some file flags can interfere with the restore by, for example, * preventing the creation of hardlinks to those files. * * Mac OS extended metadata includes ACLs, so must be deferred on dirs. * * Note that tar/cpio do not require that archives be in a particular * order; there is no way to know when the last file has been restored * within a directory, so there's no way to optimize the memory usage * here by fixing up the directory any earlier than the * end-of-archive. * * XXX TODO: Directory ACLs should be restored here, for the same * reason we set directory perms here. XXX */ static int _archive_write_disk_close(struct archive *_a) { struct archive_write_disk *a = (struct archive_write_disk *)_a; struct fixup_entry *next, *p; struct stat st; char *c; int fd, ret, openflags; archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_disk_close"); ret = _archive_write_disk_finish_entry(&a->archive); /* Sort dir list so directories are fixed up in depth-first order. */ p = sort_dir_list(a->fixup_list); while (p != NULL) { fd = -1; a->pst = NULL; /* Mark stat cache as out-of-date. */ /* We must strip trailing slashes from the path to avoid dereferencing symbolic links to directories */ c = p->name; while (*c != '\0') c++; while (c != p->name && *(c - 1) == '/') { c--; *c = '\0'; } if (p->fixup == 0) goto skip_fixup_entry; else { /* * We need to verify if the type of the file * we are going to open matches the file type * of the fixup entry. */ openflags = O_BINARY | O_NOFOLLOW | O_RDONLY | O_CLOEXEC; #if defined(O_DIRECTORY) if (p->filetype == AE_IFDIR) openflags |= O_DIRECTORY; #endif fd = open(p->name, openflags); #if defined(O_DIRECTORY) /* * If we support O_DIRECTORY and open was * successful we can skip the file type check * for directories. For other file types * we need to verify via fstat() or lstat() */ - if (fd == -1 || p->filetype != AE_IFDIR) { + if (fd < 0 || p->filetype != AE_IFDIR) { #if HAVE_FSTAT - if (fd > 0 && ( + if (fd >= 0 && ( fstat(fd, &st) != 0 || la_verify_filetype(st.st_mode, p->filetype) == 0)) { goto skip_fixup_entry; } else #endif if ( #ifdef HAVE_LSTAT lstat(p->name, &st) != 0 || #else la_stat(p->name, &st) != 0 || #endif la_verify_filetype(st.st_mode, p->filetype) == 0) { goto skip_fixup_entry; } } #else #if HAVE_FSTAT if (fd > 0 && ( fstat(fd, &st) != 0 || la_verify_filetype(st.st_mode, p->filetype) == 0)) { goto skip_fixup_entry; } else #endif if ( #ifdef HAVE_LSTAT lstat(p->name, &st) != 0 || #else la_stat(p->name, &st) != 0 || #endif la_verify_filetype(st.st_mode, p->filetype) == 0) { goto skip_fixup_entry; } #endif } if (p->fixup & TODO_TIMES) { set_times(a, fd, p->mode, p->name, p->atime, p->atime_nanos, p->birthtime, p->birthtime_nanos, p->mtime, p->mtime_nanos, p->ctime, p->ctime_nanos); } if (p->fixup & TODO_MODE_BASE) { #ifdef HAVE_FCHMOD if (fd >= 0) fchmod(fd, p->mode & 07777); else #endif #ifdef HAVE_LCHMOD lchmod(p->name, p->mode & 07777); #else chmod(p->name, p->mode & 07777); #endif } if (p->fixup & TODO_ACLS) archive_write_disk_set_acls(&a->archive, fd, p->name, &p->acl, p->mode); if (p->fixup & TODO_FFLAGS) set_fflags_platform(a, fd, p->name, p->mode, p->fflags_set, 0); if (p->fixup & TODO_MAC_METADATA) set_mac_metadata(a, p->name, p->mac_metadata, p->mac_metadata_size); skip_fixup_entry: next = p->next; archive_acl_clear(&p->acl); free(p->mac_metadata); free(p->name); if (fd >= 0) close(fd); free(p); p = next; } a->fixup_list = NULL; return (ret); } static int _archive_write_disk_free(struct archive *_a) { struct archive_write_disk *a; int ret; if (_a == NULL) return (ARCHIVE_OK); archive_check_magic(_a, ARCHIVE_WRITE_DISK_MAGIC, ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_write_disk_free"); a = (struct archive_write_disk *)_a; ret = _archive_write_disk_close(&a->archive); archive_write_disk_set_group_lookup(&a->archive, NULL, NULL, NULL); archive_write_disk_set_user_lookup(&a->archive, NULL, NULL, NULL); archive_entry_free(a->entry); archive_string_free(&a->_name_data); archive_string_free(&a->_tmpname_data); archive_string_free(&a->archive.error_string); archive_string_free(&a->path_safe); a->archive.magic = 0; __archive_clean(&a->archive); free(a->decmpfs_header_p); free(a->resource_fork); free(a->compressed_buffer); free(a->uncompressed_buffer); #if defined(__APPLE__) && defined(UF_COMPRESSED) && defined(HAVE_SYS_XATTR_H)\ && defined(HAVE_ZLIB_H) if (a->stream_valid) { switch (deflateEnd(&a->stream)) { case Z_OK: break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to clean up compressor"); ret = ARCHIVE_FATAL; break; } } #endif free(a); return (ret); } /* * Simple O(n log n) merge sort to order the fixup list. In * particular, we want to restore dir timestamps depth-first. */ static struct fixup_entry * sort_dir_list(struct fixup_entry *p) { struct fixup_entry *a, *b, *t; if (p == NULL) return (NULL); /* A one-item list is already sorted. */ if (p->next == NULL) return (p); /* Step 1: split the list. */ t = p; a = p->next->next; while (a != NULL) { /* Step a twice, t once. */ a = a->next; if (a != NULL) a = a->next; t = t->next; } /* Now, t is at the mid-point, so break the list here. */ b = t->next; t->next = NULL; a = p; /* Step 2: Recursively sort the two sub-lists. */ a = sort_dir_list(a); b = sort_dir_list(b); /* Step 3: Merge the returned lists. */ /* Pick the first element for the merged list. */ if (strcmp(a->name, b->name) > 0) { t = p = a; a = a->next; } else { t = p = b; b = b->next; } /* Always put the later element on the list first. */ while (a != NULL && b != NULL) { if (strcmp(a->name, b->name) > 0) { t->next = a; a = a->next; } else { t->next = b; b = b->next; } t = t->next; } /* Only one list is non-empty, so just splice it on. */ if (a != NULL) t->next = a; if (b != NULL) t->next = b; return (p); } /* * Returns a new, initialized fixup entry. * * TODO: Reduce the memory requirements for this list by using a tree * structure rather than a simple list of names. */ static struct fixup_entry * new_fixup(struct archive_write_disk *a, const char *pathname) { struct fixup_entry *fe; fe = calloc(1, sizeof(struct fixup_entry)); if (fe == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for a fixup"); return (NULL); } fe->next = a->fixup_list; a->fixup_list = fe; fe->fixup = 0; fe->filetype = 0; fe->name = strdup(pathname); return (fe); } /* * Returns a fixup structure for the current entry. */ static struct fixup_entry * current_fixup(struct archive_write_disk *a, const char *pathname) { if (a->current_fixup == NULL) a->current_fixup = new_fixup(a, pathname); return (a->current_fixup); } /* Error helper for new *_fsobj functions */ static void fsobj_error(int *a_eno, struct archive_string *a_estr, int err, const char *errstr, const char *path) { if (a_eno) *a_eno = err; if (a_estr) archive_string_sprintf(a_estr, "%s%s", errstr, path); } /* * TODO: Someday, integrate this with the deep dir support; they both * scan the path and both can be optimized by comparing against other * recent paths. */ /* * Checks the given path to see if any elements along it are symlinks. Returns * ARCHIVE_OK if there are none, otherwise puts an error in errmsg. */ static int check_symlinks_fsobj(char *path, int *a_eno, struct archive_string *a_estr, int flags, int checking_linkname) { #if !defined(HAVE_LSTAT) && \ !(defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT)) /* Platform doesn't have lstat, so we can't look for symlinks. */ (void)path; /* UNUSED */ (void)a_eno; /* UNUSED */ (void)a_estr; /* UNUSED */ (void)flags; /* UNUSED */ (void)checking_linkname; /* UNUSED */ return (ARCHIVE_OK); #else int res = ARCHIVE_OK; char *tail; char *head; int last; char c = '\0'; int r; struct stat st; int chdir_fd; #if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) int fd; #endif /* Nothing to do here if name is empty */ if(path[0] == '\0') return (ARCHIVE_OK); /* * Guard against symlink tricks. Reject any archive entry whose * destination would be altered by a symlink. * * Walk the filename in chunks separated by '/'. For each segment: * - if it doesn't exist, continue * - if it's symlink, abort or remove it * - if it's a directory and it's not the last chunk, cd into it * As we go: * head points to the current (relative) path * tail points to the temporary \0 terminating the segment we're * currently examining * c holds what used to be in *tail * last is 1 if this is the last tail */ chdir_fd = la_opendirat(AT_FDCWD, "."); __archive_ensure_cloexec_flag(chdir_fd); if (chdir_fd < 0) { fsobj_error(a_eno, a_estr, errno, "Could not open ", path); return (ARCHIVE_FATAL); } head = path; tail = path; last = 0; /* TODO: reintroduce a safe cache here? */ /* Skip the root directory if the path is absolute. */ if(tail == path && tail[0] == '/') ++tail; /* Keep going until we've checked the entire name. * head, tail, path all alias the same string, which is * temporarily zeroed at tail, so be careful restoring the * stashed (c=tail[0]) for error messages. * Exiting the loop with break is okay; continue is not. */ while (!last) { /* * Skip the separator we just consumed, plus any adjacent ones */ while (*tail == '/') ++tail; /* Skip the next path element. */ while (*tail != '\0' && *tail != '/') ++tail; /* is this the last path component? */ last = (tail[0] == '\0') || (tail[0] == '/' && tail[1] == '\0'); /* temporarily truncate the string here */ c = tail[0]; tail[0] = '\0'; /* Check that we haven't hit a symlink. */ #if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) r = fstatat(chdir_fd, head, &st, AT_SYMLINK_NOFOLLOW); #elif defined(HAVE_LSTAT) r = lstat(head, &st); #else r = la_stat(head, &st); #endif if (r != 0) { tail[0] = c; /* We've hit a dir that doesn't exist; stop now. */ if (errno == ENOENT) { break; } else { /* * Treat any other error as fatal - best to be * paranoid here. * Note: This effectively disables deep * directory support when security checks are * enabled. Otherwise, very long pathnames that * trigger an error here could evade the * sandbox. * TODO: We could do better, but it would * probably require merging the symlink checks * with the deep-directory editing. */ fsobj_error(a_eno, a_estr, errno, "Could not stat ", path); res = ARCHIVE_FAILED; break; } } else if (S_ISDIR(st.st_mode)) { if (!last) { #if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) fd = la_opendirat(chdir_fd, head); if (fd < 0) r = -1; else { r = 0; close(chdir_fd); chdir_fd = fd; } #else r = chdir(head); #endif if (r != 0) { tail[0] = c; fsobj_error(a_eno, a_estr, errno, "Could not chdir ", path); res = (ARCHIVE_FATAL); break; } /* Our view is now from inside this dir: */ head = tail + 1; } } else if (S_ISLNK(st.st_mode)) { if (last && checking_linkname) { #ifdef HAVE_LINKAT /* * Hardlinks to symlinks are safe to write * if linkat() is supported as it does not * follow symlinks. */ res = ARCHIVE_OK; #else /* * We return ARCHIVE_FAILED here as we are * not able to safely write hardlinks * to symlinks. */ tail[0] = c; fsobj_error(a_eno, a_estr, errno, "Cannot write hardlink to symlink ", path); res = ARCHIVE_FAILED; #endif break; } else if (last) { /* * Last element is symlink; remove it * so we can overwrite it with the * item being extracted. */ #if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) r = unlinkat(chdir_fd, head, 0); #else r = unlink(head); #endif if (r != 0) { tail[0] = c; fsobj_error(a_eno, a_estr, errno, "Could not remove symlink ", path); res = ARCHIVE_FAILED; break; } /* * Even if we did remove it, a warning * is in order. The warning is silly, * though, if we're just replacing one * symlink with another symlink. */ tail[0] = c; /* * FIXME: not sure how important this is to * restore */ /* if (!S_ISLNK(path)) { fsobj_error(a_eno, a_estr, 0, "Removing symlink ", path); } */ /* Symlink gone. No more problem! */ res = ARCHIVE_OK; break; } else if (flags & ARCHIVE_EXTRACT_UNLINK) { /* User asked us to remove problems. */ #if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) r = unlinkat(chdir_fd, head, 0); #else r = unlink(head); #endif if (r != 0) { tail[0] = c; fsobj_error(a_eno, a_estr, 0, "Cannot remove intervening " "symlink ", path); res = ARCHIVE_FAILED; break; } tail[0] = c; } else if ((flags & ARCHIVE_EXTRACT_SECURE_SYMLINKS) == 0) { /* * We are not the last element and we want to * follow symlinks if they are a directory. * * This is needed to extract hardlinks over * symlinks. */ #if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) r = fstatat(chdir_fd, head, &st, 0); #else r = la_stat(head, &st); #endif if (r != 0) { tail[0] = c; if (errno == ENOENT) { break; } else { fsobj_error(a_eno, a_estr, errno, "Could not stat ", path); res = (ARCHIVE_FAILED); break; } } else if (S_ISDIR(st.st_mode)) { #if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) fd = la_opendirat(chdir_fd, head); if (fd < 0) r = -1; else { r = 0; close(chdir_fd); chdir_fd = fd; } #else r = chdir(head); #endif if (r != 0) { tail[0] = c; fsobj_error(a_eno, a_estr, errno, "Could not chdir ", path); res = (ARCHIVE_FATAL); break; } /* * Our view is now from inside * this dir: */ head = tail + 1; } else { tail[0] = c; fsobj_error(a_eno, a_estr, 0, "Cannot extract through " "symlink ", path); res = ARCHIVE_FAILED; break; } } else { tail[0] = c; fsobj_error(a_eno, a_estr, 0, "Cannot extract through symlink ", path); res = ARCHIVE_FAILED; break; } } /* be sure to always maintain this */ tail[0] = c; if (tail[0] != '\0') tail++; /* Advance to the next segment. */ } /* Catches loop exits via break */ tail[0] = c; #if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) /* If we operate with openat(), fstatat() and unlinkat() there was * no chdir(), so just close the fd */ if (chdir_fd >= 0) close(chdir_fd); #elif HAVE_FCHDIR /* If we changed directory above, restore it here. */ if (chdir_fd >= 0) { r = fchdir(chdir_fd); if (r != 0) { fsobj_error(a_eno, a_estr, errno, "chdir() failure", ""); } close(chdir_fd); chdir_fd = -1; if (r != 0) { res = (ARCHIVE_FATAL); } } #endif /* TODO: reintroduce a safe cache here? */ return res; #endif } /* * Check a->name for symlinks, returning ARCHIVE_OK if its clean, otherwise * calls archive_set_error and returns ARCHIVE_{FATAL,FAILED} */ static int check_symlinks(struct archive_write_disk *a) { struct archive_string error_string; int error_number; int rc; archive_string_init(&error_string); rc = check_symlinks_fsobj(a->name, &error_number, &error_string, a->flags, 0); if (rc != ARCHIVE_OK) { archive_set_error(&a->archive, error_number, "%s", error_string.s); } archive_string_free(&error_string); a->pst = NULL; /* to be safe */ return rc; } #if defined(__CYGWIN__) /* * 1. Convert a path separator from '\' to '/' . * We shouldn't check multibyte character directly because some * character-set have been using the '\' character for a part of * its multibyte character code. * 2. Replace unusable characters in Windows with underscore('_'). * See also : http://msdn.microsoft.com/en-us/library/aa365247.aspx */ static void cleanup_pathname_win(char *path) { wchar_t wc; char *p; size_t alen, l; int mb, complete, utf8; alen = 0; mb = 0; complete = 1; utf8 = (strcmp(nl_langinfo(CODESET), "UTF-8") == 0)? 1: 0; for (p = path; *p != '\0'; p++) { ++alen; if (*p == '\\') { /* If previous byte is smaller than 128, * this is not second byte of multibyte characters, * so we can replace '\' with '/'. */ if (utf8 || !mb) *p = '/'; else complete = 0;/* uncompleted. */ } else if (*(unsigned char *)p > 127) mb = 1; else mb = 0; /* Rewrite the path name if its next character is unusable. */ if (*p == ':' || *p == '*' || *p == '?' || *p == '"' || *p == '<' || *p == '>' || *p == '|') *p = '_'; } if (complete) return; /* * Convert path separator in wide-character. */ p = path; while (*p != '\0' && alen) { l = mbtowc(&wc, p, alen); if (l == (size_t)-1) { while (*p != '\0') { if (*p == '\\') *p = '/'; ++p; } break; } if (l == 1 && wc == L'\\') *p = '/'; p += l; alen -= l; } } #endif /* * Canonicalize the pathname. In particular, this strips duplicate * '/' characters, '.' elements, and trailing '/'. It also raises an * error for an empty path, a trailing '..', (if _SECURE_NODOTDOT is * set) any '..' in the path or (if ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS * is set) if the path is absolute. */ static int cleanup_pathname_fsobj(char *path, int *a_eno, struct archive_string *a_estr, int flags) { char *dest, *src; char separator = '\0'; dest = src = path; if (*src == '\0') { fsobj_error(a_eno, a_estr, ARCHIVE_ERRNO_MISC, "Invalid empty ", "pathname"); return (ARCHIVE_FAILED); } #if defined(__CYGWIN__) cleanup_pathname_win(path); #endif /* Skip leading '/'. */ if (*src == '/') { if (flags & ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS) { fsobj_error(a_eno, a_estr, ARCHIVE_ERRNO_MISC, "Path is ", "absolute"); return (ARCHIVE_FAILED); } separator = *src++; } /* Scan the pathname one element at a time. */ for (;;) { /* src points to first char after '/' */ if (src[0] == '\0') { break; } else if (src[0] == '/') { /* Found '//', ignore second one. */ src++; continue; } else if (src[0] == '.') { if (src[1] == '\0') { /* Ignore trailing '.' */ break; } else if (src[1] == '/') { /* Skip './'. */ src += 2; continue; } else if (src[1] == '.') { if (src[2] == '/' || src[2] == '\0') { /* Conditionally warn about '..' */ if (flags & ARCHIVE_EXTRACT_SECURE_NODOTDOT) { fsobj_error(a_eno, a_estr, ARCHIVE_ERRNO_MISC, "Path contains ", "'..'"); return (ARCHIVE_FAILED); } } /* * Note: Under no circumstances do we * remove '..' elements. In * particular, restoring * '/foo/../bar/' should create the * 'foo' dir as a side-effect. */ } } /* Copy current element, including leading '/'. */ if (separator) *dest++ = '/'; while (*src != '\0' && *src != '/') { *dest++ = *src++; } if (*src == '\0') break; /* Skip '/' separator. */ separator = *src++; } /* * We've just copied zero or more path elements, not including the * final '/'. */ if (dest == path) { /* * Nothing got copied. The path must have been something * like '.' or '/' or './' or '/././././/./'. */ if (separator) *dest++ = '/'; else *dest++ = '.'; } /* Terminate the result. */ *dest = '\0'; return (ARCHIVE_OK); } static int cleanup_pathname(struct archive_write_disk *a) { struct archive_string error_string; int error_number; int rc; archive_string_init(&error_string); rc = cleanup_pathname_fsobj(a->name, &error_number, &error_string, a->flags); if (rc != ARCHIVE_OK) { archive_set_error(&a->archive, error_number, "%s", error_string.s); } archive_string_free(&error_string); return rc; } /* * Create the parent directory of the specified path, assuming path * is already in mutable storage. */ static int create_parent_dir(struct archive_write_disk *a, char *path) { char *slash; int r; /* Remove tail element to obtain parent name. */ slash = strrchr(path, '/'); if (slash == NULL) return (ARCHIVE_OK); *slash = '\0'; r = create_dir(a, path); *slash = '/'; return (r); } /* * Create the specified dir, recursing to create parents as necessary. * * Returns ARCHIVE_OK if the path exists when we're done here. * Otherwise, returns ARCHIVE_FAILED. * Assumes path is in mutable storage; path is unchanged on exit. */ static int create_dir(struct archive_write_disk *a, char *path) { struct stat st; struct fixup_entry *le; char *slash, *base; mode_t mode_final, mode; int r; /* Check for special names and just skip them. */ slash = strrchr(path, '/'); if (slash == NULL) base = path; else base = slash + 1; if (base[0] == '\0' || (base[0] == '.' && base[1] == '\0') || (base[0] == '.' && base[1] == '.' && base[2] == '\0')) { /* Don't bother trying to create null path, '.', or '..'. */ if (slash != NULL) { *slash = '\0'; r = create_dir(a, path); *slash = '/'; return (r); } return (ARCHIVE_OK); } /* * Yes, this should be stat() and not lstat(). Using lstat() * here loses the ability to extract through symlinks. Also note * that this should not use the a->st cache. */ if (la_stat(path, &st) == 0) { if (S_ISDIR(st.st_mode)) return (ARCHIVE_OK); if ((a->flags & ARCHIVE_EXTRACT_NO_OVERWRITE)) { archive_set_error(&a->archive, EEXIST, "Can't create directory '%s'", path); return (ARCHIVE_FAILED); } if (unlink(path) != 0) { archive_set_error(&a->archive, errno, "Can't create directory '%s': " "Conflicting file cannot be removed", path); return (ARCHIVE_FAILED); } } else if (errno != ENOENT && errno != ENOTDIR) { /* Stat failed? */ archive_set_error(&a->archive, errno, "Can't test directory '%s'", path); return (ARCHIVE_FAILED); } else if (slash != NULL) { *slash = '\0'; r = create_dir(a, path); *slash = '/'; if (r != ARCHIVE_OK) return (r); } /* * Mode we want for the final restored directory. Per POSIX, * implicitly-created dirs must be created obeying the umask. * There's no mention whether this is different for privileged * restores (which the rest of this code handles by pretending * umask=0). I've chosen here to always obey the user's umask for * implicit dirs, even if _EXTRACT_PERM was specified. */ mode_final = DEFAULT_DIR_MODE & ~a->user_umask; /* Mode we want on disk during the restore process. */ mode = mode_final; mode |= MINIMUM_DIR_MODE; mode &= MAXIMUM_DIR_MODE; if (mkdir(path, mode) == 0) { if (mode != mode_final) { le = new_fixup(a, path); if (le == NULL) return (ARCHIVE_FATAL); le->fixup |=TODO_MODE_BASE; le->mode = mode_final; } return (ARCHIVE_OK); } /* * Without the following check, a/b/../b/c/d fails at the * second visit to 'b', so 'd' can't be created. Note that we * don't add it to the fixup list here, as it's already been * added. */ if (la_stat(path, &st) == 0 && S_ISDIR(st.st_mode)) return (ARCHIVE_OK); archive_set_error(&a->archive, errno, "Failed to create dir '%s'", path); return (ARCHIVE_FAILED); } /* * Note: Although we can skip setting the user id if the desired user * id matches the current user, we cannot skip setting the group, as * many systems set the gid based on the containing directory. So * we have to perform a chown syscall if we want to set the SGID * bit. (The alternative is to stat() and then possibly chown(); it's * more efficient to skip the stat() and just always chown().) Note * that a successful chown() here clears the TODO_SGID_CHECK bit, which * allows set_mode to skip the stat() check for the GID. */ static int set_ownership(struct archive_write_disk *a) { #if !defined(__CYGWIN__) && !defined(__linux__) /* * On Linux, a process may have the CAP_CHOWN capability. * On Windows there is no 'root' user with uid 0. * Elsewhere we can skip calling chown if we are not root and the desired * user id does not match the current user. */ if (a->user_uid != 0 && a->user_uid != a->uid) { archive_set_error(&a->archive, errno, "Can't set UID=%jd", (intmax_t)a->uid); return (ARCHIVE_WARN); } #endif #ifdef HAVE_FCHOWN /* If we have an fd, we can avoid a race. */ if (a->fd >= 0 && fchown(a->fd, a->uid, a->gid) == 0) { /* We've set owner and know uid/gid are correct. */ a->todo &= ~(TODO_OWNER | TODO_SGID_CHECK | TODO_SUID_CHECK); return (ARCHIVE_OK); } #endif /* We prefer lchown() but will use chown() if that's all we have. */ /* Of course, if we have neither, this will always fail. */ #ifdef HAVE_LCHOWN if (lchown(a->name, a->uid, a->gid) == 0) { /* We've set owner and know uid/gid are correct. */ a->todo &= ~(TODO_OWNER | TODO_SGID_CHECK | TODO_SUID_CHECK); return (ARCHIVE_OK); } #elif HAVE_CHOWN if (!S_ISLNK(a->mode) && chown(a->name, a->uid, a->gid) == 0) { /* We've set owner and know uid/gid are correct. */ a->todo &= ~(TODO_OWNER | TODO_SGID_CHECK | TODO_SUID_CHECK); return (ARCHIVE_OK); } #endif archive_set_error(&a->archive, errno, "Can't set user=%jd/group=%jd for %s", (intmax_t)a->uid, (intmax_t)a->gid, a->name); return (ARCHIVE_WARN); } /* * Note: Returns 0 on success, non-zero on failure. */ static int set_time(int fd, int mode, const char *name, time_t atime, long atime_nsec, time_t mtime, long mtime_nsec) { /* Select the best implementation for this platform. */ #if defined(HAVE_UTIMENSAT) && defined(HAVE_FUTIMENS) /* * utimensat() and futimens() are defined in * POSIX.1-2008. They support ns resolution and setting times * on fds and symlinks. */ struct timespec ts[2]; (void)mode; /* UNUSED */ ts[0].tv_sec = atime; ts[0].tv_nsec = atime_nsec; ts[1].tv_sec = mtime; ts[1].tv_nsec = mtime_nsec; if (fd >= 0) return futimens(fd, ts); return utimensat(AT_FDCWD, name, ts, AT_SYMLINK_NOFOLLOW); #elif HAVE_UTIMES /* * The utimes()-family functions support µs-resolution and * setting times fds and symlinks. utimes() is documented as * LEGACY by POSIX, futimes() and lutimes() are not described * in POSIX. */ struct timeval times[2]; times[0].tv_sec = atime; times[0].tv_usec = atime_nsec / 1000; times[1].tv_sec = mtime; times[1].tv_usec = mtime_nsec / 1000; #ifdef HAVE_FUTIMES if (fd >= 0) return (futimes(fd, times)); #else (void)fd; /* UNUSED */ #endif #ifdef HAVE_LUTIMES (void)mode; /* UNUSED */ return (lutimes(name, times)); #else if (S_ISLNK(mode)) return (0); return (utimes(name, times)); #endif #elif defined(HAVE_UTIME) /* * utime() is POSIX-standard but only supports 1s resolution and * does not support fds or symlinks. */ struct utimbuf times; (void)fd; /* UNUSED */ (void)name; /* UNUSED */ (void)atime_nsec; /* UNUSED */ (void)mtime_nsec; /* UNUSED */ times.actime = atime; times.modtime = mtime; if (S_ISLNK(mode)) return (ARCHIVE_OK); return (utime(name, ×)); #else /* * We don't know how to set the time on this platform. */ (void)fd; /* UNUSED */ (void)mode; /* UNUSED */ (void)name; /* UNUSED */ (void)atime; /* UNUSED */ (void)atime_nsec; /* UNUSED */ (void)mtime; /* UNUSED */ (void)mtime_nsec; /* UNUSED */ return (ARCHIVE_WARN); #endif } #ifdef F_SETTIMES static int set_time_tru64(int fd, int mode, const char *name, time_t atime, long atime_nsec, time_t mtime, long mtime_nsec, time_t ctime, long ctime_nsec) { struct attr_timbuf tstamp; tstamp.atime.tv_sec = atime; tstamp.mtime.tv_sec = mtime; tstamp.ctime.tv_sec = ctime; #if defined (__hpux) && ( defined (__ia64) || defined (__hppa) ) tstamp.atime.tv_nsec = atime_nsec; tstamp.mtime.tv_nsec = mtime_nsec; tstamp.ctime.tv_nsec = ctime_nsec; #else tstamp.atime.tv_usec = atime_nsec / 1000; tstamp.mtime.tv_usec = mtime_nsec / 1000; tstamp.ctime.tv_usec = ctime_nsec / 1000; #endif return (fcntl(fd,F_SETTIMES,&tstamp)); } #endif /* F_SETTIMES */ static int set_times(struct archive_write_disk *a, int fd, int mode, const char *name, time_t atime, long atime_nanos, time_t birthtime, long birthtime_nanos, time_t mtime, long mtime_nanos, time_t cctime, long ctime_nanos) { /* Note: set_time doesn't use libarchive return conventions! * It uses syscall conventions. So 0 here instead of ARCHIVE_OK. */ int r1 = 0, r2 = 0; #ifdef F_SETTIMES /* * on Tru64 try own fcntl first which can restore even the * ctime, fall back to default code path below if it fails * or if we are not running as root */ if (a->user_uid == 0 && set_time_tru64(fd, mode, name, atime, atime_nanos, mtime, mtime_nanos, cctime, ctime_nanos) == 0) { return (ARCHIVE_OK); } #else /* Tru64 */ (void)cctime; /* UNUSED */ (void)ctime_nanos; /* UNUSED */ #endif /* Tru64 */ #ifdef HAVE_STRUCT_STAT_ST_BIRTHTIME /* * If you have struct stat.st_birthtime, we assume BSD * birthtime semantics, in which {f,l,}utimes() updates * birthtime to earliest mtime. So we set the time twice, * first using the birthtime, then using the mtime. If * birthtime == mtime, this isn't necessary, so we skip it. * If birthtime > mtime, then this won't work, so we skip it. */ if (birthtime < mtime || (birthtime == mtime && birthtime_nanos < mtime_nanos)) r1 = set_time(fd, mode, name, atime, atime_nanos, birthtime, birthtime_nanos); #else (void)birthtime; /* UNUSED */ (void)birthtime_nanos; /* UNUSED */ #endif r2 = set_time(fd, mode, name, atime, atime_nanos, mtime, mtime_nanos); if (r1 != 0 || r2 != 0) { archive_set_error(&a->archive, errno, "Can't restore time"); return (ARCHIVE_WARN); } return (ARCHIVE_OK); } static int set_times_from_entry(struct archive_write_disk *a) { time_t atime, birthtime, mtime, cctime; long atime_nsec, birthtime_nsec, mtime_nsec, ctime_nsec; /* Suitable defaults. */ atime = birthtime = mtime = cctime = a->start_time; atime_nsec = birthtime_nsec = mtime_nsec = ctime_nsec = 0; /* If no time was provided, we're done. */ if (!archive_entry_atime_is_set(a->entry) #if HAVE_STRUCT_STAT_ST_BIRTHTIME && !archive_entry_birthtime_is_set(a->entry) #endif && !archive_entry_mtime_is_set(a->entry)) return (ARCHIVE_OK); if (archive_entry_atime_is_set(a->entry)) { atime = archive_entry_atime(a->entry); atime_nsec = archive_entry_atime_nsec(a->entry); } if (archive_entry_birthtime_is_set(a->entry)) { birthtime = archive_entry_birthtime(a->entry); birthtime_nsec = archive_entry_birthtime_nsec(a->entry); } if (archive_entry_mtime_is_set(a->entry)) { mtime = archive_entry_mtime(a->entry); mtime_nsec = archive_entry_mtime_nsec(a->entry); } if (archive_entry_ctime_is_set(a->entry)) { cctime = archive_entry_ctime(a->entry); ctime_nsec = archive_entry_ctime_nsec(a->entry); } return set_times(a, a->fd, a->mode, a->name, atime, atime_nsec, birthtime, birthtime_nsec, mtime, mtime_nsec, cctime, ctime_nsec); } static int set_mode(struct archive_write_disk *a, int mode) { int r = ARCHIVE_OK; int r2; mode &= 07777; /* Strip off file type bits. */ if (a->todo & TODO_SGID_CHECK) { /* * If we don't know the GID is right, we must stat() * to verify it. We can't just check the GID of this * process, since systems sometimes set GID from * the enclosing dir or based on ACLs. */ if ((r = lazy_stat(a)) != ARCHIVE_OK) return (r); if (a->pst->st_gid != a->gid) { mode &= ~ S_ISGID; if (a->flags & ARCHIVE_EXTRACT_OWNER) { /* * This is only an error if you * requested owner restore. If you * didn't, we'll try to restore * sgid/suid, but won't consider it a * problem if we can't. */ archive_set_error(&a->archive, -1, "Can't restore SGID bit"); r = ARCHIVE_WARN; } } /* While we're here, double-check the UID. */ if (a->pst->st_uid != a->uid && (a->todo & TODO_SUID)) { mode &= ~ S_ISUID; if (a->flags & ARCHIVE_EXTRACT_OWNER) { archive_set_error(&a->archive, -1, "Can't restore SUID bit"); r = ARCHIVE_WARN; } } a->todo &= ~TODO_SGID_CHECK; a->todo &= ~TODO_SUID_CHECK; } else if (a->todo & TODO_SUID_CHECK) { /* * If we don't know the UID is right, we can just check * the user, since all systems set the file UID from * the process UID. */ if (a->user_uid != a->uid) { mode &= ~ S_ISUID; if (a->flags & ARCHIVE_EXTRACT_OWNER) { archive_set_error(&a->archive, -1, "Can't make file SUID"); r = ARCHIVE_WARN; } } a->todo &= ~TODO_SUID_CHECK; } if (S_ISLNK(a->mode)) { #ifdef HAVE_LCHMOD /* * If this is a symlink, use lchmod(). If the * platform doesn't support lchmod(), just skip it. A * platform that doesn't provide a way to set * permissions on symlinks probably ignores * permissions on symlinks, so a failure here has no * impact. */ if (lchmod(a->name, (mode_t)mode) != 0) { switch (errno) { case ENOTSUP: case ENOSYS: #if ENOTSUP != EOPNOTSUPP case EOPNOTSUPP: #endif /* * if lchmod is defined but the platform * doesn't support it, silently ignore * error */ break; default: archive_set_error(&a->archive, errno, "Can't set permissions to 0%o", (unsigned int)mode); r = ARCHIVE_WARN; } } #endif } else if (!S_ISDIR(a->mode)) { /* * If it's not a symlink and not a dir, then use * fchmod() or chmod(), depending on whether we have * an fd. Dirs get their perms set during the * post-extract fixup, which is handled elsewhere. */ #ifdef HAVE_FCHMOD if (a->fd >= 0) r2 = fchmod(a->fd, (mode_t)mode); else #endif /* If this platform lacks fchmod(), then * we'll just use chmod(). */ r2 = chmod(a->name, (mode_t)mode); if (r2 != 0) { archive_set_error(&a->archive, errno, "Can't set permissions to 0%o", (unsigned int)mode); r = ARCHIVE_WARN; } } return (r); } static int set_fflags(struct archive_write_disk *a) { struct fixup_entry *le; unsigned long set, clear; int r; mode_t mode = archive_entry_mode(a->entry); /* * Make 'critical_flags' hold all file flags that can't be * immediately restored. For example, on BSD systems, * SF_IMMUTABLE prevents hardlinks from being created, so * should not be set until after any hardlinks are created. To * preserve some semblance of portability, this uses #ifdef * extensively. Ugly, but it works. * * Yes, Virginia, this does create a security race. It's mitigated * somewhat by the practice of creating dirs 0700 until the extract * is done, but it would be nice if we could do more than that. * People restoring critical file systems should be wary of * other programs that might try to muck with files as they're * being restored. */ const int critical_flags = 0 #ifdef SF_IMMUTABLE | SF_IMMUTABLE #endif #ifdef UF_IMMUTABLE | UF_IMMUTABLE #endif #ifdef SF_APPEND | SF_APPEND #endif #ifdef UF_APPEND | UF_APPEND #endif #if defined(FS_APPEND_FL) | FS_APPEND_FL #elif defined(EXT2_APPEND_FL) | EXT2_APPEND_FL #endif #if defined(FS_IMMUTABLE_FL) | FS_IMMUTABLE_FL #elif defined(EXT2_IMMUTABLE_FL) | EXT2_IMMUTABLE_FL #endif #ifdef FS_JOURNAL_DATA_FL | FS_JOURNAL_DATA_FL #endif ; if (a->todo & TODO_FFLAGS) { archive_entry_fflags(a->entry, &set, &clear); /* * The first test encourages the compiler to eliminate * all of this if it's not necessary. */ if ((critical_flags != 0) && (set & critical_flags)) { le = current_fixup(a, a->name); if (le == NULL) return (ARCHIVE_FATAL); le->filetype = archive_entry_filetype(a->entry); le->fixup |= TODO_FFLAGS; le->fflags_set = set; /* Store the mode if it's not already there. */ if ((le->fixup & TODO_MODE) == 0) le->mode = mode; } else { r = set_fflags_platform(a, a->fd, a->name, mode, set, clear); if (r != ARCHIVE_OK) return (r); } } return (ARCHIVE_OK); } static int clear_nochange_fflags(struct archive_write_disk *a) { mode_t mode = archive_entry_mode(a->entry); const int nochange_flags = 0 #ifdef SF_IMMUTABLE | SF_IMMUTABLE #endif #ifdef UF_IMMUTABLE | UF_IMMUTABLE #endif #ifdef SF_APPEND | SF_APPEND #endif #ifdef UF_APPEND | UF_APPEND #endif -#ifdef EXT2_APPEND_FL +#if defined(FS_APPEND_FL) + | FS_APPEND_FL +#elif defined(EXT2_APPEND_FL) | EXT2_APPEND_FL #endif -#ifdef EXT2_IMMUTABLE_FL +#if defined(FS_IMMUTABLE_FL) + | FS_IMMUTABLE_FL +#elif defined(EXT2_IMMUTABLE_FL) | EXT2_IMMUTABLE_FL #endif ; return (set_fflags_platform(a, a->fd, a->name, mode, 0, nochange_flags)); } #if ( defined(HAVE_LCHFLAGS) || defined(HAVE_CHFLAGS) || defined(HAVE_FCHFLAGS) ) && defined(HAVE_STRUCT_STAT_ST_FLAGS) /* * BSD reads flags using stat() and sets them with one of {f,l,}chflags() */ static int set_fflags_platform(struct archive_write_disk *a, int fd, const char *name, mode_t mode, unsigned long set, unsigned long clear) { int r; const int sf_mask = 0 #ifdef SF_APPEND | SF_APPEND #endif #ifdef SF_ARCHIVED | SF_ARCHIVED #endif #ifdef SF_IMMUTABLE | SF_IMMUTABLE #endif #ifdef SF_NOUNLINK | SF_NOUNLINK #endif ; (void)mode; /* UNUSED */ if (set == 0 && clear == 0) return (ARCHIVE_OK); /* * XXX Is the stat here really necessary? Or can I just use * the 'set' flags directly? In particular, I'm not sure * about the correct approach if we're overwriting an existing * file that already has flags on it. XXX */ if ((r = lazy_stat(a)) != ARCHIVE_OK) return (r); a->st.st_flags &= ~clear; a->st.st_flags |= set; /* Only super-user may change SF_* flags */ if (a->user_uid != 0) a->st.st_flags &= ~sf_mask; #ifdef HAVE_FCHFLAGS /* If platform has fchflags() and we were given an fd, use it. */ if (fd >= 0 && fchflags(fd, a->st.st_flags) == 0) return (ARCHIVE_OK); #endif /* * If we can't use the fd to set the flags, we'll use the * pathname to set flags. We prefer lchflags() but will use * chflags() if we must. */ #ifdef HAVE_LCHFLAGS if (lchflags(name, a->st.st_flags) == 0) return (ARCHIVE_OK); #elif defined(HAVE_CHFLAGS) if (S_ISLNK(a->st.st_mode)) { archive_set_error(&a->archive, errno, "Can't set file flags on symlink."); return (ARCHIVE_WARN); } if (chflags(name, a->st.st_flags) == 0) return (ARCHIVE_OK); #endif archive_set_error(&a->archive, errno, "Failed to set file flags"); return (ARCHIVE_WARN); } #elif (defined(FS_IOC_GETFLAGS) && defined(FS_IOC_SETFLAGS) && \ defined(HAVE_WORKING_FS_IOC_GETFLAGS)) || \ (defined(EXT2_IOC_GETFLAGS) && defined(EXT2_IOC_SETFLAGS) && \ defined(HAVE_WORKING_EXT2_IOC_GETFLAGS)) /* * Linux uses ioctl() to read and write file flags. */ static int set_fflags_platform(struct archive_write_disk *a, int fd, const char *name, mode_t mode, unsigned long set, unsigned long clear) { int ret; int myfd = fd; int newflags, oldflags; /* * Linux has no define for the flags that are only settable by * the root user. This code may seem a little complex, but * there seem to be some Linux systems that lack these * defines. (?) The code below degrades reasonably gracefully * if sf_mask is incomplete. */ const int sf_mask = 0 #if defined(FS_IMMUTABLE_FL) | FS_IMMUTABLE_FL #elif defined(EXT2_IMMUTABLE_FL) | EXT2_IMMUTABLE_FL #endif #if defined(FS_APPEND_FL) | FS_APPEND_FL #elif defined(EXT2_APPEND_FL) | EXT2_APPEND_FL #endif #if defined(FS_JOURNAL_DATA_FL) | FS_JOURNAL_DATA_FL #endif ; if (set == 0 && clear == 0) return (ARCHIVE_OK); /* Only regular files and dirs can have flags. */ if (!S_ISREG(mode) && !S_ISDIR(mode)) return (ARCHIVE_OK); /* If we weren't given an fd, open it ourselves. */ if (myfd < 0) { myfd = open(name, O_RDONLY | O_NONBLOCK | O_BINARY | O_CLOEXEC | O_NOFOLLOW); __archive_ensure_cloexec_flag(myfd); } if (myfd < 0) return (ARCHIVE_OK); /* * XXX As above, this would be way simpler if we didn't have * to read the current flags from disk. XXX */ ret = ARCHIVE_OK; /* Read the current file flags. */ if (ioctl(myfd, #ifdef FS_IOC_GETFLAGS FS_IOC_GETFLAGS, #else EXT2_IOC_GETFLAGS, #endif &oldflags) < 0) goto fail; /* Try setting the flags as given. */ newflags = (oldflags & ~clear) | set; if (ioctl(myfd, #ifdef FS_IOC_SETFLAGS FS_IOC_SETFLAGS, #else EXT2_IOC_SETFLAGS, #endif &newflags) >= 0) goto cleanup; if (errno != EPERM) goto fail; /* If we couldn't set all the flags, try again with a subset. */ newflags &= ~sf_mask; oldflags &= sf_mask; newflags |= oldflags; if (ioctl(myfd, #ifdef FS_IOC_SETFLAGS FS_IOC_SETFLAGS, #else EXT2_IOC_SETFLAGS, #endif &newflags) >= 0) goto cleanup; /* We couldn't set the flags, so report the failure. */ fail: archive_set_error(&a->archive, errno, "Failed to set file flags"); ret = ARCHIVE_WARN; cleanup: if (fd < 0) close(myfd); return (ret); } #else /* * Of course, some systems have neither BSD chflags() nor Linux' flags * support through ioctl(). */ static int set_fflags_platform(struct archive_write_disk *a, int fd, const char *name, mode_t mode, unsigned long set, unsigned long clear) { (void)a; /* UNUSED */ (void)fd; /* UNUSED */ (void)name; /* UNUSED */ (void)mode; /* UNUSED */ (void)set; /* UNUSED */ (void)clear; /* UNUSED */ return (ARCHIVE_OK); } #endif /* __linux */ #ifndef HAVE_COPYFILE_H /* Default is to simply drop Mac extended metadata. */ static int set_mac_metadata(struct archive_write_disk *a, const char *pathname, const void *metadata, size_t metadata_size) { (void)a; /* UNUSED */ (void)pathname; /* UNUSED */ (void)metadata; /* UNUSED */ (void)metadata_size; /* UNUSED */ return (ARCHIVE_OK); } static int fixup_appledouble(struct archive_write_disk *a, const char *pathname) { (void)a; /* UNUSED */ (void)pathname; /* UNUSED */ return (ARCHIVE_OK); } #else /* * On Mac OS, we use copyfile() to unpack the metadata and * apply it to the target file. */ #if defined(HAVE_SYS_XATTR_H) static int copy_xattrs(struct archive_write_disk *a, int tmpfd, int dffd) { ssize_t xattr_size; char *xattr_names = NULL, *xattr_val = NULL; int ret = ARCHIVE_OK, xattr_i; xattr_size = flistxattr(tmpfd, NULL, 0, 0); if (xattr_size == -1) { archive_set_error(&a->archive, errno, "Failed to read metadata(xattr)"); ret = ARCHIVE_WARN; goto exit_xattr; } xattr_names = malloc(xattr_size); if (xattr_names == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for metadata(xattr)"); ret = ARCHIVE_FATAL; goto exit_xattr; } xattr_size = flistxattr(tmpfd, xattr_names, xattr_size, 0); if (xattr_size == -1) { archive_set_error(&a->archive, errno, "Failed to read metadata(xattr)"); ret = ARCHIVE_WARN; goto exit_xattr; } for (xattr_i = 0; xattr_i < xattr_size; xattr_i += strlen(xattr_names + xattr_i) + 1) { char *p; ssize_t s; int f; s = fgetxattr(tmpfd, xattr_names + xattr_i, NULL, 0, 0, 0); if (s == -1) { archive_set_error(&a->archive, errno, "Failed to get metadata(xattr)"); ret = ARCHIVE_WARN; goto exit_xattr; } p = realloc(xattr_val, s); if (p == NULL) { archive_set_error(&a->archive, ENOMEM, "Failed to get metadata(xattr)"); ret = ARCHIVE_WARN; goto exit_xattr; } xattr_val = p; s = fgetxattr(tmpfd, xattr_names + xattr_i, xattr_val, s, 0, 0); if (s == -1) { archive_set_error(&a->archive, errno, "Failed to get metadata(xattr)"); ret = ARCHIVE_WARN; goto exit_xattr; } f = fsetxattr(dffd, xattr_names + xattr_i, xattr_val, s, 0, 0); if (f == -1) { archive_set_error(&a->archive, errno, "Failed to get metadata(xattr)"); ret = ARCHIVE_WARN; goto exit_xattr; } } exit_xattr: free(xattr_names); free(xattr_val); return (ret); } #endif static int copy_acls(struct archive_write_disk *a, int tmpfd, int dffd) { #ifndef HAVE_SYS_ACL_H return 0; #else acl_t acl, dfacl = NULL; int acl_r, ret = ARCHIVE_OK; acl = acl_get_fd(tmpfd); if (acl == NULL) { if (errno == ENOENT) /* There are not any ACLs. */ return (ret); archive_set_error(&a->archive, errno, "Failed to get metadata(acl)"); ret = ARCHIVE_WARN; goto exit_acl; } dfacl = acl_dup(acl); acl_r = acl_set_fd(dffd, dfacl); if (acl_r == -1) { archive_set_error(&a->archive, errno, "Failed to get metadata(acl)"); ret = ARCHIVE_WARN; goto exit_acl; } exit_acl: if (acl) acl_free(acl); if (dfacl) acl_free(dfacl); return (ret); #endif } static int create_tempdatafork(struct archive_write_disk *a, const char *pathname) { struct archive_string tmpdatafork; int tmpfd; archive_string_init(&tmpdatafork); archive_strcpy(&tmpdatafork, "tar.md.XXXXXX"); tmpfd = mkstemp(tmpdatafork.s); if (tmpfd < 0) { archive_set_error(&a->archive, errno, "Failed to mkstemp"); archive_string_free(&tmpdatafork); return (-1); } if (copyfile(pathname, tmpdatafork.s, 0, COPYFILE_UNPACK | COPYFILE_NOFOLLOW | COPYFILE_ACL | COPYFILE_XATTR) < 0) { archive_set_error(&a->archive, errno, "Failed to restore metadata"); close(tmpfd); tmpfd = -1; } unlink(tmpdatafork.s); archive_string_free(&tmpdatafork); return (tmpfd); } static int copy_metadata(struct archive_write_disk *a, const char *metadata, const char *datafork, int datafork_compressed) { int ret = ARCHIVE_OK; if (datafork_compressed) { int dffd, tmpfd; tmpfd = create_tempdatafork(a, metadata); if (tmpfd == -1) return (ARCHIVE_WARN); /* * Do not open the data fork compressed by HFS+ compression * with at least a writing mode(O_RDWR or O_WRONLY). it * makes the data fork uncompressed. */ dffd = open(datafork, 0); if (dffd == -1) { archive_set_error(&a->archive, errno, "Failed to open the data fork for metadata"); close(tmpfd); return (ARCHIVE_WARN); } #if defined(HAVE_SYS_XATTR_H) ret = copy_xattrs(a, tmpfd, dffd); if (ret == ARCHIVE_OK) #endif ret = copy_acls(a, tmpfd, dffd); close(tmpfd); close(dffd); } else { if (copyfile(metadata, datafork, 0, COPYFILE_UNPACK | COPYFILE_NOFOLLOW | COPYFILE_ACL | COPYFILE_XATTR) < 0) { archive_set_error(&a->archive, errno, "Failed to restore metadata"); ret = ARCHIVE_WARN; } } return (ret); } static int set_mac_metadata(struct archive_write_disk *a, const char *pathname, const void *metadata, size_t metadata_size) { struct archive_string tmp; ssize_t written; int fd; int ret = ARCHIVE_OK; /* This would be simpler if copyfile() could just accept the * metadata as a block of memory; then we could sidestep this * silly dance of writing the data to disk just so that * copyfile() can read it back in again. */ archive_string_init(&tmp); archive_strcpy(&tmp, "tar.mmd.XXXXXX"); fd = mkstemp(tmp.s); if (fd < 0) { archive_set_error(&a->archive, errno, "Failed to restore metadata"); archive_string_free(&tmp); return (ARCHIVE_WARN); } written = write(fd, metadata, metadata_size); close(fd); if ((size_t)written != metadata_size) { archive_set_error(&a->archive, errno, "Failed to restore metadata"); ret = ARCHIVE_WARN; } else { int compressed; #if defined(UF_COMPRESSED) if ((a->todo & TODO_HFS_COMPRESSION) != 0 && (ret = lazy_stat(a)) == ARCHIVE_OK) compressed = a->st.st_flags & UF_COMPRESSED; else #endif compressed = 0; ret = copy_metadata(a, tmp.s, pathname, compressed); } unlink(tmp.s); archive_string_free(&tmp); return (ret); } static int fixup_appledouble(struct archive_write_disk *a, const char *pathname) { char buff[8]; struct stat st; const char *p; struct archive_string datafork; int fd = -1, ret = ARCHIVE_OK; archive_string_init(&datafork); /* Check if the current file name is a type of the resource * fork file. */ p = strrchr(pathname, '/'); if (p == NULL) p = pathname; else p++; if (p[0] != '.' || p[1] != '_') goto skip_appledouble; /* * Check if the data fork file exists. * * TODO: Check if this write disk object has handled it. */ archive_strncpy(&datafork, pathname, p - pathname); archive_strcat(&datafork, p + 2); if ( #ifdef HAVE_LSTAT lstat(datafork.s, &st) == -1 || #else la_stat(datafork.s, &st) == -1 || #endif (((st.st_mode & AE_IFMT) != AE_IFREG) && ((st.st_mode & AE_IFMT) != AE_IFDIR))) goto skip_appledouble; /* * Check if the file is in the AppleDouble form. */ fd = open(pathname, O_RDONLY | O_BINARY | O_CLOEXEC); __archive_ensure_cloexec_flag(fd); - if (fd == -1) { + if (fd < 0) { archive_set_error(&a->archive, errno, "Failed to open a restoring file"); ret = ARCHIVE_WARN; goto skip_appledouble; } if (read(fd, buff, 8) == -1) { archive_set_error(&a->archive, errno, "Failed to read a restoring file"); close(fd); ret = ARCHIVE_WARN; goto skip_appledouble; } close(fd); /* Check AppleDouble Magic Code. */ if (archive_be32dec(buff) != 0x00051607) goto skip_appledouble; /* Check AppleDouble Version. */ if (archive_be32dec(buff+4) != 0x00020000) goto skip_appledouble; ret = copy_metadata(a, pathname, datafork.s, #if defined(UF_COMPRESSED) st.st_flags & UF_COMPRESSED); #else 0); #endif if (ret == ARCHIVE_OK) { unlink(pathname); ret = ARCHIVE_EOF; } skip_appledouble: archive_string_free(&datafork); return (ret); } #endif #if ARCHIVE_XATTR_LINUX || ARCHIVE_XATTR_DARWIN || ARCHIVE_XATTR_AIX /* * Restore extended attributes - Linux, Darwin and AIX implementations: * AIX' ea interface is syntaxwise identical to the Linux xattr interface. */ static int set_xattrs(struct archive_write_disk *a) { struct archive_entry *entry = a->entry; struct archive_string errlist; int ret = ARCHIVE_OK; int i = archive_entry_xattr_reset(entry); short fail = 0; archive_string_init(&errlist); while (i--) { const char *name; const void *value; size_t size; int e; archive_entry_xattr_next(entry, &name, &value, &size); if (name == NULL) continue; #if ARCHIVE_XATTR_LINUX /* Linux: quietly skip POSIX.1e ACL extended attributes */ if (strncmp(name, "system.", 7) == 0 && (strcmp(name + 7, "posix_acl_access") == 0 || strcmp(name + 7, "posix_acl_default") == 0)) continue; if (strncmp(name, "trusted.SGI_", 12) == 0 && (strcmp(name + 12, "ACL_DEFAULT") == 0 || strcmp(name + 12, "ACL_FILE") == 0)) continue; /* Linux: xfsroot namespace is obsolete and unsupported */ if (strncmp(name, "xfsroot.", 8) == 0) { fail = 1; archive_strcat(&errlist, name); archive_strappend_char(&errlist, ' '); continue; } #endif if (a->fd >= 0) { #if ARCHIVE_XATTR_LINUX e = fsetxattr(a->fd, name, value, size, 0); #elif ARCHIVE_XATTR_DARWIN e = fsetxattr(a->fd, name, value, size, 0, 0); #elif ARCHIVE_XATTR_AIX e = fsetea(a->fd, name, value, size, 0); #endif } else { #if ARCHIVE_XATTR_LINUX e = lsetxattr(archive_entry_pathname(entry), name, value, size, 0); #elif ARCHIVE_XATTR_DARWIN e = setxattr(archive_entry_pathname(entry), name, value, size, 0, XATTR_NOFOLLOW); #elif ARCHIVE_XATTR_AIX e = lsetea(archive_entry_pathname(entry), name, value, size, 0); #endif } if (e == -1) { ret = ARCHIVE_WARN; archive_strcat(&errlist, name); archive_strappend_char(&errlist, ' '); if (errno != ENOTSUP && errno != ENOSYS) fail = 1; } } if (ret == ARCHIVE_WARN) { if (fail && errlist.length > 0) { errlist.length--; errlist.s[errlist.length] = '\0'; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Cannot restore extended attributes: %s", errlist.s); } else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Cannot restore extended " "attributes on this file system."); } archive_string_free(&errlist); return (ret); } #elif ARCHIVE_XATTR_FREEBSD /* * Restore extended attributes - FreeBSD implementation */ static int set_xattrs(struct archive_write_disk *a) { struct archive_entry *entry = a->entry; struct archive_string errlist; int ret = ARCHIVE_OK; int i = archive_entry_xattr_reset(entry); short fail = 0; archive_string_init(&errlist); while (i--) { const char *name; const void *value; size_t size; archive_entry_xattr_next(entry, &name, &value, &size); if (name != NULL) { ssize_t e; int namespace; namespace = EXTATTR_NAMESPACE_USER; if (strncmp(name, "user.", 5) == 0) { /* "user." attributes go to user namespace */ name += 5; namespace = EXTATTR_NAMESPACE_USER; } else if (strncmp(name, "system.", 7) == 0) { name += 7; namespace = EXTATTR_NAMESPACE_SYSTEM; if (!strcmp(name, "nfs4.acl") || !strcmp(name, "posix1e.acl_access") || !strcmp(name, "posix1e.acl_default")) continue; } else { /* Other namespaces are unsupported */ archive_strcat(&errlist, name); archive_strappend_char(&errlist, ' '); fail = 1; ret = ARCHIVE_WARN; continue; } if (a->fd >= 0) { /* * On FreeBSD, extattr_set_fd does not * return the same as * extattr_set_file. It returns zero * on success, non-zero on failure. * * We can detect the failure by * manually setting errno prior to the * call and checking after. * * If errno remains zero, fake the * return value by setting e to size. * * This is a hack for now until I * (Shawn Webb) get FreeBSD to fix the * issue, if that's even possible. */ errno = 0; e = extattr_set_fd(a->fd, namespace, name, value, size); if (e == 0 && errno == 0) { e = size; } } else { e = extattr_set_link( archive_entry_pathname(entry), namespace, name, value, size); } if (e != (ssize_t)size) { archive_strcat(&errlist, name); archive_strappend_char(&errlist, ' '); ret = ARCHIVE_WARN; if (errno != ENOTSUP && errno != ENOSYS) fail = 1; } } } if (ret == ARCHIVE_WARN) { if (fail && errlist.length > 0) { errlist.length--; errlist.s[errlist.length] = '\0'; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Cannot restore extended attributes: %s", errlist.s); } else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Cannot restore extended " "attributes on this file system."); } archive_string_free(&errlist); return (ret); } #else /* * Restore extended attributes - stub implementation for unsupported systems */ static int set_xattrs(struct archive_write_disk *a) { static int warning_done = 0; /* If there aren't any extended attributes, then it's okay not * to extract them, otherwise, issue a single warning. */ if (archive_entry_xattr_count(a->entry) != 0 && !warning_done) { warning_done = 1; archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Cannot restore extended attributes on this system"); return (ARCHIVE_WARN); } /* Warning was already emitted; suppress further warnings. */ return (ARCHIVE_OK); } #endif /* * Test if file on disk is older than entry. */ static int older(struct stat *st, struct archive_entry *entry) { /* First, test the seconds and return if we have a definite answer. */ /* Definitely older. */ if (to_int64_time(st->st_mtime) < to_int64_time(archive_entry_mtime(entry))) return (1); /* Definitely younger. */ if (to_int64_time(st->st_mtime) > to_int64_time(archive_entry_mtime(entry))) return (0); /* If this platform supports fractional seconds, try those. */ #if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC /* Definitely older. */ if (st->st_mtimespec.tv_nsec < archive_entry_mtime_nsec(entry)) return (1); #elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC /* Definitely older. */ if (st->st_mtim.tv_nsec < archive_entry_mtime_nsec(entry)) return (1); #elif HAVE_STRUCT_STAT_ST_MTIME_N /* older. */ if (st->st_mtime_n < archive_entry_mtime_nsec(entry)) return (1); #elif HAVE_STRUCT_STAT_ST_UMTIME /* older. */ if (st->st_umtime * 1000 < archive_entry_mtime_nsec(entry)) return (1); #elif HAVE_STRUCT_STAT_ST_MTIME_USEC /* older. */ if (st->st_mtime_usec * 1000 < archive_entry_mtime_nsec(entry)) return (1); #else /* This system doesn't have high-res timestamps. */ #endif /* Same age or newer, so not older. */ return (0); } #ifndef ARCHIVE_ACL_SUPPORT int archive_write_disk_set_acls(struct archive *a, int fd, const char *name, struct archive_acl *abstract_acl, __LA_MODE_T mode) { (void)a; /* UNUSED */ (void)fd; /* UNUSED */ (void)name; /* UNUSED */ (void)abstract_acl; /* UNUSED */ (void)mode; /* UNUSED */ return (ARCHIVE_OK); } #endif /* * Close the file descriptor if one is open. */ static void close_file_descriptor(struct archive_write_disk* a) { if (a->fd >= 0) { close(a->fd); a->fd = -1; } } #endif /* !_WIN32 || __CYGWIN__ */ diff --git a/contrib/libarchive/libarchive/archive_write_open_fd.c b/contrib/libarchive/libarchive/archive_write_open_fd.c index 8a3f68d0699d..ba034ed92f8a 100644 --- a/contrib/libarchive/libarchive/archive_write_open_fd.c +++ b/contrib/libarchive/libarchive/archive_write_open_fd.c @@ -1,145 +1,145 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_IO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #include "archive.h" struct write_fd_data { int fd; }; static int file_free(struct archive *, void *); static int file_open(struct archive *, void *); static ssize_t file_write(struct archive *, void *, const void *buff, size_t); int archive_write_open_fd(struct archive *a, int fd) { struct write_fd_data *mine; mine = malloc(sizeof(*mine)); if (mine == NULL) { archive_set_error(a, ENOMEM, "No memory"); return (ARCHIVE_FATAL); } mine->fd = fd; #if defined(__CYGWIN__) || defined(_WIN32) setmode(mine->fd, O_BINARY); #endif return (archive_write_open2(a, mine, file_open, file_write, NULL, file_free)); } static int file_open(struct archive *a, void *client_data) { struct write_fd_data *mine; struct stat st; mine = (struct write_fd_data *)client_data; if (fstat(mine->fd, &st) != 0) { archive_set_error(a, errno, "Couldn't stat fd %d", mine->fd); return (ARCHIVE_FATAL); } /* * If this is a regular file, don't add it to itself. */ if (S_ISREG(st.st_mode)) archive_write_set_skip_file(a, st.st_dev, st.st_ino); /* * If client hasn't explicitly set the last block handling, * then set it here. */ if (archive_write_get_bytes_in_last_block(a) < 0) { /* If the output is a block or character device, fifo, * or stdout, pad the last block, otherwise leave it * unpadded. */ if (S_ISCHR(st.st_mode) || S_ISBLK(st.st_mode) || S_ISFIFO(st.st_mode) || (mine->fd == 1)) /* Last block will be fully padded. */ archive_write_set_bytes_in_last_block(a, 0); else archive_write_set_bytes_in_last_block(a, 1); } return (ARCHIVE_OK); } static ssize_t file_write(struct archive *a, void *client_data, const void *buff, size_t length) { struct write_fd_data *mine; ssize_t bytesWritten; mine = (struct write_fd_data *)client_data; for (;;) { bytesWritten = write(mine->fd, buff, length); - if (bytesWritten <= 0) { + if (bytesWritten < 0) { if (errno == EINTR) continue; archive_set_error(a, errno, "Write error"); return (-1); } return (bytesWritten); } } static int file_free(struct archive *a, void *client_data) { struct write_fd_data *mine = (struct write_fd_data *)client_data; (void)a; /* UNUSED */ if (mine == NULL) return (ARCHIVE_OK); free(mine); return (ARCHIVE_OK); } diff --git a/contrib/libarchive/libarchive/archive_write_open_file.c b/contrib/libarchive/libarchive/archive_write_open_file.c index 4c6ebfb2269d..0b310f3da83b 100644 --- a/contrib/libarchive/libarchive/archive_write_open_file.c +++ b/contrib/libarchive/libarchive/archive_write_open_file.c @@ -1,110 +1,106 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #include "archive.h" struct write_FILE_data { FILE *f; }; static int file_free(struct archive *, void *); static int file_open(struct archive *, void *); static ssize_t file_write(struct archive *, void *, const void *buff, size_t); int archive_write_open_FILE(struct archive *a, FILE *f) { struct write_FILE_data *mine; mine = malloc(sizeof(*mine)); if (mine == NULL) { archive_set_error(a, ENOMEM, "No memory"); return (ARCHIVE_FATAL); } mine->f = f; return (archive_write_open2(a, mine, file_open, file_write, NULL, file_free)); } static int file_open(struct archive *a, void *client_data) { (void)a; /* UNUSED */ (void)client_data; /* UNUSED */ return (ARCHIVE_OK); } static ssize_t file_write(struct archive *a, void *client_data, const void *buff, size_t length) { struct write_FILE_data *mine; size_t bytesWritten; mine = client_data; - for (;;) { - bytesWritten = fwrite(buff, 1, length, mine->f); - if (bytesWritten <= 0) { - if (errno == EINTR) - continue; - archive_set_error(a, errno, "Write error"); - return (-1); - } - return (bytesWritten); + bytesWritten = fwrite(buff, 1, length, mine->f); + if (bytesWritten != length) { + archive_set_error(a, errno, "Write error"); + return (-1); } + return (bytesWritten); } static int file_free(struct archive *a, void *client_data) { struct write_FILE_data *mine = client_data; (void)a; /* UNUSED */ if (mine == NULL) return (ARCHIVE_OK); free(mine); return (ARCHIVE_OK); } diff --git a/contrib/libarchive/libarchive/archive_write_open_filename.c b/contrib/libarchive/libarchive/archive_write_open_filename.c index 34209426558c..7d0f9bde1dbb 100644 --- a/contrib/libarchive/libarchive/archive_write_open_filename.c +++ b/contrib/libarchive/libarchive/archive_write_open_filename.c @@ -1,269 +1,270 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #include "archive.h" #include "archive_private.h" #include "archive_string.h" #ifndef O_BINARY #define O_BINARY 0 #endif #ifndef O_CLOEXEC #define O_CLOEXEC 0 #endif struct write_file_data { int fd; struct archive_mstring filename; }; static int file_close(struct archive *, void *); static int file_free(struct archive *, void *); static int file_open(struct archive *, void *); static ssize_t file_write(struct archive *, void *, const void *buff, size_t); static int open_filename(struct archive *, int, const void *); int archive_write_open_file(struct archive *a, const char *filename) { return (archive_write_open_filename(a, filename)); } int archive_write_open_filename(struct archive *a, const char *filename) { if (filename == NULL || filename[0] == '\0') return (archive_write_open_fd(a, 1)); return (open_filename(a, 1, filename)); } int archive_write_open_filename_w(struct archive *a, const wchar_t *filename) { if (filename == NULL || filename[0] == L'\0') return (archive_write_open_fd(a, 1)); return (open_filename(a, 0, filename)); } static int open_filename(struct archive *a, int mbs_fn, const void *filename) { struct write_file_data *mine; int r; mine = calloc(1, sizeof(*mine)); if (mine == NULL) { archive_set_error(a, ENOMEM, "No memory"); return (ARCHIVE_FATAL); } if (mbs_fn) r = archive_mstring_copy_mbs(&mine->filename, filename); else r = archive_mstring_copy_wcs(&mine->filename, filename); if (r < 0) { + free(mine); if (errno == ENOMEM) { archive_set_error(a, ENOMEM, "No memory"); return (ARCHIVE_FATAL); } if (mbs_fn) archive_set_error(a, ARCHIVE_ERRNO_MISC, "Can't convert '%s' to WCS", (const char *)filename); else archive_set_error(a, ARCHIVE_ERRNO_MISC, "Can't convert '%ls' to MBS", (const wchar_t *)filename); return (ARCHIVE_FAILED); } mine->fd = -1; return (archive_write_open2(a, mine, file_open, file_write, file_close, file_free)); } static int file_open(struct archive *a, void *client_data) { int flags; struct write_file_data *mine; struct stat st; #if defined(_WIN32) && !defined(__CYGWIN__) wchar_t *fullpath; #endif const wchar_t *wcs; const char *mbs; mine = (struct write_file_data *)client_data; flags = O_WRONLY | O_CREAT | O_TRUNC | O_BINARY | O_CLOEXEC; /* * Open the file. */ mbs = NULL; wcs = NULL; #if defined(_WIN32) && !defined(__CYGWIN__) if (archive_mstring_get_wcs(a, &mine->filename, &wcs) != 0) { if (errno == ENOMEM) archive_set_error(a, errno, "No memory"); else { archive_mstring_get_mbs(a, &mine->filename, &mbs); archive_set_error(a, errno, "Can't convert '%s' to WCS", mbs); } return (ARCHIVE_FATAL); } fullpath = __la_win_permissive_name_w(wcs); if (fullpath != NULL) { mine->fd = _wopen(fullpath, flags, 0666); free(fullpath); } else mine->fd = _wopen(wcs, flags, 0666); #else if (archive_mstring_get_mbs(a, &mine->filename, &mbs) != 0) { if (errno == ENOMEM) archive_set_error(a, errno, "No memory"); else { archive_mstring_get_wcs(a, &mine->filename, &wcs); archive_set_error(a, errno, "Can't convert '%ls' to MBS", wcs); } return (ARCHIVE_FATAL); } mine->fd = open(mbs, flags, 0666); __archive_ensure_cloexec_flag(mine->fd); #endif if (mine->fd < 0) { if (mbs != NULL) archive_set_error(a, errno, "Failed to open '%s'", mbs); else archive_set_error(a, errno, "Failed to open '%ls'", wcs); return (ARCHIVE_FATAL); } if (fstat(mine->fd, &st) != 0) { if (mbs != NULL) archive_set_error(a, errno, "Couldn't stat '%s'", mbs); else archive_set_error(a, errno, "Couldn't stat '%ls'", wcs); return (ARCHIVE_FATAL); } /* * Set up default last block handling. */ if (archive_write_get_bytes_in_last_block(a) < 0) { if (S_ISCHR(st.st_mode) || S_ISBLK(st.st_mode) || S_ISFIFO(st.st_mode)) /* Pad last block when writing to device or FIFO. */ archive_write_set_bytes_in_last_block(a, 0); else /* Don't pad last block otherwise. */ archive_write_set_bytes_in_last_block(a, 1); } /* * If the output file is a regular file, don't add it to * itself. If it's a device file, it's okay to add the device * entry to the output archive. */ if (S_ISREG(st.st_mode)) archive_write_set_skip_file(a, st.st_dev, st.st_ino); return (ARCHIVE_OK); } static ssize_t file_write(struct archive *a, void *client_data, const void *buff, size_t length) { struct write_file_data *mine; ssize_t bytesWritten; mine = (struct write_file_data *)client_data; for (;;) { bytesWritten = write(mine->fd, buff, length); - if (bytesWritten <= 0) { + if (bytesWritten < 0) { if (errno == EINTR) continue; archive_set_error(a, errno, "Write error"); return (-1); } return (bytesWritten); } } static int file_close(struct archive *a, void *client_data) { struct write_file_data *mine = (struct write_file_data *)client_data; (void)a; /* UNUSED */ if (mine == NULL) return (ARCHIVE_FATAL); if (mine->fd >= 0) close(mine->fd); return (ARCHIVE_OK); } static int file_free(struct archive *a, void *client_data) { struct write_file_data *mine = (struct write_file_data *)client_data; (void)a; /* UNUSED */ if (mine == NULL) return (ARCHIVE_OK); archive_mstring_clean(&mine->filename); free(mine); return (ARCHIVE_OK); } diff --git a/contrib/libarchive/libarchive/archive_write_set_format_7zip.c b/contrib/libarchive/libarchive/archive_write_set_format_7zip.c index c2bce5975a46..175285da13be 100644 --- a/contrib/libarchive/libarchive/archive_write_set_format_7zip.c +++ b/contrib/libarchive/libarchive/archive_write_set_format_7zip.c @@ -1,2584 +1,2584 @@ /*- * Copyright (c) 2011-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_BZLIB_H #include #endif #if HAVE_LZMA_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #ifdef HAVE_ZSTD_H #include #endif #include "archive.h" #ifndef HAVE_ZLIB_H #include "archive_crc32.h" #endif #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_ppmd7_private.h" #include "archive_private.h" #include "archive_rb.h" #include "archive_string.h" #include "archive_time_private.h" #include "archive_write_private.h" #include "archive_write_set_format_private.h" /* * Codec ID */ #define _7Z_COPY 0 #define _7Z_LZMA1 0x030101 #define _7Z_LZMA2 0x21 #define _7Z_DEFLATE 0x040108 #define _7Z_BZIP2 0x040202 #define _7Z_PPMD 0x030401 #define _7Z_ZSTD 0x4F71101 /* Copied from https://github.com/mcmilk/7-Zip-zstd.git */ /* * 7-Zip header property IDs. */ #define kEnd 0x00 #define kHeader 0x01 #define kArchiveProperties 0x02 #define kAdditionalStreamsInfo 0x03 #define kMainStreamsInfo 0x04 #define kFilesInfo 0x05 #define kPackInfo 0x06 #define kUnPackInfo 0x07 #define kSubStreamsInfo 0x08 #define kSize 0x09 #define kCRC 0x0A #define kFolder 0x0B #define kCodersUnPackSize 0x0C #define kNumUnPackStream 0x0D #define kEmptyStream 0x0E #define kEmptyFile 0x0F #define kAnti 0x10 #define kName 0x11 #define kCTime 0x12 #define kATime 0x13 #define kMTime 0x14 #define kAttributes 0x15 #define kEncodedHeader 0x17 // Check that some windows file attribute constants are defined. // Reference: https://learn.microsoft.com/en-us/windows/win32/fileio/file-attribute-constants #ifndef FILE_ATTRIBUTE_READONLY #define FILE_ATTRIBUTE_READONLY 0x00000001 #endif #ifndef FILE_ATTRIBUTE_DIRECTORY #define FILE_ATTRIBUTE_DIRECTORY 0x00000010 #endif #ifndef FILE_ATTRIBUTE_ARCHIVE #define FILE_ATTRIBUTE_ARCHIVE 0x00000020 #endif // This value is defined in 7zip with the comment "trick for Unix". // // 7z archives created on unix have this bit set in the high 16 bits of // the attr field along with the unix permissions. #define FILE_ATTRIBUTE_UNIX_EXTENSION 0x8000 // Many systems define min or MIN, but not all. #define sevenzipmin(a,b) ((a) < (b) ? (a) : (b)) enum la_zaction { ARCHIVE_Z_FINISH, ARCHIVE_Z_RUN }; /* * A stream object of universal compressor. */ struct la_zstream { const uint8_t *next_in; size_t avail_in; uint64_t total_in; uint8_t *next_out; size_t avail_out; uint64_t total_out; uint32_t prop_size; uint8_t *props; int valid; void *real_stream; int (*code) (struct archive *a, struct la_zstream *lastrm, enum la_zaction action); int (*end)(struct archive *a, struct la_zstream *lastrm); }; #define PPMD7_DEFAULT_ORDER 6 #define PPMD7_DEFAULT_MEM_SIZE (1 << 24) struct ppmd_stream { int stat; CPpmd7 ppmd7_context; CPpmd7z_RangeEnc range_enc; IByteOut byteout; uint8_t *buff; uint8_t *buff_ptr; uint8_t *buff_end; size_t buff_bytes; }; struct coder { unsigned codec; size_t prop_size; uint8_t *props; }; struct file { struct archive_rb_node rbnode; struct file *next; unsigned name_len; uint8_t *utf16name;/* UTF16-LE name. */ uint64_t size; unsigned flg; #define MTIME_IS_SET (1<<0) #define ATIME_IS_SET (1<<1) #define CTIME_IS_SET (1<<2) #define CRC32_IS_SET (1<<3) #define HAS_STREAM (1<<4) struct { time_t time; long time_ns; } times[3]; #define MTIME 0 #define ATIME 1 #define CTIME 2 mode_t mode; uint32_t crc32; unsigned int dir:1; }; struct _7zip { int temp_fd; uint64_t temp_offset; struct file *cur_file; size_t total_number_entry; size_t total_number_nonempty_entry; size_t total_number_empty_entry; size_t total_number_dir_entry; size_t total_bytes_entry_name; size_t total_number_time_defined[3]; uint64_t total_bytes_compressed; uint64_t total_bytes_uncompressed; uint64_t entry_bytes_remaining; uint32_t entry_crc32; uint32_t precode_crc32; uint32_t encoded_crc32; int crc32flg; #define PRECODE_CRC32 1 #define ENCODED_CRC32 2 unsigned opt_compression; int opt_compression_level; int opt_zstd_compression_level; // This requires a different default value. int opt_threads; struct la_zstream stream; struct coder coder; struct archive_string_conv *sconv; /* * Compressed data buffer. */ unsigned char wbuff[512 * 20 * 6]; size_t wbuff_remaining; /* * The list of the file entries which has its contents is used to * manage struct file objects. * We use 'next' (a member of struct file) to chain. */ struct { struct file *first; struct file **last; } file_list, empty_list; struct archive_rb_tree rbtree;/* for empty files */ }; static int _7z_options(struct archive_write *, const char *, const char *); static int _7z_write_header(struct archive_write *, struct archive_entry *); static ssize_t _7z_write_data(struct archive_write *, const void *, size_t); static int _7z_finish_entry(struct archive_write *); static int _7z_close(struct archive_write *); static int _7z_free(struct archive_write *); static int file_cmp_node(const struct archive_rb_node *, const struct archive_rb_node *); static int file_cmp_key(const struct archive_rb_node *, const void *); static int file_new(struct archive_write *a, struct archive_entry *, struct file **); static void file_free(struct file *); static void file_register(struct _7zip *, struct file *); static void file_register_empty(struct _7zip *, struct file *); static void file_init_register(struct _7zip *); static void file_init_register_empty(struct _7zip *); static void file_free_register(struct _7zip *); static ssize_t compress_out(struct archive_write *, const void *, size_t , enum la_zaction); static int compression_init_encoder_copy(struct archive *, struct la_zstream *); static int compression_code_copy(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_copy(struct archive *, struct la_zstream *); static int compression_init_encoder_deflate(struct archive *, struct la_zstream *, int, int); #ifdef HAVE_ZLIB_H static int compression_code_deflate(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_deflate(struct archive *, struct la_zstream *); #endif static int compression_init_encoder_bzip2(struct archive *, struct la_zstream *, int); #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) static int compression_code_bzip2(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_bzip2(struct archive *, struct la_zstream *); #endif static int compression_init_encoder_lzma1(struct archive *, struct la_zstream *, int); static int compression_init_encoder_lzma2(struct archive *, struct la_zstream *, int); #if defined(HAVE_LZMA_H) static int compression_code_lzma(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_lzma(struct archive *, struct la_zstream *); #endif static int compression_init_encoder_ppmd(struct archive *, struct la_zstream *, uint8_t, uint32_t); static int compression_code_ppmd(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_ppmd(struct archive *, struct la_zstream *); static int _7z_compression_init_encoder(struct archive_write *, unsigned, int); static int compression_init_encoder_zstd(struct archive *, struct la_zstream *, int, int); #if defined(HAVE_ZSTD_H) static int compression_code_zstd(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_zstd(struct archive *, struct la_zstream *); #endif static int compression_code(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end(struct archive *, struct la_zstream *); static int enc_uint64(struct archive_write *, uint64_t); static int make_header(struct archive_write *, uint64_t, uint64_t, uint64_t, int, struct coder *); static int make_streamsInfo(struct archive_write *, uint64_t, uint64_t, uint64_t, int, struct coder *, int, uint32_t); static int string_to_number(const char *string, intmax_t *numberp) { char *end; if (string == NULL || *string == '\0') return (ARCHIVE_WARN); *numberp = strtoimax(string, &end, 10); if (end == string || *end != '\0' || errno == EOVERFLOW) { *numberp = 0; return (ARCHIVE_WARN); } return (ARCHIVE_OK); } int archive_write_set_format_7zip(struct archive *_a) { static const struct archive_rb_tree_ops rb_ops = { file_cmp_node, file_cmp_key }; struct archive_write *a = (struct archive_write *)_a; struct _7zip *zip; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_7zip"); /* If another format was already registered, unregister it. */ if (a->format_free != NULL) (a->format_free)(a); zip = calloc(1, sizeof(*zip)); if (zip == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate 7-Zip data"); return (ARCHIVE_FATAL); } zip->temp_fd = -1; __archive_rb_tree_init(&(zip->rbtree), &rb_ops); file_init_register(zip); file_init_register_empty(zip); /* Set default compression type and its level. */ #if HAVE_LZMA_H zip->opt_compression = _7Z_LZMA1; #elif defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) zip->opt_compression = _7Z_BZIP2; #elif defined(HAVE_ZLIB_H) zip->opt_compression = _7Z_DEFLATE; #elif HAVE_ZSTD_H zip->opt_compression = _7Z_ZSTD; #else zip->opt_compression = _7Z_COPY; #endif zip->opt_compression_level = 6; #ifdef ZSTD_CLEVEL_DEFAULT // Zstandard compression needs a different default // value than other encoders. zip->opt_zstd_compression_level = ZSTD_CLEVEL_DEFAULT; #else zip->opt_zstd_compression_level = 3; #endif zip->opt_threads = 1; a->format_data = zip; a->format_name = "7zip"; a->format_options = _7z_options; a->format_write_header = _7z_write_header; a->format_write_data = _7z_write_data; a->format_finish_entry = _7z_finish_entry; a->format_close = _7z_close; a->format_free = _7z_free; a->archive.archive_format = ARCHIVE_FORMAT_7ZIP; a->archive.archive_format_name = "7zip"; return (ARCHIVE_OK); } static int _7z_options(struct archive_write *a, const char *key, const char *value) { struct _7zip *zip; zip = (struct _7zip *)a->format_data; if (strcmp(key, "compression") == 0) { const char *name = NULL; if (value == NULL || strcmp(value, "copy") == 0 || strcmp(value, "COPY") == 0 || strcmp(value, "store") == 0 || strcmp(value, "STORE") == 0) zip->opt_compression = _7Z_COPY; else if (strcmp(value, "deflate") == 0 || strcmp(value, "DEFLATE") == 0) #if HAVE_ZLIB_H zip->opt_compression = _7Z_DEFLATE; #else name = "deflate"; #endif else if (strcmp(value, "bzip2") == 0 || strcmp(value, "BZIP2") == 0) #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) zip->opt_compression = _7Z_BZIP2; #else name = "bzip2"; #endif else if (strcmp(value, "lzma1") == 0 || strcmp(value, "LZMA1") == 0) #if HAVE_LZMA_H zip->opt_compression = _7Z_LZMA1; #else name = "lzma1"; #endif else if (strcmp(value, "lzma2") == 0 || strcmp(value, "LZMA2") == 0) #if HAVE_LZMA_H zip->opt_compression = _7Z_LZMA2; #else name = "lzma2"; #endif else if (strcmp(value, "zstd") == 0 || strcmp(value, "ZSTD") == 0) #if HAVE_ZSTD_H zip->opt_compression = _7Z_ZSTD; #else name = "zstd"; #endif else if (strcmp(value, "ppmd") == 0 || strcmp(value, "PPMD") == 0 || strcmp(value, "PPMd") == 0) zip->opt_compression = _7Z_PPMD; else { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Unknown compression name: `%s'", value); return (ARCHIVE_FAILED); } if (name != NULL) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "`%s' compression not supported " "on this platform", name); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } if (strcmp(key, "compression-level") == 0) { if (value == NULL || *value == '\0') { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Invalid compression-level option value `%s'", value); return (ARCHIVE_FAILED); } char *end = NULL; long lvl = strtol(value, &end, 10); if (end == NULL || *end != '\0') { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "parsing compression-level option value failed `%s'", value); return (ARCHIVE_FAILED); } #if HAVE_ZSTD_H && HAVE_ZSTD_compressStream && HAVE_ZSTD_minCLevel int min_level = sevenzipmin(0, ZSTD_minCLevel()); #else const int min_level = 0; #endif #if HAVE_ZSTD_H && HAVE_ZSTD_compressStream int max_level = ZSTD_maxCLevel(); #else const int max_level = 9; #endif if (lvl < min_level || lvl > max_level) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "compression-level option value `%ld' out of range", lvl); return (ARCHIVE_FAILED); } // Note: we don't know here if this value is for zstd (negative to ~22), // or zlib-style 0-9. If zstd is enabled but not in use, we will need to // validate opt_compression_level before use. zip->opt_compression_level = (int)lvl; zip->opt_zstd_compression_level = (int)lvl; return (ARCHIVE_OK); } if (strcmp(key, "threads") == 0) { intmax_t threads; if (string_to_number(value, &threads) != ARCHIVE_OK) { return (ARCHIVE_WARN); } if (threads < 0) { return (ARCHIVE_WARN); } if (threads == 0) { #if defined(HAVE_SYSCONF) && defined(_SC_NPROCESSORS_ONLN) threads = sysconf(_SC_NPROCESSORS_ONLN); #elif !defined(__CYGWIN__) && defined(_WIN32_WINNT) && \ _WIN32_WINNT >= 0x0601 /* _WIN32_WINNT_WIN7 */ DWORD winCores = GetActiveProcessorCount( ALL_PROCESSOR_GROUPS); threads = (intmax_t)winCores; #else threads = 1; #endif } zip->opt_threads = (int)threads; return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int _7z_write_header(struct archive_write *a, struct archive_entry *entry) { struct _7zip *zip; struct file *file; int r; zip = (struct _7zip *)a->format_data; zip->cur_file = NULL; zip->entry_bytes_remaining = 0; if (zip->sconv == NULL) { zip->sconv = archive_string_conversion_to_charset( &a->archive, "UTF-16LE", 1); if (zip->sconv == NULL) return (ARCHIVE_FATAL); } r = file_new(a, entry, &file); if (r < ARCHIVE_WARN) { if (file != NULL) file_free(file); return (r); } if (file->size == 0 && file->dir) { if (!__archive_rb_tree_insert_node(&(zip->rbtree), (struct archive_rb_node *)file)) { /* We have already had the same file. */ file_free(file); return (ARCHIVE_OK); } } if (file->flg & MTIME_IS_SET) zip->total_number_time_defined[MTIME]++; if (file->flg & CTIME_IS_SET) zip->total_number_time_defined[CTIME]++; if (file->flg & ATIME_IS_SET) zip->total_number_time_defined[ATIME]++; zip->total_number_entry++; zip->total_bytes_entry_name += file->name_len + 2; if (file->size == 0) { /* Count up the number of empty files. */ zip->total_number_empty_entry++; if (file->dir) zip->total_number_dir_entry++; else file_register_empty(zip, file); return (r); } /* * Init compression. */ if ((zip->total_number_entry - zip->total_number_empty_entry) == 1) { int level = zip->opt_compression_level; #if HAVE_ZSTD_H if (zip->opt_compression == _7Z_ZSTD) { level = zip->opt_zstd_compression_level; } else if (level < 0 || level > 9) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "compression-level option value `%d' out of range 0-9", level); file_free(file); return (ARCHIVE_FATAL); } #endif r = _7z_compression_init_encoder(a, zip->opt_compression, level); if (r < 0) { file_free(file); return (ARCHIVE_FATAL); } } /* Register a non-empty file. */ file_register(zip, file); /* * Set the current file to cur_file to read its contents. */ zip->cur_file = file; /* Save a offset of current file in temporary file. */ zip->entry_bytes_remaining = file->size; zip->entry_crc32 = 0; /* * Store a symbolic link name as file contents. */ if (archive_entry_filetype(entry) == AE_IFLNK) { ssize_t bytes; const void *p = (const void *)archive_entry_symlink_utf8(entry); bytes = compress_out(a, p, (size_t)file->size, ARCHIVE_Z_RUN); if (bytes < 0) return ((int)bytes); zip->entry_crc32 = crc32(zip->entry_crc32, p, (unsigned)bytes); zip->entry_bytes_remaining -= bytes; } return (r); } /* * Write data to a temporary file. */ static int write_to_temp(struct archive_write *a, const void *buff, size_t s) { struct _7zip *zip; const unsigned char *p; ssize_t ws; zip = (struct _7zip *)a->format_data; /* * Open a temporary file. */ if (zip->temp_fd == -1) { zip->temp_offset = 0; zip->temp_fd = __archive_mktemp(NULL); if (zip->temp_fd < 0) { archive_set_error(&a->archive, errno, "Couldn't create temporary file"); return (ARCHIVE_FATAL); } } p = (const unsigned char *)buff; while (s) { ws = write(zip->temp_fd, p, s); if (ws < 0) { archive_set_error(&(a->archive), errno, - "fwrite function failed"); + "write function failed"); return (ARCHIVE_FATAL); } s -= ws; p += ws; zip->temp_offset += ws; } return (ARCHIVE_OK); } static ssize_t compress_out(struct archive_write *a, const void *buff, size_t s, enum la_zaction run) { struct _7zip *zip = (struct _7zip *)a->format_data; int r; if (run == ARCHIVE_Z_FINISH && zip->stream.total_in == 0 && s == 0) return (0); if ((zip->crc32flg & PRECODE_CRC32) && s) zip->precode_crc32 = crc32(zip->precode_crc32, buff, (unsigned)s); zip->stream.next_in = (const unsigned char *)buff; zip->stream.avail_in = s; for (;;) { /* Compress file data. */ r = compression_code(&(a->archive), &(zip->stream), run); if (r != ARCHIVE_OK && r != ARCHIVE_EOF) return (ARCHIVE_FATAL); if (zip->stream.avail_out == 0) { if (write_to_temp(a, zip->wbuff, sizeof(zip->wbuff)) != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->stream.next_out = zip->wbuff; zip->stream.avail_out = sizeof(zip->wbuff); if (zip->crc32flg & ENCODED_CRC32) zip->encoded_crc32 = crc32(zip->encoded_crc32, zip->wbuff, sizeof(zip->wbuff)); if (run == ARCHIVE_Z_FINISH && r != ARCHIVE_EOF) continue; } if (zip->stream.avail_in == 0) break; } if (run == ARCHIVE_Z_FINISH) { uint64_t bytes = sizeof(zip->wbuff) - zip->stream.avail_out; if (write_to_temp(a, zip->wbuff, (size_t)bytes) != ARCHIVE_OK) return (ARCHIVE_FATAL); if ((zip->crc32flg & ENCODED_CRC32) && bytes) zip->encoded_crc32 = crc32(zip->encoded_crc32, zip->wbuff, (unsigned)bytes); } return (s); } static ssize_t _7z_write_data(struct archive_write *a, const void *buff, size_t s) { struct _7zip *zip; ssize_t bytes; zip = (struct _7zip *)a->format_data; if (s > zip->entry_bytes_remaining) s = (size_t)zip->entry_bytes_remaining; if (s == 0 || zip->cur_file == NULL) return (0); bytes = compress_out(a, buff, s, ARCHIVE_Z_RUN); if (bytes < 0) return (bytes); zip->entry_crc32 = crc32(zip->entry_crc32, buff, (unsigned)bytes); zip->entry_bytes_remaining -= bytes; return (bytes); } static int _7z_finish_entry(struct archive_write *a) { struct _7zip *zip; size_t s; ssize_t r; zip = (struct _7zip *)a->format_data; if (zip->cur_file == NULL) return (ARCHIVE_OK); while (zip->entry_bytes_remaining > 0) { s = (size_t)zip->entry_bytes_remaining; if (s > a->null_length) s = a->null_length; r = _7z_write_data(a, a->nulls, s); if (r < 0) return ((int)r); } zip->total_bytes_compressed += zip->stream.total_in; zip->total_bytes_uncompressed += zip->stream.total_out; zip->cur_file->crc32 = zip->entry_crc32; zip->cur_file = NULL; return (ARCHIVE_OK); } static int flush_wbuff(struct archive_write *a) { struct _7zip *zip; int r; size_t s; zip = (struct _7zip *)a->format_data; s = sizeof(zip->wbuff) - zip->wbuff_remaining; r = __archive_write_output(a, zip->wbuff, s); if (r != ARCHIVE_OK) return (r); zip->wbuff_remaining = sizeof(zip->wbuff); return (r); } static int copy_out(struct archive_write *a, uint64_t offset, uint64_t length) { struct _7zip *zip; int r; zip = (struct _7zip *)a->format_data; if (zip->temp_offset > 0 && lseek(zip->temp_fd, offset, SEEK_SET) < 0) { archive_set_error(&(a->archive), errno, "lseek failed"); return (ARCHIVE_FATAL); } while (length) { size_t rsize; ssize_t rs; unsigned char *wb; if (length > zip->wbuff_remaining) rsize = zip->wbuff_remaining; else rsize = (size_t)length; wb = zip->wbuff + (sizeof(zip->wbuff) - zip->wbuff_remaining); rs = read(zip->temp_fd, wb, rsize); if (rs < 0) { archive_set_error(&(a->archive), errno, "Can't read temporary file(%jd)", (intmax_t)rs); return (ARCHIVE_FATAL); } if (rs == 0) { archive_set_error(&(a->archive), 0, "Truncated 7-Zip archive"); return (ARCHIVE_FATAL); } zip->wbuff_remaining -= rs; length -= rs; if (zip->wbuff_remaining == 0) { r = flush_wbuff(a); if (r != ARCHIVE_OK) return (r); } } return (ARCHIVE_OK); } static int _7z_close(struct archive_write *a) { struct _7zip *zip; unsigned char *wb; uint64_t header_offset, header_size, header_unpacksize; uint64_t length; uint32_t header_crc32; int r; zip = (struct _7zip *)a->format_data; if (zip->total_number_entry > 0) { struct archive_rb_node *n; uint64_t data_offset, data_size, data_unpacksize; unsigned header_compression; r = (int)compress_out(a, NULL, 0, ARCHIVE_Z_FINISH); if (r < 0) return (r); data_offset = 0; data_size = zip->stream.total_out; data_unpacksize = zip->stream.total_in; zip->coder.codec = zip->opt_compression; zip->coder.prop_size = zip->stream.prop_size; zip->coder.props = zip->stream.props; zip->stream.prop_size = 0; zip->stream.props = NULL; zip->total_number_nonempty_entry = zip->total_number_entry - zip->total_number_empty_entry; /* Connect an empty file list. */ if (zip->empty_list.first != NULL) { *zip->file_list.last = zip->empty_list.first; zip->file_list.last = zip->empty_list.last; } /* Connect a directory file list. */ ARCHIVE_RB_TREE_FOREACH(n, &(zip->rbtree)) { file_register(zip, (struct file *)n); } /* * NOTE: 7z command supports just LZMA1, LZMA2 and COPY for * the compression type for encoding the header. */ #if HAVE_LZMA_H header_compression = _7Z_LZMA1; if(zip->opt_compression == _7Z_LZMA2 || zip->opt_compression == _7Z_COPY) header_compression = zip->opt_compression; /* If the stored file is only one, do not encode the header. * This is the same way 7z command does. */ if (zip->total_number_entry == 1) header_compression = _7Z_COPY; #else header_compression = _7Z_COPY; #endif int level = zip->opt_compression_level; if (level < 0) level = 0; else if (level > 9) level = 9; r = _7z_compression_init_encoder(a, header_compression, level); if (r < 0) return (r); zip->crc32flg = PRECODE_CRC32; zip->precode_crc32 = 0; r = make_header(a, data_offset, data_size, data_unpacksize, 1, &(zip->coder)); if (r < 0) return (r); r = (int)compress_out(a, NULL, 0, ARCHIVE_Z_FINISH); if (r < 0) return (r); header_offset = data_offset + data_size; header_size = zip->stream.total_out; header_crc32 = zip->precode_crc32; header_unpacksize = zip->stream.total_in; if (header_compression != _7Z_COPY) { /* * Encode the header in order to reduce the size * of the archive. */ free(zip->coder.props); zip->coder.codec = header_compression; zip->coder.prop_size = zip->stream.prop_size; zip->coder.props = zip->stream.props; zip->stream.prop_size = 0; zip->stream.props = NULL; r = _7z_compression_init_encoder(a, _7Z_COPY, 0); if (r < 0) return (r); zip->crc32flg = ENCODED_CRC32; zip->encoded_crc32 = 0; /* * Make EncodedHeader. */ r = enc_uint64(a, kEncodedHeader); if (r < 0) return (r); r = make_streamsInfo(a, header_offset, header_size, header_unpacksize, 1, &(zip->coder), 0, header_crc32); if (r < 0) return (r); r = (int)compress_out(a, NULL, 0, ARCHIVE_Z_FINISH); if (r < 0) return (r); header_offset = header_offset + header_size; header_size = zip->stream.total_out; header_crc32 = zip->encoded_crc32; } zip->crc32flg = 0; } else { header_offset = header_size = 0; header_crc32 = 0; } length = zip->temp_offset; /* * Make the zip header on wbuff(write buffer). */ wb = zip->wbuff; zip->wbuff_remaining = sizeof(zip->wbuff); memcpy(&wb[0], "7z\xBC\xAF\x27\x1C", 6); wb[6] = 0;/* Major version. */ wb[7] = 3;/* Minor version. */ archive_le64enc(&wb[12], header_offset);/* Next Header Offset */ archive_le64enc(&wb[20], header_size);/* Next Header Size */ archive_le32enc(&wb[28], header_crc32);/* Next Header CRC */ archive_le32enc(&wb[8], crc32(0, &wb[12], 20));/* Start Header CRC */ zip->wbuff_remaining -= 32; /* * Read all file contents and an encoded header from the temporary * file and write out it. */ r = copy_out(a, 0, length); if (r != ARCHIVE_OK) return (r); r = flush_wbuff(a); return (r); } /* * Encode 64 bits value into 7-Zip's encoded UINT64 value. */ static int enc_uint64(struct archive_write *a, uint64_t val) { unsigned mask = 0x80; uint8_t numdata[9]; int i; numdata[0] = 0; for (i = 1; i < (int)sizeof(numdata); i++) { if (val < mask) { numdata[0] |= (uint8_t)val; break; } numdata[i] = (uint8_t)val; val >>= 8; numdata[0] |= mask; mask >>= 1; } return ((int)compress_out(a, numdata, i, ARCHIVE_Z_RUN)); } static int make_substreamsInfo(struct archive_write *a, struct coder *coders) { struct _7zip *zip = (struct _7zip *)a->format_data; struct file *file; int r; /* * Make SubStreamsInfo. */ r = enc_uint64(a, kSubStreamsInfo); if (r < 0) return (r); if (zip->total_number_nonempty_entry > 1 && coders->codec != _7Z_COPY) { /* * Make NumUnPackStream. */ r = enc_uint64(a, kNumUnPackStream); if (r < 0) return (r); /* Write numUnpackStreams */ r = enc_uint64(a, zip->total_number_nonempty_entry); if (r < 0) return (r); /* * Make kSize. */ r = enc_uint64(a, kSize); if (r < 0) return (r); file = zip->file_list.first; for (;file != NULL; file = file->next) { if (file->next == NULL || file->next->size == 0) break; r = enc_uint64(a, file->size); if (r < 0) return (r); } } /* * Make CRC. */ r = enc_uint64(a, kCRC); if (r < 0) return (r); /* All are defined */ r = enc_uint64(a, 1); if (r < 0) return (r); file = zip->file_list.first; for (;file != NULL; file = file->next) { uint8_t crc[4]; if (file->size == 0) break; archive_le32enc(crc, file->crc32); r = (int)compress_out(a, crc, 4, ARCHIVE_Z_RUN); if (r < 0) return (r); } /* Write End. */ r = enc_uint64(a, kEnd); if (r < 0) return (r); return (ARCHIVE_OK); } static int make_streamsInfo(struct archive_write *a, uint64_t offset, uint64_t pack_size, uint64_t unpack_size, int num_coder, struct coder *coders, int substrm, uint32_t header_crc) { struct _7zip *zip = (struct _7zip *)a->format_data; uint8_t codec_buff[8]; int numFolders, fi; int codec_size; int i, r; if (coders->codec == _7Z_COPY) numFolders = (int)zip->total_number_nonempty_entry; else numFolders = 1; /* * Make PackInfo. */ r = enc_uint64(a, kPackInfo); if (r < 0) return (r); /* Write PackPos. */ r = enc_uint64(a, offset); if (r < 0) return (r); /* Write NumPackStreams. */ r = enc_uint64(a, numFolders); if (r < 0) return (r); /* Make Size. */ r = enc_uint64(a, kSize); if (r < 0) return (r); if (numFolders > 1) { struct file *file = zip->file_list.first; for (;file != NULL; file = file->next) { if (file->size == 0) break; r = enc_uint64(a, file->size); if (r < 0) return (r); } } else { /* Write size. */ r = enc_uint64(a, pack_size); if (r < 0) return (r); } r = enc_uint64(a, kEnd); if (r < 0) return (r); /* * Make UnPackInfo. */ r = enc_uint64(a, kUnPackInfo); if (r < 0) return (r); /* * Make Folder. */ r = enc_uint64(a, kFolder); if (r < 0) return (r); /* Write NumFolders. */ r = enc_uint64(a, numFolders); if (r < 0) return (r); /* Write External. */ r = enc_uint64(a, 0); if (r < 0) return (r); for (fi = 0; fi < numFolders; fi++) { /* Write NumCoders. */ r = enc_uint64(a, num_coder); if (r < 0) return (r); for (i = 0; i < num_coder; i++) { unsigned codec_id = coders[i].codec; /* Write Codec flag. */ archive_be64enc(codec_buff, codec_id); for (codec_size = 8; codec_size > 0; codec_size--) { if (codec_buff[8 - codec_size]) break; } if (codec_size == 0) codec_size = 1; if (coders[i].prop_size) r = enc_uint64(a, codec_size | 0x20); else r = enc_uint64(a, codec_size); if (r < 0) return (r); /* Write Codec ID. */ codec_size &= 0x0f; r = (int)compress_out(a, &codec_buff[8-codec_size], codec_size, ARCHIVE_Z_RUN); if (r < 0) return (r); if (coders[i].prop_size) { /* Write Codec property size. */ r = enc_uint64(a, coders[i].prop_size); if (r < 0) return (r); /* Write Codec properties. */ r = (int)compress_out(a, coders[i].props, coders[i].prop_size, ARCHIVE_Z_RUN); if (r < 0) return (r); } } } /* * Make CodersUnPackSize. */ r = enc_uint64(a, kCodersUnPackSize); if (r < 0) return (r); if (numFolders > 1) { struct file *file = zip->file_list.first; for (;file != NULL; file = file->next) { if (file->size == 0) break; r = enc_uint64(a, file->size); if (r < 0) return (r); } } else { /* Write UnPackSize. */ r = enc_uint64(a, unpack_size); if (r < 0) return (r); } if (!substrm) { uint8_t crc[4]; /* * Make CRC. */ r = enc_uint64(a, kCRC); if (r < 0) return (r); /* All are defined */ r = enc_uint64(a, 1); if (r < 0) return (r); archive_le32enc(crc, header_crc); r = (int)compress_out(a, crc, 4, ARCHIVE_Z_RUN); if (r < 0) return (r); } /* Write End. */ r = enc_uint64(a, kEnd); if (r < 0) return (r); if (substrm) { /* * Make SubStreamsInfo. */ r = make_substreamsInfo(a, coders); if (r < 0) return (r); } /* Write End. */ r = enc_uint64(a, kEnd); if (r < 0) return (r); return (ARCHIVE_OK); } static int make_time(struct archive_write *a, uint8_t type, unsigned flg, int ti) { uint8_t filetime[8]; struct _7zip *zip = (struct _7zip *)a->format_data; struct file *file; int r; uint8_t b, mask; /* * Make Time Bools. */ if (zip->total_number_time_defined[ti] == zip->total_number_entry) { /* Write Time Type. */ r = enc_uint64(a, type); if (r < 0) return (r); /* Write EmptyStream Size. */ r = enc_uint64(a, 2 + zip->total_number_entry * 8); if (r < 0) return (r); /* All are defined. */ r = enc_uint64(a, 1); if (r < 0) return (r); } else { if (zip->total_number_time_defined[ti] == 0) return (ARCHIVE_OK); /* Write Time Type. */ r = enc_uint64(a, type); if (r < 0) return (r); /* Write EmptyStream Size. */ r = enc_uint64(a, 2 + ((zip->total_number_entry + 7) >> 3) + zip->total_number_time_defined[ti] * 8); if (r < 0) return (r); /* All are not defined. */ r = enc_uint64(a, 0); if (r < 0) return (r); b = 0; mask = 0x80; file = zip->file_list.first; for (;file != NULL; file = file->next) { if (file->flg & flg) b |= mask; mask >>= 1; if (mask == 0) { r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); if (r < 0) return (r); mask = 0x80; b = 0; } } if (mask != 0x80) { r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); if (r < 0) return (r); } } /* External. */ r = enc_uint64(a, 0); if (r < 0) return (r); /* * Make Times. */ file = zip->file_list.first; for (;file != NULL; file = file->next) { if ((file->flg & flg) == 0) continue; archive_le64enc(filetime, unix_to_ntfs(file->times[ti].time, file->times[ti].time_ns)); r = (int)compress_out(a, filetime, 8, ARCHIVE_Z_RUN); if (r < 0) return (r); } return (ARCHIVE_OK); } static int make_header(struct archive_write *a, uint64_t offset, uint64_t pack_size, uint64_t unpack_size, int codernum, struct coder *coders) { struct _7zip *zip = (struct _7zip *)a->format_data; struct file *file; int r; uint8_t b, mask; /* * Make FilesInfo. */ r = enc_uint64(a, kHeader); if (r < 0) return (r); /* * If there are empty files only, do not write MainStreamInfo. */ if (zip->total_number_nonempty_entry) { /* * Make MainStreamInfo. */ r = enc_uint64(a, kMainStreamsInfo); if (r < 0) return (r); r = make_streamsInfo(a, offset, pack_size, unpack_size, codernum, coders, 1, 0); if (r < 0) return (r); } /* * Make FilesInfo. */ r = enc_uint64(a, kFilesInfo); if (r < 0) return (r); /* Write numFiles. */ r = enc_uint64(a, zip->total_number_entry); if (r < 0) return (r); if (zip->total_number_empty_entry > 0) { /* Make EmptyStream. */ r = enc_uint64(a, kEmptyStream); if (r < 0) return (r); /* Write EmptyStream Size. */ r = enc_uint64(a, (zip->total_number_entry+7)>>3); if (r < 0) return (r); b = 0; mask = 0x80; file = zip->file_list.first; for (;file != NULL; file = file->next) { if (file->size == 0) b |= mask; mask >>= 1; if (mask == 0) { r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); if (r < 0) return (r); mask = 0x80; b = 0; } } if (mask != 0x80) { r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); if (r < 0) return (r); } } if (zip->total_number_empty_entry > zip->total_number_dir_entry) { /* Make EmptyFile. */ r = enc_uint64(a, kEmptyFile); if (r < 0) return (r); /* Write EmptyFile Size. */ r = enc_uint64(a, (zip->total_number_empty_entry + 7) >> 3); if (r < 0) return (r); b = 0; mask = 0x80; file = zip->file_list.first; for (;file != NULL; file = file->next) { if (file->size) continue; if (!file->dir) b |= mask; mask >>= 1; if (mask == 0) { r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); if (r < 0) return (r); mask = 0x80; b = 0; } } if (mask != 0x80) { r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); if (r < 0) return (r); } } /* Make Name. */ r = enc_uint64(a, kName); if (r < 0) return (r); /* Write Name size. */ r = enc_uint64(a, zip->total_bytes_entry_name+1); if (r < 0) return (r); /* Write dmy byte. */ r = enc_uint64(a, 0); if (r < 0) return (r); file = zip->file_list.first; for (;file != NULL; file = file->next) { r = (int)compress_out(a, file->utf16name, file->name_len+2, ARCHIVE_Z_RUN); if (r < 0) return (r); } /* Make MTime. */ r = make_time(a, kMTime, MTIME_IS_SET, MTIME); if (r < 0) return (r); /* Make CTime. */ r = make_time(a, kCTime, CTIME_IS_SET, CTIME); if (r < 0) return (r); /* Make ATime. */ r = make_time(a, kATime, ATIME_IS_SET, ATIME); if (r < 0) return (r); /* Make Attributes. */ r = enc_uint64(a, kAttributes); if (r < 0) return (r); /* Write Attributes size. */ r = enc_uint64(a, 2 + zip->total_number_entry * 4); if (r < 0) return (r); /* Write "All Are Defined". */ r = enc_uint64(a, 1); if (r < 0) return (r); /* Write dmy byte. */ r = enc_uint64(a, 0); if (r < 0) return (r); file = zip->file_list.first; for (;file != NULL; file = file->next) { /* * High 16bits is unix mode. * Low 16bits is Windows attributes. */ uint32_t encattr, attr = 0; if (file->dir) attr |= FILE_ATTRIBUTE_DIRECTORY; else attr |= FILE_ATTRIBUTE_ARCHIVE; if ((file->mode & 0222) == 0) attr |= FILE_ATTRIBUTE_READONLY; attr |= FILE_ATTRIBUTE_UNIX_EXTENSION; attr |= ((uint32_t)file->mode) << 16; archive_le32enc(&encattr, attr); r = (int)compress_out(a, &encattr, 4, ARCHIVE_Z_RUN); if (r < 0) return (r); } /* Write End. */ r = enc_uint64(a, kEnd); if (r < 0) return (r); /* Write End. */ r = enc_uint64(a, kEnd); if (r < 0) return (r); return (ARCHIVE_OK); } static int _7z_free(struct archive_write *a) { struct _7zip *zip = (struct _7zip *)a->format_data; /* Close the temporary file. */ if (zip->temp_fd >= 0) close(zip->temp_fd); file_free_register(zip); compression_end(&(a->archive), &(zip->stream)); free(zip->coder.props); free(zip); return (ARCHIVE_OK); } static int file_cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct file *f1 = (const struct file *)n1; const struct file *f2 = (const struct file *)n2; if (f1->name_len == f2->name_len) return (memcmp(f1->utf16name, f2->utf16name, f1->name_len)); return (f1->name_len > f2->name_len)?1:-1; } static int file_cmp_key(const struct archive_rb_node *n, const void *key) { const struct file *f = (const struct file *)n; return (f->name_len - *(const char *)key); } static int file_new(struct archive_write *a, struct archive_entry *entry, struct file **newfile) { struct _7zip *zip; struct file *file; const char *u16; size_t u16len; int ret = ARCHIVE_OK; zip = (struct _7zip *)a->format_data; *newfile = NULL; file = calloc(1, sizeof(*file)); if (file == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } if (0 > archive_entry_pathname_l(entry, &u16, &u16len, zip->sconv)) { if (errno == ENOMEM) { free(file); archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for UTF-16LE"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "A filename cannot be converted to UTF-16LE;" "You should disable making Joliet extension"); ret = ARCHIVE_WARN; } file->utf16name = malloc(u16len + 2); if (file->utf16name == NULL) { free(file); archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Name"); return (ARCHIVE_FATAL); } memcpy(file->utf16name, u16, u16len); file->utf16name[u16len+0] = 0; file->utf16name[u16len+1] = 0; file->name_len = (unsigned)u16len; file->mode = archive_entry_mode(entry); if (archive_entry_filetype(entry) == AE_IFREG) file->size = archive_entry_size(entry); else archive_entry_set_size(entry, 0); if (archive_entry_filetype(entry) == AE_IFDIR) file->dir = 1; else if (archive_entry_filetype(entry) == AE_IFLNK) { const char* linkpath; linkpath = archive_entry_symlink_utf8(entry); if (linkpath == NULL) { free(file); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "symlink path could not be converted to UTF-8"); return (ARCHIVE_FAILED); } else file->size = strlen(linkpath); } if (archive_entry_mtime_is_set(entry)) { file->flg |= MTIME_IS_SET; file->times[MTIME].time = archive_entry_mtime(entry); file->times[MTIME].time_ns = archive_entry_mtime_nsec(entry); } if (archive_entry_atime_is_set(entry)) { file->flg |= ATIME_IS_SET; file->times[ATIME].time = archive_entry_atime(entry); file->times[ATIME].time_ns = archive_entry_atime_nsec(entry); } if (archive_entry_ctime_is_set(entry)) { file->flg |= CTIME_IS_SET; file->times[CTIME].time = archive_entry_ctime(entry); file->times[CTIME].time_ns = archive_entry_ctime_nsec(entry); } *newfile = file; return (ret); } static void file_free(struct file *file) { free(file->utf16name); free(file); } static void file_register(struct _7zip *zip, struct file *file) { file->next = NULL; *zip->file_list.last = file; zip->file_list.last = &(file->next); } static void file_init_register(struct _7zip *zip) { zip->file_list.first = NULL; zip->file_list.last = &(zip->file_list.first); } static void file_free_register(struct _7zip *zip) { struct file *file, *file_next; file = zip->file_list.first; while (file != NULL) { file_next = file->next; file_free(file); file = file_next; } } static void file_register_empty(struct _7zip *zip, struct file *file) { file->next = NULL; *zip->empty_list.last = file; zip->empty_list.last = &(file->next); } static void file_init_register_empty(struct _7zip *zip) { zip->empty_list.first = NULL; zip->empty_list.last = &(zip->empty_list.first); } #if !defined(HAVE_ZLIB_H) || !defined(HAVE_BZLIB_H) ||\ !defined(BZ_CONFIG_ERROR) || !defined(HAVE_LZMA_H) ||\ !(HAVE_ZSTD_H && HAVE_ZSTD_compressStream) static int compression_unsupported_encoder(struct archive *a, struct la_zstream *lastrm, const char *name) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "%s compression not supported on this platform", name); lastrm->valid = 0; lastrm->real_stream = NULL; return (ARCHIVE_FAILED); } #endif /* * _7_COPY compressor. */ static int compression_init_encoder_copy(struct archive *a, struct la_zstream *lastrm) { if (lastrm->valid) compression_end(a, lastrm); lastrm->valid = 1; lastrm->code = compression_code_copy; lastrm->end = compression_end_copy; return (ARCHIVE_OK); } static int compression_code_copy(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { size_t bytes; (void)a; /* UNUSED */ if (lastrm->avail_out > lastrm->avail_in) bytes = lastrm->avail_in; else bytes = lastrm->avail_out; if (bytes) { memcpy(lastrm->next_out, lastrm->next_in, bytes); lastrm->next_in += bytes; lastrm->avail_in -= bytes; lastrm->total_in += bytes; lastrm->next_out += bytes; lastrm->avail_out -= bytes; lastrm->total_out += bytes; } if (action == ARCHIVE_Z_FINISH && lastrm->avail_in == 0) return (ARCHIVE_EOF); return (ARCHIVE_OK); } static int compression_end_copy(struct archive *a, struct la_zstream *lastrm) { (void)a; /* UNUSED */ lastrm->valid = 0; return (ARCHIVE_OK); } /* * _7_DEFLATE compressor. */ #ifdef HAVE_ZLIB_H static int compression_init_encoder_deflate(struct archive *a, struct la_zstream *lastrm, int level, int withheader) { z_stream *strm; if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm)); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for gzip stream"); return (ARCHIVE_FATAL); } /* zlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (Bytef *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = (uInt)lastrm->avail_in; strm->total_in = (uLong)lastrm->total_in; strm->next_out = lastrm->next_out; strm->avail_out = (uInt)lastrm->avail_out; strm->total_out = (uLong)lastrm->total_out; if (deflateInit2(strm, level, Z_DEFLATED, (withheader)?15:-15, 8, Z_DEFAULT_STRATEGY) != Z_OK) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_deflate; lastrm->end = compression_end_deflate; return (ARCHIVE_OK); } static int compression_code_deflate(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { z_stream *strm; int r; strm = (z_stream *)lastrm->real_stream; /* zlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (Bytef *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = (uInt)lastrm->avail_in; strm->total_in = (uLong)lastrm->total_in; strm->next_out = lastrm->next_out; strm->avail_out = (uInt)lastrm->avail_out; strm->total_out = (uLong)lastrm->total_out; r = deflate(strm, (action == ARCHIVE_Z_FINISH)? Z_FINISH: Z_NO_FLUSH); lastrm->next_in = strm->next_in; lastrm->avail_in = strm->avail_in; lastrm->total_in = strm->total_in; lastrm->next_out = strm->next_out; lastrm->avail_out = strm->avail_out; lastrm->total_out = strm->total_out; switch (r) { case Z_OK: return (ARCHIVE_OK); case Z_STREAM_END: return (ARCHIVE_EOF); default: archive_set_error(a, ARCHIVE_ERRNO_MISC, "GZip compression failed:" " deflate() call returned status %d", r); return (ARCHIVE_FATAL); } } static int compression_end_deflate(struct archive *a, struct la_zstream *lastrm) { z_stream *strm; int r; strm = (z_stream *)lastrm->real_stream; r = deflateEnd(strm); free(strm); lastrm->real_stream = NULL; lastrm->valid = 0; if (r != Z_OK) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "Failed to clean up compressor"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } #else static int compression_init_encoder_deflate(struct archive *a, struct la_zstream *lastrm, int level, int withheader) { (void) level; /* UNUSED */ (void) withheader; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "deflate")); } #endif /* * _7_BZIP2 compressor. */ #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) static int compression_init_encoder_bzip2(struct archive *a, struct la_zstream *lastrm, int level) { bz_stream *strm; if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm)); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for bzip2 stream"); return (ARCHIVE_FATAL); } /* bzlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (char *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = (uint32_t)lastrm->avail_in; strm->total_in_lo32 = (uint32_t)(lastrm->total_in & 0xffffffff); strm->total_in_hi32 = (uint32_t)(lastrm->total_in >> 32); strm->next_out = (char *)lastrm->next_out; strm->avail_out = (uint32_t)lastrm->avail_out; strm->total_out_lo32 = (uint32_t)(lastrm->total_out & 0xffffffff); strm->total_out_hi32 = (uint32_t)(lastrm->total_out >> 32); if (BZ2_bzCompressInit(strm, level, 0, 30) != BZ_OK) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_bzip2; lastrm->end = compression_end_bzip2; return (ARCHIVE_OK); } static int compression_code_bzip2(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { bz_stream *strm; int r; strm = (bz_stream *)lastrm->real_stream; /* bzlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (char *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = (uint32_t)lastrm->avail_in; strm->total_in_lo32 = (uint32_t)(lastrm->total_in & 0xffffffff); strm->total_in_hi32 = (uint32_t)(lastrm->total_in >> 32); strm->next_out = (char *)lastrm->next_out; strm->avail_out = (uint32_t)lastrm->avail_out; strm->total_out_lo32 = (uint32_t)(lastrm->total_out & 0xffffffff); strm->total_out_hi32 = (uint32_t)(lastrm->total_out >> 32); r = BZ2_bzCompress(strm, (action == ARCHIVE_Z_FINISH)? BZ_FINISH: BZ_RUN); lastrm->next_in = (const unsigned char *)strm->next_in; lastrm->avail_in = strm->avail_in; lastrm->total_in = (((uint64_t)(uint32_t)strm->total_in_hi32) << 32) + (uint64_t)(uint32_t)strm->total_in_lo32; lastrm->next_out = (unsigned char *)strm->next_out; lastrm->avail_out = strm->avail_out; lastrm->total_out = (((uint64_t)(uint32_t)strm->total_out_hi32) << 32) + (uint64_t)(uint32_t)strm->total_out_lo32; switch (r) { case BZ_RUN_OK: /* Non-finishing */ case BZ_FINISH_OK: /* Finishing: There's more work to do */ return (ARCHIVE_OK); case BZ_STREAM_END: /* Finishing: all done */ /* Only occurs in finishing case */ return (ARCHIVE_EOF); default: /* Any other return value indicates an error */ archive_set_error(a, ARCHIVE_ERRNO_MISC, "Bzip2 compression failed:" " BZ2_bzCompress() call returned status %d", r); return (ARCHIVE_FATAL); } } static int compression_end_bzip2(struct archive *a, struct la_zstream *lastrm) { bz_stream *strm; int r; strm = (bz_stream *)lastrm->real_stream; r = BZ2_bzCompressEnd(strm); free(strm); lastrm->real_stream = NULL; lastrm->valid = 0; if (r != BZ_OK) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "Failed to clean up compressor"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } #else static int compression_init_encoder_bzip2(struct archive *a, struct la_zstream *lastrm, int level) { (void) level; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "bzip2")); } #endif /* * _7_LZMA1, _7_LZMA2 compressor. */ #if defined(HAVE_LZMA_H) static int compression_init_encoder_lzma(struct archive *a, struct la_zstream *lastrm, int level, uint64_t filter_id) { static const lzma_stream lzma_init_data = LZMA_STREAM_INIT; lzma_stream *strm; lzma_filter *lzmafilters; lzma_options_lzma lzma_opt; int r; if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm) + sizeof(*lzmafilters) * 2); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for lzma stream"); return (ARCHIVE_FATAL); } lzmafilters = (lzma_filter *)(strm+1); if (level > 9) level = 9; if (lzma_lzma_preset(&lzma_opt, level)) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } lzmafilters[0].id = filter_id; lzmafilters[0].options = &lzma_opt; lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */ r = lzma_properties_size(&(lastrm->prop_size), lzmafilters); if (r != LZMA_OK) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "lzma_properties_size failed"); return (ARCHIVE_FATAL); } if (lastrm->prop_size) { lastrm->props = malloc(lastrm->prop_size); if (lastrm->props == NULL) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Cannot allocate memory"); return (ARCHIVE_FATAL); } r = lzma_properties_encode(lzmafilters, lastrm->props); if (r != LZMA_OK) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "lzma_properties_encode failed"); return (ARCHIVE_FATAL); } } *strm = lzma_init_data; r = lzma_raw_encoder(strm, lzmafilters); switch (r) { case LZMA_OK: lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_lzma; lastrm->end = compression_end_lzma; r = ARCHIVE_OK; break; case LZMA_MEM_ERROR: free(strm); lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Internal error initializing compression library: " "Cannot allocate memory"); r = ARCHIVE_FATAL; break; default: free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library: " "It's a bug in liblzma"); r = ARCHIVE_FATAL; break; } return (r); } static int compression_init_encoder_lzma1(struct archive *a, struct la_zstream *lastrm, int level) { return compression_init_encoder_lzma(a, lastrm, level, LZMA_FILTER_LZMA1); } static int compression_init_encoder_lzma2(struct archive *a, struct la_zstream *lastrm, int level) { return compression_init_encoder_lzma(a, lastrm, level, LZMA_FILTER_LZMA2); } static int compression_code_lzma(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { lzma_stream *strm; int r; strm = (lzma_stream *)lastrm->real_stream; strm->next_in = lastrm->next_in; strm->avail_in = lastrm->avail_in; strm->total_in = lastrm->total_in; strm->next_out = lastrm->next_out; strm->avail_out = lastrm->avail_out; strm->total_out = lastrm->total_out; r = lzma_code(strm, (action == ARCHIVE_Z_FINISH)? LZMA_FINISH: LZMA_RUN); lastrm->next_in = strm->next_in; lastrm->avail_in = strm->avail_in; lastrm->total_in = strm->total_in; lastrm->next_out = strm->next_out; lastrm->avail_out = strm->avail_out; lastrm->total_out = strm->total_out; switch (r) { case LZMA_OK: /* Non-finishing case */ return (ARCHIVE_OK); case LZMA_STREAM_END: /* This return can only occur in finishing case. */ return (ARCHIVE_EOF); case LZMA_MEMLIMIT_ERROR: archive_set_error(a, ENOMEM, "lzma compression error:" " %ju MiB would have been needed", (uintmax_t)((lzma_memusage(strm) + 1024 * 1024 -1) / (1024 * 1024))); return (ARCHIVE_FATAL); default: /* Any other return value indicates an error */ archive_set_error(a, ARCHIVE_ERRNO_MISC, "lzma compression failed:" " lzma_code() call returned status %d", r); return (ARCHIVE_FATAL); } } static int compression_end_lzma(struct archive *a, struct la_zstream *lastrm) { lzma_stream *strm; (void)a; /* UNUSED */ strm = (lzma_stream *)lastrm->real_stream; lzma_end(strm); free(strm); lastrm->valid = 0; lastrm->real_stream = NULL; return (ARCHIVE_OK); } #else static int compression_init_encoder_lzma1(struct archive *a, struct la_zstream *lastrm, int level) { (void) level; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "lzma")); } static int compression_init_encoder_lzma2(struct archive *a, struct la_zstream *lastrm, int level) { (void) level; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "lzma")); } #endif /* * _7_PPMD compressor. */ static void ppmd_write(void *p, Byte b) { struct archive_write *a = ((IByteOut *)p)->a; struct _7zip *zip = (struct _7zip *)(a->format_data); struct la_zstream *lastrm = &(zip->stream); struct ppmd_stream *strm; if (lastrm->avail_out) { *lastrm->next_out++ = b; lastrm->avail_out--; lastrm->total_out++; return; } strm = (struct ppmd_stream *)lastrm->real_stream; if (strm->buff_ptr < strm->buff_end) { *strm->buff_ptr++ = b; strm->buff_bytes++; } } static int compression_init_encoder_ppmd(struct archive *a, struct la_zstream *lastrm, uint8_t maxOrder, uint32_t msize) { struct ppmd_stream *strm; uint8_t *props; int r; if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm)); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for PPMd"); return (ARCHIVE_FATAL); } strm->buff = malloc(32); if (strm->buff == NULL) { free(strm); archive_set_error(a, ENOMEM, "Can't allocate memory for PPMd"); return (ARCHIVE_FATAL); } strm->buff_ptr = strm->buff; strm->buff_end = strm->buff + 32; props = malloc(1+4); if (props == NULL) { free(strm->buff); free(strm); archive_set_error(a, ENOMEM, "Coludn't allocate memory for PPMd"); return (ARCHIVE_FATAL); } props[0] = maxOrder; archive_le32enc(props+1, msize); __archive_ppmd7_functions.Ppmd7_Construct(&strm->ppmd7_context); r = __archive_ppmd7_functions.Ppmd7_Alloc( &strm->ppmd7_context, msize); if (r == 0) { free(strm->buff); free(strm); free(props); archive_set_error(a, ENOMEM, "Coludn't allocate memory for PPMd"); return (ARCHIVE_FATAL); } __archive_ppmd7_functions.Ppmd7_Init(&(strm->ppmd7_context), maxOrder); strm->byteout.a = (struct archive_write *)a; strm->byteout.Write = ppmd_write; strm->range_enc.Stream = &(strm->byteout); __archive_ppmd7_functions.Ppmd7z_RangeEnc_Init(&(strm->range_enc)); strm->stat = 0; lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_ppmd; lastrm->end = compression_end_ppmd; lastrm->prop_size = 5; lastrm->props = props; return (ARCHIVE_OK); } static int compression_code_ppmd(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { struct ppmd_stream *strm; (void)a; /* UNUSED */ strm = (struct ppmd_stream *)lastrm->real_stream; /* Copy encoded data if there are remaining bytes from previous call. */ if (strm->buff_bytes) { uint8_t *p = strm->buff_ptr - strm->buff_bytes; while (lastrm->avail_out && strm->buff_bytes) { *lastrm->next_out++ = *p++; lastrm->avail_out--; lastrm->total_out++; strm->buff_bytes--; } if (strm->buff_bytes) return (ARCHIVE_OK); if (strm->stat == 1) return (ARCHIVE_EOF); strm->buff_ptr = strm->buff; } while (lastrm->avail_in && lastrm->avail_out) { __archive_ppmd7_functions.Ppmd7_EncodeSymbol( &(strm->ppmd7_context), &(strm->range_enc), *lastrm->next_in++); lastrm->avail_in--; lastrm->total_in++; } if (lastrm->avail_in == 0 && action == ARCHIVE_Z_FINISH) { __archive_ppmd7_functions.Ppmd7z_RangeEnc_FlushData( &(strm->range_enc)); strm->stat = 1; /* Return EOF if there are no remaining bytes. */ if (strm->buff_bytes == 0) return (ARCHIVE_EOF); } return (ARCHIVE_OK); } static int compression_end_ppmd(struct archive *a, struct la_zstream *lastrm) { struct ppmd_stream *strm; (void)a; /* UNUSED */ strm = (struct ppmd_stream *)lastrm->real_stream; __archive_ppmd7_functions.Ppmd7_Free(&strm->ppmd7_context); free(strm->buff); free(strm); lastrm->real_stream = NULL; lastrm->valid = 0; return (ARCHIVE_OK); } #if HAVE_ZSTD_H && HAVE_ZSTD_compressStream static int compression_init_encoder_zstd(struct archive *a, struct la_zstream *lastrm, int level, int threads) { if (lastrm->valid) compression_end(a, lastrm); ZSTD_CStream *strm = ZSTD_createCStream(); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for zstd stream"); return (ARCHIVE_FATAL); } if (ZSTD_isError(ZSTD_initCStream(strm, level))) { ZSTD_freeCStream(strm); archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing zstd compressor object"); return (ARCHIVE_FATAL); } ZSTD_CCtx_setParameter(strm, ZSTD_c_nbWorkers, threads); // p7zip-zstd fails to unpack archives that don't have prop_size 5. // 7-Zip-zstd fails to unpack archives that don't have prop_size 3 or 5. // So let's use 5... lastrm->prop_size = 5; lastrm->props = calloc(5, 1); if (lastrm->props == NULL) { ZSTD_freeCStream(strm); archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing zstd compressor properties"); return (ARCHIVE_FATAL); } // Refer to the DProps struct in 7-Zip-zstd's ZstdDecoder.h: // https://github.com/mcmilk/7-Zip-zstd/blob/79b2c78e9e7735ddf90147129b75cf2797ff6522/CPP/7zip/Compress/ZstdDecoder.h#L34S lastrm->props[0] = ZSTD_VERSION_MAJOR; lastrm->props[1] = ZSTD_VERSION_MINOR; lastrm->props[2] = level; // lastrm->props[3] and lastrm->props[4] are reserved. Leave them as 0. lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_zstd; lastrm->end = compression_end_zstd; return (ARCHIVE_OK); } static int compression_code_zstd(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { ZSTD_CStream *strm = (ZSTD_CStream *)lastrm->real_stream; ZSTD_outBuffer out = { .dst = lastrm->next_out, .size = lastrm->avail_out, .pos = 0 }; ZSTD_inBuffer in = { .src = lastrm->next_in, .size = lastrm->avail_in, .pos = 0 }; size_t zret; ZSTD_EndDirective mode = (action == ARCHIVE_Z_RUN) ? ZSTD_e_continue : ZSTD_e_end; zret = ZSTD_compressStream2(strm, &out, &in, mode); if (ZSTD_isError(zret)) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "zstd compression failed, ZSTD_compressStream2 returned: %s", ZSTD_getErrorName(zret)); return (ARCHIVE_FATAL); } lastrm->next_in += in.pos; lastrm->avail_in -= in.pos; lastrm->total_in += in.pos; lastrm->next_out += out.pos; lastrm->avail_out -= out.pos; lastrm->total_out += out.pos; if (action == ARCHIVE_Z_FINISH && zret == 0) return (ARCHIVE_EOF); // All done. return (ARCHIVE_OK); // More work to do. } static int compression_end_zstd(struct archive *a, struct la_zstream *lastrm) { ZSTD_CStream *strm; (void)a; /* UNUSED */ strm = (ZSTD_CStream *)lastrm->real_stream; ZSTD_freeCStream(strm); lastrm->valid = 0; lastrm->real_stream = NULL; return (ARCHIVE_OK); } #else static int compression_init_encoder_zstd(struct archive *a, struct la_zstream *lastrm, int level, int threads) { (void) level; /* UNUSED */ (void) threads; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "zstd")); } #endif /* * Universal compressor initializer. */ static int _7z_compression_init_encoder(struct archive_write *a, unsigned compression, int compression_level) { struct _7zip *zip; int r; zip = (struct _7zip *)a->format_data; switch (compression) { case _7Z_DEFLATE: r = compression_init_encoder_deflate( &(a->archive), &(zip->stream), compression_level, 0); break; case _7Z_BZIP2: r = compression_init_encoder_bzip2( &(a->archive), &(zip->stream), compression_level); break; case _7Z_LZMA1: r = compression_init_encoder_lzma1( &(a->archive), &(zip->stream), compression_level); break; case _7Z_LZMA2: r = compression_init_encoder_lzma2( &(a->archive), &(zip->stream), compression_level); break; case _7Z_PPMD: r = compression_init_encoder_ppmd( &(a->archive), &(zip->stream), PPMD7_DEFAULT_ORDER, PPMD7_DEFAULT_MEM_SIZE); break; case _7Z_ZSTD: r = compression_init_encoder_zstd( &(a->archive), &(zip->stream), compression_level, zip->opt_threads); break; case _7Z_COPY: default: r = compression_init_encoder_copy( &(a->archive), &(zip->stream)); break; } if (r == ARCHIVE_OK) { zip->stream.total_in = 0; zip->stream.next_out = zip->wbuff; zip->stream.avail_out = sizeof(zip->wbuff); zip->stream.total_out = 0; } return (r); } static int compression_code(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { if (lastrm->valid) return (lastrm->code(a, lastrm, action)); return (ARCHIVE_OK); } static int compression_end(struct archive *a, struct la_zstream *lastrm) { if (lastrm->valid) { lastrm->prop_size = 0; free(lastrm->props); lastrm->props = NULL; return (lastrm->end(a, lastrm)); } return (ARCHIVE_OK); } diff --git a/contrib/libarchive/libarchive/archive_write_set_format_mtree.c b/contrib/libarchive/libarchive/archive_write_set_format_mtree.c index 02fbb2d2f555..8131574c8da2 100644 --- a/contrib/libarchive/libarchive/archive_write_set_format_mtree.c +++ b/contrib/libarchive/libarchive/archive_write_set_format_mtree.c @@ -1,2299 +1,2296 @@ /*- * Copyright (c) 2008 Joerg Sonnenberger * Copyright (c) 2009-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_SYS_TYPES_H #include #endif #include #include #include #include "archive.h" #include "archive_digest_private.h" #include "archive_entry.h" #include "archive_entry_private.h" #include "archive_private.h" #include "archive_rb.h" #include "archive_string.h" #include "archive_write_private.h" #define INDENTNAMELEN 15 #define MAXLINELEN 80 #define SET_KEYS \ (F_FLAGS | F_GID | F_GNAME | F_MODE | F_TYPE | F_UID | F_UNAME) struct attr_counter { struct attr_counter *prev; struct attr_counter *next; struct mtree_entry *m_entry; int count; }; struct attr_counter_set { struct attr_counter *uid_list; struct attr_counter *gid_list; struct attr_counter *mode_list; struct attr_counter *flags_list; }; struct mtree_chain { struct mtree_entry *first; struct mtree_entry **last; }; /* * The Data only for a directory file. */ struct dir_info { struct archive_rb_tree rbtree; struct mtree_chain children; struct mtree_entry *chnext; int virtual; }; /* * The Data only for a regular file. */ struct reg_info { int compute_sum; uint32_t crc; uint_least32_t mset_digest; struct ae_digest digest; }; struct mtree_entry { struct archive_rb_node rbnode; struct mtree_entry *next; struct mtree_entry *parent; struct dir_info *dir_info; struct reg_info *reg_info; struct archive_string parentdir; struct archive_string basename; struct archive_string pathname; struct archive_string symlink; struct archive_string uname; struct archive_string gname; struct archive_string fflags_text; unsigned int nlink; mode_t filetype; mode_t mode; int64_t size; int64_t uid; int64_t gid; time_t mtime; long mtime_nsec; unsigned long fflags_set; unsigned long fflags_clear; dev_t rdevmajor; dev_t rdevminor; dev_t devmajor; dev_t devminor; int64_t ino; }; struct mtree_writer { struct mtree_entry *mtree_entry; struct mtree_entry *root; struct mtree_entry *cur_dirent; struct archive_string cur_dirstr; struct mtree_chain file_list; struct archive_string ebuf; struct archive_string buf; int first; uint64_t entry_bytes_remaining; /* * Set global value. */ struct { int processing; mode_t type; int keys; int64_t uid; int64_t gid; mode_t mode; unsigned long fflags_set; unsigned long fflags_clear; } set; struct attr_counter_set acs; int classic; int depth; /* check sum */ int compute_sum; uint32_t crc; uint64_t crc_len; #ifdef ARCHIVE_HAS_MD5 archive_md5_ctx md5ctx; #endif #ifdef ARCHIVE_HAS_RMD160 archive_rmd160_ctx rmd160ctx; #endif #ifdef ARCHIVE_HAS_SHA1 archive_sha1_ctx sha1ctx; #endif #ifdef ARCHIVE_HAS_SHA256 archive_sha256_ctx sha256ctx; #endif #ifdef ARCHIVE_HAS_SHA384 archive_sha384_ctx sha384ctx; #endif #ifdef ARCHIVE_HAS_SHA512 archive_sha512_ctx sha512ctx; #endif /* Keyword options */ int keys; #define F_CKSUM 0x00000001 /* checksum */ #define F_DEV 0x00000002 /* device type */ #define F_DONE 0x00000004 /* directory done */ #define F_FLAGS 0x00000008 /* file flags */ #define F_GID 0x00000010 /* gid */ #define F_GNAME 0x00000020 /* group name */ #define F_IGN 0x00000040 /* ignore */ #define F_MAGIC 0x00000080 /* name has magic chars */ #define F_MD5 0x00000100 /* MD5 digest */ #define F_MODE 0x00000200 /* mode */ #define F_NLINK 0x00000400 /* number of links */ #define F_NOCHANGE 0x00000800 /* If owner/mode "wrong", do * not change */ #define F_OPT 0x00001000 /* existence optional */ #define F_RMD160 0x00002000 /* RIPEMD160 digest */ #define F_SHA1 0x00004000 /* SHA-1 digest */ #define F_SIZE 0x00008000 /* size */ #define F_SLINK 0x00010000 /* symbolic link */ #define F_TAGS 0x00020000 /* tags */ #define F_TIME 0x00040000 /* modification time */ #define F_TYPE 0x00080000 /* file type */ #define F_UID 0x00100000 /* uid */ #define F_UNAME 0x00200000 /* user name */ #define F_VISIT 0x00400000 /* file visited */ #define F_SHA256 0x00800000 /* SHA-256 digest */ #define F_SHA384 0x01000000 /* SHA-384 digest */ #define F_SHA512 0x02000000 /* SHA-512 digest */ #define F_INO 0x04000000 /* inode number */ #define F_RESDEV 0x08000000 /* device ID on which the * entry resides */ /* Options */ int dironly; /* If it is set, ignore all files except * directory files, like mtree(8) -d option. */ int indent; /* If it is set, indent output data. */ int output_global_set; /* If it is set, use /set keyword to set * global values. When generating mtree * classic format, it is set by default. */ }; #define DEFAULT_KEYS (F_DEV | F_FLAGS | F_GID | F_GNAME | F_SLINK | F_MODE\ | F_NLINK | F_SIZE | F_TIME | F_TYPE | F_UID\ | F_UNAME) #define attr_counter_set_reset attr_counter_set_free static void attr_counter_free(struct attr_counter **); static int attr_counter_inc(struct attr_counter **, struct attr_counter *, struct attr_counter *, struct mtree_entry *); static struct attr_counter * attr_counter_new(struct mtree_entry *, struct attr_counter *); static int attr_counter_set_collect(struct mtree_writer *, struct mtree_entry *); static void attr_counter_set_free(struct mtree_writer *); static int get_global_set_keys(struct mtree_writer *, struct mtree_entry *); static int mtree_entry_add_child_tail(struct mtree_entry *, struct mtree_entry *); static int mtree_entry_create_virtual_dir(struct archive_write *, const char *, struct mtree_entry **); static int mtree_entry_cmp_node(const struct archive_rb_node *, const struct archive_rb_node *); static int mtree_entry_cmp_key(const struct archive_rb_node *, const void *); static int mtree_entry_exchange_same_entry(struct archive_write *, struct mtree_entry *, struct mtree_entry *); static void mtree_entry_free(struct mtree_entry *); static int mtree_entry_new(struct archive_write *, struct archive_entry *, struct mtree_entry **); static void mtree_entry_register_free(struct mtree_writer *); static void mtree_entry_register_init(struct mtree_writer *); static int mtree_entry_setup_filenames(struct archive_write *, struct mtree_entry *, struct archive_entry *); static int mtree_entry_tree_add(struct archive_write *, struct mtree_entry **); static void sum_init(struct mtree_writer *); static void sum_update(struct mtree_writer *, const void *, size_t); static void sum_final(struct mtree_writer *, struct reg_info *); static void sum_write(struct archive_string *, struct reg_info *); static int write_mtree_entry(struct archive_write *, struct mtree_entry *); static int write_dot_dot_entry(struct archive_write *, struct mtree_entry *); #define COMPUTE_CRC(var, ch) (var) = (var) << 8 ^ crctab[(var) >> 24 ^ (ch)] static const uint32_t crctab[] = { 0x0, 0x04c11db7, 0x09823b6e, 0x0d4326d9, 0x130476dc, 0x17c56b6b, 0x1a864db2, 0x1e475005, 0x2608edb8, 0x22c9f00f, 0x2f8ad6d6, 0x2b4bcb61, 0x350c9b64, 0x31cd86d3, 0x3c8ea00a, 0x384fbdbd, 0x4c11db70, 0x48d0c6c7, 0x4593e01e, 0x4152fda9, 0x5f15adac, 0x5bd4b01b, 0x569796c2, 0x52568b75, 0x6a1936c8, 0x6ed82b7f, 0x639b0da6, 0x675a1011, 0x791d4014, 0x7ddc5da3, 0x709f7b7a, 0x745e66cd, 0x9823b6e0, 0x9ce2ab57, 0x91a18d8e, 0x95609039, 0x8b27c03c, 0x8fe6dd8b, 0x82a5fb52, 0x8664e6e5, 0xbe2b5b58, 0xbaea46ef, 0xb7a96036, 0xb3687d81, 0xad2f2d84, 0xa9ee3033, 0xa4ad16ea, 0xa06c0b5d, 0xd4326d90, 0xd0f37027, 0xddb056fe, 0xd9714b49, 0xc7361b4c, 0xc3f706fb, 0xceb42022, 0xca753d95, 0xf23a8028, 0xf6fb9d9f, 0xfbb8bb46, 0xff79a6f1, 0xe13ef6f4, 0xe5ffeb43, 0xe8bccd9a, 0xec7dd02d, 0x34867077, 0x30476dc0, 0x3d044b19, 0x39c556ae, 0x278206ab, 0x23431b1c, 0x2e003dc5, 0x2ac12072, 0x128e9dcf, 0x164f8078, 0x1b0ca6a1, 0x1fcdbb16, 0x018aeb13, 0x054bf6a4, 0x0808d07d, 0x0cc9cdca, 0x7897ab07, 0x7c56b6b0, 0x71159069, 0x75d48dde, 0x6b93dddb, 0x6f52c06c, 0x6211e6b5, 0x66d0fb02, 0x5e9f46bf, 0x5a5e5b08, 0x571d7dd1, 0x53dc6066, 0x4d9b3063, 0x495a2dd4, 0x44190b0d, 0x40d816ba, 0xaca5c697, 0xa864db20, 0xa527fdf9, 0xa1e6e04e, 0xbfa1b04b, 0xbb60adfc, 0xb6238b25, 0xb2e29692, 0x8aad2b2f, 0x8e6c3698, 0x832f1041, 0x87ee0df6, 0x99a95df3, 0x9d684044, 0x902b669d, 0x94ea7b2a, 0xe0b41de7, 0xe4750050, 0xe9362689, 0xedf73b3e, 0xf3b06b3b, 0xf771768c, 0xfa325055, 0xfef34de2, 0xc6bcf05f, 0xc27dede8, 0xcf3ecb31, 0xcbffd686, 0xd5b88683, 0xd1799b34, 0xdc3abded, 0xd8fba05a, 0x690ce0ee, 0x6dcdfd59, 0x608edb80, 0x644fc637, 0x7a089632, 0x7ec98b85, 0x738aad5c, 0x774bb0eb, 0x4f040d56, 0x4bc510e1, 0x46863638, 0x42472b8f, 0x5c007b8a, 0x58c1663d, 0x558240e4, 0x51435d53, 0x251d3b9e, 0x21dc2629, 0x2c9f00f0, 0x285e1d47, 0x36194d42, 0x32d850f5, 0x3f9b762c, 0x3b5a6b9b, 0x0315d626, 0x07d4cb91, 0x0a97ed48, 0x0e56f0ff, 0x1011a0fa, 0x14d0bd4d, 0x19939b94, 0x1d528623, 0xf12f560e, 0xf5ee4bb9, 0xf8ad6d60, 0xfc6c70d7, 0xe22b20d2, 0xe6ea3d65, 0xeba91bbc, 0xef68060b, 0xd727bbb6, 0xd3e6a601, 0xdea580d8, 0xda649d6f, 0xc423cd6a, 0xc0e2d0dd, 0xcda1f604, 0xc960ebb3, 0xbd3e8d7e, 0xb9ff90c9, 0xb4bcb610, 0xb07daba7, 0xae3afba2, 0xaafbe615, 0xa7b8c0cc, 0xa379dd7b, 0x9b3660c6, 0x9ff77d71, 0x92b45ba8, 0x9675461f, 0x8832161a, 0x8cf30bad, 0x81b02d74, 0x857130c3, 0x5d8a9099, 0x594b8d2e, 0x5408abf7, 0x50c9b640, 0x4e8ee645, 0x4a4ffbf2, 0x470cdd2b, 0x43cdc09c, 0x7b827d21, 0x7f436096, 0x7200464f, 0x76c15bf8, 0x68860bfd, 0x6c47164a, 0x61043093, 0x65c52d24, 0x119b4be9, 0x155a565e, 0x18197087, 0x1cd86d30, 0x029f3d35, 0x065e2082, 0x0b1d065b, 0x0fdc1bec, 0x3793a651, 0x3352bbe6, 0x3e119d3f, 0x3ad08088, 0x2497d08d, 0x2056cd3a, 0x2d15ebe3, 0x29d4f654, 0xc5a92679, 0xc1683bce, 0xcc2b1d17, 0xc8ea00a0, 0xd6ad50a5, 0xd26c4d12, 0xdf2f6bcb, 0xdbee767c, 0xe3a1cbc1, 0xe760d676, 0xea23f0af, 0xeee2ed18, 0xf0a5bd1d, 0xf464a0aa, 0xf9278673, 0xfde69bc4, 0x89b8fd09, 0x8d79e0be, 0x803ac667, 0x84fbdbd0, 0x9abc8bd5, 0x9e7d9662, 0x933eb0bb, 0x97ffad0c, 0xafb010b1, 0xab710d06, 0xa6322bdf, 0xa2f33668, 0xbcb4666d, 0xb8757bda, 0xb5365d03, 0xb1f740b4 }; static const unsigned char safe_char[256] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ /* !"$%&'()*+,-./ EXCLUSION:0x20( ) 0x23(#) */ 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 20 - 2F */ /* 0123456789:;<>? EXCLUSION:0x3d(=) */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, /* 30 - 3F */ /* @ABCDEFGHIJKLMNO */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 40 - 4F */ /* PQRSTUVWXYZ[]^_ EXCLUSION:0x5c(\) */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, /* 50 - 5F */ /* `abcdefghijklmno */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 60 - 6F */ /* pqrstuvwxyz{|}~ */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, /* 70 - 7F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ }; static void mtree_quote(struct archive_string *s, const char *str) { const char *start; char buf[4]; unsigned char c; for (start = str; *str != '\0'; ++str) { if (safe_char[*(const unsigned char *)str]) continue; if (start != str) archive_strncat(s, start, str - start); c = (unsigned char)*str; buf[0] = '\\'; buf[1] = (c / 64) + '0'; buf[2] = (c / 8 % 8) + '0'; buf[3] = (c % 8) + '0'; archive_strncat(s, buf, 4); start = str + 1; } if (start != str) archive_strncat(s, start, str - start); } /* * Indent a line as the mtree utility does so it is readable for people. */ static void mtree_indent(struct mtree_writer *mtree) { int i, fn, nd, pd; const char *r, *s, *x; if (mtree->classic) { if (mtree->indent) { nd = 0; pd = mtree->depth * 4; } else { nd = mtree->depth?4:0; pd = 0; } } else nd = pd = 0; fn = 1; s = r = mtree->ebuf.s; x = NULL; while (*r == ' ') r++; while ((r = strchr(r, ' ')) != NULL) { if (fn) { fn = 0; for (i = 0; i < nd + pd; i++) archive_strappend_char(&mtree->buf, ' '); archive_strncat(&mtree->buf, s, r - s); if (nd + (r -s) > INDENTNAMELEN) { archive_strncat(&mtree->buf, " \\\n", 3); for (i = 0; i < (INDENTNAMELEN + 1 + pd); i++) archive_strappend_char(&mtree->buf, ' '); } else { for (i = (int)(r -s + nd); i < (INDENTNAMELEN + 1); i++) archive_strappend_char(&mtree->buf, ' '); } s = ++r; x = NULL; continue; } if (pd + (r - s) <= MAXLINELEN - 3 - INDENTNAMELEN) x = r++; else { if (x == NULL) x = r; archive_strncat(&mtree->buf, s, x - s); archive_strncat(&mtree->buf, " \\\n", 3); for (i = 0; i < (INDENTNAMELEN + 1 + pd); i++) archive_strappend_char(&mtree->buf, ' '); s = r = ++x; x = NULL; } } if (fn) { for (i = 0; i < nd + pd; i++) archive_strappend_char(&mtree->buf, ' '); archive_strcat(&mtree->buf, s); s += strlen(s); } if (x != NULL && pd + strlen(s) > MAXLINELEN - 3 - INDENTNAMELEN) { /* Last keyword is longer. */ archive_strncat(&mtree->buf, s, x - s); archive_strncat(&mtree->buf, " \\\n", 3); for (i = 0; i < (INDENTNAMELEN + 1 + pd); i++) archive_strappend_char(&mtree->buf, ' '); s = ++x; } archive_strcat(&mtree->buf, s); archive_string_empty(&mtree->ebuf); } /* * Write /set keyword. * Set the most used value of uid, gid, mode and fflags, which are * collected by the attr_counter_set_collect() function. */ static void write_global(struct mtree_writer *mtree) { struct archive_string setstr; struct archive_string unsetstr; struct attr_counter_set *acs; int keys, oldkeys, effkeys; archive_string_init(&setstr); archive_string_init(&unsetstr); keys = mtree->keys & SET_KEYS; oldkeys = mtree->set.keys; effkeys = keys; acs = &mtree->acs; if (mtree->set.processing) { /* * Check if the global data needs updating. */ effkeys &= ~F_TYPE; if (acs->uid_list == NULL) effkeys &= ~(F_UNAME | F_UID); else if (oldkeys & (F_UNAME | F_UID)) { if (acs->uid_list->count < 2 || mtree->set.uid == acs->uid_list->m_entry->uid) effkeys &= ~(F_UNAME | F_UID); } if (acs->gid_list == NULL) effkeys &= ~(F_GNAME | F_GID); else if (oldkeys & (F_GNAME | F_GID)) { if (acs->gid_list->count < 2 || mtree->set.gid == acs->gid_list->m_entry->gid) effkeys &= ~(F_GNAME | F_GID); } if (acs->mode_list == NULL) effkeys &= ~F_MODE; else if (oldkeys & F_MODE) { if (acs->mode_list->count < 2 || mtree->set.mode == acs->mode_list->m_entry->mode) effkeys &= ~F_MODE; } if (acs->flags_list == NULL) effkeys &= ~F_FLAGS; else if ((oldkeys & F_FLAGS) != 0) { if (acs->flags_list->count < 2 || (acs->flags_list->m_entry->fflags_set == mtree->set.fflags_set && acs->flags_list->m_entry->fflags_clear == mtree->set.fflags_clear)) effkeys &= ~F_FLAGS; } } else { if (acs->uid_list == NULL) keys &= ~(F_UNAME | F_UID); if (acs->gid_list == NULL) keys &= ~(F_GNAME | F_GID); if (acs->mode_list == NULL) keys &= ~F_MODE; if (acs->flags_list == NULL) keys &= ~F_FLAGS; } if ((keys & effkeys & F_TYPE) != 0) { if (mtree->dironly) { archive_strcat(&setstr, " type=dir"); mtree->set.type = AE_IFDIR; } else { archive_strcat(&setstr, " type=file"); mtree->set.type = AE_IFREG; } } if ((keys & effkeys & F_UNAME) != 0) { if (archive_strlen(&(acs->uid_list->m_entry->uname)) > 0) { archive_strcat(&setstr, " uname="); mtree_quote(&setstr, acs->uid_list->m_entry->uname.s); } else { keys &= ~F_UNAME; if ((oldkeys & F_UNAME) != 0) archive_strcat(&unsetstr, " uname"); } } if ((keys & effkeys & F_UID) != 0) { mtree->set.uid = acs->uid_list->m_entry->uid; archive_string_sprintf(&setstr, " uid=%jd", (intmax_t)mtree->set.uid); } if ((keys & effkeys & F_GNAME) != 0) { if (archive_strlen(&(acs->gid_list->m_entry->gname)) > 0) { archive_strcat(&setstr, " gname="); mtree_quote(&setstr, acs->gid_list->m_entry->gname.s); } else { keys &= ~F_GNAME; if ((oldkeys & F_GNAME) != 0) archive_strcat(&unsetstr, " gname"); } } if ((keys & effkeys & F_GID) != 0) { mtree->set.gid = acs->gid_list->m_entry->gid; archive_string_sprintf(&setstr, " gid=%jd", (intmax_t)mtree->set.gid); } if ((keys & effkeys & F_MODE) != 0) { mtree->set.mode = acs->mode_list->m_entry->mode; archive_string_sprintf(&setstr, " mode=%o", (unsigned int)mtree->set.mode); } if ((keys & effkeys & F_FLAGS) != 0) { if (archive_strlen( &(acs->flags_list->m_entry->fflags_text)) > 0) { archive_strcat(&setstr, " flags="); mtree_quote(&setstr, acs->flags_list->m_entry->fflags_text.s); mtree->set.fflags_set = acs->flags_list->m_entry->fflags_set; mtree->set.fflags_clear = acs->flags_list->m_entry->fflags_clear; } else { keys &= ~F_FLAGS; if ((oldkeys & F_FLAGS) != 0) archive_strcat(&unsetstr, " flags"); } } if (unsetstr.length > 0) archive_string_sprintf(&mtree->buf, "/unset%s\n", unsetstr.s); archive_string_free(&unsetstr); if (setstr.length > 0) archive_string_sprintf(&mtree->buf, "/set%s\n", setstr.s); archive_string_free(&setstr); mtree->set.keys = keys; mtree->set.processing = 1; } static struct attr_counter * attr_counter_new(struct mtree_entry *me, struct attr_counter *prev) { struct attr_counter *ac; ac = malloc(sizeof(*ac)); if (ac != NULL) { ac->prev = prev; ac->next = NULL; ac->count = 1; ac->m_entry = me; } return (ac); } static void attr_counter_free(struct attr_counter **top) { struct attr_counter *ac, *tac; if (*top == NULL) return; ac = *top; while (ac != NULL) { tac = ac->next; free(ac); ac = tac; } *top = NULL; } static int attr_counter_inc(struct attr_counter **top, struct attr_counter *ac, struct attr_counter *last, struct mtree_entry *me) { struct attr_counter *pac; if (ac != NULL) { ac->count++; if (*top == ac || ac->prev->count >= ac->count) return (0); for (pac = ac->prev; pac; pac = pac->prev) { if (pac->count >= ac->count) break; } ac->prev->next = ac->next; if (ac->next != NULL) ac->next->prev = ac->prev; if (pac != NULL) { ac->prev = pac; ac->next = pac->next; pac->next = ac; if (ac->next != NULL) ac->next->prev = ac; } else { ac->prev = NULL; ac->next = *top; *top = ac; ac->next->prev = ac; } } else if (last != NULL) { ac = attr_counter_new(me, last); if (ac == NULL) return (-1); last->next = ac; } return (0); } /* * Tabulate uid, gid, mode and fflags of a entry in order to be used for /set. */ static int attr_counter_set_collect(struct mtree_writer *mtree, struct mtree_entry *me) { struct attr_counter *ac, *last; struct attr_counter_set *acs = &mtree->acs; int keys = mtree->keys; if (keys & (F_UNAME | F_UID)) { if (acs->uid_list == NULL) { acs->uid_list = attr_counter_new(me, NULL); if (acs->uid_list == NULL) return (-1); } else { last = NULL; for (ac = acs->uid_list; ac; ac = ac->next) { if (ac->m_entry->uid == me->uid) break; last = ac; } if (attr_counter_inc(&acs->uid_list, ac, last, me) < 0) return (-1); } } if (keys & (F_GNAME | F_GID)) { if (acs->gid_list == NULL) { acs->gid_list = attr_counter_new(me, NULL); if (acs->gid_list == NULL) return (-1); } else { last = NULL; for (ac = acs->gid_list; ac; ac = ac->next) { if (ac->m_entry->gid == me->gid) break; last = ac; } if (attr_counter_inc(&acs->gid_list, ac, last, me) < 0) return (-1); } } if (keys & F_MODE) { if (acs->mode_list == NULL) { acs->mode_list = attr_counter_new(me, NULL); if (acs->mode_list == NULL) return (-1); } else { last = NULL; for (ac = acs->mode_list; ac; ac = ac->next) { if (ac->m_entry->mode == me->mode) break; last = ac; } if (attr_counter_inc(&acs->mode_list, ac, last, me) < 0) return (-1); } } if (keys & F_FLAGS) { if (acs->flags_list == NULL) { acs->flags_list = attr_counter_new(me, NULL); if (acs->flags_list == NULL) return (-1); } else { last = NULL; for (ac = acs->flags_list; ac; ac = ac->next) { if (ac->m_entry->fflags_set == me->fflags_set && ac->m_entry->fflags_clear == me->fflags_clear) break; last = ac; } if (attr_counter_inc(&acs->flags_list, ac, last, me) < 0) return (-1); } } return (0); } static void attr_counter_set_free(struct mtree_writer *mtree) { struct attr_counter_set *acs = &mtree->acs; attr_counter_free(&acs->uid_list); attr_counter_free(&acs->gid_list); attr_counter_free(&acs->mode_list); attr_counter_free(&acs->flags_list); } static int get_global_set_keys(struct mtree_writer *mtree, struct mtree_entry *me) { int keys; keys = mtree->keys; /* * If a keyword has been set by /set, we do not need to * output it. */ if (mtree->set.keys == 0) return (keys);/* /set is not used. */ if ((mtree->set.keys & (F_GNAME | F_GID)) != 0 && mtree->set.gid == me->gid) keys &= ~(F_GNAME | F_GID); if ((mtree->set.keys & (F_UNAME | F_UID)) != 0 && mtree->set.uid == me->uid) keys &= ~(F_UNAME | F_UID); if (mtree->set.keys & F_FLAGS) { if (mtree->set.fflags_set == me->fflags_set && mtree->set.fflags_clear == me->fflags_clear) keys &= ~F_FLAGS; } if ((mtree->set.keys & F_MODE) != 0 && mtree->set.mode == me->mode) keys &= ~F_MODE; switch (me->filetype) { case AE_IFLNK: case AE_IFSOCK: case AE_IFCHR: case AE_IFBLK: case AE_IFIFO: break; case AE_IFDIR: if ((mtree->set.keys & F_TYPE) != 0 && mtree->set.type == AE_IFDIR) keys &= ~F_TYPE; break; case AE_IFREG: default: /* Handle unknown file types as regular files. */ if ((mtree->set.keys & F_TYPE) != 0 && mtree->set.type == AE_IFREG) keys &= ~F_TYPE; break; } return (keys); } static int mtree_entry_new(struct archive_write *a, struct archive_entry *entry, struct mtree_entry **m_entry) { struct mtree_entry *me; const char *s; int r; static const struct archive_rb_tree_ops rb_ops = { mtree_entry_cmp_node, mtree_entry_cmp_key }; me = calloc(1, sizeof(*me)); if (me == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for a mtree entry"); *m_entry = NULL; return (ARCHIVE_FATAL); } r = mtree_entry_setup_filenames(a, me, entry); if (r < ARCHIVE_WARN) { mtree_entry_free(me); *m_entry = NULL; return (r); } if ((s = archive_entry_symlink(entry)) != NULL) archive_strcpy(&me->symlink, s); me->nlink = archive_entry_nlink(entry); me->filetype = archive_entry_filetype(entry); me->mode = archive_entry_mode(entry) & 07777; me->uid = archive_entry_uid(entry); me->gid = archive_entry_gid(entry); if ((s = archive_entry_uname(entry)) != NULL) archive_strcpy(&me->uname, s); if ((s = archive_entry_gname(entry)) != NULL) archive_strcpy(&me->gname, s); if ((s = archive_entry_fflags_text(entry)) != NULL) archive_strcpy(&me->fflags_text, s); archive_entry_fflags(entry, &me->fflags_set, &me->fflags_clear); me->mtime = archive_entry_mtime(entry); me->mtime_nsec = archive_entry_mtime_nsec(entry); me->rdevmajor = archive_entry_rdevmajor(entry); me->rdevminor = archive_entry_rdevminor(entry); me->devmajor = archive_entry_devmajor(entry); me->devminor = archive_entry_devminor(entry); me->ino = archive_entry_ino(entry); me->size = archive_entry_size(entry); if (me->filetype == AE_IFDIR) { me->dir_info = calloc(1, sizeof(*me->dir_info)); if (me->dir_info == NULL) { mtree_entry_free(me); archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for a mtree entry"); *m_entry = NULL; return (ARCHIVE_FATAL); } __archive_rb_tree_init(&me->dir_info->rbtree, &rb_ops); me->dir_info->children.first = NULL; me->dir_info->children.last = &(me->dir_info->children.first); me->dir_info->chnext = NULL; } else if (me->filetype == AE_IFREG) { me->reg_info = calloc(1, sizeof(*me->reg_info)); if (me->reg_info == NULL) { mtree_entry_free(me); archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for a mtree entry"); *m_entry = NULL; return (ARCHIVE_FATAL); } me->reg_info->compute_sum = 0; } *m_entry = me; return (ARCHIVE_OK); } static void mtree_entry_free(struct mtree_entry *me) { archive_string_free(&me->parentdir); archive_string_free(&me->basename); archive_string_free(&me->pathname); archive_string_free(&me->symlink); archive_string_free(&me->uname); archive_string_free(&me->gname); archive_string_free(&me->fflags_text); free(me->dir_info); free(me->reg_info); free(me); } static void mtree_copy_ae_digests(struct reg_info *reg, struct archive_entry *entry, int compute_sum) { reg->compute_sum = compute_sum; reg->mset_digest = entry->mset_digest; if ((reg->compute_sum & F_MD5) && (reg->mset_digest & AE_MSET_DIGEST_MD5)) { memcpy(®->digest.md5, entry->digest.md5, sizeof(reg->digest.md5)); } if ((reg->compute_sum & F_RMD160) && (reg->mset_digest & AE_MSET_DIGEST_RMD160)) { memcpy(®->digest.rmd160, entry->digest.rmd160, sizeof(reg->digest.rmd160)); } if ((reg->compute_sum & F_SHA1) && (reg->mset_digest & AE_MSET_DIGEST_SHA1)) { memcpy(®->digest.sha1, entry->digest.sha1, sizeof(reg->digest.sha1)); } if ((reg->compute_sum & F_SHA256) && (reg->mset_digest & AE_MSET_DIGEST_SHA256)) { memcpy(®->digest.sha256, entry->digest.sha256, sizeof(reg->digest.sha256)); } if ((reg->compute_sum & F_SHA384) && (reg->mset_digest & AE_MSET_DIGEST_SHA384)) { memcpy(®->digest.sha384, entry->digest.sha384, sizeof(reg->digest.sha384)); } if ((reg->compute_sum & F_SHA512) && (reg->mset_digest & AE_MSET_DIGEST_SHA512)) { memcpy(®->digest.sha512, entry->digest.sha512, sizeof(reg->digest.sha512)); } } static int archive_write_mtree_header(struct archive_write *a, struct archive_entry *entry) { struct mtree_writer *mtree= a->format_data; struct mtree_entry *mtree_entry; int r, r2; if (mtree->first) { mtree->first = 0; archive_strcat(&mtree->buf, "#mtree\n"); if ((mtree->keys & SET_KEYS) == 0) mtree->output_global_set = 0;/* Disabled. */ } mtree->entry_bytes_remaining = archive_entry_size(entry); /* While directory only mode, we do not handle non directory files. */ if (mtree->dironly && archive_entry_filetype(entry) != AE_IFDIR) return (ARCHIVE_OK); r2 = mtree_entry_new(a, entry, &mtree_entry); if (r2 < ARCHIVE_WARN) return (r2); r = mtree_entry_tree_add(a, &mtree_entry); if (r < ARCHIVE_WARN) { mtree_entry_free(mtree_entry); return (r); } mtree->mtree_entry = mtree_entry; /* If the current file is a regular file, we have to * compute the sum of its content. * Initialize a bunch of checksum context. */ if (mtree_entry->reg_info) { sum_init(mtree); /* honor archive_entry_set_digest() calls. These values will be * overwritten if archive_write_mtree_data() is called */ mtree_copy_ae_digests(mtree_entry->reg_info, entry, mtree->compute_sum); } return (r2); } static int write_mtree_entry(struct archive_write *a, struct mtree_entry *me) { struct mtree_writer *mtree = a->format_data; struct archive_string *str; int keys, ret; if (me->dir_info) { if (mtree->classic) { /* * Output a comment line to describe the full * pathname of the entry as mtree utility does * while generating classic format. */ if (!mtree->dironly) archive_strappend_char(&mtree->buf, '\n'); if (me->parentdir.s) archive_string_sprintf(&mtree->buf, "# %s/%s\n", me->parentdir.s, me->basename.s); else archive_string_sprintf(&mtree->buf, "# %s\n", me->basename.s); } if (mtree->output_global_set) write_global(mtree); } archive_string_empty(&mtree->ebuf); str = (mtree->indent || mtree->classic)? &mtree->ebuf : &mtree->buf; if (!mtree->classic && me->parentdir.s) { /* * If generating format is not classic one(v1), output * a full pathname. */ mtree_quote(str, me->parentdir.s); archive_strappend_char(str, '/'); } mtree_quote(str, me->basename.s); keys = get_global_set_keys(mtree, me); if ((keys & F_NLINK) != 0 && me->nlink != 1 && me->filetype != AE_IFDIR) archive_string_sprintf(str, " nlink=%u", me->nlink); if ((keys & F_GNAME) != 0 && archive_strlen(&me->gname) > 0) { archive_strcat(str, " gname="); mtree_quote(str, me->gname.s); } if ((keys & F_UNAME) != 0 && archive_strlen(&me->uname) > 0) { archive_strcat(str, " uname="); mtree_quote(str, me->uname.s); } if ((keys & F_FLAGS) != 0) { if (archive_strlen(&me->fflags_text) > 0) { archive_strcat(str, " flags="); mtree_quote(str, me->fflags_text.s); } else if (mtree->set.processing && (mtree->set.keys & F_FLAGS) != 0) /* Overwrite the global parameter. */ archive_strcat(str, " flags=none"); } if ((keys & F_TIME) != 0) archive_string_sprintf(str, " time=%jd.%jd", (intmax_t)me->mtime, (intmax_t)me->mtime_nsec); if ((keys & F_MODE) != 0) archive_string_sprintf(str, " mode=%o", (unsigned int)me->mode); if ((keys & F_GID) != 0) archive_string_sprintf(str, " gid=%jd", (intmax_t)me->gid); if ((keys & F_UID) != 0) archive_string_sprintf(str, " uid=%jd", (intmax_t)me->uid); if ((keys & F_INO) != 0) archive_string_sprintf(str, " inode=%jd", (intmax_t)me->ino); if ((keys & F_RESDEV) != 0) { archive_string_sprintf(str, " resdevice=native,%ju,%ju", (uintmax_t)me->devmajor, (uintmax_t)me->devminor); } switch (me->filetype) { case AE_IFLNK: if ((keys & F_TYPE) != 0) archive_strcat(str, " type=link"); if ((keys & F_SLINK) != 0) { archive_strcat(str, " link="); mtree_quote(str, me->symlink.s); } break; case AE_IFSOCK: if ((keys & F_TYPE) != 0) archive_strcat(str, " type=socket"); break; case AE_IFCHR: if ((keys & F_TYPE) != 0) archive_strcat(str, " type=char"); if ((keys & F_DEV) != 0) { archive_string_sprintf(str, " device=native,%ju,%ju", (uintmax_t)me->rdevmajor, (uintmax_t)me->rdevminor); } break; case AE_IFBLK: if ((keys & F_TYPE) != 0) archive_strcat(str, " type=block"); if ((keys & F_DEV) != 0) { archive_string_sprintf(str, " device=native,%ju,%ju", (uintmax_t)me->rdevmajor, (uintmax_t)me->rdevminor); } break; case AE_IFDIR: if ((keys & F_TYPE) != 0) archive_strcat(str, " type=dir"); break; case AE_IFIFO: if ((keys & F_TYPE) != 0) archive_strcat(str, " type=fifo"); break; case AE_IFREG: default: /* Handle unknown file types as regular files. */ if ((keys & F_TYPE) != 0) archive_strcat(str, " type=file"); if ((keys & F_SIZE) != 0) archive_string_sprintf(str, " size=%jd", (intmax_t)me->size); break; } /* Write a bunch of sum. */ if (me->reg_info) sum_write(str, me->reg_info); archive_strappend_char(str, '\n'); if (mtree->indent || mtree->classic) mtree_indent(mtree); if (mtree->buf.length > 32768) { ret = __archive_write_output( a, mtree->buf.s, mtree->buf.length); archive_string_empty(&mtree->buf); } else ret = ARCHIVE_OK; return (ret); } static int write_dot_dot_entry(struct archive_write *a, struct mtree_entry *n) { struct mtree_writer *mtree = a->format_data; int ret; if (n->parentdir.s) { if (mtree->indent) { int i, pd = mtree->depth * 4; for (i = 0; i < pd; i++) archive_strappend_char(&mtree->buf, ' '); } archive_string_sprintf(&mtree->buf, "# %s/%s\n", n->parentdir.s, n->basename.s); } if (mtree->indent) { archive_string_empty(&mtree->ebuf); archive_strncat(&mtree->ebuf, "..\n\n", (mtree->dironly)?3:4); mtree_indent(mtree); } else archive_strncat(&mtree->buf, "..\n\n", (mtree->dironly)?3:4); if (mtree->buf.length > 32768) { ret = __archive_write_output( a, mtree->buf.s, mtree->buf.length); archive_string_empty(&mtree->buf); } else ret = ARCHIVE_OK; return (ret); } /* * Write mtree entries saved at attr_counter_set_collect() function. */ static int write_mtree_entry_tree(struct archive_write *a) { struct mtree_writer *mtree = a->format_data; struct mtree_entry *np = mtree->root; struct archive_rb_node *n; int ret; do { if (mtree->output_global_set) { /* * Collect attribute information to know which value * is frequently used among the children. */ attr_counter_set_reset(mtree); ARCHIVE_RB_TREE_FOREACH(n, &(np->dir_info->rbtree)) { struct mtree_entry *e = (struct mtree_entry *)n; if (attr_counter_set_collect(mtree, e) < 0) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } } } if (!np->dir_info->virtual || mtree->classic) { ret = write_mtree_entry(a, np); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } else { /* Whenever output_global_set is enabled * output global value(/set keywords) * even if the directory entry is not allowed * to be written because the global values * can be used for the children. */ if (mtree->output_global_set) write_global(mtree); } /* * Output the attribute of all files except directory files. */ mtree->depth++; ARCHIVE_RB_TREE_FOREACH(n, &(np->dir_info->rbtree)) { struct mtree_entry *e = (struct mtree_entry *)n; if (e->dir_info) mtree_entry_add_child_tail(np, e); else { ret = write_mtree_entry(a, e); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } } mtree->depth--; if (np->dir_info->children.first != NULL) { /* * Descend the tree. */ np = np->dir_info->children.first; if (mtree->indent) mtree->depth++; continue; } else if (mtree->classic) { /* * While printing mtree classic, if there are not * any directory files(except "." and "..") in the * directory, output two dots ".." as returning * the parent directory. */ ret = write_dot_dot_entry(a, np); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } while (np != np->parent) { if (np->dir_info->chnext == NULL) { /* * Ascend the tree; go back to the parent. */ if (mtree->indent) mtree->depth--; if (mtree->classic) { ret = write_dot_dot_entry(a, np->parent); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } np = np->parent; } else { /* * Switch to next mtree entry in the directory. */ np = np->dir_info->chnext; break; } } } while (np != np->parent); return (ARCHIVE_OK); } static int archive_write_mtree_finish_entry(struct archive_write *a) { struct mtree_writer *mtree = a->format_data; struct mtree_entry *me; if ((me = mtree->mtree_entry) == NULL) return (ARCHIVE_OK); mtree->mtree_entry = NULL; if (me->reg_info) sum_final(mtree, me->reg_info); return (ARCHIVE_OK); } static int archive_write_mtree_close(struct archive_write *a) { struct mtree_writer *mtree= a->format_data; int ret; if (mtree->root != NULL) { ret = write_mtree_entry_tree(a); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } archive_write_set_bytes_in_last_block(&a->archive, 1); return __archive_write_output(a, mtree->buf.s, mtree->buf.length); } static ssize_t archive_write_mtree_data(struct archive_write *a, const void *buff, size_t n) { struct mtree_writer *mtree= a->format_data; if (n > mtree->entry_bytes_remaining) n = (size_t)mtree->entry_bytes_remaining; mtree->entry_bytes_remaining -= n; /* We don't need to compute a regular file sum */ if (mtree->mtree_entry == NULL) return (n); if (mtree->mtree_entry->filetype == AE_IFREG) sum_update(mtree, buff, n); return (n); } static int archive_write_mtree_free(struct archive_write *a) { struct mtree_writer *mtree= a->format_data; if (mtree == NULL) return (ARCHIVE_OK); /* Make sure we do not leave any entries. */ mtree_entry_register_free(mtree); archive_string_free(&mtree->cur_dirstr); archive_string_free(&mtree->ebuf); archive_string_free(&mtree->buf); attr_counter_set_free(mtree); free(mtree); a->format_data = NULL; return (ARCHIVE_OK); } static int archive_write_mtree_options(struct archive_write *a, const char *key, const char *value) { struct mtree_writer *mtree= a->format_data; int keybit = 0; switch (key[0]) { case 'a': if (strcmp(key, "all") == 0) keybit = ~0; break; case 'c': if (strcmp(key, "cksum") == 0) keybit = F_CKSUM; break; case 'd': if (strcmp(key, "device") == 0) keybit = F_DEV; else if (strcmp(key, "dironly") == 0) { mtree->dironly = (value != NULL)? 1: 0; return (ARCHIVE_OK); } break; case 'f': if (strcmp(key, "flags") == 0) keybit = F_FLAGS; break; case 'g': if (strcmp(key, "gid") == 0) keybit = F_GID; else if (strcmp(key, "gname") == 0) keybit = F_GNAME; break; case 'i': if (strcmp(key, "indent") == 0) { mtree->indent = (value != NULL)? 1: 0; return (ARCHIVE_OK); } else if (strcmp(key, "inode") == 0) { keybit = F_INO; } break; case 'l': if (strcmp(key, "link") == 0) keybit = F_SLINK; break; case 'm': if (strcmp(key, "md5") == 0 || strcmp(key, "md5digest") == 0) keybit = F_MD5; if (strcmp(key, "mode") == 0) keybit = F_MODE; break; case 'n': if (strcmp(key, "nlink") == 0) keybit = F_NLINK; break; case 'r': if (strcmp(key, "resdevice") == 0) { keybit = F_RESDEV; } else if (strcmp(key, "ripemd160digest") == 0 || strcmp(key, "rmd160") == 0 || strcmp(key, "rmd160digest") == 0) keybit = F_RMD160; break; case 's': if (strcmp(key, "sha1") == 0 || strcmp(key, "sha1digest") == 0) keybit = F_SHA1; if (strcmp(key, "sha256") == 0 || strcmp(key, "sha256digest") == 0) keybit = F_SHA256; if (strcmp(key, "sha384") == 0 || strcmp(key, "sha384digest") == 0) keybit = F_SHA384; if (strcmp(key, "sha512") == 0 || strcmp(key, "sha512digest") == 0) keybit = F_SHA512; if (strcmp(key, "size") == 0) keybit = F_SIZE; break; case 't': if (strcmp(key, "time") == 0) keybit = F_TIME; else if (strcmp(key, "type") == 0) keybit = F_TYPE; break; case 'u': if (strcmp(key, "uid") == 0) keybit = F_UID; else if (strcmp(key, "uname") == 0) keybit = F_UNAME; else if (strcmp(key, "use-set") == 0) { mtree->output_global_set = (value != NULL)? 1: 0; return (ARCHIVE_OK); } break; } if (keybit != 0) { if (value != NULL) mtree->keys |= keybit; else mtree->keys &= ~keybit; return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int archive_write_set_format_mtree_default(struct archive *_a, const char *fn) { struct archive_write *a = (struct archive_write *)_a; struct mtree_writer *mtree; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, fn); if (a->format_free != NULL) (a->format_free)(a); if ((mtree = calloc(1, sizeof(*mtree))) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate mtree data"); return (ARCHIVE_FATAL); } mtree->mtree_entry = NULL; mtree->first = 1; memset(&(mtree->set), 0, sizeof(mtree->set)); mtree->keys = DEFAULT_KEYS; mtree->dironly = 0; mtree->indent = 0; archive_string_init(&mtree->ebuf); archive_string_init(&mtree->buf); mtree_entry_register_init(mtree); a->format_data = mtree; a->format_free = archive_write_mtree_free; a->format_name = "mtree"; a->format_options = archive_write_mtree_options; a->format_write_header = archive_write_mtree_header; a->format_close = archive_write_mtree_close; a->format_write_data = archive_write_mtree_data; a->format_finish_entry = archive_write_mtree_finish_entry; a->archive.archive_format = ARCHIVE_FORMAT_MTREE; a->archive.archive_format_name = "mtree"; return (ARCHIVE_OK); } int archive_write_set_format_mtree(struct archive *_a) { return archive_write_set_format_mtree_default(_a, "archive_write_set_format_mtree"); } int archive_write_set_format_mtree_classic(struct archive *_a) { int r; r = archive_write_set_format_mtree_default(_a, "archive_write_set_format_mtree_classic"); if (r == ARCHIVE_OK) { struct archive_write *a = (struct archive_write *)_a; struct mtree_writer *mtree; mtree = (struct mtree_writer *)a->format_data; /* Set to output a mtree archive in classic format. */ mtree->classic = 1; /* Basically, mtree classic format uses '/set' global * value. */ mtree->output_global_set = 1; } return (r); } static void sum_init(struct mtree_writer *mtree) { mtree->compute_sum = 0; if (mtree->keys & F_CKSUM) { mtree->compute_sum |= F_CKSUM; mtree->crc = 0; mtree->crc_len = 0; } #ifdef ARCHIVE_HAS_MD5 if (mtree->keys & F_MD5) { if (archive_md5_init(&mtree->md5ctx) == ARCHIVE_OK) mtree->compute_sum |= F_MD5; else mtree->keys &= ~F_MD5;/* Not supported. */ } #endif #ifdef ARCHIVE_HAS_RMD160 if (mtree->keys & F_RMD160) { if (archive_rmd160_init(&mtree->rmd160ctx) == ARCHIVE_OK) mtree->compute_sum |= F_RMD160; else mtree->keys &= ~F_RMD160;/* Not supported. */ } #endif #ifdef ARCHIVE_HAS_SHA1 if (mtree->keys & F_SHA1) { if (archive_sha1_init(&mtree->sha1ctx) == ARCHIVE_OK) mtree->compute_sum |= F_SHA1; else mtree->keys &= ~F_SHA1;/* Not supported. */ } #endif #ifdef ARCHIVE_HAS_SHA256 if (mtree->keys & F_SHA256) { if (archive_sha256_init(&mtree->sha256ctx) == ARCHIVE_OK) mtree->compute_sum |= F_SHA256; else mtree->keys &= ~F_SHA256;/* Not supported. */ } #endif #ifdef ARCHIVE_HAS_SHA384 if (mtree->keys & F_SHA384) { if (archive_sha384_init(&mtree->sha384ctx) == ARCHIVE_OK) mtree->compute_sum |= F_SHA384; else mtree->keys &= ~F_SHA384;/* Not supported. */ } #endif #ifdef ARCHIVE_HAS_SHA512 if (mtree->keys & F_SHA512) { if (archive_sha512_init(&mtree->sha512ctx) == ARCHIVE_OK) mtree->compute_sum |= F_SHA512; else mtree->keys &= ~F_SHA512;/* Not supported. */ } #endif } static void sum_update(struct mtree_writer *mtree, const void *buff, size_t n) { if (mtree->compute_sum & F_CKSUM) { /* * Compute a POSIX 1003.2 checksum */ const unsigned char *p; size_t nn; for (nn = n, p = buff; nn--; ++p) COMPUTE_CRC(mtree->crc, *p); mtree->crc_len += n; } #ifdef ARCHIVE_HAS_MD5 if (mtree->compute_sum & F_MD5) { archive_md5_update(&mtree->md5ctx, buff, n); mtree->mtree_entry->reg_info->mset_digest &= ~AE_MSET_DIGEST_MD5; } #endif #ifdef ARCHIVE_HAS_RMD160 if (mtree->compute_sum & F_RMD160) { archive_rmd160_update(&mtree->rmd160ctx, buff, n); mtree->mtree_entry->reg_info->mset_digest &= ~AE_MSET_DIGEST_RMD160; } #endif #ifdef ARCHIVE_HAS_SHA1 if (mtree->compute_sum & F_SHA1) { archive_sha1_update(&mtree->sha1ctx, buff, n); mtree->mtree_entry->reg_info->mset_digest &= ~AE_MSET_DIGEST_SHA1; } #endif #ifdef ARCHIVE_HAS_SHA256 if (mtree->compute_sum & F_SHA256) { archive_sha256_update(&mtree->sha256ctx, buff, n); mtree->mtree_entry->reg_info->mset_digest &= ~AE_MSET_DIGEST_SHA256; } #endif #ifdef ARCHIVE_HAS_SHA384 if (mtree->compute_sum & F_SHA384) { archive_sha384_update(&mtree->sha384ctx, buff, n); mtree->mtree_entry->reg_info->mset_digest &= ~AE_MSET_DIGEST_SHA384; } #endif #ifdef ARCHIVE_HAS_SHA512 if (mtree->compute_sum & F_SHA512) { archive_sha512_update(&mtree->sha512ctx, buff, n); mtree->mtree_entry->reg_info->mset_digest &= ~AE_MSET_DIGEST_SHA512; } #endif } static void sum_final(struct mtree_writer *mtree, struct reg_info *reg) { if (mtree->compute_sum & F_CKSUM) { uint64_t len; /* Include the length of the file. */ for (len = mtree->crc_len; len != 0; len >>= 8) COMPUTE_CRC(mtree->crc, len & 0xff); reg->crc = ~mtree->crc; } #ifdef ARCHIVE_HAS_MD5 if ((mtree->compute_sum & F_MD5) && !(reg->mset_digest & AE_MSET_DIGEST_MD5)) archive_md5_final(&mtree->md5ctx, reg->digest.md5); #endif #ifdef ARCHIVE_HAS_RMD160 if ((mtree->compute_sum & F_RMD160) && !(reg->mset_digest & AE_MSET_DIGEST_RMD160)) archive_rmd160_final(&mtree->rmd160ctx, reg->digest.rmd160); #endif #ifdef ARCHIVE_HAS_SHA1 if ((mtree->compute_sum & F_SHA1) && !(reg->mset_digest & AE_MSET_DIGEST_SHA1)) archive_sha1_final(&mtree->sha1ctx, reg->digest.sha1); #endif #ifdef ARCHIVE_HAS_SHA256 if ((mtree->compute_sum & F_SHA256) && !(reg->mset_digest & AE_MSET_DIGEST_SHA256)) archive_sha256_final(&mtree->sha256ctx, reg->digest.sha256); #endif #ifdef ARCHIVE_HAS_SHA384 if ((mtree->compute_sum & F_SHA384) && !(reg->mset_digest & AE_MSET_DIGEST_SHA384)) archive_sha384_final(&mtree->sha384ctx, reg->digest.sha384); #endif #ifdef ARCHIVE_HAS_SHA512 if ((mtree->compute_sum & F_SHA512) && !(reg->mset_digest & AE_MSET_DIGEST_SHA512)) archive_sha512_final(&mtree->sha512ctx, reg->digest.sha512); #endif /* Save what types of sum are computed. */ reg->compute_sum = mtree->compute_sum; } #if defined(ARCHIVE_HAS_MD5) || defined(ARCHIVE_HAS_RMD160) || \ defined(ARCHIVE_HAS_SHA1) || defined(ARCHIVE_HAS_SHA256) || \ defined(ARCHIVE_HAS_SHA384) || defined(ARCHIVE_HAS_SHA512) static void strappend_bin(struct archive_string *s, const unsigned char *bin, int n) { static const char hex[] = "0123456789abcdef"; int i; for (i = 0; i < n; i++) { archive_strappend_char(s, hex[bin[i] >> 4]); archive_strappend_char(s, hex[bin[i] & 0x0f]); } } #endif static void sum_write(struct archive_string *str, struct reg_info *reg) { if (reg->compute_sum & F_CKSUM) { archive_string_sprintf(str, " cksum=%ju", (uintmax_t)reg->crc); } #define append_digest(_s, _r, _t) \ strappend_bin(_s, _r->digest._t, sizeof(_r->digest._t)) #ifdef ARCHIVE_HAS_MD5 if (reg->compute_sum & F_MD5) { archive_strcat(str, " md5digest="); append_digest(str, reg, md5); } #endif #ifdef ARCHIVE_HAS_RMD160 if (reg->compute_sum & F_RMD160) { archive_strcat(str, " rmd160digest="); append_digest(str, reg, rmd160); } #endif #ifdef ARCHIVE_HAS_SHA1 if (reg->compute_sum & F_SHA1) { archive_strcat(str, " sha1digest="); append_digest(str, reg, sha1); } #endif #ifdef ARCHIVE_HAS_SHA256 if (reg->compute_sum & F_SHA256) { archive_strcat(str, " sha256digest="); append_digest(str, reg, sha256); } #endif #ifdef ARCHIVE_HAS_SHA384 if (reg->compute_sum & F_SHA384) { archive_strcat(str, " sha384digest="); append_digest(str, reg, sha384); } #endif #ifdef ARCHIVE_HAS_SHA512 if (reg->compute_sum & F_SHA512) { archive_strcat(str, " sha512digest="); append_digest(str, reg, sha512); } #endif #undef append_digest } static int mtree_entry_cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct mtree_entry *e1 = (const struct mtree_entry *)n1; const struct mtree_entry *e2 = (const struct mtree_entry *)n2; return (strcmp(e2->basename.s, e1->basename.s)); } static int mtree_entry_cmp_key(const struct archive_rb_node *n, const void *key) { const struct mtree_entry *e = (const struct mtree_entry *)n; return (strcmp((const char *)key, e->basename.s)); } #if defined(_WIN32) || defined(__CYGWIN__) static int cleanup_backslash_1(char *p) { int mb, dos; mb = dos = 0; while (*p) { if (*(unsigned char *)p > 127) mb = 1; if (*p == '\\') { /* If we have not met any multi-byte characters, * we can replace '\' with '/'. */ if (!mb) *p = '/'; dos = 1; } p++; } if (!mb || !dos) return (0); return (-1); } static void cleanup_backslash_2(wchar_t *p) { /* Convert a path-separator from '\' to '/' */ while (*p != L'\0') { if (*p == L'\\') *p = L'/'; p++; } } #endif /* * Generate a parent directory name and a base name from a pathname. */ static int mtree_entry_setup_filenames(struct archive_write *a, struct mtree_entry *file, struct archive_entry *entry) { const char *pathname; char *p, *dirname, *slash; size_t len; int ret = ARCHIVE_OK; archive_strcpy(&file->pathname, archive_entry_pathname(entry)); #if defined(_WIN32) || defined(__CYGWIN__) /* * Convert a path-separator from '\' to '/' */ if (cleanup_backslash_1(file->pathname.s) != 0) { const wchar_t *wp = archive_entry_pathname_w(entry); struct archive_wstring ws; if (wp != NULL) { int r; archive_string_init(&ws); archive_wstrcpy(&ws, wp); cleanup_backslash_2(ws.s); archive_string_empty(&(file->pathname)); r = archive_string_append_from_wcs(&(file->pathname), ws.s, ws.length); archive_wstring_free(&ws); if (r < 0 && errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } } } #else (void)a; /* UNUSED */ #endif pathname = file->pathname.s; if (strcmp(pathname, ".") == 0) { archive_strcpy(&file->basename, "."); return (ARCHIVE_OK); } archive_strcpy(&(file->parentdir), pathname); len = file->parentdir.length; p = dirname = file->parentdir.s; /* * Remove leading '/' and '../' elements */ while (*p) { if (p[0] == '/') { p++; len--; } else if (p[0] != '.') break; else if (p[1] == '.' && p[2] == '/') { p += 3; len -= 3; } else break; } if (p != dirname) { memmove(dirname, p, len+1); p = dirname; } /* * Remove "/","/." and "/.." elements from tail. */ while (len > 0) { size_t ll = len; if (len > 0 && p[len-1] == '/') { p[len-1] = '\0'; len--; } if (len > 1 && p[len-2] == '/' && p[len-1] == '.') { p[len-2] = '\0'; len -= 2; } if (len > 2 && p[len-3] == '/' && p[len-2] == '.' && p[len-1] == '.') { p[len-3] = '\0'; len -= 3; } if (ll == len) break; } while (*p) { if (p[0] == '/') { if (p[1] == '/') /* Convert '//' --> '/' */ memmove(p, p+1, strlen(p+1) + 1); else if (p[1] == '.' && p[2] == '/') /* Convert '/./' --> '/' */ memmove(p, p+2, strlen(p+2) + 1); else if (p[1] == '.' && p[2] == '.' && p[3] == '/') { /* Convert 'dir/dir1/../dir2/' * --> 'dir/dir2/' */ char *rp = p -1; while (rp >= dirname) { if (*rp == '/') break; --rp; } if (rp > dirname) { strcpy(rp, p+3); p = rp; } else { strcpy(dirname, p+4); p = dirname; } } else p++; } else p++; } p = dirname; len = strlen(p); /* * Add "./" prefix. * NOTE: If the pathname does not have a path separator, we have * to add "./" to the head of the pathname because mtree reader * will suppose that it is v1(a.k.a classic) mtree format and * change the directory unexpectedly and so it will make a wrong * path. */ if (strcmp(p, ".") != 0 && strncmp(p, "./", 2) != 0) { struct archive_string as; archive_string_init(&as); archive_strcpy(&as, "./"); archive_strncat(&as, p, len); archive_string_empty(&file->parentdir); archive_string_concat(&file->parentdir, &as); archive_string_free(&as); p = file->parentdir.s; len = archive_strlen(&file->parentdir); } /* - * Find out the position which points the last position of + * Find out the position which points to the last position of * path separator('/'). */ slash = NULL; for (; *p != '\0'; p++) { if (*p == '/') slash = p; } if (slash == NULL) { /* The pathname doesn't have a parent directory. */ file->parentdir.length = len; archive_string_copy(&(file->basename), &(file->parentdir)); archive_string_empty(&(file->parentdir)); *file->parentdir.s = '\0'; return (ret); } /* Make a basename from file->parentdir.s and slash */ *slash = '\0'; file->parentdir.length = slash - file->parentdir.s; archive_strcpy(&(file->basename), slash + 1); return (ret); } static int mtree_entry_create_virtual_dir(struct archive_write *a, const char *pathname, struct mtree_entry **m_entry) { struct archive_entry *entry; struct mtree_entry *file; int r; entry = archive_entry_new(); if (entry == NULL) { *m_entry = NULL; archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } archive_entry_copy_pathname(entry, pathname); archive_entry_set_mode(entry, AE_IFDIR | 0755); archive_entry_set_mtime(entry, time(NULL), 0); r = mtree_entry_new(a, entry, &file); archive_entry_free(entry); if (r < ARCHIVE_WARN) { *m_entry = NULL; archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } file->dir_info->virtual = 1; *m_entry = file; return (ARCHIVE_OK); } static void mtree_entry_register_add(struct mtree_writer *mtree, struct mtree_entry *file) { file->next = NULL; *mtree->file_list.last = file; mtree->file_list.last = &(file->next); } static void mtree_entry_register_init(struct mtree_writer *mtree) { mtree->file_list.first = NULL; mtree->file_list.last = &(mtree->file_list.first); } static void mtree_entry_register_free(struct mtree_writer *mtree) { struct mtree_entry *file, *file_next; file = mtree->file_list.first; while (file != NULL) { file_next = file->next; mtree_entry_free(file); file = file_next; } } static int mtree_entry_add_child_tail(struct mtree_entry *parent, struct mtree_entry *child) { child->dir_info->chnext = NULL; *parent->dir_info->children.last = child; parent->dir_info->children.last = &(child->dir_info->chnext); return (1); } /* - * Find a entry from a parent entry with the name. + * Find an entry from a parent entry with given name. */ static struct mtree_entry * mtree_entry_find_child(struct mtree_entry *parent, const char *child_name) { struct mtree_entry *np; if (parent == NULL) return (NULL); np = (struct mtree_entry *)__archive_rb_tree_find_node( &(parent->dir_info->rbtree), child_name); return (np); } static int get_path_component(char *name, size_t n, const char *fn) { char *p; size_t l; p = strchr(fn, '/'); if (p == NULL) { if ((l = strlen(fn)) == 0) return (0); } else l = p - fn; if (l > n -1) return (-1); memcpy(name, fn, l); name[l] = '\0'; return ((int)l); } /* * Add a new entry into the tree. */ static int mtree_entry_tree_add(struct archive_write *a, struct mtree_entry **filep) { #if defined(_WIN32) && !defined(__CYGWIN__) char name[_MAX_FNAME];/* Included null terminator size. */ #elif defined(NAME_MAX) && NAME_MAX >= 255 char name[NAME_MAX+1]; #else char name[256]; #endif struct mtree_writer *mtree = (struct mtree_writer *)a->format_data; struct mtree_entry *dent, *file, *np; const char *fn, *p; int l, r; file = *filep; if (file->parentdir.length == 0 && file->basename.length == 1 && file->basename.s[0] == '.') { file->parent = file; if (mtree->root != NULL) { np = mtree->root; goto same_entry; } mtree->root = file; mtree_entry_register_add(mtree, file); return (ARCHIVE_OK); } if (file->parentdir.length == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Internal programming error " "in generating canonical name for %s", file->pathname.s); return (ARCHIVE_FAILED); } fn = p = file->parentdir.s; /* * If the path of the parent directory of `file' entry is * the same as the path of `cur_dirent', add `file' entry to * `cur_dirent'. */ if (archive_strlen(&(mtree->cur_dirstr)) == archive_strlen(&(file->parentdir)) && strcmp(mtree->cur_dirstr.s, fn) == 0) { if (!__archive_rb_tree_insert_node( &(mtree->cur_dirent->dir_info->rbtree), (struct archive_rb_node *)file)) { /* There is the same name in the tree. */ np = (struct mtree_entry *)__archive_rb_tree_find_node( &(mtree->cur_dirent->dir_info->rbtree), file->basename.s); goto same_entry; } file->parent = mtree->cur_dirent; mtree_entry_register_add(mtree, file); return (ARCHIVE_OK); } dent = mtree->root; for (;;) { l = get_path_component(name, sizeof(name), fn); if (l == 0) { np = NULL; break; } if (l < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "A name buffer is too small"); return (ARCHIVE_FATAL); } if (l == 1 && name[0] == '.' && dent != NULL && dent == mtree->root) { fn += l; if (fn[0] == '/') fn++; continue; } np = mtree_entry_find_child(dent, name); if (np == NULL || fn[0] == '\0') break; /* Find next sub directory. */ if (!np->dir_info) { - /* NOT Directory! */ + /* NOT a directory! */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, - "`%s' is not directory, we cannot insert `%s' ", + "`%s' is not a directory, we cannot insert `%s' ", np->pathname.s, file->pathname.s); return (ARCHIVE_FAILED); } fn += l; if (fn[0] == '/') fn++; dent = np; } if (np == NULL) { /* * Create virtual parent directories. */ while (fn[0] != '\0') { struct mtree_entry *vp; struct archive_string as; archive_string_init(&as); archive_strncat(&as, p, fn - p + l); if (as.s[as.length-1] == '/') { as.s[as.length-1] = '\0'; as.length--; } r = mtree_entry_create_virtual_dir(a, as.s, &vp); archive_string_free(&as); if (r < ARCHIVE_WARN) return (r); if (strcmp(vp->pathname.s, ".") == 0) { vp->parent = vp; mtree->root = vp; } else { __archive_rb_tree_insert_node( &(dent->dir_info->rbtree), (struct archive_rb_node *)vp); vp->parent = dent; } mtree_entry_register_add(mtree, vp); np = vp; fn += l; if (fn[0] == '/') fn++; l = get_path_component(name, sizeof(name), fn); if (l < 0) { archive_string_free(&as); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "A name buffer is too small"); return (ARCHIVE_FATAL); } dent = np; } /* Found out the parent directory where `file' can be * inserted. */ mtree->cur_dirent = dent; archive_string_empty(&(mtree->cur_dirstr)); if (archive_string_ensure(&(mtree->cur_dirstr), archive_strlen(&(dent->parentdir)) + archive_strlen(&(dent->basename)) + 2) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } if (archive_strlen(&(dent->parentdir)) + archive_strlen(&(dent->basename)) == 0) mtree->cur_dirstr.s[0] = 0; else { if (archive_strlen(&(dent->parentdir)) > 0) { archive_string_copy(&(mtree->cur_dirstr), &(dent->parentdir)); archive_strappend_char( &(mtree->cur_dirstr), '/'); } archive_string_concat(&(mtree->cur_dirstr), &(dent->basename)); } if (!__archive_rb_tree_insert_node( &(dent->dir_info->rbtree), (struct archive_rb_node *)file)) { np = (struct mtree_entry *)__archive_rb_tree_find_node( &(dent->dir_info->rbtree), file->basename.s); goto same_entry; } file->parent = dent; mtree_entry_register_add(mtree, file); return (ARCHIVE_OK); } same_entry: - /* - * We have already has the entry the filename of which is - * the same. - */ + /* We already have an entry with same filename. */ r = mtree_entry_exchange_same_entry(a, np, file); if (r < ARCHIVE_WARN) return (r); if (np->dir_info) np->dir_info->virtual = 0; *filep = np; mtree_entry_free(file); return (ARCHIVE_WARN); } static int mtree_entry_exchange_same_entry(struct archive_write *a, struct mtree_entry *np, struct mtree_entry *file) { if ((np->mode & AE_IFMT) != (file->mode & AE_IFMT)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, - "Found duplicate entries `%s' and its file type is " - "different", + "Found duplicate entries for `%s' with " + "differing file types.", np->pathname.s); return (ARCHIVE_FAILED); } - /* Update the existent mtree entry's attributes by the new one's. */ + /* Update the existing mtree entry's attributes by the new one's. */ archive_string_empty(&np->symlink); archive_string_concat(&np->symlink, &file->symlink); archive_string_empty(&np->uname); archive_string_concat(&np->uname, &file->uname); archive_string_empty(&np->gname); archive_string_concat(&np->gname, &file->gname); archive_string_empty(&np->fflags_text); archive_string_concat(&np->fflags_text, &file->fflags_text); np->nlink = file->nlink; np->filetype = file->filetype; np->mode = file->mode; np->size = file->size; np->uid = file->uid; np->gid = file->gid; np->fflags_set = file->fflags_set; np->fflags_clear = file->fflags_clear; np->mtime = file->mtime; np->mtime_nsec = file->mtime_nsec; np->rdevmajor = file->rdevmajor; np->rdevminor = file->rdevminor; np->devmajor = file->devmajor; np->devminor = file->devminor; np->ino = file->ino; return (ARCHIVE_WARN); } diff --git a/contrib/libarchive/libarchive/archive_write_set_format_xar.c b/contrib/libarchive/libarchive/archive_write_set_format_xar.c index 3775e9f5819a..9921f1032be5 100644 --- a/contrib/libarchive/libarchive/archive_write_set_format_xar.c +++ b/contrib/libarchive/libarchive/archive_write_set_format_xar.c @@ -1,3698 +1,3698 @@ /*- * Copyright (c) 2010-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #include #if HAVE_LIBXML_XMLWRITER_H #include #if defined(LIBXML_VERSION) && LIBXML_VERSION >= 20703 #define XAR_WRITER_HAS_XML #endif /* LIBXML_VERSION */ #elif HAVE_XMLLITE_H #include #include #include #define XAR_WRITER_HAS_XML #endif #ifdef HAVE_BZLIB_H #include #endif #if HAVE_LZMA_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #include "archive.h" #include "archive_digest_private.h" #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_rb.h" #include "archive_string.h" #include "archive_write_private.h" /* * Differences to xar utility. * - Subdocument is not supported yet. * - ACL is not supported yet. * - When writing an XML element , * which is a file type a symbolic link is referencing is always marked * as "broken". Xar utility uses stat(2) to get the file type, but, in * libarchive format writer, we should not use it; if it is needed, we * should get about it at archive_read_disk.c. * - It is possible to appear both and elements. * Xar utility generates on BSD platform and on Linux * platform. * */ #if !defined(XAR_WRITER_HAS_XML) ||\ !defined(HAVE_ZLIB_H) || \ !defined(ARCHIVE_HAS_MD5) || !defined(ARCHIVE_HAS_SHA1) /* * xar needs several external libraries. * o libxml2 or xmllite (on Windows) * o openssl or MD5/SHA1 hash function * o zlib * o bzlib2 (option) * o liblzma (option) */ int archive_write_set_format_xar(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Xar not supported on this platform"); return (ARCHIVE_WARN); } #else /* Support xar format */ struct xml_writer; static int xml_writer_create(struct xml_writer **pctx); static int xml_writer_start_document(struct xml_writer *ctx); static int xml_writer_end_document(struct xml_writer *ctx); static int xml_writer_set_indent(struct xml_writer *ctx, unsigned int indent); static int xml_writer_start_element(struct xml_writer *ctx, const char *localName); static int xml_writer_write_attribute(struct xml_writer *ctx, const char *key, const char *value); static int xml_writer_write_attributef(struct xml_writer *ctx, const char *key, const char *format, ...); static int xml_writer_write_string(struct xml_writer *ctx, const char *string); static int xml_writer_write_base64(struct xml_writer* ctx, const char *data, size_t start, size_t len); static int xml_writer_end_element(struct xml_writer *ctx); static int xml_writer_get_final_content_and_length(struct xml_writer *ctx, const char **out, size_t *size); static int xml_writer_destroy(struct xml_writer *ctx); /*#define DEBUG_PRINT_TOC 1 */ #define HEADER_MAGIC 0x78617221 #define HEADER_SIZE 28 #define HEADER_VERSION 1 enum sumalg { CKSUM_NONE = 0, CKSUM_SHA1 = 1, CKSUM_MD5 = 2 }; #define MD5_SIZE 16 #define SHA1_SIZE 20 #define MAX_SUM_SIZE 20 #define MD5_NAME "md5" #define SHA1_NAME "sha1" enum enctype { NONE, GZIP, BZIP2, LZMA, XZ, }; struct chksumwork { enum sumalg alg; #ifdef ARCHIVE_HAS_MD5 archive_md5_ctx md5ctx; #endif #ifdef ARCHIVE_HAS_SHA1 archive_sha1_ctx sha1ctx; #endif }; enum la_zaction { ARCHIVE_Z_FINISH, ARCHIVE_Z_RUN }; /* * Universal zstream. */ struct la_zstream { const unsigned char *next_in; size_t avail_in; uint64_t total_in; unsigned char *next_out; size_t avail_out; uint64_t total_out; int valid; void *real_stream; int (*code) (struct archive *a, struct la_zstream *lastrm, enum la_zaction action); int (*end)(struct archive *a, struct la_zstream *lastrm); }; struct chksumval { enum sumalg alg; size_t len; unsigned char val[MAX_SUM_SIZE]; }; struct heap_data { int id; struct heap_data *next; uint64_t temp_offset; uint64_t length; /* archived size. */ uint64_t size; /* extracted size. */ enum enctype compression; struct chksumval a_sum; /* archived checksum. */ struct chksumval e_sum; /* extracted checksum. */ }; struct file { struct archive_rb_node rbnode; int id; struct archive_entry *entry; struct archive_rb_tree rbtree; struct file *next; struct file *chnext; struct file *hlnext; /* For hardlinked files. * Use only when archive_entry_nlink() > 1 */ struct file *hardlink_target; struct file *parent; /* parent directory entry */ /* * To manage sub directory files. * We use 'chnext' (a member of struct file) to chain. */ struct { struct file *first; struct file **last; } children; /* For making a directory tree. */ struct archive_string parentdir; struct archive_string basename; struct archive_string symlink; int ea_idx; struct { struct heap_data *first; struct heap_data **last; } xattr; struct heap_data data; struct archive_string script; unsigned int virtual:1; unsigned int dir:1; }; struct hardlink { struct archive_rb_node rbnode; int nlink; struct { struct file *first; struct file **last; } file_list; }; struct xar { int temp_fd; uint64_t temp_offset; int file_idx; struct file *root; struct file *cur_dirent; struct archive_string cur_dirstr; struct file *cur_file; uint64_t bytes_remaining; struct archive_string tstr; struct archive_string vstr; enum sumalg opt_toc_sumalg; enum sumalg opt_sumalg; enum enctype opt_compression; int opt_compression_level; uint32_t opt_threads; struct chksumwork a_sumwrk; /* archived checksum. */ struct chksumwork e_sumwrk; /* extracted checksum. */ struct la_zstream stream; struct archive_string_conv *sconv; /* * Compressed data buffer. */ unsigned char wbuff[1024 * 64]; size_t wbuff_remaining; struct heap_data toc; /* * The list of all file entries is used to manage struct file * objects. * We use 'next' (a member of struct file) to chain. */ struct { struct file *first; struct file **last; } file_list; /* * The list of hard-linked file entries. * We use 'hlnext' (a member of struct file) to chain. */ struct archive_rb_tree hardlink_rbtree; }; static int xar_options(struct archive_write *, const char *, const char *); static int xar_write_header(struct archive_write *, struct archive_entry *); static ssize_t xar_write_data(struct archive_write *, const void *, size_t); static int xar_finish_entry(struct archive_write *); static int xar_close(struct archive_write *); static int xar_free(struct archive_write *); static struct file *file_new(struct archive_write *a, struct archive_entry *); static void file_free(struct file *); static struct file *file_create_virtual_dir(struct archive_write *a, struct xar *, const char *); static int file_add_child_tail(struct file *, struct file *); static struct file *file_find_child(struct file *, const char *); static int file_gen_utility_names(struct archive_write *, struct file *); static int get_path_component(char *, int, const char *); static int file_tree(struct archive_write *, struct file **); static void file_register(struct xar *, struct file *); static void file_init_register(struct xar *); static void file_free_register(struct xar *); static int file_register_hardlink(struct archive_write *, struct file *); static void file_connect_hardlink_files(struct xar *); static void file_init_hardlinks(struct xar *); static void file_free_hardlinks(struct xar *); static void checksum_init(struct chksumwork *, enum sumalg); static void checksum_update(struct chksumwork *, const void *, size_t); static void checksum_final(struct chksumwork *, struct chksumval *); static int compression_init_encoder_gzip(struct archive *, struct la_zstream *, int, int); static int compression_code_gzip(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_gzip(struct archive *, struct la_zstream *); static int compression_init_encoder_bzip2(struct archive *, struct la_zstream *, int); #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) static int compression_code_bzip2(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_bzip2(struct archive *, struct la_zstream *); #endif static int compression_init_encoder_lzma(struct archive *, struct la_zstream *, int); static int compression_init_encoder_xz(struct archive *, struct la_zstream *, int, int); #if defined(HAVE_LZMA_H) static int compression_code_lzma(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_lzma(struct archive *, struct la_zstream *); #endif static int xar_compression_init_encoder(struct archive_write *); static int compression_code(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end(struct archive *, struct la_zstream *); static int save_xattrs(struct archive_write *, struct file *); static int getalgsize(enum sumalg); static const char *getalgname(enum sumalg); int archive_write_set_format_xar(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct xar *xar; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_xar"); /* If another format was already registered, unregister it. */ if (a->format_free != NULL) (a->format_free)(a); xar = calloc(1, sizeof(*xar)); if (xar == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate xar data"); return (ARCHIVE_FATAL); } xar->temp_fd = -1; file_init_register(xar); file_init_hardlinks(xar); archive_string_init(&(xar->tstr)); archive_string_init(&(xar->vstr)); /* * Create the root directory. */ xar->root = file_create_virtual_dir(a, xar, ""); if (xar->root == NULL) { free(xar); archive_set_error(&a->archive, ENOMEM, "Can't allocate xar data"); return (ARCHIVE_FATAL); } xar->root->parent = xar->root; file_register(xar, xar->root); xar->cur_dirent = xar->root; archive_string_init(&(xar->cur_dirstr)); archive_string_ensure(&(xar->cur_dirstr), 1); xar->cur_dirstr.s[0] = 0; /* * Initialize option. */ /* Set default checksum type. */ xar->opt_toc_sumalg = CKSUM_SHA1; xar->opt_sumalg = CKSUM_SHA1; /* Set default compression type, level, and number of threads. */ xar->opt_compression = GZIP; xar->opt_compression_level = 6; xar->opt_threads = 1; a->format_data = xar; a->format_name = "xar"; a->format_options = xar_options; a->format_write_header = xar_write_header; a->format_write_data = xar_write_data; a->format_finish_entry = xar_finish_entry; a->format_close = xar_close; a->format_free = xar_free; a->archive.archive_format = ARCHIVE_FORMAT_XAR; a->archive.archive_format_name = "xar"; return (ARCHIVE_OK); } static int xar_options(struct archive_write *a, const char *key, const char *value) { struct xar *xar; xar = (struct xar *)a->format_data; if (strcmp(key, "checksum") == 0) { if (value == NULL) xar->opt_sumalg = CKSUM_NONE; else if (strcmp(value, "none") == 0) xar->opt_sumalg = CKSUM_NONE; else if (strcmp(value, "sha1") == 0) xar->opt_sumalg = CKSUM_SHA1; else if (strcmp(value, "md5") == 0) xar->opt_sumalg = CKSUM_MD5; else { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Unknown checksum name: `%s'", value); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } if (strcmp(key, "compression") == 0) { const char *name = NULL; if (value == NULL) xar->opt_compression = NONE; else if (strcmp(value, "none") == 0) xar->opt_compression = NONE; else if (strcmp(value, "gzip") == 0) xar->opt_compression = GZIP; else if (strcmp(value, "bzip2") == 0) #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) xar->opt_compression = BZIP2; #else name = "bzip2"; #endif else if (strcmp(value, "lzma") == 0) #if HAVE_LZMA_H xar->opt_compression = LZMA; #else name = "lzma"; #endif else if (strcmp(value, "xz") == 0) #if HAVE_LZMA_H xar->opt_compression = XZ; #else name = "xz"; #endif else { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Unknown compression name: `%s'", value); return (ARCHIVE_FAILED); } if (name != NULL) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "`%s' compression not supported " "on this platform", name); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } if (strcmp(key, "compression-level") == 0) { if (value == NULL || !(value[0] >= '0' && value[0] <= '9') || value[1] != '\0') { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Illegal value `%s'", value); return (ARCHIVE_FAILED); } xar->opt_compression_level = value[0] - '0'; return (ARCHIVE_OK); } if (strcmp(key, "toc-checksum") == 0) { if (value == NULL) xar->opt_toc_sumalg = CKSUM_NONE; else if (strcmp(value, "none") == 0) xar->opt_toc_sumalg = CKSUM_NONE; else if (strcmp(value, "sha1") == 0) xar->opt_toc_sumalg = CKSUM_SHA1; else if (strcmp(value, "md5") == 0) xar->opt_toc_sumalg = CKSUM_MD5; else { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Unknown checksum name: `%s'", value); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } if (strcmp(key, "threads") == 0) { char *endptr; if (value == NULL) return (ARCHIVE_FAILED); errno = 0; xar->opt_threads = (int)strtoul(value, &endptr, 10); if (errno != 0 || *endptr != '\0') { xar->opt_threads = 1; archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Illegal value `%s'", value); return (ARCHIVE_FAILED); } if (xar->opt_threads == 0) { #ifdef HAVE_LZMA_STREAM_ENCODER_MT xar->opt_threads = lzma_cputhreads(); #else xar->opt_threads = 1; #endif } } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int xar_write_header(struct archive_write *a, struct archive_entry *entry) { struct xar *xar; struct file *file; struct archive_entry *file_entry; int r, r2; xar = (struct xar *)a->format_data; xar->cur_file = NULL; xar->bytes_remaining = 0; if (xar->sconv == NULL) { xar->sconv = archive_string_conversion_to_charset( &a->archive, "UTF-8", 1); if (xar->sconv == NULL) return (ARCHIVE_FATAL); } file = file_new(a, entry); if (file == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate data"); return (ARCHIVE_FATAL); } r2 = file_gen_utility_names(a, file); if (r2 < ARCHIVE_WARN) return (r2); /* * Ignore a path which looks like the top of directory name * since we have already made the root directory of an Xar archive. */ if (archive_strlen(&(file->parentdir)) == 0 && archive_strlen(&(file->basename)) == 0) { file_free(file); return (r2); } /* Add entry into tree */ file_entry = file->entry; r = file_tree(a, &file); if (r != ARCHIVE_OK) return (r); /* There is the same file in tree and * the current file is older than the file in tree. * So we don't need the current file data anymore. */ if (file->entry != file_entry) return (r2); if (file->id == 0) file_register(xar, file); /* A virtual file, which is a directory, does not have * any contents and we won't store it into a archive * file other than its name. */ if (file->virtual) return (r2); /* * Prepare to save the contents of the file. */ if (xar->temp_fd == -1) { int algsize; xar->temp_offset = 0; xar->temp_fd = __archive_mktemp(NULL); if (xar->temp_fd < 0) { archive_set_error(&a->archive, errno, "Couldn't create temporary file"); return (ARCHIVE_FATAL); } algsize = getalgsize(xar->opt_toc_sumalg); if (algsize > 0) { if (lseek(xar->temp_fd, algsize, SEEK_SET) < 0) { archive_set_error(&(a->archive), errno, "lseek failed"); return (ARCHIVE_FATAL); } xar->temp_offset = algsize; } } if (archive_entry_hardlink(file->entry) == NULL) { r = save_xattrs(a, file); if (r != ARCHIVE_OK) return (ARCHIVE_FATAL); } /* Non regular files contents are unneeded to be saved to * a temporary file. */ if (archive_entry_filetype(file->entry) != AE_IFREG) return (r2); /* * Set the current file to cur_file to read its contents. */ xar->cur_file = file; if (archive_entry_nlink(file->entry) > 1) { r = file_register_hardlink(a, file); if (r != ARCHIVE_OK) return (r); if (archive_entry_hardlink(file->entry) != NULL) { archive_entry_unset_size(file->entry); return (r2); } } /* Save a offset of current file in temporary file. */ file->data.temp_offset = xar->temp_offset; file->data.size = archive_entry_size(file->entry); file->data.compression = xar->opt_compression; xar->bytes_remaining = archive_entry_size(file->entry); checksum_init(&(xar->a_sumwrk), xar->opt_sumalg); checksum_init(&(xar->e_sumwrk), xar->opt_sumalg); r = xar_compression_init_encoder(a); if (r != ARCHIVE_OK) return (r); else return (r2); } static int write_to_temp(struct archive_write *a, const void *buff, size_t s) { struct xar *xar; const unsigned char *p; ssize_t ws; xar = (struct xar *)a->format_data; p = (const unsigned char *)buff; while (s) { ws = write(xar->temp_fd, p, s); if (ws < 0) { archive_set_error(&(a->archive), errno, - "fwrite function failed"); + "write function failed"); return (ARCHIVE_FATAL); } s -= ws; p += ws; xar->temp_offset += ws; } return (ARCHIVE_OK); } static ssize_t xar_write_data(struct archive_write *a, const void *buff, size_t s) { struct xar *xar; enum la_zaction run; size_t size = 0; size_t rsize; int r; xar = (struct xar *)a->format_data; if (s > xar->bytes_remaining) s = (size_t)xar->bytes_remaining; if (s == 0 || xar->cur_file == NULL) return (0); if (xar->cur_file->data.compression == NONE) { checksum_update(&(xar->e_sumwrk), buff, s); checksum_update(&(xar->a_sumwrk), buff, s); size = rsize = s; } else { xar->stream.next_in = (const unsigned char *)buff; xar->stream.avail_in = s; if (xar->bytes_remaining > s) run = ARCHIVE_Z_RUN; else run = ARCHIVE_Z_FINISH; /* Compress file data. */ for (;;) { r = compression_code(&(a->archive), &(xar->stream), run); if (r != ARCHIVE_OK && r != ARCHIVE_EOF) return (ARCHIVE_FATAL); if (xar->stream.avail_out == 0 || run == ARCHIVE_Z_FINISH) { size = sizeof(xar->wbuff) - xar->stream.avail_out; checksum_update(&(xar->a_sumwrk), xar->wbuff, size); xar->cur_file->data.length += size; if (write_to_temp(a, xar->wbuff, size) != ARCHIVE_OK) return (ARCHIVE_FATAL); if (r == ARCHIVE_OK) { /* Output buffer was full */ xar->stream.next_out = xar->wbuff; xar->stream.avail_out = sizeof(xar->wbuff); } else { /* ARCHIVE_EOF - We are done */ break; } } else { /* Compressor wants more input */ break; } } rsize = s - xar->stream.avail_in; checksum_update(&(xar->e_sumwrk), buff, rsize); } #if !defined(_WIN32) || defined(__CYGWIN__) if (xar->bytes_remaining == (uint64_t)archive_entry_size(xar->cur_file->entry)) { /* * Get the path of a shell script if so. */ const unsigned char *b = (const unsigned char *)buff; archive_string_empty(&(xar->cur_file->script)); if (rsize > 2 && b[0] == '#' && b[1] == '!') { size_t i, end, off; off = 2; if (b[off] == ' ') off++; #ifdef PATH_MAX if ((rsize - off) > PATH_MAX) end = off + PATH_MAX; else #endif end = rsize; /* Find the end of a script path. */ for (i = off; i < end && b[i] != '\0' && b[i] != '\n' && b[i] != '\r' && b[i] != ' ' && b[i] != '\t'; i++) ; archive_strncpy(&(xar->cur_file->script), b + off, i - off); } } #endif if (xar->cur_file->data.compression == NONE) { if (write_to_temp(a, buff, size) != ARCHIVE_OK) return (ARCHIVE_FATAL); xar->cur_file->data.length += size; } xar->bytes_remaining -= rsize; return (rsize); } static int xar_finish_entry(struct archive_write *a) { struct xar *xar; struct file *file; size_t s; ssize_t w; xar = (struct xar *)a->format_data; if (xar->cur_file == NULL) return (ARCHIVE_OK); while (xar->bytes_remaining > 0) { s = (size_t)xar->bytes_remaining; if (s > a->null_length) s = a->null_length; w = xar_write_data(a, a->nulls, s); if (w > 0) xar->bytes_remaining -= w; else return ((int)w); } file = xar->cur_file; checksum_final(&(xar->e_sumwrk), &(file->data.e_sum)); checksum_final(&(xar->a_sumwrk), &(file->data.a_sum)); xar->cur_file = NULL; return (ARCHIVE_OK); } static int xmlwrite_string_attr(struct archive_write *a, struct xml_writer *writer, const char *key, const char *value, const char *attrkey, const char *attrvalue) { int r; r = xml_writer_start_element(writer, key); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() failed: %d", r); return (ARCHIVE_FATAL); } if (attrkey != NULL && attrvalue != NULL) { r = xml_writer_write_attribute(writer, attrkey, attrvalue); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_write_attribute() failed: %d", r); return (ARCHIVE_FATAL); } } if (value != NULL) { r = xml_writer_write_string(writer, value); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_write_string() failed: %d", r); return (ARCHIVE_FATAL); } } r = xml_writer_end_element(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_end_element() failed: %d", r); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } static int xmlwrite_string(struct archive_write *a, struct xml_writer *writer, const char *key, const char *value) { int r; if (value == NULL) return (ARCHIVE_OK); r = xml_writer_start_element(writer, key); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() failed: %d", r); return (ARCHIVE_FATAL); } if (value != NULL) { r = xml_writer_write_string(writer, value); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_write_string() failed: %d", r); return (ARCHIVE_FATAL); } } r = xml_writer_end_element(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_end_element() failed: %d", r); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } static int xmlwrite_fstring(struct archive_write *a, struct xml_writer *writer, const char *key, const char *fmt, ...) { struct xar *xar; va_list ap; xar = (struct xar *)a->format_data; va_start(ap, fmt); archive_string_empty(&xar->vstr); archive_string_vsprintf(&xar->vstr, fmt, ap); va_end(ap); return (xmlwrite_string(a, writer, key, xar->vstr.s)); } static int xmlwrite_time(struct archive_write *a, struct xml_writer *writer, const char *key, time_t t, int z) { char timestr[100]; struct tm tm; #if defined(HAVE_GMTIME_S) gmtime_s(&tm, &t); #elif defined(HAVE_GMTIME_R) gmtime_r(&t, &tm); #else memcpy(&tm, gmtime(&t), sizeof(tm)); #endif memset(×tr, 0, sizeof(timestr)); /* Do not use %F and %T for portability. */ strftime(timestr, sizeof(timestr), "%Y-%m-%dT%H:%M:%S", &tm); if (z) strcat(timestr, "Z"); return (xmlwrite_string(a, writer, key, timestr)); } static int xmlwrite_mode(struct archive_write *a, struct xml_writer *writer, const char *key, mode_t mode) { char ms[5]; ms[0] = '0'; ms[1] = '0' + ((mode >> 6) & 07); ms[2] = '0' + ((mode >> 3) & 07); ms[3] = '0' + (mode & 07); ms[4] = '\0'; return (xmlwrite_string(a, writer, key, ms)); } static int xmlwrite_sum(struct archive_write *a, struct xml_writer *writer, const char *key, struct chksumval *sum) { const char *algname; int algsize; char buff[MAX_SUM_SIZE*2 + 1]; char *p; unsigned char *s; int i, r; if (sum->len > 0) { algname = getalgname(sum->alg); algsize = getalgsize(sum->alg); if (algname != NULL) { const char *hex = "0123456789abcdef"; p = buff; s = sum->val; for (i = 0; i < algsize; i++) { *p++ = hex[(*s >> 4)]; *p++ = hex[(*s & 0x0f)]; s++; } *p = '\0'; r = xmlwrite_string_attr(a, writer, key, buff, "style", algname); if (r < 0) return (ARCHIVE_FATAL); } } return (ARCHIVE_OK); } static int xmlwrite_heap(struct archive_write *a, struct xml_writer *writer, struct heap_data *heap) { const char *encname; int r; r = xmlwrite_fstring(a, writer, "length", "%ju", heap->length); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_fstring(a, writer, "offset", "%ju", heap->temp_offset); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_fstring(a, writer, "size", "%ju", heap->size); if (r < 0) return (ARCHIVE_FATAL); switch (heap->compression) { case GZIP: encname = "application/x-gzip"; break; case BZIP2: encname = "application/x-bzip2"; break; case LZMA: encname = "application/x-lzma"; break; case XZ: encname = "application/x-xz"; break; default: encname = "application/octet-stream"; break; } r = xmlwrite_string_attr(a, writer, "encoding", NULL, "style", encname); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_sum(a, writer, "archived-checksum", &(heap->a_sum)); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_sum(a, writer, "extracted-checksum", &(heap->e_sum)); if (r < 0) return (ARCHIVE_FATAL); return (ARCHIVE_OK); } /* * xar utility records fflags as following xml elements: * * * ..... * * or * * * ..... * * If xar is running on BSD platform, records ..; * if xar is running on linux platform, records ..; * otherwise does not record. * * Our implements records both and if it's necessary. */ static int make_fflags_entry(struct archive_write *a, struct xml_writer *writer, const char *element, const char *fflags_text) { static const struct flagentry { const char *name; const char *xarname; } flagbsd[] = { { "sappnd", "SystemAppend"}, { "sappend", "SystemAppend"}, { "arch", "SystemArchived"}, { "archived", "SystemArchived"}, { "schg", "SystemImmutable"}, { "schange", "SystemImmutable"}, { "simmutable", "SystemImmutable"}, { "nosunlnk", "SystemNoUnlink"}, { "nosunlink", "SystemNoUnlink"}, { "snapshot", "SystemSnapshot"}, { "uappnd", "UserAppend"}, { "uappend", "UserAppend"}, { "uchg", "UserImmutable"}, { "uchange", "UserImmutable"}, { "uimmutable", "UserImmutable"}, { "nodump", "UserNoDump"}, { "noopaque", "UserOpaque"}, { "nouunlnk", "UserNoUnlink"}, { "nouunlink", "UserNoUnlink"}, { NULL, NULL} }, flagext2[] = { { "sappnd", "AppendOnly"}, { "sappend", "AppendOnly"}, { "schg", "Immutable"}, { "schange", "Immutable"}, { "simmutable", "Immutable"}, { "nodump", "NoDump"}, { "nouunlnk", "Undelete"}, { "nouunlink", "Undelete"}, { "btree", "BTree"}, { "comperr", "CompError"}, { "compress", "Compress"}, { "noatime", "NoAtime"}, { "compdirty", "CompDirty"}, { "comprblk", "CompBlock"}, { "dirsync", "DirSync"}, { "hashidx", "HashIndexed"}, { "imagic", "iMagic"}, { "journal", "Journaled"}, { "securedeletion", "SecureDeletion"}, { "sync", "Synchronous"}, { "notail", "NoTail"}, { "topdir", "TopDir"}, { "reserved", "Reserved"}, { NULL, NULL} }; const struct flagentry *fe, *flagentry; #define FLAGENTRY_MAXSIZE ((sizeof(flagbsd)+sizeof(flagext2))/sizeof(flagbsd)) const struct flagentry *avail[FLAGENTRY_MAXSIZE]; const char *p; int i, n, r; if (strcmp(element, "ext2") == 0) flagentry = flagext2; else flagentry = flagbsd; n = 0; p = fflags_text; do { const char *cp; cp = strchr(p, ','); if (cp == NULL) cp = p + strlen(p); for (fe = flagentry; fe->name != NULL; fe++) { if (fe->name[cp - p] != '\0' || p[0] != fe->name[0]) continue; if (strncmp(p, fe->name, cp - p) == 0) { avail[n++] = fe; break; } } if (*cp == ',') p = cp + 1; else p = NULL; } while (p != NULL); if (n > 0) { r = xml_writer_start_element(writer, element); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() failed: %d", r); return (ARCHIVE_FATAL); } for (i = 0; i < n; i++) { r = xmlwrite_string(a, writer, avail[i]->xarname, NULL); if (r != ARCHIVE_OK) return (r); } r = xml_writer_end_element(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_end_element() failed: %d", r); return (ARCHIVE_FATAL); } } return (ARCHIVE_OK); } /* * This function determines whether a UTF-8 string contains * only codepoints that are convertible to Latin-1. Strings * beyond Latin-1 are stored base64-encoded in the XAR TOC. */ static int is_u8_zstring_latin1(const char *in) { unsigned int c; while (*in) { c = *in++; if (c < 0x80) continue; /* * Filter out non-continuation, any continuation of 2-3 * bytes, and any continuation of 1 byte whose high 3 bits * are non-zero. Recall, 1-byte continuations can store 11 * bits whereas Latin-1 codepoints are only 8 bits wide. */ if ((c & 0xFC) != 0xC0) return (0); c = *in++; /* * If we get any non-continuation byte (including 0x00!), * the string is not valid UTF-8. */ if ((c & 0xC0) != 0x80) return (0); /* invalid unicode */ } return (1); } static int make_file_entry(struct archive_write *a, struct xml_writer *writer, struct file *file) { struct xar *xar; const char *filetype, *filelink, *fflags; struct archive_string linkto; struct heap_data *heap; const char *p; size_t len; int r, r2; xar = (struct xar *)a->format_data; r2 = ARCHIVE_OK; /* * Make a file name entry, "". */ if (!is_u8_zstring_latin1(file->basename.s)) { r = xml_writer_start_element(writer, "name"); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() failed: %d", r); return (ARCHIVE_FATAL); } r = xml_writer_write_attribute(writer, "enctype", "base64"); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_write_attribute() failed: %d", r); return (ARCHIVE_FATAL); } r = xml_writer_write_base64(writer, file->basename.s, 0, (int)archive_strlen(&(file->basename))); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_write_base64() failed: %d", r); return (ARCHIVE_FATAL); } r = xml_writer_end_element(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_end_element() failed: %d", r); return (ARCHIVE_FATAL); } } else { r = xmlwrite_string(a, writer, "name", file->basename.s); if (r < 0) return (ARCHIVE_FATAL); } /* * Make a file type entry, "". */ filelink = NULL; archive_string_init(&linkto); switch (archive_entry_filetype(file->entry)) { case AE_IFDIR: filetype = "directory"; break; case AE_IFLNK: filetype = "symlink"; break; case AE_IFCHR: filetype = "character special"; break; case AE_IFBLK: filetype = "block special"; break; case AE_IFSOCK: filetype = "socket"; break; case AE_IFIFO: filetype = "fifo"; break; case AE_IFREG: default: if (file->hardlink_target != NULL) { filetype = "hardlink"; filelink = "link"; if (file->hardlink_target == file) archive_strcpy(&linkto, "original"); else archive_string_sprintf(&linkto, "%d", file->hardlink_target->id); } else filetype = "file"; break; } r = xmlwrite_string_attr(a, writer, "type", filetype, filelink, linkto.s); archive_string_free(&linkto); if (r < 0) return (ARCHIVE_FATAL); /* * On a virtual directory, we record "name" and "type" only. */ if (file->virtual) return (ARCHIVE_OK); switch (archive_entry_filetype(file->entry)) { case AE_IFLNK: /* * xar utility has checked a file type, which * a symbolic-link file has referenced. * For example: * ../ref/ * The symlink target file is "../ref/" and its * file type is a directory. * * ../f * The symlink target file is "../f" and its * file type is a regular file. * * But our implementation cannot do it, and then we * always record that a attribute "type" is "broken", * for example: * foo/bar * It means "foo/bar" is not reachable. */ r = xmlwrite_string_attr(a, writer, "link", file->symlink.s, "type", "broken"); if (r < 0) return (ARCHIVE_FATAL); break; case AE_IFCHR: case AE_IFBLK: r = xml_writer_start_element(writer, "device"); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() failed: %d", r); return (ARCHIVE_FATAL); } r = xmlwrite_fstring(a, writer, "major", "%d", archive_entry_rdevmajor(file->entry)); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_fstring(a, writer, "minor", "%d", archive_entry_rdevminor(file->entry)); if (r < 0) return (ARCHIVE_FATAL); r = xml_writer_end_element(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_end_element() failed: %d", r); return (ARCHIVE_FATAL); } break; default: break; } /* * Make a inode entry, "". */ r = xmlwrite_fstring(a, writer, "inode", "%jd", archive_entry_ino64(file->entry)); if (r < 0) return (ARCHIVE_FATAL); if (archive_entry_dev(file->entry) != 0) { r = xmlwrite_fstring(a, writer, "deviceno", "%d", archive_entry_dev(file->entry)); if (r < 0) return (ARCHIVE_FATAL); } /* * Make a file mode entry, "". */ r = xmlwrite_mode(a, writer, "mode", archive_entry_mode(file->entry)); if (r < 0) return (ARCHIVE_FATAL); /* * Make a user entry, "" and ". */ r = xmlwrite_fstring(a, writer, "uid", "%d", archive_entry_uid(file->entry)); if (r < 0) return (ARCHIVE_FATAL); r = archive_entry_uname_l(file->entry, &p, &len, xar->sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Uname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate uname '%s' to UTF-8", archive_entry_uname(file->entry)); r2 = ARCHIVE_WARN; } if (len > 0) { r = xmlwrite_string(a, writer, "user", p); if (r < 0) return (ARCHIVE_FATAL); } /* * Make a group entry, "" and ". */ r = xmlwrite_fstring(a, writer, "gid", "%d", archive_entry_gid(file->entry)); if (r < 0) return (ARCHIVE_FATAL); r = archive_entry_gname_l(file->entry, &p, &len, xar->sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Gname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate gname '%s' to UTF-8", archive_entry_gname(file->entry)); r2 = ARCHIVE_WARN; } if (len > 0) { r = xmlwrite_string(a, writer, "group", p); if (r < 0) return (ARCHIVE_FATAL); } /* * Make a ctime entry, "". */ if (archive_entry_ctime_is_set(file->entry)) { r = xmlwrite_time(a, writer, "ctime", archive_entry_ctime(file->entry), 1); if (r < 0) return (ARCHIVE_FATAL); } /* * Make a mtime entry, "". */ if (archive_entry_mtime_is_set(file->entry)) { r = xmlwrite_time(a, writer, "mtime", archive_entry_mtime(file->entry), 1); if (r < 0) return (ARCHIVE_FATAL); } /* * Make a atime entry, "". */ if (archive_entry_atime_is_set(file->entry)) { r = xmlwrite_time(a, writer, "atime", archive_entry_atime(file->entry), 1); if (r < 0) return (ARCHIVE_FATAL); } /* * Make fflags entries, "" and "". */ fflags = archive_entry_fflags_text(file->entry); if (fflags != NULL) { r = make_fflags_entry(a, writer, "flags", fflags); if (r < 0) return (r); r = make_fflags_entry(a, writer, "ext2", fflags); if (r < 0) return (r); } /* * Make extended attribute entries, "". */ archive_entry_xattr_reset(file->entry); for (heap = file->xattr.first; heap != NULL; heap = heap->next) { const char *name; const void *value; size_t size; archive_entry_xattr_next(file->entry, &name, &value, &size); r = xml_writer_start_element(writer, "ea"); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() failed: %d", r); return (ARCHIVE_FATAL); } r = xml_writer_write_attributef(writer, "id", "%d", heap->id); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_write_attributef() failed: %d", r); return (ARCHIVE_FATAL); } r = xmlwrite_heap(a, writer, heap); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_string(a, writer, "name", name); if (r < 0) return (ARCHIVE_FATAL); r = xml_writer_end_element(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_end_element() failed: %d", r); return (ARCHIVE_FATAL); } } /* * Make a file data entry, "". */ if (file->data.length > 0) { r = xml_writer_start_element(writer, "data"); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() failed: %d", r); return (ARCHIVE_FATAL); } r = xmlwrite_heap(a, writer, &(file->data)); if (r < 0) return (ARCHIVE_FATAL); r = xml_writer_end_element(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_end_element() failed: %d", r); return (ARCHIVE_FATAL); } } if (archive_strlen(&file->script) > 0) { r = xml_writer_start_element(writer, "content"); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() failed: %d", r); return (ARCHIVE_FATAL); } r = xmlwrite_string(a, writer, "interpreter", file->script.s); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_string(a, writer, "type", "script"); if (r < 0) return (ARCHIVE_FATAL); r = xml_writer_end_element(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_end_element() failed: %d", r); return (ARCHIVE_FATAL); } } return (r2); } /* * Make the TOC */ static int make_toc(struct archive_write *a) { struct xar *xar; struct file *np; struct xml_writer *writer; const char* content; size_t use; int algsize; int r, ret; xar = (struct xar *)a->format_data; ret = ARCHIVE_FATAL; /* * Initialize xml writer. */ writer = NULL; r = xml_writer_create(&writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_create() failed: %d", r); goto exit_toc; } r = xml_writer_set_indent(writer, 4); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_set_indent() failed: %d", r); goto exit_toc; } r = xml_writer_start_document(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_document() failed: %d", r); goto exit_toc; } /* * Start recording TOC */ r = xml_writer_start_element(writer, "xar"); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() failed: %d", r); goto exit_toc; } r = xml_writer_start_element(writer, "toc"); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() failed: %d", r); goto exit_toc; } /* * Record the creation time of the archive file. */ r = xmlwrite_time(a, writer, "creation-time", time(NULL), 0); if (r < 0) goto exit_toc; /* * Record the checksum value of TOC */ algsize = getalgsize(xar->opt_toc_sumalg); if (algsize) { /* * Record TOC checksum */ r = xml_writer_start_element(writer, "checksum"); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() failed: %d", r); goto exit_toc; } r = xml_writer_write_attribute(writer, "style", getalgname(xar->opt_toc_sumalg)); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_write_attribute() failed: %d", r); goto exit_toc; } /* * Record the offset of the value of checksum of TOC */ r = xmlwrite_string(a, writer, "offset", "0"); if (r < 0) goto exit_toc; /* * Record the size of the value of checksum of TOC */ r = xmlwrite_fstring(a, writer, "size", "%d", algsize); if (r < 0) goto exit_toc; r = xml_writer_end_element(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_end_element() failed: %d", r); goto exit_toc; } } np = xar->root; do { if (np != np->parent) { r = make_file_entry(a, writer, np); if (r != ARCHIVE_OK) goto exit_toc; } if (np->dir && np->children.first != NULL) { /* Enter to sub directories. */ np = np->children.first; r = xml_writer_start_element(writer, "file"); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() " "failed: %d", r); goto exit_toc; } r = xml_writer_write_attributef( writer, "id", "%d", np->id); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_write_attributef() " "failed: %d", r); goto exit_toc; } continue; } while (np != np->parent) { r = xml_writer_end_element(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_end_element() " "failed: %d", r); goto exit_toc; } if (np->chnext == NULL) { /* Return to the parent directory. */ np = np->parent; } else { np = np->chnext; r = xml_writer_start_element(writer, "file"); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_start_element() " "failed: %d", r); goto exit_toc; } r = xml_writer_write_attributef( writer, "id", "%d", np->id); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_write_attributef() " "failed: %d", r); goto exit_toc; } break; } } } while (np != np->parent); r = xml_writer_end_document(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_end_document() failed: %d", r); goto exit_toc; } r = xml_writer_get_final_content_and_length(writer, &content, &use); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xml_writer_get_final_content_and_length() failed: %d", r); goto exit_toc; } #if DEBUG_PRINT_TOC fprintf(stderr, "\n---TOC-- %d bytes --\n%s\n", (int)strlen(content), content); #endif /* * Compress the TOC and calculate the sum of the TOC. */ xar->toc.temp_offset = xar->temp_offset; xar->toc.size = (uint64_t)use; checksum_init(&(xar->a_sumwrk), xar->opt_toc_sumalg); r = compression_init_encoder_gzip(&(a->archive), &(xar->stream), 6, 1); if (r != ARCHIVE_OK) goto exit_toc; xar->stream.next_in = (const unsigned char *)content; xar->stream.avail_in = use; xar->stream.total_in = 0; xar->stream.next_out = xar->wbuff; xar->stream.avail_out = sizeof(xar->wbuff); xar->stream.total_out = 0; for (;;) { size_t size; r = compression_code(&(a->archive), &(xar->stream), ARCHIVE_Z_FINISH); if (r != ARCHIVE_OK && r != ARCHIVE_EOF) goto exit_toc; size = sizeof(xar->wbuff) - xar->stream.avail_out; checksum_update(&(xar->a_sumwrk), xar->wbuff, size); if (write_to_temp(a, xar->wbuff, size) != ARCHIVE_OK) goto exit_toc; if (r == ARCHIVE_EOF) break; xar->stream.next_out = xar->wbuff; xar->stream.avail_out = sizeof(xar->wbuff); } r = compression_end(&(a->archive), &(xar->stream)); if (r != ARCHIVE_OK) goto exit_toc; xar->toc.length = xar->stream.total_out; xar->toc.compression = GZIP; checksum_final(&(xar->a_sumwrk), &(xar->toc.a_sum)); ret = ARCHIVE_OK; exit_toc: if (writer) xml_writer_destroy(writer); return (ret); } static int flush_wbuff(struct archive_write *a) { struct xar *xar; int r; size_t s; xar = (struct xar *)a->format_data; s = sizeof(xar->wbuff) - xar->wbuff_remaining; r = __archive_write_output(a, xar->wbuff, s); if (r != ARCHIVE_OK) return (r); xar->wbuff_remaining = sizeof(xar->wbuff); return (r); } static int copy_out(struct archive_write *a, uint64_t offset, uint64_t length) { struct xar *xar; int r; xar = (struct xar *)a->format_data; if (lseek(xar->temp_fd, offset, SEEK_SET) < 0) { archive_set_error(&(a->archive), errno, "lseek failed"); return (ARCHIVE_FATAL); } while (length) { size_t rsize; ssize_t rs; unsigned char *wb; if (length > xar->wbuff_remaining) rsize = xar->wbuff_remaining; else rsize = (size_t)length; wb = xar->wbuff + (sizeof(xar->wbuff) - xar->wbuff_remaining); rs = read(xar->temp_fd, wb, rsize); if (rs < 0) { archive_set_error(&(a->archive), errno, "Can't read temporary file(%jd)", (intmax_t)rs); return (ARCHIVE_FATAL); } if (rs == 0) { archive_set_error(&(a->archive), 0, "Truncated xar archive"); return (ARCHIVE_FATAL); } xar->wbuff_remaining -= rs; length -= rs; if (xar->wbuff_remaining == 0) { r = flush_wbuff(a); if (r != ARCHIVE_OK) return (r); } } return (ARCHIVE_OK); } static int xar_close(struct archive_write *a) { struct xar *xar; unsigned char *wb; uint64_t length; int r; xar = (struct xar *)a->format_data; /* Empty! */ if (xar->root->children.first == NULL) return (ARCHIVE_OK); /* Save the length of all file extended attributes and contents. */ length = xar->temp_offset; /* Connect hardlinked files */ file_connect_hardlink_files(xar); /* Make the TOC */ r = make_toc(a); if (r != ARCHIVE_OK) return (r); /* * Make the xar header on wbuff(write buffer). */ wb = xar->wbuff; xar->wbuff_remaining = sizeof(xar->wbuff); archive_be32enc(&wb[0], HEADER_MAGIC); archive_be16enc(&wb[4], HEADER_SIZE); archive_be16enc(&wb[6], HEADER_VERSION); archive_be64enc(&wb[8], xar->toc.length); archive_be64enc(&wb[16], xar->toc.size); archive_be32enc(&wb[24], xar->toc.a_sum.alg); xar->wbuff_remaining -= HEADER_SIZE; /* * Write the TOC */ r = copy_out(a, xar->toc.temp_offset, xar->toc.length); if (r != ARCHIVE_OK) return (r); /* Write the checksum value of the TOC. */ if (xar->toc.a_sum.len) { if (xar->wbuff_remaining < xar->toc.a_sum.len) { r = flush_wbuff(a); if (r != ARCHIVE_OK) return (r); } wb = xar->wbuff + (sizeof(xar->wbuff) - xar->wbuff_remaining); memcpy(wb, xar->toc.a_sum.val, xar->toc.a_sum.len); xar->wbuff_remaining -= xar->toc.a_sum.len; } /* * Write all file extended attributes and contents. */ r = copy_out(a, xar->toc.a_sum.len, length); if (r != ARCHIVE_OK) return (r); r = flush_wbuff(a); return (r); } static int xar_free(struct archive_write *a) { struct xar *xar; xar = (struct xar *)a->format_data; /* Close the temporary file. */ if (xar->temp_fd >= 0) close(xar->temp_fd); archive_string_free(&(xar->cur_dirstr)); archive_string_free(&(xar->tstr)); archive_string_free(&(xar->vstr)); file_free_hardlinks(xar); file_free_register(xar); compression_end(&(a->archive), &(xar->stream)); free(xar); return (ARCHIVE_OK); } static int file_cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct file *f1 = (const struct file *)n1; const struct file *f2 = (const struct file *)n2; return (strcmp(f1->basename.s, f2->basename.s)); } static int file_cmp_key(const struct archive_rb_node *n, const void *key) { const struct file *f = (const struct file *)n; return (strcmp(f->basename.s, (const char *)key)); } static struct file * file_new(struct archive_write *a, struct archive_entry *entry) { struct file *file; static const struct archive_rb_tree_ops rb_ops = { file_cmp_node, file_cmp_key }; file = calloc(1, sizeof(*file)); if (file == NULL) return (NULL); if (entry != NULL) file->entry = archive_entry_clone(entry); else file->entry = archive_entry_new2(&a->archive); if (file->entry == NULL) { free(file); return (NULL); } __archive_rb_tree_init(&(file->rbtree), &rb_ops); file->children.first = NULL; file->children.last = &(file->children.first); file->xattr.first = NULL; file->xattr.last = &(file->xattr.first); archive_string_init(&(file->parentdir)); archive_string_init(&(file->basename)); archive_string_init(&(file->symlink)); archive_string_init(&(file->script)); if (entry != NULL && archive_entry_filetype(entry) == AE_IFDIR) file->dir = 1; return (file); } static void file_free(struct file *file) { struct heap_data *heap, *next_heap; heap = file->xattr.first; while (heap != NULL) { next_heap = heap->next; free(heap); heap = next_heap; } archive_string_free(&(file->parentdir)); archive_string_free(&(file->basename)); archive_string_free(&(file->symlink)); archive_string_free(&(file->script)); archive_entry_free(file->entry); free(file); } static struct file * file_create_virtual_dir(struct archive_write *a, struct xar *xar, const char *pathname) { struct file *file; (void)xar; /* UNUSED */ file = file_new(a, NULL); if (file == NULL) return (NULL); archive_entry_set_pathname(file->entry, pathname); archive_entry_set_mode(file->entry, 0555 | AE_IFDIR); file->dir = 1; file->virtual = 1; return (file); } static int file_add_child_tail(struct file *parent, struct file *child) { if (!__archive_rb_tree_insert_node( &(parent->rbtree), (struct archive_rb_node *)child)) return (0); child->chnext = NULL; *parent->children.last = child; parent->children.last = &(child->chnext); child->parent = parent; return (1); } /* * Find a entry from `parent' */ static struct file * file_find_child(struct file *parent, const char *child_name) { struct file *np; np = (struct file *)__archive_rb_tree_find_node( &(parent->rbtree), child_name); return (np); } #if defined(_WIN32) || defined(__CYGWIN__) static void cleanup_backslash(char *utf8, size_t len) { /* Convert a path-separator from '\' to '/' */ while (*utf8 != '\0' && len) { if (*utf8 == '\\') *utf8 = '/'; ++utf8; --len; } } #else #define cleanup_backslash(p, len) /* nop */ #endif /* * Generate a parent directory name and a base name from a pathname. */ static int file_gen_utility_names(struct archive_write *a, struct file *file) { struct xar *xar; const char *pp; char *p, *dirname, *slash; size_t len; int r = ARCHIVE_OK; xar = (struct xar *)a->format_data; archive_string_empty(&(file->parentdir)); archive_string_empty(&(file->basename)); archive_string_empty(&(file->symlink)); if (file->parent == file)/* virtual root */ return (ARCHIVE_OK); if (archive_entry_pathname_l(file->entry, &pp, &len, xar->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate pathname '%s' to UTF-8", archive_entry_pathname(file->entry)); r = ARCHIVE_WARN; } archive_strncpy(&(file->parentdir), pp, len); len = file->parentdir.length; p = dirname = file->parentdir.s; /* * Convert a path-separator from '\' to '/' */ cleanup_backslash(p, len); /* * Remove leading '/', '../' and './' elements */ while (*p) { if (p[0] == '/') { p++; len--; } else if (p[0] != '.') break; else if (p[1] == '.' && p[2] == '/') { p += 3; len -= 3; } else if (p[1] == '/' || (p[1] == '.' && p[2] == '\0')) { p += 2; len -= 2; } else if (p[1] == '\0') { p++; len--; } else break; } if (p != dirname) { memmove(dirname, p, len+1); p = dirname; } /* * Remove "/","/." and "/.." elements from tail. */ while (len > 0) { size_t ll = len; if (p[len-1] == '/') { p[len-1] = '\0'; len--; } if (len > 1 && p[len-2] == '/' && p[len-1] == '.') { p[len-2] = '\0'; len -= 2; } if (len > 2 && p[len-3] == '/' && p[len-2] == '.' && p[len-1] == '.') { p[len-3] = '\0'; len -= 3; } if (ll == len) break; } while (*p) { if (p[0] == '/') { if (p[1] == '/') /* Convert '//' --> '/' */ memmove(p, p+1, strlen(p+1) + 1); else if (p[1] == '.' && p[2] == '/') /* Convert '/./' --> '/' */ memmove(p, p+2, strlen(p+2) + 1); else if (p[1] == '.' && p[2] == '.' && p[3] == '/') { /* Convert 'dir/dir1/../dir2/' * --> 'dir/dir2/' */ char *rp = p -1; while (rp >= dirname) { if (*rp == '/') break; --rp; } if (rp > dirname) { strcpy(rp, p+3); p = rp; } else { strcpy(dirname, p+4); p = dirname; } } else p++; } else p++; } p = dirname; len = strlen(p); if (archive_entry_filetype(file->entry) == AE_IFLNK) { size_t len2; /* Convert symlink name too. */ if (archive_entry_symlink_l(file->entry, &pp, &len2, xar->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate symlink '%s' to UTF-8", archive_entry_symlink(file->entry)); r = ARCHIVE_WARN; } archive_strncpy(&(file->symlink), pp, len2); cleanup_backslash(file->symlink.s, file->symlink.length); } /* * - Count up directory elements. * - Find out the position which points the last position of * path separator('/'). */ slash = NULL; for (; *p != '\0'; p++) if (*p == '/') slash = p; if (slash == NULL) { /* The pathname doesn't have a parent directory. */ file->parentdir.length = len; archive_string_copy(&(file->basename), &(file->parentdir)); archive_string_empty(&(file->parentdir)); *file->parentdir.s = '\0'; return (r); } /* Make a basename from dirname and slash */ *slash = '\0'; file->parentdir.length = slash - dirname; archive_strcpy(&(file->basename), slash + 1); return (r); } static int get_path_component(char *name, int n, const char *fn) { char *p; int l; p = strchr(fn, '/'); if (p == NULL) { if ((l = (int)strlen(fn)) == 0) return (0); } else l = (int)(p - fn); if (l > n -1) return (-1); memcpy(name, fn, l); name[l] = '\0'; return (l); } /* * Add a new entry into the tree. */ static int file_tree(struct archive_write *a, struct file **filepp) { #if defined(_WIN32) && !defined(__CYGWIN__) char name[_MAX_FNAME];/* Included null terminator size. */ #elif defined(NAME_MAX) && NAME_MAX >= 255 char name[NAME_MAX+1]; #else char name[256]; #endif struct xar *xar = (struct xar *)a->format_data; struct file *dent, *file, *np; struct archive_entry *ent; const char *fn, *p; int l; file = *filepp; dent = xar->root; if (file->parentdir.length > 0) fn = p = file->parentdir.s; else fn = p = ""; /* * If the path of the parent directory of `file' entry is * the same as the path of `cur_dirent', add isoent to * `cur_dirent'. */ if (archive_strlen(&(xar->cur_dirstr)) == archive_strlen(&(file->parentdir)) && strcmp(xar->cur_dirstr.s, fn) == 0) { if (!file_add_child_tail(xar->cur_dirent, file)) { np = (struct file *)__archive_rb_tree_find_node( &(xar->cur_dirent->rbtree), file->basename.s); goto same_entry; } return (ARCHIVE_OK); } for (;;) { l = get_path_component(name, sizeof(name), fn); if (l == 0) { np = NULL; break; } if (l < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "A name buffer is too small"); file_free(file); *filepp = NULL; return (ARCHIVE_FATAL); } np = file_find_child(dent, name); if (np == NULL || fn[0] == '\0') break; /* Find next subdirectory. */ if (!np->dir) { /* NOT Directory! */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "`%s' is not directory, we cannot insert `%s' ", archive_entry_pathname(np->entry), archive_entry_pathname(file->entry)); file_free(file); *filepp = NULL; return (ARCHIVE_FAILED); } fn += l; if (fn[0] == '/') fn++; dent = np; } if (np == NULL) { /* * Create virtual parent directories. */ while (fn[0] != '\0') { struct file *vp; struct archive_string as; archive_string_init(&as); archive_strncat(&as, p, fn - p + l); if (as.s[as.length-1] == '/') { as.s[as.length-1] = '\0'; as.length--; } vp = file_create_virtual_dir(a, xar, as.s); if (vp == NULL) { archive_string_free(&as); archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); file_free(file); *filepp = NULL; return (ARCHIVE_FATAL); } archive_string_free(&as); if (file_gen_utility_names(a, vp) <= ARCHIVE_FAILED) return (ARCHIVE_FATAL); file_add_child_tail(dent, vp); file_register(xar, vp); np = vp; fn += l; if (fn[0] == '/') fn++; l = get_path_component(name, sizeof(name), fn); if (l < 0) { archive_string_free(&as); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "A name buffer is too small"); file_free(file); *filepp = NULL; return (ARCHIVE_FATAL); } dent = np; } /* Found out the parent directory where isoent can be * inserted. */ xar->cur_dirent = dent; archive_string_empty(&(xar->cur_dirstr)); archive_string_ensure(&(xar->cur_dirstr), archive_strlen(&(dent->parentdir)) + archive_strlen(&(dent->basename)) + 2); if (archive_strlen(&(dent->parentdir)) + archive_strlen(&(dent->basename)) == 0) xar->cur_dirstr.s[0] = 0; else { if (archive_strlen(&(dent->parentdir)) > 0) { archive_string_copy(&(xar->cur_dirstr), &(dent->parentdir)); archive_strappend_char(&(xar->cur_dirstr), '/'); } archive_string_concat(&(xar->cur_dirstr), &(dent->basename)); } if (!file_add_child_tail(dent, file)) { np = (struct file *)__archive_rb_tree_find_node( &(dent->rbtree), file->basename.s); goto same_entry; } return (ARCHIVE_OK); } same_entry: /* * We have already has the entry the filename of which is * the same. */ if (archive_entry_filetype(np->entry) != archive_entry_filetype(file->entry)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Found duplicate entries `%s' and its file type is " "different", archive_entry_pathname(np->entry)); file_free(file); *filepp = NULL; return (ARCHIVE_FAILED); } /* Swap files. */ ent = np->entry; np->entry = file->entry; file->entry = ent; np->virtual = 0; file_free(file); *filepp = np; return (ARCHIVE_OK); } static void file_register(struct xar *xar, struct file *file) { file->id = xar->file_idx++; file->next = NULL; *xar->file_list.last = file; xar->file_list.last = &(file->next); } static void file_init_register(struct xar *xar) { xar->file_list.first = NULL; xar->file_list.last = &(xar->file_list.first); } static void file_free_register(struct xar *xar) { struct file *file, *file_next; file = xar->file_list.first; while (file != NULL) { file_next = file->next; file_free(file); file = file_next; } } /* * Register entry to get a hardlink target. */ static int file_register_hardlink(struct archive_write *a, struct file *file) { struct xar *xar = (struct xar *)a->format_data; struct hardlink *hl; const char *pathname; archive_entry_set_nlink(file->entry, 1); pathname = archive_entry_hardlink(file->entry); if (pathname == NULL) { /* This `file` is a hardlink target. */ hl = malloc(sizeof(*hl)); if (hl == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } hl->nlink = 1; /* A hardlink target must be the first position. */ file->hlnext = NULL; hl->file_list.first = file; hl->file_list.last = &(file->hlnext); __archive_rb_tree_insert_node(&(xar->hardlink_rbtree), (struct archive_rb_node *)hl); } else { hl = (struct hardlink *)__archive_rb_tree_find_node( &(xar->hardlink_rbtree), pathname); if (hl != NULL) { /* Insert `file` entry into the tail. */ file->hlnext = NULL; *hl->file_list.last = file; hl->file_list.last = &(file->hlnext); hl->nlink++; } archive_entry_unset_size(file->entry); } return (ARCHIVE_OK); } /* * Hardlinked files have to have the same location of extent. * We have to find out hardlink target entries for entries which * have a hardlink target name. */ static void file_connect_hardlink_files(struct xar *xar) { struct archive_rb_node *n; struct hardlink *hl; struct file *target, *nf; ARCHIVE_RB_TREE_FOREACH(n, &(xar->hardlink_rbtree)) { hl = (struct hardlink *)n; /* The first entry must be a hardlink target. */ target = hl->file_list.first; archive_entry_set_nlink(target->entry, hl->nlink); if (hl->nlink > 1) /* It means this file is a hardlink * target itself. */ target->hardlink_target = target; for (nf = target->hlnext; nf != NULL; nf = nf->hlnext) { nf->hardlink_target = target; archive_entry_set_nlink(nf->entry, hl->nlink); } } } static int file_hd_cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct hardlink *h1 = (const struct hardlink *)n1; const struct hardlink *h2 = (const struct hardlink *)n2; return (strcmp(archive_entry_pathname(h1->file_list.first->entry), archive_entry_pathname(h2->file_list.first->entry))); } static int file_hd_cmp_key(const struct archive_rb_node *n, const void *key) { const struct hardlink *h = (const struct hardlink *)n; return (strcmp(archive_entry_pathname(h->file_list.first->entry), (const char *)key)); } static void file_init_hardlinks(struct xar *xar) { static const struct archive_rb_tree_ops rb_ops = { file_hd_cmp_node, file_hd_cmp_key, }; __archive_rb_tree_init(&(xar->hardlink_rbtree), &rb_ops); } static void file_free_hardlinks(struct xar *xar) { struct archive_rb_node *n, *tmp; ARCHIVE_RB_TREE_FOREACH_SAFE(n, &(xar->hardlink_rbtree), tmp) { __archive_rb_tree_remove_node(&(xar->hardlink_rbtree), n); free(n); } } static void checksum_init(struct chksumwork *sumwrk, enum sumalg sum_alg) { sumwrk->alg = sum_alg; switch (sum_alg) { case CKSUM_NONE: break; case CKSUM_SHA1: archive_sha1_init(&(sumwrk->sha1ctx)); break; case CKSUM_MD5: archive_md5_init(&(sumwrk->md5ctx)); break; } } static void checksum_update(struct chksumwork *sumwrk, const void *buff, size_t size) { switch (sumwrk->alg) { case CKSUM_NONE: break; case CKSUM_SHA1: archive_sha1_update(&(sumwrk->sha1ctx), buff, size); break; case CKSUM_MD5: archive_md5_update(&(sumwrk->md5ctx), buff, size); break; } } static void checksum_final(struct chksumwork *sumwrk, struct chksumval *sumval) { switch (sumwrk->alg) { case CKSUM_NONE: sumval->len = 0; break; case CKSUM_SHA1: archive_sha1_final(&(sumwrk->sha1ctx), sumval->val); sumval->len = SHA1_SIZE; break; case CKSUM_MD5: archive_md5_final(&(sumwrk->md5ctx), sumval->val); sumval->len = MD5_SIZE; break; } sumval->alg = sumwrk->alg; } #if !defined(HAVE_BZLIB_H) || !defined(BZ_CONFIG_ERROR) || !defined(HAVE_LZMA_H) static int compression_unsupported_encoder(struct archive *a, struct la_zstream *lastrm, const char *name) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "%s compression not supported on this platform", name); lastrm->valid = 0; lastrm->real_stream = NULL; return (ARCHIVE_FAILED); } #endif static int compression_init_encoder_gzip(struct archive *a, struct la_zstream *lastrm, int level, int withheader) { z_stream *strm; if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm)); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for gzip stream"); return (ARCHIVE_FATAL); } /* zlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (Bytef *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = (uInt)lastrm->avail_in; strm->total_in = (uLong)lastrm->total_in; strm->next_out = lastrm->next_out; strm->avail_out = (uInt)lastrm->avail_out; strm->total_out = (uLong)lastrm->total_out; if (deflateInit2(strm, level, Z_DEFLATED, (withheader)?15:-15, 8, Z_DEFAULT_STRATEGY) != Z_OK) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_gzip; lastrm->end = compression_end_gzip; return (ARCHIVE_OK); } static int compression_code_gzip(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { z_stream *strm; int r; strm = (z_stream *)lastrm->real_stream; /* zlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (Bytef *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = (uInt)lastrm->avail_in; strm->total_in = (uLong)lastrm->total_in; strm->next_out = lastrm->next_out; strm->avail_out = (uInt)lastrm->avail_out; strm->total_out = (uLong)lastrm->total_out; r = deflate(strm, (action == ARCHIVE_Z_FINISH)? Z_FINISH: Z_NO_FLUSH); lastrm->next_in = strm->next_in; lastrm->avail_in = strm->avail_in; lastrm->total_in = strm->total_in; lastrm->next_out = strm->next_out; lastrm->avail_out = strm->avail_out; lastrm->total_out = strm->total_out; switch (r) { case Z_OK: return (ARCHIVE_OK); case Z_STREAM_END: return (ARCHIVE_EOF); default: archive_set_error(a, ARCHIVE_ERRNO_MISC, "GZip compression failed:" " deflate() call returned status %d", r); return (ARCHIVE_FATAL); } } static int compression_end_gzip(struct archive *a, struct la_zstream *lastrm) { z_stream *strm; int r; strm = (z_stream *)lastrm->real_stream; r = deflateEnd(strm); free(strm); lastrm->real_stream = NULL; lastrm->valid = 0; if (r != Z_OK) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "Failed to clean up compressor"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) static int compression_init_encoder_bzip2(struct archive *a, struct la_zstream *lastrm, int level) { bz_stream *strm; if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm)); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for bzip2 stream"); return (ARCHIVE_FATAL); } /* bzlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (char *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = (unsigned int)lastrm->avail_in; strm->total_in_lo32 = (uint32_t)(lastrm->total_in & 0xffffffff); strm->total_in_hi32 = (uint32_t)(lastrm->total_in >> 32); strm->next_out = (char *)lastrm->next_out; strm->avail_out = (unsigned int)lastrm->avail_out; strm->total_out_lo32 = (uint32_t)(lastrm->total_out & 0xffffffff); strm->total_out_hi32 = (uint32_t)(lastrm->total_out >> 32); if (BZ2_bzCompressInit(strm, level, 0, 30) != BZ_OK) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_bzip2; lastrm->end = compression_end_bzip2; return (ARCHIVE_OK); } static int compression_code_bzip2(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { bz_stream *strm; int r; strm = (bz_stream *)lastrm->real_stream; /* bzlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (char *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = (unsigned int)lastrm->avail_in; strm->total_in_lo32 = (uint32_t)(lastrm->total_in & 0xffffffff); strm->total_in_hi32 = (uint32_t)(lastrm->total_in >> 32); strm->next_out = (char *)lastrm->next_out; strm->avail_out = (unsigned int)lastrm->avail_out; strm->total_out_lo32 = (uint32_t)(lastrm->total_out & 0xffffffff); strm->total_out_hi32 = (uint32_t)(lastrm->total_out >> 32); r = BZ2_bzCompress(strm, (action == ARCHIVE_Z_FINISH)? BZ_FINISH: BZ_RUN); lastrm->next_in = (const unsigned char *)strm->next_in; lastrm->avail_in = strm->avail_in; lastrm->total_in = (((uint64_t)(uint32_t)strm->total_in_hi32) << 32) + (uint64_t)(uint32_t)strm->total_in_lo32; lastrm->next_out = (unsigned char *)strm->next_out; lastrm->avail_out = strm->avail_out; lastrm->total_out = (((uint64_t)(uint32_t)strm->total_out_hi32) << 32) + (uint64_t)(uint32_t)strm->total_out_lo32; switch (r) { case BZ_RUN_OK: /* Non-finishing */ case BZ_FINISH_OK: /* Finishing: There's more work to do */ return (ARCHIVE_OK); case BZ_STREAM_END: /* Finishing: all done */ /* Only occurs in finishing case */ return (ARCHIVE_EOF); default: /* Any other return value indicates an error */ archive_set_error(a, ARCHIVE_ERRNO_MISC, "Bzip2 compression failed:" " BZ2_bzCompress() call returned status %d", r); return (ARCHIVE_FATAL); } } static int compression_end_bzip2(struct archive *a, struct la_zstream *lastrm) { bz_stream *strm; int r; strm = (bz_stream *)lastrm->real_stream; r = BZ2_bzCompressEnd(strm); free(strm); lastrm->real_stream = NULL; lastrm->valid = 0; if (r != BZ_OK) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "Failed to clean up compressor"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } #else static int compression_init_encoder_bzip2(struct archive *a, struct la_zstream *lastrm, int level) { (void) level; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "bzip2")); } #endif #if defined(HAVE_LZMA_H) static int compression_init_encoder_lzma(struct archive *a, struct la_zstream *lastrm, int level) { static const lzma_stream lzma_init_data = LZMA_STREAM_INIT; lzma_stream *strm; lzma_options_lzma lzma_opt; int r; if (lastrm->valid) compression_end(a, lastrm); if (lzma_lzma_preset(&lzma_opt, level)) { lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } strm = calloc(1, sizeof(*strm)); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for lzma stream"); return (ARCHIVE_FATAL); } *strm = lzma_init_data; r = lzma_alone_encoder(strm, &lzma_opt); switch (r) { case LZMA_OK: lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_lzma; lastrm->end = compression_end_lzma; r = ARCHIVE_OK; break; case LZMA_MEM_ERROR: free(strm); lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Internal error initializing compression library: " "Cannot allocate memory"); r = ARCHIVE_FATAL; break; default: free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library: " "It's a bug in liblzma"); r = ARCHIVE_FATAL; break; } return (r); } static int compression_init_encoder_xz(struct archive *a, struct la_zstream *lastrm, int level, int threads) { static const lzma_stream lzma_init_data = LZMA_STREAM_INIT; lzma_stream *strm; lzma_filter *lzmafilters; lzma_options_lzma lzma_opt; int r; #ifdef HAVE_LZMA_STREAM_ENCODER_MT lzma_mt mt_options; #endif (void)threads; /* UNUSED (if multi-threaded LZMA library not avail) */ if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm) + sizeof(*lzmafilters) * 2); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for xz stream"); return (ARCHIVE_FATAL); } lzmafilters = (lzma_filter *)(strm+1); if (level > 9) level = 9; if (lzma_lzma_preset(&lzma_opt, level)) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } lzmafilters[0].id = LZMA_FILTER_LZMA2; lzmafilters[0].options = &lzma_opt; lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */ *strm = lzma_init_data; #ifdef HAVE_LZMA_STREAM_ENCODER_MT if (threads > 1) { memset(&mt_options, 0, sizeof(mt_options)); mt_options.threads = threads; mt_options.timeout = 300; mt_options.filters = lzmafilters; mt_options.check = LZMA_CHECK_CRC64; r = lzma_stream_encoder_mt(strm, &mt_options); } else #endif r = lzma_stream_encoder(strm, lzmafilters, LZMA_CHECK_CRC64); switch (r) { case LZMA_OK: lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_lzma; lastrm->end = compression_end_lzma; r = ARCHIVE_OK; break; case LZMA_MEM_ERROR: free(strm); lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Internal error initializing compression library: " "Cannot allocate memory"); r = ARCHIVE_FATAL; break; default: free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library: " "It's a bug in liblzma"); r = ARCHIVE_FATAL; break; } return (r); } static int compression_code_lzma(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { lzma_stream *strm; int r; strm = (lzma_stream *)lastrm->real_stream; strm->next_in = lastrm->next_in; strm->avail_in = lastrm->avail_in; strm->total_in = lastrm->total_in; strm->next_out = lastrm->next_out; strm->avail_out = lastrm->avail_out; strm->total_out = lastrm->total_out; r = lzma_code(strm, (action == ARCHIVE_Z_FINISH)? LZMA_FINISH: LZMA_RUN); lastrm->next_in = strm->next_in; lastrm->avail_in = strm->avail_in; lastrm->total_in = strm->total_in; lastrm->next_out = strm->next_out; lastrm->avail_out = strm->avail_out; lastrm->total_out = strm->total_out; switch (r) { case LZMA_OK: /* Non-finishing case */ return (ARCHIVE_OK); case LZMA_STREAM_END: /* This return can only occur in finishing case. */ return (ARCHIVE_EOF); case LZMA_MEMLIMIT_ERROR: archive_set_error(a, ENOMEM, "lzma compression error:" " %ju MiB would have been needed", (uintmax_t)((lzma_memusage(strm) + 1024 * 1024 -1) / (1024 * 1024))); return (ARCHIVE_FATAL); default: /* Any other return value indicates an error */ archive_set_error(a, ARCHIVE_ERRNO_MISC, "lzma compression failed:" " lzma_code() call returned status %d", r); return (ARCHIVE_FATAL); } } static int compression_end_lzma(struct archive *a, struct la_zstream *lastrm) { lzma_stream *strm; (void)a; /* UNUSED */ strm = (lzma_stream *)lastrm->real_stream; lzma_end(strm); free(strm); lastrm->valid = 0; lastrm->real_stream = NULL; return (ARCHIVE_OK); } #else static int compression_init_encoder_lzma(struct archive *a, struct la_zstream *lastrm, int level) { (void) level; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "lzma")); } static int compression_init_encoder_xz(struct archive *a, struct la_zstream *lastrm, int level, int threads) { (void) level; /* UNUSED */ (void) threads; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "xz")); } #endif static int xar_compression_init_encoder(struct archive_write *a) { struct xar *xar; int r; xar = (struct xar *)a->format_data; switch (xar->opt_compression) { case GZIP: r = compression_init_encoder_gzip( &(a->archive), &(xar->stream), xar->opt_compression_level, 1); break; case BZIP2: r = compression_init_encoder_bzip2( &(a->archive), &(xar->stream), xar->opt_compression_level); break; case LZMA: r = compression_init_encoder_lzma( &(a->archive), &(xar->stream), xar->opt_compression_level); break; case XZ: r = compression_init_encoder_xz( &(a->archive), &(xar->stream), xar->opt_compression_level, xar->opt_threads); break; default: r = ARCHIVE_OK; break; } if (r == ARCHIVE_OK) { xar->stream.total_in = 0; xar->stream.next_out = xar->wbuff; xar->stream.avail_out = sizeof(xar->wbuff); xar->stream.total_out = 0; } return (r); } static int compression_code(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { if (lastrm->valid) return (lastrm->code(a, lastrm, action)); return (ARCHIVE_OK); } static int compression_end(struct archive *a, struct la_zstream *lastrm) { if (lastrm->valid) return (lastrm->end(a, lastrm)); return (ARCHIVE_OK); } static int save_xattrs(struct archive_write *a, struct file *file) { struct xar *xar; const char *name; const void *value; struct heap_data *heap; size_t size; int count, r; xar = (struct xar *)a->format_data; count = archive_entry_xattr_reset(file->entry); if (count == 0) return (ARCHIVE_OK); while (count--) { archive_entry_xattr_next(file->entry, &name, &value, &size); checksum_init(&(xar->a_sumwrk), xar->opt_sumalg); checksum_init(&(xar->e_sumwrk), xar->opt_sumalg); heap = calloc(1, sizeof(*heap)); if (heap == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for xattr"); return (ARCHIVE_FATAL); } heap->id = file->ea_idx++; heap->temp_offset = xar->temp_offset; heap->size = size;/* save a extracted size */ heap->compression = xar->opt_compression; /* Get a extracted sumcheck value. */ checksum_update(&(xar->e_sumwrk), value, size); checksum_final(&(xar->e_sumwrk), &(heap->e_sum)); /* * Not compression to xattr is simple way. */ if (heap->compression == NONE) { checksum_update(&(xar->a_sumwrk), value, size); checksum_final(&(xar->a_sumwrk), &(heap->a_sum)); if (write_to_temp(a, value, size) != ARCHIVE_OK) { free(heap); return (ARCHIVE_FATAL); } heap->length = size; /* Add heap to the tail of file->xattr. */ heap->next = NULL; *file->xattr.last = heap; file->xattr.last = &(heap->next); /* Next xattr */ continue; } /* * Init compression library. */ r = xar_compression_init_encoder(a); if (r != ARCHIVE_OK) { free(heap); return (ARCHIVE_FATAL); } xar->stream.next_in = (const unsigned char *)value; xar->stream.avail_in = size; for (;;) { r = compression_code(&(a->archive), &(xar->stream), ARCHIVE_Z_FINISH); if (r != ARCHIVE_OK && r != ARCHIVE_EOF) { free(heap); return (ARCHIVE_FATAL); } size = sizeof(xar->wbuff) - xar->stream.avail_out; checksum_update(&(xar->a_sumwrk), xar->wbuff, size); if (write_to_temp(a, xar->wbuff, size) != ARCHIVE_OK) { free(heap); return (ARCHIVE_FATAL); } if (r == ARCHIVE_OK) { xar->stream.next_out = xar->wbuff; xar->stream.avail_out = sizeof(xar->wbuff); } else { checksum_final(&(xar->a_sumwrk), &(heap->a_sum)); heap->length = xar->stream.total_out; /* Add heap to the tail of file->xattr. */ heap->next = NULL; *file->xattr.last = heap; file->xattr.last = &(heap->next); break; } } /* Clean up compression library. */ r = compression_end(&(a->archive), &(xar->stream)); if (r != ARCHIVE_OK) return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } static int getalgsize(enum sumalg sumalg) { switch (sumalg) { default: case CKSUM_NONE: return (0); case CKSUM_SHA1: return (SHA1_SIZE); case CKSUM_MD5: return (MD5_SIZE); } } static const char * getalgname(enum sumalg sumalg) { switch (sumalg) { default: case CKSUM_NONE: return (NULL); case CKSUM_SHA1: return (SHA1_NAME); case CKSUM_MD5: return (MD5_NAME); } } #if HAVE_LIBXML_XMLWRITER_H #define BAD_CAST_CONST (const xmlChar *) struct xml_writer { xmlTextWriterPtr writer; xmlBufferPtr bp; unsigned int indent; }; static int xml_writer_create(struct xml_writer **pctx) { struct xml_writer *ctx = calloc(1, sizeof(struct xml_writer)); if (ctx == NULL) { return (-1); } ctx->bp = xmlBufferCreate(); if (ctx->bp == NULL) { free(ctx); return (-1); } ctx->writer = xmlNewTextWriterMemory(ctx->bp, 0); if (ctx->writer == NULL) { xmlBufferFree(ctx->bp); free(ctx); return (-1); } *pctx = ctx; return (0); } static int xml_writer_destroy(struct xml_writer *ctx) { xmlFreeTextWriter(ctx->writer); xmlBufferFree(ctx->bp); free(ctx); return (0); } static int xml_writer_start_document(struct xml_writer *ctx) { int r; r = xmlTextWriterStartDocument(ctx->writer, "1.0", "UTF-8", NULL); if (r < 0) { return (r); } r = xmlTextWriterSetIndent(ctx->writer, (int)ctx->indent); return (r); } static int xml_writer_end_document(struct xml_writer *ctx) { return (xmlTextWriterEndDocument(ctx->writer)); } static int xml_writer_set_indent(struct xml_writer *ctx, unsigned int indent) { /* libxml2 only lets you set the indent after starting the document */ ctx->indent = indent; return (0); } static int xml_writer_start_element(struct xml_writer *ctx, const char *localName) { return (xmlTextWriterStartElement(ctx->writer, BAD_CAST_CONST(localName))); } static int xml_writer_write_attribute(struct xml_writer *ctx, const char *key, const char *value) { return (xmlTextWriterWriteAttribute(ctx->writer, BAD_CAST_CONST(key), BAD_CAST_CONST(value))); } static int xml_writer_write_attributef(struct xml_writer *ctx, const char *key, const char *format, ...) { va_list ap; int ret; va_start(ap, format); ret = xmlTextWriterWriteVFormatAttribute(ctx->writer, BAD_CAST_CONST(key), format, ap); va_end(ap); return (ret); } static int xml_writer_write_string(struct xml_writer *ctx, const char *string) { return (xmlTextWriterWriteString(ctx->writer, BAD_CAST_CONST(string))); } static int xml_writer_write_base64(struct xml_writer* ctx, const char *data, size_t start, size_t len) { return (xmlTextWriterWriteBase64(ctx->writer, data, (int)start, (int)len)); } static int xml_writer_end_element(struct xml_writer *ctx) { return (xmlTextWriterEndElement(ctx->writer)); } static int xml_writer_get_final_content_and_length(struct xml_writer *ctx, const char **out, size_t *size) { - *out = (const char*)ctx->bp->content; - *size = (size_t)ctx->bp->use; + *out = (const char*)xmlBufferContent(ctx->bp); + *size = (size_t)xmlBufferLength(ctx->bp); return (0); } #elif HAVE_XMLLITE_H struct xml_writer { IXmlWriter *writer; IStream *stream; HGLOBAL global; }; static int xml_writer_create(struct xml_writer **pctx) { struct xml_writer *ctx; HRESULT hr; ctx = calloc(1, sizeof(struct xml_writer)); if (ctx == NULL) { return (E_OUTOFMEMORY); } hr = CreateStreamOnHGlobal(NULL, TRUE, &ctx->stream); if (FAILED(hr)) { free(ctx); return (hr); } hr = CreateXmlWriter(&IID_IXmlWriter, (void **)&ctx->writer, NULL); if (FAILED(hr)) { ctx->stream->lpVtbl->Release(ctx->stream); free(ctx); return (hr); } hr = ctx->writer->lpVtbl->SetOutput(ctx->writer, (IUnknown *)ctx->stream); if (FAILED(hr)) { ctx->writer->lpVtbl->Release(ctx->writer); ctx->stream->lpVtbl->Release(ctx->stream); free(ctx); return (hr); } *pctx = ctx; return (S_OK); } static int xml_writer_destroy(struct xml_writer *ctx) { if (ctx->global) GlobalUnlock(ctx->global); ctx->writer->lpVtbl->Release(ctx->writer); /* Destroys only writer */ ctx->stream->lpVtbl->Release(ctx->stream); /* Destroys stream, global */ free(ctx); return (S_OK); } static int xml_writer_start_document(struct xml_writer *ctx) { return ctx->writer->lpVtbl->WriteStartDocument(ctx->writer, XmlStandalone_Omit); } static int xml_writer_end_document(struct xml_writer *ctx) { return ctx->writer->lpVtbl->WriteEndDocument(ctx->writer); } static int xml_writer_set_indent(struct xml_writer *ctx, unsigned int indent) { /* Windows' xmllite does not support indent sizes; will always be 2 */ (void)indent; return ctx->writer->lpVtbl->SetProperty(ctx->writer, XmlWriterProperty_Indent, (LONG_PTR)TRUE); } static int xml_writer_start_element(struct xml_writer *ctx, const char *localName) { struct archive_wstring as; HRESULT hr; archive_string_init(&as); if (archive_wstring_append_from_mbs(&as, localName, strlen(localName))) { hr = E_OUTOFMEMORY; goto exit_hr; } hr = ctx->writer->lpVtbl->WriteStartElement(ctx->writer, NULL, as.s, NULL); exit_hr: archive_wstring_free(&as); return hr; } static int xml_writer_write_attribute(struct xml_writer *ctx, const char *key, const char *value) { struct archive_wstring ask, asv; HRESULT hr; archive_string_init(&ask); archive_string_init(&asv); if (archive_wstring_append_from_mbs(&ask, key, strlen(key))) { hr = E_OUTOFMEMORY; goto exit_hr; } if (archive_wstring_append_from_mbs(&asv, value, strlen(value))) { hr = E_OUTOFMEMORY; goto exit_hr; } hr = ctx->writer->lpVtbl->WriteAttributeString(ctx->writer, NULL, ask.s, NULL, asv.s); exit_hr: archive_wstring_free(&asv); archive_wstring_free(&ask); return hr; } static int xml_writer_write_attributef(struct xml_writer *ctx, const char *key, const char *format, ...) { struct archive_wstring ask, asv; struct archive_string asf; HRESULT hr; va_list ap; va_start(ap, format); archive_string_init(&ask); archive_string_init(&asv); archive_string_init(&asf); if (archive_wstring_append_from_mbs(&ask, key, strlen(key))) { hr = E_OUTOFMEMORY; goto exit_hr; } archive_string_vsprintf(&asf, format, ap); if (archive_wstring_append_from_mbs(&asv, asf.s, asf.length)) { hr = E_OUTOFMEMORY; goto exit_hr; } hr = ctx->writer->lpVtbl->WriteAttributeString(ctx->writer, NULL, ask.s, NULL, asv.s); exit_hr: archive_string_free(&asf); archive_wstring_free(&asv); archive_wstring_free(&ask); va_end(ap); return hr; } static int xml_writer_write_string(struct xml_writer *ctx, const char *string) { struct archive_wstring as; HRESULT hr; archive_string_init(&as); if (archive_wstring_append_from_mbs(&as, string, strlen(string))) { hr = E_OUTOFMEMORY; goto exit_hr; } hr = ctx->writer->lpVtbl->WriteString(ctx->writer, as.s); exit_hr: archive_wstring_free(&as); return hr; } static const wchar_t base64[] = { L'A', L'B', L'C', L'D', L'E', L'F', L'G', L'H', L'I', L'J', L'K', L'L', L'M', L'N', L'O', L'P', L'Q', L'R', L'S', L'T', L'U', L'V', L'W', L'X', L'Y', L'Z', L'a', L'b', L'c', L'd', L'e', L'f', L'g', L'h', L'i', L'j', L'k', L'l', L'm', L'n', L'o', L'p', L'q', L'r', L's', L't', L'u', L'v', L'w', L'x', L'y', L'z', L'0', L'1', L'2', L'3', L'4', L'5', L'6', L'7', L'8', L'9', L'+', L'/' }; static void la_b64_wencode(struct archive_wstring *as, const unsigned char *p, size_t len) { int c; for (; len >= 3; p += 3, len -= 3) { c = p[0] >> 2; archive_wstrappend_wchar(as, base64[c]); c = ((p[0] & 0x03) << 4) | ((p[1] & 0xf0) >> 4); archive_wstrappend_wchar(as, base64[c]); c = ((p[1] & 0x0f) << 2) | ((p[2] & 0xc0) >> 6); archive_wstrappend_wchar(as, base64[c]); c = p[2] & 0x3f; archive_wstrappend_wchar(as, base64[c]); } if (len > 0) { c = p[0] >> 2; archive_wstrappend_wchar(as, base64[c]); c = (p[0] & 0x03) << 4; if (len == 1) { archive_wstrappend_wchar(as, base64[c]); archive_wstrappend_wchar(as, '='); archive_wstrappend_wchar(as, '='); } else { c |= (p[1] & 0xf0) >> 4; archive_wstrappend_wchar(as, base64[c]); c = (p[1] & 0x0f) << 2; archive_wstrappend_wchar(as, base64[c]); archive_wstrappend_wchar(as, '='); } } } static int xml_writer_write_base64(struct xml_writer* ctx, const char *data, size_t start, size_t len) { struct archive_wstring as; HRESULT hr; archive_string_init(&as); la_b64_wencode(&as, (const unsigned char *)data + start, len - start); hr = ctx->writer->lpVtbl->WriteString(ctx->writer, as.s); archive_wstring_free(&as); return hr; } static int xml_writer_end_element(struct xml_writer *ctx) { return ctx->writer->lpVtbl->WriteEndElement(ctx->writer); } static int xml_writer_get_final_content_and_length(struct xml_writer *ctx, const char **out, size_t *size) { HGLOBAL gbl; HRESULT hr; hr = ctx->writer->lpVtbl->Flush(ctx->writer); if (FAILED(hr)) { return (hr); } hr = GetHGlobalFromStream(ctx->stream, &gbl); if (FAILED(hr)) { return (hr); } *out = (const char *)GlobalLock(gbl); if (*out == NULL) { hr = HRESULT_FROM_WIN32(GetLastError()); return (hr); } /* GlobalUnlock is called in * xml_writer_destroy. */ *size = (size_t)GlobalSize(gbl); ctx->global = gbl; return (hr); } #endif /* HAVE_LIBXML_XMLWRITER_H */ #endif /* Support xar format */ diff --git a/contrib/libarchive/libarchive/archive_write_set_format_zip.c b/contrib/libarchive/libarchive/archive_write_set_format_zip.c index 3630b9f2b3a3..19121b519148 100644 --- a/contrib/libarchive/libarchive/archive_write_set_format_zip.c +++ b/contrib/libarchive/libarchive/archive_write_set_format_zip.c @@ -1,2504 +1,2523 @@ /*- * Copyright (c) 2008 Anselm Strauss * Copyright (c) 2009 Joerg Sonnenberger * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /* * Development supported by Google Summer of Code 2008. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LANGINFO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #ifdef HAVE_LZMA_H #include #endif #ifdef HAVE_BZLIB_H #include #endif #ifdef HAVE_ZSTD_H #include #endif #include "archive.h" #include "archive_cryptor_private.h" #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_hmac_private.h" #include "archive_private.h" #include "archive_random_private.h" #include "archive_time_private.h" #include "archive_write_private.h" #include "archive_write_set_format_private.h" #ifndef HAVE_ZLIB_H #include "archive_crc32.h" #endif #define ZIP_ENTRY_FLAG_ENCRYPTED (1 << 0) #define ZIP_ENTRY_FLAG_LZMA_EOPM (1 << 1) #define ZIP_ENTRY_FLAG_DEFLATE_MAX (1 << 1) /* i.e. compression levels 8 & 9 */ #define ZIP_ENTRY_FLAG_DEFLATE_FAST (1 << 2) /* i.e. compression levels 3 & 4 */ #define ZIP_ENTRY_FLAG_DEFLATE_SUPER_FAST (1 << 1) | (1 << 2) /* i.e. compression levels 1 & 2 */ #define ZIP_ENTRY_FLAG_LENGTH_AT_END (1 << 3) #define ZIP_ENTRY_FLAG_UTF8_NAME (1 << 11) #define ZIP_4GB_MAX ARCHIVE_LITERAL_LL(0xffffffff) #define ZIP_4GB_MAX_UNCOMPRESSED ARCHIVE_LITERAL_LL(0xff000000) enum compression { COMPRESSION_UNSPECIFIED = -1, COMPRESSION_STORE = 0, COMPRESSION_DEFLATE = 8, COMPRESSION_BZIP2 = 12, COMPRESSION_LZMA = 14, COMPRESSION_ZSTD = 93, COMPRESSION_XZ = 95 }; #ifdef HAVE_ZLIB_H #define COMPRESSION_DEFAULT COMPRESSION_DEFLATE #else #define COMPRESSION_DEFAULT COMPRESSION_STORE #endif enum encryption { ENCRYPTION_NONE = 0, ENCRYPTION_TRADITIONAL, /* Traditional PKWARE encryption. */ ENCRYPTION_WINZIP_AES128, /* WinZIP AES-128 encryption. */ ENCRYPTION_WINZIP_AES256, /* WinZIP AES-256 encryption. */ }; #define TRAD_HEADER_SIZE 12 /* * See "WinZip - AES Encryption Information" * http://www.winzip.com/aes_info.htm */ /* Value used in compression method. */ #define WINZIP_AES_ENCRYPTION 99 /* A WinZip AES header size which is stored at the beginning of * file contents. */ #define WINZIP_AES128_HEADER_SIZE (8 + 2) #define WINZIP_AES256_HEADER_SIZE (16 + 2) /* AES vendor version. */ #define AES_VENDOR_AE_1 0x0001 #define AES_VENDOR_AE_2 0x0002 /* Authentication code size. */ #define AUTH_CODE_SIZE 10 /**/ #define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) struct cd_segment { struct cd_segment *next; size_t buff_size; unsigned char *buff; unsigned char *p; }; struct trad_enc_ctx { uint32_t keys[3]; }; struct zip { int64_t entry_offset; int64_t entry_compressed_size; int64_t entry_uncompressed_size; int64_t entry_compressed_written; int64_t entry_uncompressed_written; int64_t entry_uncompressed_limit; struct archive_entry *entry; uint32_t entry_crc32; enum compression entry_compression; enum encryption entry_encryption; int entry_flags; int experiments; struct trad_enc_ctx tctx; char tctx_valid; unsigned char trad_chkdat; unsigned aes_vendor; archive_crypto_ctx cctx; char cctx_valid; archive_hmac_sha1_ctx hctx; char hctx_valid; unsigned char *file_header; size_t file_header_extra_offset; unsigned long (*crc32func)(unsigned long crc, const void *buff, size_t len); struct cd_segment *central_directory; struct cd_segment *central_directory_last; size_t central_directory_bytes; size_t central_directory_entries; int64_t written_bytes; /* Overall position in file. */ struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv_default; enum compression requested_compression; short compression_level; int init_default_conversion; enum encryption encryption_type; short threads; #define ZIP_FLAG_AVOID_ZIP64 1 #define ZIP_FLAG_FORCE_ZIP64 2 #define ZIP_FLAG_EXPERIMENT_xl 4 int flags; #if defined(HAVE_LZMA_H) || defined(HAVE_ZLIB_H) || defined(HAVE_BZLIB_H) || defined(HAVE_ZSTD_H) union { #ifdef HAVE_LZMA_H /* ZIP's XZ format (id 95) is easy enough: copy Deflate, mutatis * mutandis the library changes. ZIP's LZMA format (id 14), * however, is rather more involved, starting here: it being a * modified LZMA Alone format requires a bit more * book-keeping. */ struct { char headers_to_write; lzma_options_lzma options; lzma_stream context; } lzma; #endif #ifdef HAVE_ZLIB_H z_stream deflate; #endif #ifdef HAVE_BZLIB_H bz_stream bzip2; #endif #if defined(HAVE_ZSTD_H) && HAVE_ZSTD_compressStream struct { /* Libzstd's init function gives a pointer to a memory area * it manages rather than asking for memory to initialise. */ ZSTD_CStream* context; ZSTD_inBuffer in; ZSTD_outBuffer out; } zstd; #endif } stream; #endif size_t len_buf; unsigned char *buf; }; /* Don't call this min or MIN, since those are already defined on lots of platforms (but not all). */ #define zipmin(a, b) ((a) > (b) ? (b) : (a)) static ssize_t archive_write_zip_data(struct archive_write *, const void *buff, size_t s); static int archive_write_zip_close(struct archive_write *); static int archive_write_zip_free(struct archive_write *); static int archive_write_zip_finish_entry(struct archive_write *); static int archive_write_zip_header(struct archive_write *, struct archive_entry *); static int archive_write_zip_options(struct archive_write *, const char *, const char *); static size_t path_length(struct archive_entry *); static int write_path(struct archive_entry *, struct archive_write *); static void copy_path(struct archive_entry *, unsigned char *); static struct archive_string_conv *get_sconv(struct archive_write *, struct zip *); static int trad_enc_init(struct trad_enc_ctx *, const char *, size_t); static unsigned trad_enc_encrypt_update(struct trad_enc_ctx *, const uint8_t *, size_t, uint8_t *, size_t); static int init_traditional_pkware_encryption(struct archive_write *); static int is_traditional_pkware_encryption_supported(void); static int init_winzip_aes_encryption(struct archive_write *); static int is_winzip_aes_encryption_supported(int encryption); #ifdef HAVE_LZMA_H /* ZIP's LZMA format requires the use of a alas not exposed in LibLZMA * function to write the ZIP header. Given our internal version never * fails, no need for a non-void return type. */ static void lzma_lzma_props_encode(const lzma_options_lzma* options, uint8_t* out) { out[0] = (options->pb * 5 + options->lp) * 9 + options->lc; archive_le32enc(out + 1, options->dict_size); } #endif #if defined(HAVE_LZMA_H) && !defined(HAVE_LZMA_STREAM_ENCODER_MT) /* Dummy mt declarations, to avoid spaghetti includes below. Defined with * macros being renamed afterwards to shadow liblzma's types in order to * avoid some compiler errors. */ #define lzma_stream_encoder_mt(str, opt) dummy_mt(str, opt) #define lzma_mt dummy_options typedef struct { void* filters; uint32_t preset; lzma_check check; short threads; char flags; char block_size; char timeout; } dummy_options; static inline lzma_ret dummy_mt(lzma_stream* stream, const lzma_mt* options) { (void)stream; /* UNUSED */ (void)options; /* UNUSED */ return LZMA_PROG_ERROR; } #endif static unsigned char * cd_alloc(struct zip *zip, size_t length) { unsigned char *p; if (zip->central_directory == NULL || (zip->central_directory_last->p + length > zip->central_directory_last->buff + zip->central_directory_last->buff_size)) { struct cd_segment *segment = calloc(1, sizeof(*segment)); if (segment == NULL) return NULL; segment->buff_size = 64 * 1024; segment->buff = malloc(segment->buff_size); if (segment->buff == NULL) { free(segment); return NULL; } segment->p = segment->buff; if (zip->central_directory == NULL) { zip->central_directory = zip->central_directory_last = segment; } else { zip->central_directory_last->next = segment; zip->central_directory_last = segment; } } p = zip->central_directory_last->p; zip->central_directory_last->p += length; zip->central_directory_bytes += length; return (p); } static unsigned long real_crc32(unsigned long crc, const void *buff, size_t len) { return crc32(crc, buff, (unsigned int)len); } static unsigned long fake_crc32(unsigned long crc, const void *buff, size_t len) { (void)crc; /* UNUSED */ (void)buff; /* UNUSED */ (void)len; /* UNUSED */ return 0; } static int archive_write_zip_options(struct archive_write *a, const char *key, const char *val) { struct zip *zip = a->format_data; int ret = ARCHIVE_FAILED; if (strcmp(key, "compression") == 0) { /* * Set compression to use on all future entries. * This only affects regular files. */ if (val == NULL || val[0] == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: compression option needs a compression name", a->format_name); } else if (strcmp(val, "deflate") == 0) { #ifdef HAVE_ZLIB_H zip->requested_compression = COMPRESSION_DEFLATE; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "deflate compression not supported"); #endif } else if (strcmp(val, "store") == 0) { zip->requested_compression = COMPRESSION_STORE; ret = ARCHIVE_OK; } else if (strcmp(val, "bzip2") == 0) { #ifdef HAVE_BZLIB_H zip->requested_compression = COMPRESSION_BZIP2; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "bzip2 compression not supported"); #endif } else if (strcmp(val, "lzma") == 0) { #ifdef HAVE_LZMA_H zip->requested_compression = COMPRESSION_LZMA; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "lzma compression not supported"); #endif } else if (strcmp(val, "xz") == 0) { #ifdef HAVE_LZMA_H zip->requested_compression = COMPRESSION_XZ; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xz compression not supported"); #endif } else if (strcmp(val, "zstd") == 0) { #if defined(HAVE_ZSTD_H) && HAVE_ZSTD_compressStream zip->requested_compression = COMPRESSION_ZSTD; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "zstd compression not supported"); #endif } return (ret); } else if (strcmp(key, "compression-level") == 0) { char *endptr; if (val == NULL) return (ARCHIVE_WARN); errno = 0; zip->compression_level = (short)strtoul(val, &endptr, 10); if (errno != 0 || *endptr != '\0' || zip->compression_level < 0 || zip->compression_level > 9) { zip->compression_level = 6; // set to default return (ARCHIVE_WARN); } if (zip->compression_level == 0) { zip->requested_compression = COMPRESSION_STORE; return ARCHIVE_OK; } else { #if defined(HAVE_ZLIB_H) || defined(HAVE_LZMA_H) || defined(HAVE_BZLIB_H) || (defined(HAVE_ZSTD_H) && HAVE_ZSTD_compressStream) // Not forcing an already specified compression algorithm if (zip->requested_compression == COMPRESSION_UNSPECIFIED) { #ifdef HAVE_ZLIB_H zip->requested_compression = COMPRESSION_DEFLATE; #elif defined(HAVE_BZLIB_H) zip->requested_compression = COMPRESSION_BZIP2; #elif defined(HAVE_LZMA_H) // Arbitrarily choosing LZMA of the two LZMA methods zip->requested_compression = COMPRESSION_LZMA; #else zip->requested_compression = COMPRESSION_ZSTD; #endif } return ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "compression not supported"); #endif } } else if (strcmp(key, "threads") == 0) { char *endptr; if (val == NULL) return (ARCHIVE_FAILED); errno = 0; zip->threads = (short)strtoul(val, &endptr, 10); if (errno != 0 || *endptr != '\0') { zip->threads = 1; archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Illegal value `%s'", val); return (ARCHIVE_FAILED); } if (zip->threads == 0) { #ifdef HAVE_LZMA_STREAM_ENCODER_MT zip->threads = lzma_cputhreads(); #elif defined(HAVE_SYSCONF) && defined(_SC_NPROCESSORS_ONLN) zip->threads = sysconf(_SC_NPROCESSORS_ONLN); #elif !defined(__CYGWIN__) && defined(_WIN32_WINNT) && _WIN32_WINNT >= 0x0601 /* Windows 7 and up */ DWORD activeProcs = GetActiveProcessorCount(ALL_PROCESSOR_GROUPS); zip->threads = activeProcs <= SHRT_MAX ? (short)activeProcs : SHRT_MAX; #else zip->threads = 1; #endif } return (ARCHIVE_OK); } else if (strcmp(key, "encryption") == 0) { if (val == NULL) { zip->encryption_type = ENCRYPTION_NONE; ret = ARCHIVE_OK; } else if (val[0] == '1' || strcmp(val, "traditional") == 0 || strcmp(val, "zipcrypt") == 0 || strcmp(val, "ZipCrypt") == 0) { if (is_traditional_pkware_encryption_supported()) { zip->encryption_type = ENCRYPTION_TRADITIONAL; ret = ARCHIVE_OK; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "encryption not supported"); } } else if (strcmp(val, "aes128") == 0) { if (is_winzip_aes_encryption_supported( ENCRYPTION_WINZIP_AES128)) { zip->encryption_type = ENCRYPTION_WINZIP_AES128; ret = ARCHIVE_OK; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "encryption not supported"); } } else if (strcmp(val, "aes256") == 0) { if (is_winzip_aes_encryption_supported( ENCRYPTION_WINZIP_AES256)) { zip->encryption_type = ENCRYPTION_WINZIP_AES256; ret = ARCHIVE_OK; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "encryption not supported"); } } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: unknown encryption '%s'", a->format_name, val); } return (ret); } else if (strcmp(key, "experimental") == 0) { if (val == NULL || val[0] == 0) { zip->flags &= ~ ZIP_FLAG_EXPERIMENT_xl; } else { zip->flags |= ZIP_FLAG_EXPERIMENT_xl; } return (ARCHIVE_OK); } else if (strcmp(key, "fakecrc32") == 0) { /* * FOR TESTING ONLY: disable CRC calculation to speed up * certain complex tests. */ if (val == NULL || val[0] == 0) { zip->crc32func = real_crc32; } else { zip->crc32func = fake_crc32; } return (ARCHIVE_OK); } else if (strcmp(key, "hdrcharset") == 0) { /* * Set the character set used in translating filenames. */ if (val == NULL || val[0] == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: hdrcharset option needs a character-set name", a->format_name); } else { zip->opt_sconv = archive_string_conversion_to_charset( &a->archive, val, 0); if (zip->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } else if (strcmp(key, "zip64") == 0) { /* * Bias decisions about Zip64: force them to be * generated in certain cases where they are not * forbidden or avoid them in certain cases where they * are not strictly required. */ if (val != NULL && *val != '\0') { zip->flags |= ZIP_FLAG_FORCE_ZIP64; zip->flags &= ~ZIP_FLAG_AVOID_ZIP64; } else { zip->flags &= ~ZIP_FLAG_FORCE_ZIP64; zip->flags |= ZIP_FLAG_AVOID_ZIP64; } return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } int archive_write_zip_set_compression_deflate(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int ret = ARCHIVE_FAILED; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_zip_set_compression_deflate"); if (a->archive.archive_format != ARCHIVE_FORMAT_ZIP) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can only use archive_write_zip_set_compression_deflate" " with zip format"); ret = ARCHIVE_FATAL; } else { #ifdef HAVE_ZLIB_H struct zip *zip = a->format_data; zip->requested_compression = COMPRESSION_DEFLATE; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "deflate compression not supported"); ret = ARCHIVE_FAILED; #endif } return (ret); } int archive_write_zip_set_compression_bzip2(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int ret = ARCHIVE_FAILED; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_zip_set_compression_bzip2"); if (a->archive.archive_format != ARCHIVE_FORMAT_ZIP) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can only use archive_write_zip_set_compression_bzip2" " with zip format"); ret = ARCHIVE_FATAL; } else { #ifdef HAVE_BZLIB_H struct zip *zip = a->format_data; zip->requested_compression = COMPRESSION_BZIP2; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "bzip2 compression not supported"); ret = ARCHIVE_FAILED; #endif } return (ret); } int archive_write_zip_set_compression_zstd(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int ret = ARCHIVE_FAILED; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_zip_set_compression_zstd"); if (a->archive.archive_format != ARCHIVE_FORMAT_ZIP) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can only use archive_write_zip_set_compression_zstd" " with zip format"); ret = ARCHIVE_FATAL; } else { #if defined(HAVE_ZSTD_H) && HAVE_ZSTD_compressStream struct zip *zip = a->format_data; zip->requested_compression = COMPRESSION_ZSTD; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "zstd compression not supported"); ret = ARCHIVE_FAILED; #endif } return (ret); } int archive_write_zip_set_compression_lzma(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int ret = ARCHIVE_FAILED; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_zip_set_compression_lzma"); if (a->archive.archive_format != ARCHIVE_FORMAT_ZIP) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can only use archive_write_zip_set_compression_lzma" " with zip format"); ret = ARCHIVE_FATAL; } else { #ifdef HAVE_LZMA_H struct zip *zip = a->format_data; zip->requested_compression = COMPRESSION_LZMA; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "lzma compression not supported"); ret = ARCHIVE_FAILED; #endif } return (ret); } int archive_write_zip_set_compression_xz(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int ret = ARCHIVE_FAILED; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_zip_set_compression_xz"); if (a->archive.archive_format != ARCHIVE_FORMAT_ZIP) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can only use archive_write_zip_set_compression_xz" " with zip format"); ret = ARCHIVE_FATAL; } else { #ifdef HAVE_LZMA_H struct zip *zip = a->format_data; zip->requested_compression = COMPRESSION_XZ; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xz compression not supported"); ret = ARCHIVE_FAILED; #endif } return (ret); } int archive_write_zip_set_compression_store(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct zip *zip = a->format_data; int ret = ARCHIVE_FAILED; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_zip_set_compression_store"); if (a->archive.archive_format != ARCHIVE_FORMAT_ZIP) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can only use archive_write_zip_set_compression_store" " with zip format"); ret = ARCHIVE_FATAL; } else { zip->requested_compression = COMPRESSION_STORE; ret = ARCHIVE_OK; } return (ret); } int archive_write_set_format_zip(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct zip *zip; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_zip"); /* If another format was already registered, unregister it. */ if (a->format_free != NULL) (a->format_free)(a); zip = calloc(1, sizeof(*zip)); if (zip == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip data"); return (ARCHIVE_FATAL); } /* "Unspecified" lets us choose the appropriate compression. */ zip->requested_compression = COMPRESSION_UNSPECIFIED; /* Following the 7-zip write support's lead, setting the default * compression level explicitly to 6 no matter what. */ zip->compression_level = 6; /* Following the xar write support's lead, the default number of * threads is 1 (i.e. the xz compression, the only one caring about * that, not being multi-threaded even if the multi-threaded encoder * were available) */ zip->threads = 1; zip->crc32func = real_crc32; /* A buffer used for both compression and encryption. */ zip->len_buf = 65536; zip->buf = malloc(zip->len_buf); if (zip->buf == NULL) { free(zip); archive_set_error(&a->archive, ENOMEM, "Can't allocate compression buffer"); return (ARCHIVE_FATAL); } a->format_data = zip; a->format_name = "zip"; a->format_options = archive_write_zip_options; a->format_write_header = archive_write_zip_header; a->format_write_data = archive_write_zip_data; a->format_finish_entry = archive_write_zip_finish_entry; a->format_close = archive_write_zip_close; a->format_free = archive_write_zip_free; a->archive.archive_format = ARCHIVE_FORMAT_ZIP; a->archive.archive_format_name = "ZIP"; return (ARCHIVE_OK); } static int is_all_ascii(const char *p) { const unsigned char *pp = (const unsigned char *)p; while (*pp) { if (*pp++ > 127) return (0); } return (1); } static int archive_write_zip_header(struct archive_write *a, struct archive_entry *entry) { unsigned char local_header[32]; unsigned char local_extra[144]; struct zip *zip = a->format_data; unsigned char *e; unsigned char *cd_extra; size_t filename_length; const char *slink = NULL; size_t slink_size = 0; struct archive_string_conv *sconv = get_sconv(a, zip); int ret, ret2 = ARCHIVE_OK; mode_t type; int version_needed = 10; #define MIN_VERSION_NEEDED(x) do { if (version_needed < x) { version_needed = x; } } while (0) /* Ignore types of entries that we don't support. */ type = archive_entry_filetype(entry); if (type != AE_IFREG && type != AE_IFDIR && type != AE_IFLNK) { __archive_write_entry_filetype_unsupported( &a->archive, entry, "zip"); return ARCHIVE_FAILED; } /* If we're not using Zip64, reject large files. */ if (zip->flags & ZIP_FLAG_AVOID_ZIP64) { /* Reject entries over 4GB. */ if (archive_entry_size_is_set(entry) && (archive_entry_size(entry) > ZIP_4GB_MAX)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Files > 4GB require Zip64 extensions"); return ARCHIVE_FAILED; } /* Reject entries if archive is > 4GB. */ if (zip->written_bytes > ZIP_4GB_MAX) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Archives > 4GB require Zip64 extensions"); return ARCHIVE_FAILED; } } /* Only regular files can have size > 0. */ if (type != AE_IFREG) archive_entry_set_size(entry, 0); /* Reset information from last entry. */ zip->entry_offset = zip->written_bytes; zip->entry_uncompressed_limit = INT64_MAX; /* Zero size values implies that we're using a trailing data descriptor */ zip->entry_compressed_size = 0; zip->entry_uncompressed_size = 0; zip->entry_compressed_written = 0; zip->entry_uncompressed_written = 0; zip->entry_flags = 0; zip->entry_crc32 = zip->crc32func(0, NULL, 0); zip->entry_encryption = 0; archive_entry_free(zip->entry); zip->entry = NULL; if (zip->cctx_valid) archive_encrypto_aes_ctr_release(&zip->cctx); if (zip->hctx_valid) archive_hmac_sha1_cleanup(&zip->hctx); zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; if (type == AE_IFREG &&(!archive_entry_size_is_set(entry) || archive_entry_size(entry) > 0)) { switch (zip->encryption_type) { case ENCRYPTION_TRADITIONAL: case ENCRYPTION_WINZIP_AES128: case ENCRYPTION_WINZIP_AES256: zip->entry_flags |= ZIP_ENTRY_FLAG_ENCRYPTED; zip->entry_encryption = zip->encryption_type; break; case ENCRYPTION_NONE: default: break; } } #if defined(_WIN32) && !defined(__CYGWIN__) /* Make sure the path separators in pathname, hardlink and symlink * are all slash '/', not the Windows path separator '\'. */ zip->entry = __la_win_entry_in_posix_pathseparator(entry); if (zip->entry == entry) zip->entry = archive_entry_clone(entry); #else zip->entry = archive_entry_clone(entry); #endif if (zip->entry == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip header data"); return (ARCHIVE_FATAL); } if (sconv != NULL) { const char *p; size_t len; if (archive_entry_pathname_l(zip->entry, &p, &len, sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate Pathname '%s' to %s", archive_entry_pathname(zip->entry), archive_string_conversion_charset_name(sconv)); ret2 = ARCHIVE_WARN; } if (len > 0) archive_entry_set_pathname(zip->entry, p); /* * There is no standard for symlink handling; we convert * it using the same character-set translation that we use * for filename. */ if (type == AE_IFLNK) { if (archive_entry_symlink_l(zip->entry, &p, &len, sconv)) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory " " for Symlink"); return (ARCHIVE_FATAL); } /* No error if we can't convert. */ } else if (len > 0) archive_entry_set_symlink(zip->entry, p); } } /* If filename isn't ASCII and we can use UTF-8, set the UTF-8 flag. */ if (!is_all_ascii(archive_entry_pathname(zip->entry))) { if (zip->opt_sconv != NULL) { if (strcmp(archive_string_conversion_charset_name( zip->opt_sconv), "UTF-8") == 0) zip->entry_flags |= ZIP_ENTRY_FLAG_UTF8_NAME; #if HAVE_NL_LANGINFO } else if (strcmp(nl_langinfo(CODESET), "UTF-8") == 0) { zip->entry_flags |= ZIP_ENTRY_FLAG_UTF8_NAME; #endif } } filename_length = path_length(zip->entry); /* Determine appropriate compression and size for this entry. */ if (type == AE_IFLNK) { slink = archive_entry_symlink(zip->entry); if (slink != NULL) slink_size = strlen(slink); else slink_size = 0; zip->entry_uncompressed_limit = slink_size; zip->entry_compressed_size = slink_size; zip->entry_uncompressed_size = slink_size; zip->entry_crc32 = zip->crc32func(zip->entry_crc32, (const unsigned char *)slink, slink_size); zip->entry_compression = COMPRESSION_STORE; MIN_VERSION_NEEDED(20); } else if (type != AE_IFREG) { zip->entry_compression = COMPRESSION_STORE; zip->entry_uncompressed_limit = 0; MIN_VERSION_NEEDED(20); } else if (archive_entry_size_is_set(zip->entry)) { int64_t size = archive_entry_size(zip->entry); int64_t additional_size = 0; zip->entry_uncompressed_limit = size; zip->entry_compression = zip->requested_compression; if (zip->entry_compression == COMPRESSION_UNSPECIFIED) { zip->entry_compression = COMPRESSION_DEFAULT; } switch (zip->entry_compression) { case COMPRESSION_STORE: zip->entry_compressed_size = size; zip->entry_uncompressed_size = size; MIN_VERSION_NEEDED(10); break; case COMPRESSION_ZSTD: zip->entry_uncompressed_size = size; MIN_VERSION_NEEDED(63); break; case COMPRESSION_LZMA: zip->entry_uncompressed_size = size; zip->entry_flags |= ZIP_ENTRY_FLAG_LZMA_EOPM; MIN_VERSION_NEEDED(63); break; case COMPRESSION_XZ: zip->entry_uncompressed_size = size; MIN_VERSION_NEEDED(63); break; case COMPRESSION_BZIP2: zip->entry_uncompressed_size = size; MIN_VERSION_NEEDED(46); break; default: // i.e. deflate compression zip->entry_uncompressed_size = size; switch (zip->compression_level) { case 1: case 2: zip->entry_flags |= ZIP_ENTRY_FLAG_DEFLATE_SUPER_FAST; break; case 3: case 4: zip->entry_flags |= ZIP_ENTRY_FLAG_DEFLATE_FAST; break; case 8: case 9: zip->entry_flags |= ZIP_ENTRY_FLAG_DEFLATE_MAX; break; default: break; } MIN_VERSION_NEEDED(20); break; } if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { switch (zip->entry_encryption) { case ENCRYPTION_TRADITIONAL: additional_size = TRAD_HEADER_SIZE; MIN_VERSION_NEEDED(20); break; case ENCRYPTION_WINZIP_AES128: additional_size = WINZIP_AES128_HEADER_SIZE + AUTH_CODE_SIZE; MIN_VERSION_NEEDED(20); break; case ENCRYPTION_WINZIP_AES256: additional_size = WINZIP_AES256_HEADER_SIZE + AUTH_CODE_SIZE; MIN_VERSION_NEEDED(20); break; case ENCRYPTION_NONE: default: break; } if (zip->entry_compression == COMPRESSION_STORE) zip->entry_compressed_size += additional_size; } /* * Set Zip64 extension in any of the following cases * (this was suggested by discussion on info-zip-dev * mailing list): * = Zip64 is being forced by user * = File is over 4GiB uncompressed * (including encryption header, if any) * = File is close to 4GiB and is being compressed * (compression might make file larger) */ if ((zip->flags & ZIP_FLAG_FORCE_ZIP64) || (zip->entry_uncompressed_size + additional_size > ZIP_4GB_MAX) || (zip->entry_uncompressed_size > ZIP_4GB_MAX_UNCOMPRESSED && zip->entry_compression != COMPRESSION_STORE)) { MIN_VERSION_NEEDED(45); } /* We may know the size, but never the CRC. */ zip->entry_flags |= ZIP_ENTRY_FLAG_LENGTH_AT_END; } else { /* We don't know the size. Use the default * compression unless specified otherwise. */ zip->entry_compression = zip->requested_compression; if (zip->entry_compression == COMPRESSION_UNSPECIFIED) { zip->entry_compression = COMPRESSION_DEFAULT; } zip->entry_flags |= ZIP_ENTRY_FLAG_LENGTH_AT_END; if ((zip->flags & ZIP_FLAG_AVOID_ZIP64) == 0) { /* We might use zip64 extensions, so require 4.5 */ MIN_VERSION_NEEDED(45); } switch (zip->entry_compression) { case COMPRESSION_STORE: MIN_VERSION_NEEDED(10); break; case COMPRESSION_ZSTD: MIN_VERSION_NEEDED(63); break; case COMPRESSION_LZMA: zip->entry_flags |= ZIP_ENTRY_FLAG_LZMA_EOPM; MIN_VERSION_NEEDED(63); break; case COMPRESSION_XZ: MIN_VERSION_NEEDED(63); break; case COMPRESSION_BZIP2: MIN_VERSION_NEEDED(46); break; default: // i.e. deflate compression switch (zip->compression_level) { case 1: case 2: zip->entry_flags |= ZIP_ENTRY_FLAG_DEFLATE_SUPER_FAST; break; case 3: case 4: zip->entry_flags |= ZIP_ENTRY_FLAG_DEFLATE_FAST; break; case 8: case 9: zip->entry_flags |= ZIP_ENTRY_FLAG_DEFLATE_MAX; break; default: break; } MIN_VERSION_NEEDED(20); break; } if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { switch (zip->entry_encryption) { case ENCRYPTION_TRADITIONAL: case ENCRYPTION_WINZIP_AES128: case ENCRYPTION_WINZIP_AES256: MIN_VERSION_NEEDED(20); break; case ENCRYPTION_NONE: default: break; } } } /* Format the local header. */ memset(local_header, 0, sizeof(local_header)); memcpy(local_header, "PK\003\004", 4); archive_le16enc(local_header + 4, version_needed); archive_le16enc(local_header + 6, zip->entry_flags); if (zip->entry_encryption == ENCRYPTION_WINZIP_AES128 || zip->entry_encryption == ENCRYPTION_WINZIP_AES256) archive_le16enc(local_header + 8, WINZIP_AES_ENCRYPTION); else archive_le16enc(local_header + 8, zip->entry_compression); archive_le32enc(local_header + 10, unix_to_dos(archive_entry_mtime(zip->entry))); if ((zip->entry_flags & ZIP_ENTRY_FLAG_LENGTH_AT_END) == 0) { archive_le32enc(local_header + 14, zip->entry_crc32); archive_le32enc(local_header + 18, (uint32_t)zip->entry_compressed_size); archive_le32enc(local_header + 22, (uint32_t)zip->entry_uncompressed_size); } archive_le16enc(local_header + 26, (uint16_t)filename_length); if (zip->entry_encryption == ENCRYPTION_TRADITIONAL) { if (zip->entry_flags & ZIP_ENTRY_FLAG_LENGTH_AT_END) zip->trad_chkdat = local_header[11]; else zip->trad_chkdat = local_header[17]; } /* Format as much of central directory file header as we can: */ zip->file_header = cd_alloc(zip, 46); /* If (zip->file_header == NULL) XXXX */ ++zip->central_directory_entries; memset(zip->file_header, 0, 46); memcpy(zip->file_header, "PK\001\002", 4); /* "Made by PKZip 2.0 on Unix." */ archive_le16enc(zip->file_header + 4, 3 * 256 + version_needed); archive_le16enc(zip->file_header + 6, version_needed); archive_le16enc(zip->file_header + 8, zip->entry_flags); if (zip->entry_encryption == ENCRYPTION_WINZIP_AES128 || zip->entry_encryption == ENCRYPTION_WINZIP_AES256) archive_le16enc(zip->file_header + 10, WINZIP_AES_ENCRYPTION); else archive_le16enc(zip->file_header + 10, zip->entry_compression); archive_le32enc(zip->file_header + 12, unix_to_dos(archive_entry_mtime(zip->entry))); archive_le16enc(zip->file_header + 28, (uint16_t)filename_length); /* Following Info-Zip, store mode in the "external attributes" field. */ archive_le32enc(zip->file_header + 38, ((uint32_t)archive_entry_mode(zip->entry)) << 16); e = cd_alloc(zip, filename_length); /* If (e == NULL) XXXX */ copy_path(zip->entry, e); /* Format extra data. */ memset(local_extra, 0, sizeof(local_extra)); e = local_extra; /* First, extra blocks that are the same between * the local file header and the central directory. * We format them once and then duplicate them. */ /* ux Unix extra data, length 11, version 1 */ if (archive_entry_uid_is_set(entry) || archive_entry_gid_is_set(entry)) { /* TODO: If uid < 64k, use 2 bytes, ditto for gid. */ memcpy(e, "ux\013\000\001", 5); e += 5; *e++ = 4; /* Length of following UID */ archive_le32enc(e, (uint32_t)archive_entry_uid(entry)); e += 4; *e++ = 4; /* Length of following GID */ archive_le32enc(e, (uint32_t)archive_entry_gid(entry)); e += 4; } /* AES extra data field: WinZIP AES information, ID=0x9901 */ if ((zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) && (zip->entry_encryption == ENCRYPTION_WINZIP_AES128 || zip->entry_encryption == ENCRYPTION_WINZIP_AES256)) { memcpy(e, "\001\231\007\000\001\000AE", 8); /* AES vendor version AE-2 does not store a CRC. * WinZip 11 uses AE-1, which does store the CRC, * but it does not store the CRC when the file size * is less than 20 bytes. So we simulate what * WinZip 11 does. * NOTE: WinZip 9.0 and 10.0 uses AE-2 by default. */ if (archive_entry_size_is_set(zip->entry) && archive_entry_size(zip->entry) < 20) { archive_le16enc(e+4, AES_VENDOR_AE_2); zip->aes_vendor = AES_VENDOR_AE_2;/* no CRC. */ } else zip->aes_vendor = AES_VENDOR_AE_1; e += 8; /* AES encryption strength. */ *e++ = (zip->entry_encryption == ENCRYPTION_WINZIP_AES128)?1:3; /* Actual compression method. */ archive_le16enc(e, zip->entry_compression); e += 2; } /* Copy ux, AES-extra into central directory as well. */ zip->file_header_extra_offset = zip->central_directory_bytes; cd_extra = cd_alloc(zip, e - local_extra); memcpy(cd_extra, local_extra, e - local_extra); /* * Following extra blocks vary between local header and * central directory. These are the local header versions. * Central directory versions get formatted in * archive_write_zip_finish_entry() below. */ /* UT timestamp: length depends on what timestamps are set. * This header appears in the Central Directory also, but * according to Info-Zip specification, the CD form * only holds mtime, so we format it separately. */ if (archive_entry_mtime_is_set(entry) || archive_entry_atime_is_set(entry) || archive_entry_ctime_is_set(entry)) { unsigned char *ut = e; memcpy(e, "UT\000\000", 4); e += 4; *e++ = (archive_entry_mtime_is_set(entry) ? 1 : 0) | (archive_entry_atime_is_set(entry) ? 2 : 0) | (archive_entry_ctime_is_set(entry) ? 4 : 0); if (archive_entry_mtime_is_set(entry)) { archive_le32enc(e, (uint32_t)archive_entry_mtime(entry)); e += 4; } if (archive_entry_atime_is_set(entry)) { archive_le32enc(e, (uint32_t)archive_entry_atime(entry)); e += 4; } if (archive_entry_ctime_is_set(entry)) { archive_le32enc(e, (uint32_t)archive_entry_ctime(entry)); e += 4; } archive_le16enc(ut + 2, (uint16_t)(e - ut - 4)); } /* * Note about Zip64 Extended Information Extra Field: * Because libarchive always writes in a streaming * fashion, we never know the CRC when we're writing * the local header. So we have to use length-at-end, which * prevents us from putting size information into a Zip64 * extra field. However, apparently some readers find it * a helpful clue to have an empty such field so they * can expect a 64-bit length-at-end marker. */ if (archive_entry_size_is_set(zip->entry) && (zip->entry_uncompressed_size > ZIP_4GB_MAX || zip->entry_compressed_size > ZIP_4GB_MAX)) { /* Header ID 0x0001, size 0 */ memcpy(e, "\001\000\000\000", 4); e += 4; } if (zip->flags & ZIP_FLAG_EXPERIMENT_xl) { /* Experimental 'xl' extension to improve streaming. */ unsigned char *external_info = e; int included = 7; memcpy(e, "xl\000\000", 4); // 0x6c65 + 2-byte length e += 4; e[0] = included; /* bitmap of included fields */ e += 1; if (included & 1) { archive_le16enc(e, /* "Version created by" */ 3 * 256 + version_needed); e += 2; } if (included & 2) { archive_le16enc(e, 0); /* internal file attributes */ e += 2; } if (included & 4) { archive_le32enc(e, /* external file attributes */ ((uint32_t)archive_entry_mode(zip->entry)) << 16); e += 4; } if (included & 8) { // Libarchive does not currently support file comments. } archive_le16enc(external_info + 2, (uint16_t)(e - (external_info + 4))); } /* Update local header with size of extra data and write it all out: */ archive_le16enc(local_header + 28, (uint16_t)(e - local_extra)); ret = __archive_write_output(a, local_header, 30); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += 30; ret = write_path(zip->entry, a); if (ret <= ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += ret; ret = __archive_write_output(a, local_extra, e - local_extra); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += e - local_extra; /* For symlinks, write the body now. */ if (slink != NULL) { ret = __archive_write_output(a, slink, slink_size); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->entry_compressed_written += slink_size; zip->entry_uncompressed_written += slink_size; zip->written_bytes += slink_size; } switch (zip->entry_compression) { #ifdef HAVE_ZLIB_H case COMPRESSION_DEFLATE: zip->stream.deflate.zalloc = Z_NULL; zip->stream.deflate.zfree = Z_NULL; zip->stream.deflate.opaque = Z_NULL; zip->stream.deflate.next_out = zip->buf; zip->stream.deflate.avail_out = (uInt)zip->len_buf; if (deflateInit2(&zip->stream.deflate, zip->compression_level, Z_DEFLATED, -15, 8, Z_DEFAULT_STRATEGY) != Z_OK) { archive_set_error(&a->archive, ENOMEM, "Can't init deflate compressor"); return (ARCHIVE_FATAL); } break; #endif #ifdef HAVE_BZLIB_H case COMPRESSION_BZIP2: memset(&zip->stream.bzip2, 0, sizeof(bz_stream)); zip->stream.bzip2.next_out = (char*)zip->buf; zip->stream.bzip2.avail_out = (unsigned int)zip->len_buf; if (BZ2_bzCompressInit(&zip->stream.bzip2, zip->compression_level, 0, 0) != BZ_OK) { archive_set_error(&a->archive, ENOMEM, "Can't init bzip2 compressor"); return (ARCHIVE_FATAL); } break; #endif #if defined(HAVE_ZSTD_H) && HAVE_ZSTD_compressStream case COMPRESSION_ZSTD: {/* Libzstd, contrary to many compression libraries, doesn't use * zlib's 0 to 9 scale and its negative scale is way bigger than * its positive one. So setting 1 as the lowest allowed compression * level and rescaling to 2 to 9 to libzstd's positive scale. */ int zstd_compression_level = zip->compression_level == 1 ? ZSTD_minCLevel() // ZSTD_minCLevel is negative ! : (zip->compression_level - 1) * ZSTD_maxCLevel() / 8; zip->stream.zstd.context = ZSTD_createCStream(); size_t zret = ZSTD_initCStream(zip->stream.zstd.context, zstd_compression_level); if (ZSTD_isError(zret)) { archive_set_error(&a->archive, ENOMEM, "Can't init zstd compressor"); return (ARCHIVE_FATAL); } /* Asking for the multi-threaded compressor is a no-op in zstd if * it's not supported, so no need to explicitly check for it */ ZSTD_CCtx_setParameter(zip->stream.zstd.context, ZSTD_c_nbWorkers, zip->threads); zip->stream.zstd.out.dst = zip->buf; zip->stream.zstd.out.size = zip->len_buf; zip->stream.zstd.out.pos = 0; break;} #endif #ifdef HAVE_LZMA_H case COMPRESSION_LZMA: {/* Set compression level 9 as the no-holds barred one */ uint32_t lzma_compression_level = zip->compression_level == 9 ? LZMA_PRESET_EXTREME | zip->compression_level : (uint32_t)zip->compression_level; /* Forcibly setting up the encoder to use the LZMA1 variant, as * it is the one LZMA Alone uses. */ lzma_filter filters[2] = { { .id = LZMA_FILTER_LZMA1, .options = &zip->stream.lzma.options }, { .id = LZMA_VLI_UNKNOWN } }; memset(&zip->stream.lzma.context, 0, sizeof(lzma_stream)); lzma_lzma_preset(&zip->stream.lzma.options, lzma_compression_level); zip->stream.lzma.headers_to_write = 1; /* We'll be writing the headers ourselves, so using the raw * encoder */ if (lzma_raw_encoder(&zip->stream.lzma.context, filters) != LZMA_OK) { archive_set_error(&a->archive, ENOMEM, "Can't init lzma compressor"); return (ARCHIVE_FATAL); } zip->stream.lzma.context.next_out = zip->buf; zip->stream.lzma.context.avail_out = (unsigned int)zip->len_buf; break;} case COMPRESSION_XZ: {/* Set compression level 9 as the no-holds barred one */ uint32_t lzma_compression_level = zip->compression_level == 9 ? LZMA_PRESET_EXTREME | zip->compression_level : (uint32_t)zip->compression_level; lzma_ret retval; #ifndef HAVE_LZMA_STREAM_ENCODER_MT /* Force the number of threads to one, and thus to a mono-threaded * encoder in case we don't have the multi-threaded one */ zip->threads = 1; #endif memset(&zip->stream.lzma.context, 0, sizeof(lzma_stream)); /* The XZ check will be arbitrarily set to none: ZIP already has * a CRC-32 check of its own */ if (zip->threads == 1) { /* XZ uses LZMA2. */ lzma_filter filters[2] = { { .id = LZMA_FILTER_LZMA2, .options = &zip->stream.lzma.options }, { .id = LZMA_VLI_UNKNOWN } }; /* Might as well use the lzma_options we already allocated, * even if we'll never use it after the initialisation */ lzma_lzma_preset(&zip->stream.lzma.options, lzma_compression_level); /* 1 thread requested, so non multi-threaded encoder */ retval = lzma_stream_encoder(&zip->stream.lzma.context, filters, LZMA_CHECK_NONE); } else { lzma_mt options = { .flags = 0, .block_size = 0, .timeout = 0, .filters = NULL, .check = LZMA_CHECK_NONE, .preset = lzma_compression_level, .threads = zip->threads }; /* More than 1 thread requested, so multi-threaded encoder * which always outputs XZ */ retval = lzma_stream_encoder_mt(&zip->stream.lzma.context, &options); } if (retval != LZMA_OK) { archive_set_error(&a->archive, ENOMEM, "Can't init xz compressor"); return (ARCHIVE_FATAL); } zip->stream.lzma.context.next_out = zip->buf; zip->stream.lzma.context.avail_out = (unsigned int)zip->len_buf; break;} #endif default: break; } return (ret2); } static ssize_t archive_write_zip_data(struct archive_write *a, const void *buff, size_t s) { int ret; struct zip *zip = a->format_data; if ((int64_t)s > zip->entry_uncompressed_limit) s = (size_t)zip->entry_uncompressed_limit; zip->entry_uncompressed_written += s; if (s == 0) return 0; if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { switch (zip->entry_encryption) { case ENCRYPTION_TRADITIONAL: /* Initialize traditional PKWARE encryption context. */ if (!zip->tctx_valid) { ret = init_traditional_pkware_encryption(a); if (ret != ARCHIVE_OK) return (ret); zip->tctx_valid = 1; } break; case ENCRYPTION_WINZIP_AES128: case ENCRYPTION_WINZIP_AES256: if (!zip->cctx_valid) { ret = init_winzip_aes_encryption(a); if (ret != ARCHIVE_OK) return (ret); zip->cctx_valid = zip->hctx_valid = 1; } break; case ENCRYPTION_NONE: default: break; } } switch (zip->entry_compression) { case COMPRESSION_STORE: if (zip->tctx_valid || zip->cctx_valid) { const uint8_t *rb = (const uint8_t *)buff; const uint8_t * const re = rb + s; while (rb < re) { size_t l; if (zip->tctx_valid) { l = trad_enc_encrypt_update(&zip->tctx, rb, re - rb, zip->buf, zip->len_buf); } else { l = zip->len_buf; ret = archive_encrypto_aes_ctr_update( &zip->cctx, rb, re - rb, zip->buf, &l); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, l); } ret = __archive_write_output(a, zip->buf, l); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += l; zip->written_bytes += l; rb += l; } } else { ret = __archive_write_output(a, buff, s); if (ret != ARCHIVE_OK) return (ret); zip->written_bytes += s; zip->entry_compressed_written += s; } break; #ifdef HAVE_ZLIB_H case COMPRESSION_DEFLATE: zip->stream.deflate.next_in = (unsigned char*)(uintptr_t)buff; zip->stream.deflate.avail_in = (uInt)s; do { ret = deflate(&zip->stream.deflate, Z_NO_FLUSH); if (ret == Z_STREAM_ERROR) return (ARCHIVE_FATAL); if (zip->stream.deflate.avail_out == 0) { if (zip->tctx_valid) { trad_enc_encrypt_update(&zip->tctx, zip->buf, zip->len_buf, zip->buf, zip->len_buf); } else if (zip->cctx_valid) { size_t outl = zip->len_buf; ret = archive_encrypto_aes_ctr_update( &zip->cctx, zip->buf, zip->len_buf, zip->buf, &outl); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, zip->len_buf); } ret = __archive_write_output(a, zip->buf, zip->len_buf); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += zip->len_buf; zip->written_bytes += zip->len_buf; zip->stream.deflate.next_out = zip->buf; zip->stream.deflate.avail_out = (uInt)zip->len_buf; } } while (zip->stream.deflate.avail_in != 0); break; #endif #if defined(HAVE_ZSTD_H) && HAVE_ZSTD_compressStream case COMPRESSION_ZSTD: zip->stream.zstd.in.src = buff; zip->stream.zstd.in.size = s; zip->stream.zstd.in.pos = 0; do { size_t zret = ZSTD_compressStream(zip->stream.zstd.context, &zip->stream.zstd.out, &zip->stream.zstd.in); if (ZSTD_isError(zret)) return (ARCHIVE_FATAL); if (zip->stream.zstd.out.pos == zip->stream.zstd.out.size) { if (zip->tctx_valid) { trad_enc_encrypt_update(&zip->tctx, zip->buf, zip->len_buf, zip->buf, zip->len_buf); } else if (zip->cctx_valid) { size_t outl = zip->len_buf; ret = archive_encrypto_aes_ctr_update( &zip->cctx, zip->buf, zip->len_buf, zip->buf, &outl); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, zip->len_buf); } ret = __archive_write_output(a, zip->buf, zip->len_buf); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += zip->len_buf; zip->written_bytes += zip->len_buf; zip->stream.zstd.out.dst = zip->buf; zip->stream.zstd.out.size = zip->len_buf; zip->stream.zstd.out.pos = 0; } } while (zip->stream.zstd.in.pos != zip->stream.zstd.in.size); break; #endif #ifdef HAVE_BZLIB_H case COMPRESSION_BZIP2: zip->stream.bzip2.next_in = (char*)(uintptr_t)buff; zip->stream.bzip2.avail_in = (unsigned int)s; do { ret = BZ2_bzCompress(&zip->stream.bzip2, BZ_RUN); if (ret != BZ_RUN_OK) return (ARCHIVE_FATAL); if (zip->stream.bzip2.avail_out == 0) { if (zip->tctx_valid) { trad_enc_encrypt_update(&zip->tctx, zip->buf, zip->len_buf, zip->buf, zip->len_buf); } else if (zip->cctx_valid) { size_t outl = zip->len_buf; ret = archive_encrypto_aes_ctr_update( &zip->cctx, zip->buf, zip->len_buf, zip->buf, &outl); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, zip->len_buf); } ret = __archive_write_output(a, zip->buf, zip->len_buf); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += zip->len_buf; zip->written_bytes += zip->len_buf; zip->stream.bzip2.next_out = (char*)zip->buf; zip->stream.bzip2.avail_out = (unsigned int)zip->len_buf; } } while (zip->stream.bzip2.avail_in != 0); break; #endif #ifdef HAVE_LZMA_H case COMPRESSION_LZMA: if (zip->stream.lzma.headers_to_write) { /* LZMA Alone and ZIP's LZMA format (i.e. id 14) are almost * the same. Here's an example of a structure of LZMA Alone: * * $ cat /bin/ls | lzma | xxd | head -n 1 * 00000000: 5d00 0080 00ff ffff ffff ffff ff00 2814 * * 5 bytes 8 bytes n bytes * * * lzma_params is a 5-byte blob that has to be decoded to * extract parameters of this LZMA stream. The * uncompressed_size field is an uint64_t value that contains * information about the size of the uncompressed file, or * UINT64_MAX if this value is unknown. The part is * the actual LZMA-compressed data stream. * * Now here's the structure of ZIP's LZMA format: * * $ cat stream_inside_zipx | xxd | head -n 1 * 00000000: 0914 0500 5d00 8000 0000 2814 .... .... * * 2byte 2byte 5 bytes n bytes * * * This means that ZIP's LZMA format contains an additional * magic1 and magic2 headers, the lzma_params field contains * the same parameter set as in LZMA Alone, and the * field is the same as in LZMA Alone as well. However, note * that ZIP's format is missing the uncompressed_size field. * * So we need to write a raw LZMA stream, set up for LZMA1 * (i.e. the algorithm variant LZMA Alone uses), which was * done above in the initialisation but first we need to * write ZIP's LZMA header, as if it were Stored data. Then * we can use the raw stream as if it were any other. magic1 * being version numbers and magic2 being lzma_params's size, * they get written in without further ado but lzma_params * requires to use other functions than the usual lzma_stream * manipulating ones, hence the additional book-keeping * required alongside the lzma_stream. */ uint8_t buf[9] = { LZMA_VERSION_MAJOR, LZMA_VERSION_MINOR, 5, 0 }; lzma_lzma_props_encode(&zip->stream.lzma.options, buf + 4); const size_t sh = 9; if (zip->tctx_valid || zip->cctx_valid) { uint8_t* header = buf; const uint8_t * const rh = header + sh; while (header < rh) { size_t l; if (zip->tctx_valid) { l = trad_enc_encrypt_update(&zip->tctx, header, rh - header, zip->buf, zip->len_buf); } else { l = zip->len_buf; ret = archive_encrypto_aes_ctr_update( &zip->cctx, header, rh - header, zip->buf, &l); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, l); } ret = __archive_write_output(a, zip->buf, l); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += l; zip->written_bytes += l; header += l; } } else { ret = __archive_write_output(a, buf, sh); if (ret != ARCHIVE_OK) return (ret); zip->written_bytes += sh; zip->entry_compressed_written += sh; } zip->stream.lzma.headers_to_write = 0; } /* FALLTHROUGH */ case COMPRESSION_XZ: zip->stream.lzma.context.next_in = (unsigned char*)(uintptr_t)buff; zip->stream.lzma.context.avail_in = (unsigned int)s; do { ret = lzma_code(&zip->stream.lzma.context, LZMA_RUN); if (ret == LZMA_MEM_ERROR) return (ARCHIVE_FATAL); if (zip->stream.lzma.context.avail_out == 0) { if (zip->tctx_valid) { trad_enc_encrypt_update(&zip->tctx, zip->buf, zip->len_buf, zip->buf, zip->len_buf); } else if (zip->cctx_valid) { size_t outl = zip->len_buf; ret = archive_encrypto_aes_ctr_update( &zip->cctx, zip->buf, zip->len_buf, zip->buf, &outl); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, zip->len_buf); } ret = __archive_write_output(a, zip->buf, zip->len_buf); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += zip->len_buf; zip->written_bytes += zip->len_buf; zip->stream.lzma.context.next_out = zip->buf; zip->stream.lzma.context.avail_out = (unsigned int)zip->len_buf; } } while (zip->stream.lzma.context.avail_in != 0); break; #endif case COMPRESSION_UNSPECIFIED: default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid ZIP compression type"); return ARCHIVE_FATAL; } zip->entry_uncompressed_limit -= s; if (!zip->cctx_valid || zip->aes_vendor != AES_VENDOR_AE_2) zip->entry_crc32 = zip->crc32func(zip->entry_crc32, buff, (unsigned)s); return (s); } static int archive_write_zip_finish_entry(struct archive_write *a) { struct zip *zip = a->format_data; int ret; #if defined(HAVE_BZLIB_H) || (defined(HAVE_ZSTD_H) && HAVE_ZSTD_compressStream) || HAVE_LZMA_H char finishing; #endif switch (zip->entry_compression) { #ifdef HAVE_ZLIB_H case COMPRESSION_DEFLATE: for (;;) { size_t remainder; ret = deflate(&zip->stream.deflate, Z_FINISH); if (ret == Z_STREAM_ERROR) return (ARCHIVE_FATAL); remainder = zip->len_buf - zip->stream.deflate.avail_out; if (zip->tctx_valid) { trad_enc_encrypt_update(&zip->tctx, zip->buf, remainder, zip->buf, remainder); } else if (zip->cctx_valid) { size_t outl = remainder; ret = archive_encrypto_aes_ctr_update( &zip->cctx, zip->buf, remainder, zip->buf, &outl); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, remainder); } ret = __archive_write_output(a, zip->buf, remainder); if (ret != ARCHIVE_OK) + { + deflateEnd(&zip->stream.deflate); return (ret); + } zip->entry_compressed_written += remainder; zip->written_bytes += remainder; zip->stream.deflate.next_out = zip->buf; if (zip->stream.deflate.avail_out != 0) break; zip->stream.deflate.avail_out = (uInt)zip->len_buf; } deflateEnd(&zip->stream.deflate); break; #endif #ifdef HAVE_BZLIB_H case COMPRESSION_BZIP2: finishing = 1; do { size_t remainder; ret = BZ2_bzCompress(&zip->stream.bzip2, BZ_FINISH); if (ret == BZ_STREAM_END) finishing = 0; else if (ret != BZ_RUN_OK && ret != BZ_FINISH_OK) return (ARCHIVE_FATAL); remainder = zip->len_buf - zip->stream.bzip2.avail_out; if (zip->tctx_valid) { trad_enc_encrypt_update(&zip->tctx, zip->buf, remainder, zip->buf, remainder); } else if (zip->cctx_valid) { size_t outl = remainder; ret = archive_encrypto_aes_ctr_update( &zip->cctx, zip->buf, remainder, zip->buf, &outl); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, remainder); } ret = __archive_write_output(a, zip->buf, remainder); if (ret != ARCHIVE_OK) + { + BZ2_bzCompressEnd(&zip->stream.bzip2); return (ret); + } zip->entry_compressed_written += remainder; zip->written_bytes += remainder; zip->stream.bzip2.next_out = (char*)zip->buf; if (zip->stream.bzip2.avail_out != 0) finishing = 0; zip->stream.bzip2.avail_out = (unsigned int)zip->len_buf; } while (finishing); BZ2_bzCompressEnd(&zip->stream.bzip2); break; #endif #if defined(HAVE_ZSTD_H) && HAVE_ZSTD_compressStream case COMPRESSION_ZSTD: finishing = 1; do { size_t remainder; size_t zret = ZSTD_endStream(zip->stream.zstd.context, &zip->stream.zstd.out); if (zret == 0) finishing = 0; else if (ZSTD_isError(zret)) return (ARCHIVE_FATAL); remainder = zip->len_buf - (zip->stream.zstd.out.size - zip->stream.zstd.out.pos); if (zip->tctx_valid) { trad_enc_encrypt_update(&zip->tctx, zip->buf, remainder, zip->buf, remainder); } else if (zip->cctx_valid) { size_t outl = remainder; ret = archive_encrypto_aes_ctr_update( &zip->cctx, zip->buf, remainder, zip->buf, &outl); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, remainder); } ret = __archive_write_output(a, zip->buf, remainder); if (ret != ARCHIVE_OK) + { + ZSTD_freeCStream(zip->stream.zstd.context); return (ret); + } zip->entry_compressed_written += remainder; zip->written_bytes += remainder; - zip->stream.zstd.out.dst = zip->buf; if (zip->stream.zstd.out.pos != zip->stream.zstd.out.size) finishing = 0; + zip->stream.zstd.out.dst = zip->buf; zip->stream.zstd.out.size = zip->len_buf; + zip->stream.zstd.out.pos = 0; } while (finishing); ZSTD_freeCStream(zip->stream.zstd.context); break; #endif #ifdef HAVE_LZMA_H /* XZ and LZMA share clean-up code */ case COMPRESSION_LZMA: case COMPRESSION_XZ: finishing = 1; do { size_t remainder; ret = lzma_code(&zip->stream.lzma.context, LZMA_FINISH); if (ret == LZMA_STREAM_END) finishing = 0; else if (ret == LZMA_MEM_ERROR) return (ARCHIVE_FATAL); remainder = zip->len_buf - zip->stream.lzma.context.avail_out; if (zip->tctx_valid) { trad_enc_encrypt_update(&zip->tctx, zip->buf, remainder, zip->buf, remainder); } else if (zip->cctx_valid) { size_t outl = remainder; ret = archive_encrypto_aes_ctr_update( &zip->cctx, zip->buf, remainder, zip->buf, &outl); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, remainder); } ret = __archive_write_output(a, zip->buf, remainder); if (ret != ARCHIVE_OK) + { + lzma_end(&zip->stream.lzma.context); return (ret); + } zip->entry_compressed_written += remainder; zip->written_bytes += remainder; zip->stream.lzma.context.next_out = zip->buf; if (zip->stream.lzma.context.avail_out != 0) finishing = 0; zip->stream.lzma.context.avail_out = (unsigned int)zip->len_buf; } while (finishing); lzma_end(&zip->stream.lzma.context); break; #endif default: break; } if (zip->hctx_valid) { uint8_t hmac[20]; size_t hmac_len = 20; archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); ret = __archive_write_output(a, hmac, AUTH_CODE_SIZE); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += AUTH_CODE_SIZE; zip->written_bytes += AUTH_CODE_SIZE; } /* Write trailing data descriptor. */ if ((zip->entry_flags & ZIP_ENTRY_FLAG_LENGTH_AT_END) != 0) { char d[24]; memcpy(d, "PK\007\010", 4); if (zip->cctx_valid && zip->aes_vendor == AES_VENDOR_AE_2) archive_le32enc(d + 4, 0);/* no CRC.*/ else archive_le32enc(d + 4, zip->entry_crc32); if (zip->entry_compressed_written > ZIP_4GB_MAX || zip->entry_uncompressed_written > ZIP_4GB_MAX || zip->flags & ZIP_FLAG_FORCE_ZIP64) { archive_le64enc(d + 8, (uint64_t)zip->entry_compressed_written); archive_le64enc(d + 16, (uint64_t)zip->entry_uncompressed_written); ret = __archive_write_output(a, d, 24); zip->written_bytes += 24; } else { archive_le32enc(d + 8, (uint32_t)zip->entry_compressed_written); archive_le32enc(d + 12, (uint32_t)zip->entry_uncompressed_written); ret = __archive_write_output(a, d, 16); zip->written_bytes += 16; } if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } /* UT timestamp: Info-Zip specifies that _only_ the mtime should * be recorded here; ctime and atime are also included in the * local file descriptor. */ if (archive_entry_mtime_is_set(zip->entry)) { unsigned char ut[9]; unsigned char *u = ut, *ud; memcpy(u, "UT\005\000\001", 5); u += 5; archive_le32enc(u, (uint32_t)archive_entry_mtime(zip->entry)); u += 4; ud = cd_alloc(zip, u - ut); if (ud == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip data"); return (ARCHIVE_FATAL); } memcpy(ud, ut, u - ut); } /* Fill in size information in the central directory entry. */ /* Fix up central directory file header. */ if (zip->cctx_valid && zip->aes_vendor == AES_VENDOR_AE_2) archive_le32enc(zip->file_header + 16, 0);/* no CRC.*/ else archive_le32enc(zip->file_header + 16, zip->entry_crc32); /* Truncate to 32 bits; we'll fix up below. */ archive_le32enc(zip->file_header + 20, (uint32_t)zip->entry_compressed_written); archive_le32enc(zip->file_header + 24, (uint32_t)zip->entry_uncompressed_written); archive_le16enc(zip->file_header + 30, (uint16_t)(zip->central_directory_bytes - zip->file_header_extra_offset)); archive_le32enc(zip->file_header + 42, (uint32_t)zip->entry_offset); /* If any of the values immediately above are too large, we'll * need to put the corresponding value in a Zip64 extra field * and set the central directory value to 0xffffffff as a flag. */ if (zip->entry_compressed_written >= ZIP_4GB_MAX || zip->entry_uncompressed_written >= ZIP_4GB_MAX || zip->entry_offset > ZIP_4GB_MAX) { unsigned char zip64[32]; unsigned char *z = zip64, *zd; memcpy(z, "\001\000\000\000", 4); z += 4; if (zip->entry_uncompressed_written >= ZIP_4GB_MAX) { archive_le32enc(zip->file_header + 24, ZIP_4GB_MAX); archive_le64enc(z, zip->entry_uncompressed_written); z += 8; } if (zip->entry_compressed_written >= ZIP_4GB_MAX) { archive_le32enc(zip->file_header + 20, ZIP_4GB_MAX); archive_le64enc(z, zip->entry_compressed_written); z += 8; } if (zip->entry_offset >= ZIP_4GB_MAX) { archive_le32enc(zip->file_header + 42, ZIP_4GB_MAX); archive_le64enc(z, zip->entry_offset); z += 8; } archive_le16enc(zip64 + 2, (uint16_t)(z - (zip64 + 4))); zd = cd_alloc(zip, z - zip64); if (zd == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip data"); return (ARCHIVE_FATAL); } memcpy(zd, zip64, z - zip64); /* Zip64 means version needs to be set to at least 4.5 */ if (archive_le16dec(zip->file_header + 6) < 45) archive_le16enc(zip->file_header + 6, 45); } /* Fix up central directory file header. */ if (zip->cctx_valid && zip->aes_vendor == AES_VENDOR_AE_2) archive_le32enc(zip->file_header + 16, 0);/* no CRC.*/ else archive_le32enc(zip->file_header + 16, zip->entry_crc32); archive_le32enc(zip->file_header + 20, (uint32_t)zipmin(zip->entry_compressed_written, ZIP_4GB_MAX)); archive_le32enc(zip->file_header + 24, (uint32_t)zipmin(zip->entry_uncompressed_written, ZIP_4GB_MAX)); archive_le16enc(zip->file_header + 30, (uint16_t)(zip->central_directory_bytes - zip->file_header_extra_offset)); archive_le32enc(zip->file_header + 42, (uint32_t)zipmin(zip->entry_offset, ZIP_4GB_MAX)); return (ARCHIVE_OK); } static int archive_write_zip_close(struct archive_write *a) { uint8_t buff[64]; int64_t offset_start, offset_end; struct zip *zip = a->format_data; struct cd_segment *segment; int ret; offset_start = zip->written_bytes; segment = zip->central_directory; while (segment != NULL) { ret = __archive_write_output(a, segment->buff, segment->p - segment->buff); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += segment->p - segment->buff; segment = segment->next; } offset_end = zip->written_bytes; /* If central dir info is too large, write Zip64 end-of-cd */ if (offset_end - offset_start > ZIP_4GB_MAX || offset_start > ZIP_4GB_MAX || zip->central_directory_entries > 0xffffUL || (zip->flags & ZIP_FLAG_FORCE_ZIP64)) { /* Zip64 end-of-cd record */ memset(buff, 0, 56); memcpy(buff, "PK\006\006", 4); archive_le64enc(buff + 4, 44); archive_le16enc(buff + 12, 45); archive_le16enc(buff + 14, 45); /* This is disk 0 of 0. */ archive_le64enc(buff + 24, zip->central_directory_entries); archive_le64enc(buff + 32, zip->central_directory_entries); archive_le64enc(buff + 40, offset_end - offset_start); archive_le64enc(buff + 48, offset_start); ret = __archive_write_output(a, buff, 56); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += 56; /* Zip64 end-of-cd locator record. */ memset(buff, 0, 20); memcpy(buff, "PK\006\007", 4); archive_le32enc(buff + 4, 0); archive_le64enc(buff + 8, offset_end); archive_le32enc(buff + 16, 1); ret = __archive_write_output(a, buff, 20); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += 20; } /* Format and write end of central directory. */ memset(buff, 0, sizeof(buff)); memcpy(buff, "PK\005\006", 4); archive_le16enc(buff + 8, (uint16_t)zipmin(0xffffU, zip->central_directory_entries)); archive_le16enc(buff + 10, (uint16_t)zipmin(0xffffU, zip->central_directory_entries)); archive_le32enc(buff + 12, (uint32_t)zipmin(ZIP_4GB_MAX, (offset_end - offset_start))); archive_le32enc(buff + 16, (uint32_t)zipmin(ZIP_4GB_MAX, offset_start)); ret = __archive_write_output(a, buff, 22); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += 22; return (ARCHIVE_OK); } static int archive_write_zip_free(struct archive_write *a) { struct zip *zip; struct cd_segment *segment; zip = a->format_data; while (zip->central_directory != NULL) { segment = zip->central_directory; zip->central_directory = segment->next; free(segment->buff); free(segment); } free(zip->buf); archive_entry_free(zip->entry); if (zip->cctx_valid) archive_encrypto_aes_ctr_release(&zip->cctx); if (zip->hctx_valid) archive_hmac_sha1_cleanup(&zip->hctx); /* TODO: Free opt_sconv, sconv_default */ free(zip); a->format_data = NULL; return (ARCHIVE_OK); } static size_t path_length(struct archive_entry *entry) { mode_t type; const char *path; size_t len; type = archive_entry_filetype(entry); path = archive_entry_pathname(entry); if (path == NULL) return (0); len = strlen(path); if (type == AE_IFDIR && (path[0] == '\0' || path[len - 1] != '/')) ++len; /* Space for the trailing / */ return len; } static int write_path(struct archive_entry *entry, struct archive_write *archive) { int ret; const char *path; mode_t type; size_t written_bytes; path = archive_entry_pathname(entry); type = archive_entry_filetype(entry); written_bytes = 0; if (path == NULL) return (ARCHIVE_FATAL); ret = __archive_write_output(archive, path, strlen(path)); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); written_bytes += strlen(path); /* Folders are recognized by a trailing slash. */ if ((type == AE_IFDIR) & (path[strlen(path) - 1] != '/')) { ret = __archive_write_output(archive, "/", 1); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); written_bytes += 1; } return ((int)written_bytes); } static void copy_path(struct archive_entry *entry, unsigned char *p) { const char *path; size_t pathlen; mode_t type; path = archive_entry_pathname(entry); pathlen = strlen(path); type = archive_entry_filetype(entry); memcpy(p, path, pathlen); /* Folders are recognized by a trailing slash. */ if ((type == AE_IFDIR) && (path[pathlen - 1] != '/')) p[pathlen] = '/'; } static struct archive_string_conv * get_sconv(struct archive_write *a, struct zip *zip) { if (zip->opt_sconv != NULL) return (zip->opt_sconv); if (!zip->init_default_conversion) { zip->sconv_default = archive_string_default_conversion_for_write(&(a->archive)); zip->init_default_conversion = 1; } return (zip->sconv_default); } /* Traditional PKWARE Decryption functions. */ static void trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) { uint8_t t; #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) ctx->keys[0] = CRC32(ctx->keys[0], c); ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; t = (ctx->keys[1] >> 24) & 0xff; ctx->keys[2] = CRC32(ctx->keys[2], t); #undef CRC32 } static uint8_t trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) { unsigned temp = ctx->keys[2] | 2; return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; } static unsigned trad_enc_encrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, size_t in_len, uint8_t *out, size_t out_len) { unsigned i, max; max = (unsigned)((in_len < out_len)? in_len: out_len); for (i = 0; i < max; i++) { uint8_t t = in[i]; out[i] = t ^ trad_enc_decrypt_byte(ctx); trad_enc_update_keys(ctx, t); } return i; } static int trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len) { ctx->keys[0] = 305419896L; ctx->keys[1] = 591751049L; ctx->keys[2] = 878082192L; for (;pw_len; --pw_len) trad_enc_update_keys(ctx, *pw++); return 0; } static int is_traditional_pkware_encryption_supported(void) { uint8_t key[TRAD_HEADER_SIZE]; if (archive_random(key, sizeof(key)-1) != ARCHIVE_OK) return (0); return (1); } static int init_traditional_pkware_encryption(struct archive_write *a) { struct zip *zip = a->format_data; const char *passphrase; uint8_t key[TRAD_HEADER_SIZE]; uint8_t key_encrypted[TRAD_HEADER_SIZE]; int ret; passphrase = __archive_write_get_passphrase(a); if (passphrase == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Encryption needs passphrase"); return ARCHIVE_FAILED; } if (archive_random(key, sizeof(key)-1) != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't generate random number for encryption"); return ARCHIVE_FATAL; } trad_enc_init(&zip->tctx, passphrase, strlen(passphrase)); /* Set the last key code which will be used as a check code * for verifying passphrase in decryption. */ key[TRAD_HEADER_SIZE-1] = zip->trad_chkdat; trad_enc_encrypt_update(&zip->tctx, key, TRAD_HEADER_SIZE, key_encrypted, TRAD_HEADER_SIZE); /* Write encrypted keys in the top of the file content. */ ret = __archive_write_output(a, key_encrypted, TRAD_HEADER_SIZE); if (ret != ARCHIVE_OK) return (ret); zip->written_bytes += TRAD_HEADER_SIZE; zip->entry_compressed_written += TRAD_HEADER_SIZE; return (ret); } static int init_winzip_aes_encryption(struct archive_write *a) { struct zip *zip = a->format_data; const char *passphrase; size_t key_len, salt_len; uint8_t salt[16 + 2]; uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; int ret; passphrase = __archive_write_get_passphrase(a); if (passphrase == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Encryption needs passphrase"); return (ARCHIVE_FAILED); } if (zip->entry_encryption == ENCRYPTION_WINZIP_AES128) { salt_len = 8; key_len = 16; } else { /* AES 256 */ salt_len = 16; key_len = 32; } if (archive_random(salt, salt_len) != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't generate random number for encryption"); return (ARCHIVE_FATAL); } - archive_pbkdf2_sha1(passphrase, strlen(passphrase), + ret = archive_pbkdf2_sha1(passphrase, strlen(passphrase), salt, salt_len, 1000, derived_key, key_len * 2 + 2); + if (ret != 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + ret == CRYPTOR_STUB_FUNCTION ? "Encryption is unsupported due to " + "lack of crypto library" : "Failed to process passphrase"); + return (ARCHIVE_FAILED); + } ret = archive_encrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); if (ret != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, - "Decryption is unsupported due to lack of crypto library"); + "Failed to initialize AES CTR mode"); return (ARCHIVE_FAILED); } ret = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); if (ret != 0) { archive_encrypto_aes_ctr_release(&zip->cctx); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to initialize HMAC-SHA1"); return (ARCHIVE_FAILED); } /* Set a password verification value after the 'salt'. */ salt[salt_len] = derived_key[key_len * 2]; salt[salt_len + 1] = derived_key[key_len * 2 + 1]; /* Write encrypted keys in the top of the file content. */ ret = __archive_write_output(a, salt, salt_len + 2); if (ret != ARCHIVE_OK) return (ret); zip->written_bytes += salt_len + 2; zip->entry_compressed_written += salt_len + 2; return (ARCHIVE_OK); } static int is_winzip_aes_encryption_supported(int encryption) { size_t key_len, salt_len; uint8_t salt[16 + 2]; uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; archive_crypto_ctx cctx; archive_hmac_sha1_ctx hctx; int ret; if (encryption == ENCRYPTION_WINZIP_AES128) { salt_len = 8; key_len = 16; } else { /* AES 256 */ salt_len = 16; key_len = 32; } if (archive_random(salt, salt_len) != ARCHIVE_OK) return (0); ret = archive_pbkdf2_sha1("p", 1, salt, salt_len, 1000, derived_key, key_len * 2 + 2); if (ret != 0) return (0); ret = archive_encrypto_aes_ctr_init(&cctx, derived_key, key_len); if (ret != 0) return (0); ret = archive_hmac_sha1_init(&hctx, derived_key + key_len, key_len); archive_encrypto_aes_ctr_release(&cctx); if (ret != 0) return (0); archive_hmac_sha1_cleanup(&hctx); return (1); } diff --git a/contrib/libarchive/libarchive/filter_fork_posix.c b/contrib/libarchive/libarchive/filter_fork_posix.c index c895c08e59b3..7c48519336ff 100644 --- a/contrib/libarchive/libarchive/filter_fork_posix.c +++ b/contrib/libarchive/libarchive/filter_fork_posix.c @@ -1,238 +1,269 @@ /*- * Copyright (c) 2007 Joerg Sonnenberger - * Copyright (c) 2012 Michihiro NAKAJIMA + * Copyright (c) 2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" /* This capability is only available on POSIX systems. */ #if defined(HAVE_PIPE) && defined(HAVE_FCNTL) && \ (defined(HAVE_FORK) || defined(HAVE_VFORK) || defined(HAVE_POSIX_SPAWNP)) #if defined(HAVE_SYS_TYPES_H) # include #endif #ifdef HAVE_ERRNO_H # include #endif #ifdef HAVE_STRING_H # include #endif #if defined(HAVE_POLL) && (defined(HAVE_POLL_H) || defined(HAVE_SYS_POLL_H)) # if defined(HAVE_POLL_H) # include # elif defined(HAVE_SYS_POLL_H) # include # endif #elif defined(HAVE_SELECT) # if defined(HAVE_SYS_SELECT_H) # include # elif defined(HAVE_UNISTD_H) # include # endif #endif #ifdef HAVE_FCNTL_H # include #endif #ifdef HAVE_SPAWN_H # include #endif #ifdef HAVE_STDLIB_H # include #endif #ifdef HAVE_UNISTD_H # include #endif #include "archive.h" #include "archive_cmdline_private.h" #include "filter_fork.h" int __archive_create_child(const char *cmd, int *child_stdin, int *child_stdout, pid_t *out_child) { pid_t child = -1; int stdin_pipe[2], stdout_pipe[2], tmp; + +#if !defined(POSIX_SPAWN_CLOEXEC_DEFAULT) && \ + (HAVE_FORK || HAVE_VFORK) && \ + (HAVE_CLOSEFROM || HAVE_CLOSE_RANGE || defined(_SC_OPEN_MAX)) +#undef HAVE_POSIX_SPAWNP +#endif + #if HAVE_POSIX_SPAWNP + posix_spawnattr_t attr; posix_spawn_file_actions_t actions; int r; #endif struct archive_cmdline *cmdline; cmdline = __archive_cmdline_allocate(); if (cmdline == NULL) goto state_allocated; if (__archive_cmdline_parse(cmdline, cmd) != ARCHIVE_OK) goto state_allocated; if (pipe(stdin_pipe) == -1) goto state_allocated; if (stdin_pipe[0] == 1 /* stdout */) { if ((tmp = dup(stdin_pipe[0])) == -1) goto stdin_opened; close(stdin_pipe[0]); stdin_pipe[0] = tmp; } if (pipe(stdout_pipe) == -1) goto stdin_opened; if (stdout_pipe[1] == 0 /* stdin */) { if ((tmp = dup(stdout_pipe[1])) == -1) goto stdout_opened; close(stdout_pipe[1]); stdout_pipe[1] = tmp; } #if HAVE_POSIX_SPAWNP - r = posix_spawn_file_actions_init(&actions); + r = posix_spawnattr_init(&attr); if (r != 0) { errno = r; goto stdout_opened; } + r = posix_spawn_file_actions_init(&actions); + if (r != 0) { + errno = r; + goto attr_inited; + } +#ifdef POSIX_SPAWN_CLOEXEC_DEFAULT + r = posix_spawnattr_setflags(&attr, POSIX_SPAWN_CLOEXEC_DEFAULT); + if (r != 0) + goto actions_inited; +#endif r = posix_spawn_file_actions_addclose(&actions, stdin_pipe[1]); if (r != 0) goto actions_inited; r = posix_spawn_file_actions_addclose(&actions, stdout_pipe[0]); if (r != 0) goto actions_inited; /* Setup for stdin. */ r = posix_spawn_file_actions_adddup2(&actions, stdin_pipe[0], 0); if (r != 0) goto actions_inited; if (stdin_pipe[0] != 0 /* stdin */) { r = posix_spawn_file_actions_addclose(&actions, stdin_pipe[0]); if (r != 0) goto actions_inited; } /* Setup for stdout. */ r = posix_spawn_file_actions_adddup2(&actions, stdout_pipe[1], 1); if (r != 0) goto actions_inited; if (stdout_pipe[1] != 1 /* stdout */) { r = posix_spawn_file_actions_addclose(&actions, stdout_pipe[1]); if (r != 0) goto actions_inited; } - r = posix_spawnp(&child, cmdline->path, &actions, NULL, + r = posix_spawnp(&child, cmdline->path, &actions, &attr, cmdline->argv, NULL); if (r != 0) goto actions_inited; posix_spawn_file_actions_destroy(&actions); + posix_spawnattr_destroy(&attr); #else /* HAVE_POSIX_SPAWNP */ #if HAVE_VFORK child = vfork(); #else child = fork(); #endif if (child == -1) goto stdout_opened; if (child == 0) { close(stdin_pipe[1]); close(stdout_pipe[0]); if (dup2(stdin_pipe[0], 0 /* stdin */) == -1) _exit(254); if (stdin_pipe[0] != 0 /* stdin */) close(stdin_pipe[0]); if (dup2(stdout_pipe[1], 1 /* stdout */) == -1) _exit(254); if (stdout_pipe[1] != 1 /* stdout */) close(stdout_pipe[1]); + +#if HAVE_CLOSEFROM + closefrom(3); +#elif HAVE_CLOSE_RANGE + close_range(3, ~0U, 0); +#elif defined(_SC_OPEN_MAX) + for (int i = sysconf(_SC_OPEN_MAX); i > 3;) + close(--i); +#endif + execvp(cmdline->path, cmdline->argv); _exit(254); } #endif /* HAVE_POSIX_SPAWNP */ close(stdin_pipe[0]); close(stdout_pipe[1]); *child_stdin = stdin_pipe[1]; fcntl(*child_stdin, F_SETFL, O_NONBLOCK); *child_stdout = stdout_pipe[0]; fcntl(*child_stdout, F_SETFL, O_NONBLOCK); __archive_cmdline_free(cmdline); *out_child = child; return ARCHIVE_OK; #if HAVE_POSIX_SPAWNP actions_inited: errno = r; posix_spawn_file_actions_destroy(&actions); +attr_inited: + posix_spawnattr_destroy(&attr); #endif stdout_opened: close(stdout_pipe[0]); close(stdout_pipe[1]); stdin_opened: close(stdin_pipe[0]); close(stdin_pipe[1]); state_allocated: __archive_cmdline_free(cmdline); return ARCHIVE_FAILED; } void __archive_check_child(int in, int out) { #if defined(HAVE_POLL) && (defined(HAVE_POLL_H) || defined(HAVE_SYS_POLL_H)) struct pollfd fds[2]; int idx; idx = 0; if (in != -1) { fds[idx].fd = in; fds[idx].events = POLLOUT; ++idx; } if (out != -1) { fds[idx].fd = out; fds[idx].events = POLLIN; ++idx; } poll(fds, idx, -1); /* -1 == INFTIM, wait forever */ #elif defined(HAVE_SELECT) fd_set fds_in, fds_out, fds_error; FD_ZERO(&fds_in); FD_ZERO(&fds_out); FD_ZERO(&fds_error); if (out != -1) { FD_SET(out, &fds_in); FD_SET(out, &fds_error); } if (in != -1) { FD_SET(in, &fds_out); FD_SET(in, &fds_error); } select(in < out ? out + 1 : in + 1, &fds_in, &fds_out, &fds_error, NULL); #else sleep(1); #endif } #endif /* defined(HAVE_PIPE) && defined(HAVE_VFORK) && defined(HAVE_FCNTL) */ diff --git a/contrib/libarchive/libarchive/test/test_acl_nfs4.c b/contrib/libarchive/libarchive/test/test_acl_nfs4.c index 98d39689df69..050c0a063654 100644 --- a/contrib/libarchive/libarchive/test/test_acl_nfs4.c +++ b/contrib/libarchive/libarchive/test/test_acl_nfs4.c @@ -1,207 +1,214 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" /* * Exercise the system-independent portion of the ACL support. * Check that archive_entry objects can save and restore NFS4 ACL data. * * This should work on all systems, regardless of whether local * filesystems support ACLs or not. */ static struct archive_test_acl_t acls1[] = { { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_USER_OBJ, -1, "" }, { ARCHIVE_ENTRY_ACL_TYPE_DENY, ARCHIVE_ENTRY_ACL_READ_DATA, ARCHIVE_ENTRY_ACL_USER, 77, "user77" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_READ_DATA, ARCHIVE_ENTRY_ACL_GROUP_OBJ, -1, "" }, { ARCHIVE_ENTRY_ACL_TYPE_DENY, ARCHIVE_ENTRY_ACL_WRITE_DATA, ARCHIVE_ENTRY_ACL_EVERYONE, -1, "" }, }; static struct archive_test_acl_t acls2[] = { /* An entry for each type. */ { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, 0, ARCHIVE_ENTRY_ACL_USER, 108, "user108" }, { ARCHIVE_ENTRY_ACL_TYPE_DENY, 0, ARCHIVE_ENTRY_ACL_USER, 109, "user109" }, { ARCHIVE_ENTRY_ACL_TYPE_AUDIT, 0, ARCHIVE_ENTRY_ACL_USER, 110, "user110" }, { ARCHIVE_ENTRY_ACL_TYPE_ALARM, 0, ARCHIVE_ENTRY_ACL_USER, 111, "user111" }, /* An entry for each permission. */ { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_USER, 112, "user112" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_READ_DATA, ARCHIVE_ENTRY_ACL_USER, 113, "user113" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_LIST_DIRECTORY, ARCHIVE_ENTRY_ACL_USER, 114, "user114" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_WRITE_DATA, ARCHIVE_ENTRY_ACL_USER, 115, "user115" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_ADD_FILE, ARCHIVE_ENTRY_ACL_USER, 116, "user116" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_APPEND_DATA, ARCHIVE_ENTRY_ACL_USER, 117, "user117" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY, ARCHIVE_ENTRY_ACL_USER, 118, "user118" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS, ARCHIVE_ENTRY_ACL_USER, 119, "user119" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS, ARCHIVE_ENTRY_ACL_USER, 120, "user120" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_DELETE_CHILD, ARCHIVE_ENTRY_ACL_USER, 121, "user121" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES, ARCHIVE_ENTRY_ACL_USER, 122, "user122" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES, ARCHIVE_ENTRY_ACL_USER, 123, "user123" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_DELETE, ARCHIVE_ENTRY_ACL_USER, 124, "user124" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_READ_ACL, ARCHIVE_ENTRY_ACL_USER, 125, "user125" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_WRITE_ACL, ARCHIVE_ENTRY_ACL_USER, 126, "user126" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_WRITE_OWNER, ARCHIVE_ENTRY_ACL_USER, 127, "user127" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_SYNCHRONIZE, ARCHIVE_ENTRY_ACL_USER, 128, "user128" }, /* One entry with each inheritance value. */ { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_READ_DATA | ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, ARCHIVE_ENTRY_ACL_USER, 129, "user129" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_READ_DATA | ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT, ARCHIVE_ENTRY_ACL_USER, 130, "user130" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_READ_DATA | ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT, ARCHIVE_ENTRY_ACL_USER, 131, "user131" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_READ_DATA | ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ARCHIVE_ENTRY_ACL_USER, 132, "user132" }, { ARCHIVE_ENTRY_ACL_TYPE_AUDIT, ARCHIVE_ENTRY_ACL_READ_DATA | ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS, ARCHIVE_ENTRY_ACL_USER, 133, "user133" }, { ARCHIVE_ENTRY_ACL_TYPE_AUDIT, ARCHIVE_ENTRY_ACL_READ_DATA | ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS, ARCHIVE_ENTRY_ACL_USER, 134, "user134" }, /* One entry for each qualifier. */ { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_USER, 135, "user135" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_USER_OBJ, -1, "" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_GROUP, 136, "group136" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_GROUP_OBJ, -1, "" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_EVERYONE, -1, "" }, }; /* * Entries that should be rejected when we attempt to set them * on an ACL that already has NFS4 entries. */ static struct archive_test_acl_t acls_bad[] = { /* POSIX.1e ACL types */ { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_USER, 78, "" }, { ARCHIVE_ENTRY_ACL_TYPE_DEFAULT, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_USER, 78, "" }, /* POSIX.1e tags */ { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_OTHER, -1, "" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_MASK, -1, "" }, /* POSIX.1e permissions */ { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_EVERYONE, -1, "" }, { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_WRITE, ARCHIVE_ENTRY_ACL_EVERYONE, -1, "" }, + + /* Multiple types */ + { ARCHIVE_ENTRY_ACL_TYPE_ALLOW | ARCHIVE_ENTRY_ACL_TYPE_AUDIT, + ARCHIVE_ENTRY_ACL_EXECUTE, + ARCHIVE_ENTRY_ACL_EVERYONE, -1, "" }, + { ARCHIVE_ENTRY_ACL_TYPE_NFS4, ARCHIVE_ENTRY_ACL_EXECUTE, + ARCHIVE_ENTRY_ACL_EVERYONE, -1, "" }, }; DEFINE_TEST(test_acl_nfs4) { struct archive_entry *ae; int i; /* Create a simple archive_entry. */ assert((ae = archive_entry_new()) != NULL); archive_entry_set_pathname(ae, "file"); archive_entry_set_mode(ae, S_IFREG | 0777); /* Store and read back some basic ACL entries. */ assertEntrySetAcls(ae, acls1, sizeof(acls1)/sizeof(acls1[0])); /* Check that entry contains only NFSv4 types */ assert((archive_entry_acl_types(ae) & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) == 0); assert((archive_entry_acl_types(ae) & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0); assertEqualInt(4, archive_entry_acl_reset(ae, ARCHIVE_ENTRY_ACL_TYPE_NFS4)); assertEntryCompareAcls(ae, acls1, sizeof(acls1)/sizeof(acls1[0]), ARCHIVE_ENTRY_ACL_TYPE_NFS4, 0); /* A more extensive set of ACLs. */ assertEntrySetAcls(ae, acls2, sizeof(acls2)/sizeof(acls2[0])); assertEqualInt(32, archive_entry_acl_reset(ae, ARCHIVE_ENTRY_ACL_TYPE_NFS4)); assertEntryCompareAcls(ae, acls2, sizeof(acls2)/sizeof(acls2[0]), ARCHIVE_ENTRY_ACL_TYPE_NFS4, 0); /* * Check that clearing ACLs gets rid of them all by repeating * the first test. */ assertEntrySetAcls(ae, acls1, sizeof(acls1)/sizeof(acls1[0])); failure("Basic ACLs shouldn't be stored as extended ACLs"); assertEqualInt(4, archive_entry_acl_reset(ae, ARCHIVE_ENTRY_ACL_TYPE_NFS4)); /* * Different types of malformed ACL entries that should * fail when added to existing NFS4 ACLs. */ assertEntrySetAcls(ae, acls2, sizeof(acls2)/sizeof(acls2[0])); for (i = 0; i < (int)(sizeof(acls_bad)/sizeof(acls_bad[0])); ++i) { struct archive_test_acl_t *p = &acls_bad[i]; failure("Malformed ACL test #%d", i); assertEqualInt(ARCHIVE_FAILED, archive_entry_acl_add_entry(ae, p->type, p->permset, p->tag, p->qual, p->name)); failure("Malformed ACL test #%d", i); assertEqualInt(32, archive_entry_acl_reset(ae, ARCHIVE_ENTRY_ACL_TYPE_NFS4)); } archive_entry_free(ae); } diff --git a/contrib/libarchive/libarchive/test/test_acl_posix1e.c b/contrib/libarchive/libarchive/test/test_acl_posix1e.c index 025ef6afd102..f9b6ffeb6ab7 100644 --- a/contrib/libarchive/libarchive/test/test_acl_posix1e.c +++ b/contrib/libarchive/libarchive/test/test_acl_posix1e.c @@ -1,179 +1,184 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" /* * Exercise the system-independent portion of the ACL support. * Check that archive_entry objects can save and restore POSIX.1e-style * ACL data. * * This should work on all systems, regardless of whether local * filesystems support ACLs or not. */ static struct archive_test_acl_t acls0[] = { { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_USER_OBJ, 0, "" }, { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_GROUP_OBJ, 0, "" }, { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_WRITE, ARCHIVE_ENTRY_ACL_OTHER, 0, "" }, }; static struct archive_test_acl_t acls1[] = { { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_USER_OBJ, -1, "" }, { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_USER, 77, "user77" }, { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_GROUP_OBJ, -1, "" }, { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_WRITE, ARCHIVE_ENTRY_ACL_OTHER, -1, "" }, }; static struct archive_test_acl_t acls2[] = { { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_EXECUTE | ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_USER_OBJ, -1, "" }, { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_USER, 77, "user77" }, { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, 0, ARCHIVE_ENTRY_ACL_USER, 78, "user78" }, { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_GROUP_OBJ, -1, "" }, { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, 0007, ARCHIVE_ENTRY_ACL_GROUP, 78, "group78" }, { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_WRITE | ARCHIVE_ENTRY_ACL_EXECUTE, ARCHIVE_ENTRY_ACL_OTHER, -1, "" }, }; /* * NFS4 entry types; attempts to set these on top of POSIX.1e * attributes should fail. */ static struct archive_test_acl_t acls_nfs4[] = { /* NFS4 types */ { ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_USER, 78, "" }, { ARCHIVE_ENTRY_ACL_TYPE_DENY, ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_USER, 78, "" }, { ARCHIVE_ENTRY_ACL_TYPE_AUDIT, ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_USER, 78, "" }, { ARCHIVE_ENTRY_ACL_TYPE_ALARM, ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_USER, 78, "" }, /* NFS4 tags */ { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_EVERYONE, -1, "" }, /* NFS4 inheritance markers */ { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_READ | ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, ARCHIVE_ENTRY_ACL_USER, 79, "" }, { ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_READ | ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, ARCHIVE_ENTRY_ACL_USER_OBJ, -1, "" }, + + /* Invalid type codes */ + { ARCHIVE_ENTRY_ACL_TYPE_ACCESS | ARCHIVE_ENTRY_ACL_TYPE_DEFAULT, + ARCHIVE_ENTRY_ACL_READ, + ARCHIVE_ENTRY_ACL_GROUP_OBJ, -1, "" }, }; DEFINE_TEST(test_acl_posix1e) { struct archive_entry *ae; int i; /* Create a simple archive_entry. */ assert((ae = archive_entry_new()) != NULL); archive_entry_set_pathname(ae, "file"); archive_entry_set_mode(ae, S_IFREG | 0777); /* Basic owner/owning group should just update mode bits. */ /* * Note: This features of libarchive's ACL implementation * shouldn't be relied on and should probably be removed. It * was done to identify trivial ACLs so we could avoid * triggering unnecessary extensions. It's better to identify * trivial ACLs at the point they are being read from disk. */ assertEntrySetAcls(ae, acls0, sizeof(acls0)/sizeof(acls0[0])); failure("Basic ACLs shouldn't be stored as extended ACLs"); assert(0 == archive_entry_acl_reset(ae, ARCHIVE_ENTRY_ACL_TYPE_ACCESS)); failure("Basic ACLs should set mode to 0142, not %04o", (unsigned int)archive_entry_mode(ae)&0777); assert((archive_entry_mode(ae) & 0777) == 0142); /* With any extended ACL entry, we should read back a full set. */ assertEntrySetAcls(ae, acls1, sizeof(acls1)/sizeof(acls1[0])); failure("One extended ACL should flag all ACLs to be returned."); /* Check that entry contains only POSIX.1e types */ assert((archive_entry_acl_types(ae) & ARCHIVE_ENTRY_ACL_TYPE_NFS4) == 0); assert((archive_entry_acl_types(ae) & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0); assert(4 == archive_entry_acl_reset(ae, ARCHIVE_ENTRY_ACL_TYPE_ACCESS)); assertEntryCompareAcls(ae, acls1, sizeof(acls1)/sizeof(acls1[0]), ARCHIVE_ENTRY_ACL_TYPE_ACCESS, 0142); failure("Basic ACLs should set mode to 0142, not %04o", (unsigned int)archive_entry_mode(ae)&0777); assert((archive_entry_mode(ae) & 0777) == 0142); /* A more extensive set of ACLs. */ assertEntrySetAcls(ae, acls2, sizeof(acls2)/sizeof(acls2[0])); assertEqualInt(6, archive_entry_acl_reset(ae, ARCHIVE_ENTRY_ACL_TYPE_ACCESS)); assertEntryCompareAcls(ae, acls2, sizeof(acls2)/sizeof(acls2[0]), ARCHIVE_ENTRY_ACL_TYPE_ACCESS, 0543); failure("Basic ACLs should set mode to 0543, not %04o", (unsigned int)archive_entry_mode(ae)&0777); assert((archive_entry_mode(ae) & 0777) == 0543); /* * Check that clearing ACLs gets rid of them all by repeating * the first test. */ assertEntrySetAcls(ae, acls0, sizeof(acls0)/sizeof(acls0[0])); failure("Basic ACLs shouldn't be stored as extended ACLs"); assert(0 == archive_entry_acl_reset(ae, ARCHIVE_ENTRY_ACL_TYPE_ACCESS)); failure("Basic ACLs should set mode to 0142, not %04o", (unsigned int)archive_entry_mode(ae)&0777); assert((archive_entry_mode(ae) & 0777) == 0142); /* * Different types of malformed ACL entries that should * fail when added to existing POSIX.1e ACLs. */ assertEntrySetAcls(ae, acls2, sizeof(acls2)/sizeof(acls2[0])); for (i = 0; i < (int)(sizeof(acls_nfs4)/sizeof(acls_nfs4[0])); ++i) { struct archive_test_acl_t *p = &acls_nfs4[i]; failure("Malformed ACL test #%d", i); assertEqualInt(ARCHIVE_FAILED, archive_entry_acl_add_entry(ae, p->type, p->permset, p->tag, p->qual, p->name)); assertEqualInt(6, archive_entry_acl_reset(ae, ARCHIVE_ENTRY_ACL_TYPE_ACCESS)); } archive_entry_free(ae); } diff --git a/contrib/libarchive/libarchive/test/test_archive_parse_date.c b/contrib/libarchive/libarchive/test/test_archive_parse_date.c index 0a70971b1da6..5251b3393999 100644 --- a/contrib/libarchive/libarchive/test/test_archive_parse_date.c +++ b/contrib/libarchive/libarchive/test/test_archive_parse_date.c @@ -1,97 +1,99 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" #include /* * Verify that the archive_parse_date() function works. */ #define get_date archive_parse_date DEFINE_TEST(test_archive_parse_date) { time_t now = time(NULL); assertEqualInt(get_date(now, "Jan 1, 1970 UTC"), 0); assertEqualInt(get_date(now, "7:12:18-0530 4 May 1983"), 420900138); assertEqualInt(get_date(now, "2004/01/29 513 mest"), 1075345980); + assertEqualInt(get_date(now, "2038-06-01 00:01:02 UTC"), + sizeof(time_t) <= 4 ? -1 : 2158963262); assertEqualInt(get_date(now, "99/02/17 7pm utc"), 919278000); assertEqualInt(get_date(now, "02/17/99 7:11am est"), 919253460); assertEqualInt(get_date(now, "now - 2 hours"), get_date(now, "2 hours ago")); assertEqualInt(get_date(now, "2 hours ago"), get_date(now, "+2 hours ago")); assertEqualInt(get_date(now, "now - 2 hours"), get_date(now, "-2 hours")); /* It's important that we handle ctime() format. */ assertEqualInt(get_date(now, "Sun Feb 22 17:38:26 PST 2009"), 1235353106); /* Basic relative offsets. */ /* If we use the actual current time as the reference, then * these tests break around DST changes, so it's actually * important to use a specific reference time here. */ assertEqualInt(get_date(0, "tomorrow"), 24 * 60 * 60); assertEqualInt(get_date(0, "yesterday"), - 24 * 60 * 60); assertEqualInt(get_date(0, "now + 1 hour"), 60 * 60); assertEqualInt(get_date(0, "now + 1 hour + 1 minute"), 60 * 60 + 60); /* Repeat the above for a different start time. */ now = 1231113600; /* Jan 5, 2009 00:00 UTC */ assertEqualInt(get_date(0, "Jan 5, 2009 00:00 UTC"), now); assertEqualInt(get_date(now, "tomorrow"), now + 24 * 60 * 60); assertEqualInt(get_date(now, "yesterday"), now - 24 * 60 * 60); assertEqualInt(get_date(now, "now + 1 hour"), now + 60 * 60); assertEqualInt(get_date(now, "now + 1 hour + 1 minute"), now + 60 * 60 + 60); assertEqualInt(get_date(now, "tomorrow 5:16am UTC"), now + 24 * 60 * 60 + 5 * 60 * 60 + 16 * 60); assertEqualInt(get_date(now, "UTC 5:16am tomorrow"), now + 24 * 60 * 60 + 5 * 60 * 60 + 16 * 60); /* Jan 5, 2009 was a Monday. */ assertEqualInt(get_date(now, "monday UTC"), now); assertEqualInt(get_date(now, "sunday UTC"), now + 6 * 24 * 60 * 60); assertEqualInt(get_date(now, "tuesday UTC"), now + 24 * 60 * 60); /* "next tuesday" is one week after "tuesday" */ assertEqualInt(get_date(now, "UTC next tuesday"), now + 8 * 24 * 60 * 60); /* "last tuesday" is one week before "tuesday" */ assertEqualInt(get_date(now, "last tuesday UTC"), now - 6 * 24 * 60 * 60); /* Unix epoch timestamps */ assertEqualInt(get_date(now, "@0"), 0); assertEqualInt(get_date(now, "@100"), 100); assertEqualInt(get_date(now, "@+100"), 100); assertEqualInt(get_date(now, "@"), -1); assertEqualInt(get_date(now, "@-"), -1); assertEqualInt(get_date(now, "@+"), -1); assertEqualInt(get_date(now, "@tenth"), -1); assertEqualInt(get_date(now, "@100 tomorrow"), -1); /* TODO: Lots more tests here. */ } diff --git a/contrib/libarchive/libarchive/test/test_archive_string_conversion.c b/contrib/libarchive/libarchive/test/test_archive_string_conversion.c index 12976f3e70ec..055bddc75068 100644 --- a/contrib/libarchive/libarchive/test/test_archive_string_conversion.c +++ b/contrib/libarchive/libarchive/test/test_archive_string_conversion.c @@ -1,1019 +1,1029 @@ /*- * Copyright (c) 2011-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" #include #define __LIBARCHIVE_TEST #include "archive_string.h" /* Execute the following to rebuild the data for this program: tail -n +36 test_archive_string_conversion.c | /bin/sh # # This requires http://unicode.org/Public/6.0.0/ucd/NormalizationTest.txt # if="NormalizationTest.txt" if [ ! -f ${if} ]; then echo "Not found: \"${if}\"" exit 0 fi of=test_archive_string_conversion.txt.Z awk -F ';' '$0 ~/^[0-9A-F]+/ {printf "%s;%s\n", $2, $3}' ${if} | compress | uuencode ${of} > ${of}.uu exit 1 */ static int unicode_to_utf8(char *p, uint32_t uc) { char *_p = p; /* Translate code point to UTF8 */ if (uc <= 0x7f) { *p++ = (char)uc; } else if (uc <= 0x7ff) { *p++ = 0xc0 | ((uc >> 6) & 0x1f); *p++ = 0x80 | (uc & 0x3f); } else if (uc <= 0xffff) { *p++ = 0xe0 | ((uc >> 12) & 0x0f); *p++ = 0x80 | ((uc >> 6) & 0x3f); *p++ = 0x80 | (uc & 0x3f); } else { *p++ = 0xf0 | ((uc >> 18) & 0x07); *p++ = 0x80 | ((uc >> 12) & 0x3f); *p++ = 0x80 | ((uc >> 6) & 0x3f); *p++ = 0x80 | (uc & 0x3f); } return ((int)(p - _p)); } static void archive_be16enc(void *pp, uint16_t u) { unsigned char *p = (unsigned char *)pp; p[0] = (u >> 8) & 0xff; p[1] = u & 0xff; } static int unicode_to_utf16be(char *p, uint32_t uc) { char *utf16 = p; if (uc > 0xffff) { /* We have a code point that won't fit into a * wchar_t; convert it to a surrogate pair. */ uc -= 0x10000; archive_be16enc(utf16, ((uc >> 10) & 0x3ff) + 0xD800); archive_be16enc(utf16+2, (uc & 0x3ff) + 0xDC00); return (4); } else { archive_be16enc(utf16, (uint16_t)uc); return (2); } } static void archive_le16enc(void *pp, uint16_t u) { unsigned char *p = (unsigned char *)pp; p[0] = u & 0xff; p[1] = (u >> 8) & 0xff; } static size_t unicode_to_utf16le(char *p, uint32_t uc) { char *utf16 = p; if (uc > 0xffff) { /* We have a code point that won't fit into a * wchar_t; convert it to a surrogate pair. */ uc -= 0x10000; archive_le16enc(utf16, ((uc >> 10) & 0x3ff) + 0xD800); archive_le16enc(utf16+2, (uc & 0x3ff) + 0xDC00); return (4); } else { archive_le16enc(utf16, (uint16_t)uc); return (2); } } static int wc_size(void) { return (sizeof(wchar_t)); } static int unicode_to_wc(wchar_t *wp, uint32_t uc) { if (wc_size() == 4) { *wp = (wchar_t)uc; return (1); } if (uc > 0xffff) { /* We have a code point that won't fit into a * wchar_t; convert it to a surrogate pair. */ uc -= 0x10000; *wp++ = (wchar_t)(((uc >> 10) & 0x3ff) + 0xD800); *wp = (wchar_t)((uc & 0x3ff) + 0xDC00); return (2); } else { *wp = (wchar_t)uc; return (1); } } /* * Note: U+2000 - U+2FFF, U+F900 - U+FAFF and U+2F800 - U+2FAFF are not * converted to NFD on Mac OS. * see also http://developer.apple.com/library/mac/#qa/qa2001/qa1173.html */ static int scan_unicode_pattern(char *out, wchar_t *wout, char *u16be, char *u16le, const char *pattern, int mac_nfd) { unsigned uc = 0; const char *p = pattern; char *op = out; wchar_t *owp = wout; char *op16be = u16be; char *op16le = u16le; int ret = 0; for (;;) { if (*p >= '0' && *p <= '9') uc = (uc << 4) + (*p - '0'); else if (*p >= 'A' && *p <= 'F') uc = (uc << 4) + (*p - 'A' + 0x0a); else { if (mac_nfd && op == out) { /* * These are not converted to NFD on Mac OS. * U+2000 - U+2FFF * U+F900 - U+FAFF * U+2F800 - U+2FAFF */ switch (uc) { case 0x2194: case 0x219A: case 0x219B: case 0x21AE: case 0x21CD: case 0x21CE: case 0x21CF: case 0x2204: case 0x2209: case 0x220C: case 0x2224: case 0x2226: case 0x2241: case 0x2244: case 0x2247: case 0x2249: case 0x2260: case 0x2262: case 0x226D: case 0x226E: case 0x226F: case 0x2270: case 0x2271: case 0x2274: case 0x2275: case 0x2276: case 0x2278: case 0x2279: case 0x227A: case 0x227B: case 0x2280: case 0x2281: case 0x2284: case 0x2285: case 0x2288: case 0x2289: case 0x22AC: case 0x22AD: case 0x22AE: case 0x22AF: case 0x22E0: case 0x22E1: case 0x22E2: case 0x22E3: case 0x22EA: case 0x22EB: case 0x22EC: case 0x22ED: /* * Those code points are not converted to * NFD on Mac OS. I do not know the reason * because it is undocumented. * NFC NFD * 1109A ==> 11099 110BA * 1109C ==> 1109B 110BA * 110AB ==> 110A5 110BA */ case 0x1109A: case 0x1109C: case 0x110AB: ret = 1; break; } } op16be += unicode_to_utf16be(op16be, uc); op16le += unicode_to_utf16le(op16le, uc); owp += unicode_to_wc(owp, uc); op += unicode_to_utf8(op, uc); if (!*p) { *op16be++ = 0; *op16be = 0; *op16le++ = 0; *op16le = 0; *owp = L'\0'; *op = '\0'; break; } uc = 0; } p++; } return (ret); } static int is_wc_unicode(void) { #if defined(_WIN32) && !defined(__CYGWIN__) return (1); #else return (0); #endif } /* * A conversion test that we correctly normalize UTF-8 and UTF-16BE characters. * On Mac OS, the characters to be Form D. * On other platforms, the characters to be Form C. */ static void test_archive_string_normalization_nfc(const char *testdata) { struct archive *a, *a2; struct archive_string utf8; struct archive_mstring mstr; struct archive_string_conv *f_sconv8, *t_sconv8; struct archive_string_conv *f_sconv16be, *f_sconv16le; FILE *fp; char buff[512]; int line = 0; int locale_is_utf8, wc_is_unicode; int sconv_opt = SCONV_SET_OPT_NORMALIZATION_C; locale_is_utf8 = (NULL != setlocale(LC_ALL, "en_US.UTF-8")); wc_is_unicode = is_wc_unicode(); /* If it doesn't exist, just warn and return. */ if (!locale_is_utf8 && !wc_is_unicode) { skipping("A test of string normalization for NFC requires " "a suitable locale; en_US.UTF-8 not available on this " "system"); return; } archive_string_init(&utf8); memset(&mstr, 0, sizeof(mstr)); /* * Create string conversion objects. */ assert((a = archive_read_new()) != NULL); assertA(NULL != (f_sconv8 = archive_string_conversion_from_charset(a, "UTF-8", 0))); assertA(NULL != (f_sconv16be = archive_string_conversion_from_charset(a, "UTF-16BE", 0))); assertA(NULL != (f_sconv16le = archive_string_conversion_from_charset(a, "UTF-16LE", 0))); assert((a2 = archive_write_new()) != NULL); assertA(NULL != (t_sconv8 = archive_string_conversion_to_charset(a2, "UTF-8", 0))); if (f_sconv8 == NULL || f_sconv16be == NULL || f_sconv16le == NULL || t_sconv8 == NULL) { /* We cannot continue this test. */ assertEqualInt(ARCHIVE_OK, archive_read_free(a)); return; } archive_string_conversion_set_opt(f_sconv8, sconv_opt); archive_string_conversion_set_opt(f_sconv16be, sconv_opt); archive_string_conversion_set_opt(f_sconv16le, sconv_opt); archive_string_conversion_set_opt(t_sconv8, sconv_opt); /* Open a test pattern file. */ assert((fp = fopen(testdata, "r")) != NULL); /* * Read test data. * Test data format: * ';' '\n' * Unicode pattern format: * [0-9A-F]{4,5}([ ][0-9A-F]{4,5}){0,} */ while (fgets(buff, sizeof(buff), fp) != NULL) { char nfc[80], nfd[80]; char utf8_nfc[80], utf8_nfd[80]; char utf16be_nfc[80], utf16be_nfd[80]; char utf16le_nfc[80], utf16le_nfd[80]; wchar_t wc_nfc[40], wc_nfd[40]; char *e, *p; const wchar_t *wp; const char *mp; size_t mplen; line++; if (buff[0] == '#') continue; p = strchr(buff, ';'); if (p == NULL) continue; *p++ = '\0'; /* Copy an NFC pattern */ strncpy(nfc, buff, sizeof(nfc)-1); nfc[sizeof(nfc)-1] = '\0'; e = p; p = strchr(p, '\n'); if (p == NULL) continue; *p = '\0'; /* Copy an NFD pattern */ strncpy(nfd, e, sizeof(nfd)-1); nfd[sizeof(nfd)-1] = '\0'; /* * Get an NFC patterns. */ scan_unicode_pattern(utf8_nfc, wc_nfc, utf16be_nfc, utf16le_nfc, nfc, 0); /* * Get an NFD patterns. */ scan_unicode_pattern(utf8_nfd, wc_nfd, utf16be_nfd, utf16le_nfd, nfd, 0); if (locale_is_utf8) { /* * Normalize an NFD string for import. */ assertEqualInt(0, archive_strcpy_l( &utf8, utf8_nfd, f_sconv8)); failure("NFD(%s) should be converted to NFC(%s):%d", nfd, nfc, line); assertEqualUTF8String(utf8_nfc, utf8.s); /* * Normalize an NFC string for import. */ assertEqualInt(0, archive_strcpy_l( &utf8, utf8_nfc, f_sconv8)); failure("NFC(%s) should not be any changed:%d", nfc, line); assertEqualUTF8String(utf8_nfc, utf8.s); /* * Copy an NFC string for export. */ assertEqualInt(0, archive_strcpy_l( &utf8, utf8_nfc, t_sconv8)); failure("NFC(%s) should not be any changed:%d", nfc, line); assertEqualUTF8String(utf8_nfc, utf8.s); /* * Normalize an NFD string in UTF-16BE for import. */ assertEqualInt(0, archive_strncpy_l( &utf8, utf16be_nfd, 100000, f_sconv16be)); failure("NFD(%s) should be converted to NFC(%s):%d", nfd, nfc, line); assertEqualUTF8String(utf8_nfc, utf8.s); /* * Normalize an NFD string in UTF-16LE for import. */ assertEqualInt(0, archive_strncpy_l( &utf8, utf16le_nfd, 100000, f_sconv16le)); failure("NFD(%s) should be converted to NFC(%s):%d", nfd, nfc, line); assertEqualUTF8String(utf8_nfc, utf8.s); } /* * Test for archive_mstring interface. * In specific, Windows platform UTF-16BE is directly * converted to/from wide-character to avoid the effect of * current locale since windows platform cannot make * locale UTF-8. */ if (locale_is_utf8 || wc_is_unicode) { /* * Normalize an NFD string in UTF-8 for import. */ assertEqualInt(0, archive_mstring_copy_mbs_len_l( &mstr, utf8_nfd, 100000, f_sconv8)); assertEqualInt(0, archive_mstring_get_wcs(a, &mstr, &wp)); failure("UTF-8 NFD(%s) should be converted " "to WCS NFC(%s):%d", nfd, nfc, line); assertEqualWString(wc_nfc, wp); /* * Normalize an NFD string in UTF-16BE for import. */ assertEqualInt(0, archive_mstring_copy_mbs_len_l( &mstr, utf16be_nfd, 100000, f_sconv16be)); assertEqualInt(0, archive_mstring_get_wcs(a, &mstr, &wp)); failure("UTF-8 NFD(%s) should be converted " "to WCS NFC(%s):%d", nfd, nfc, line); assertEqualWString(wc_nfc, wp); /* * Normalize an NFD string in UTF-16LE for import. */ assertEqualInt(0, archive_mstring_copy_mbs_len_l( &mstr, utf16le_nfd, 100000, f_sconv16le)); assertEqualInt(0, archive_mstring_get_wcs(a, &mstr, &wp)); failure("UTF-8 NFD(%s) should be converted " "to WCS NFC(%s):%d", nfd, nfc, line); assertEqualWString(wc_nfc, wp); /* * Copy an NFC wide-string for export. */ assertEqualInt(0, archive_mstring_copy_wcs(&mstr, wc_nfc)); assertEqualInt(0, archive_mstring_get_mbs_l( a, &mstr, &mp, &mplen, t_sconv8)); failure("WCS NFC(%s) should be UTF-8 NFC:%d" ,nfc, line); assertEqualUTF8String(utf8_nfc, mp); } } archive_string_free(&utf8); archive_mstring_clean(&mstr); fclose(fp); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a2)); } static void test_archive_string_normalization_mac_nfd(const char *testdata) { struct archive *a, *a2; struct archive_string utf8; struct archive_mstring mstr; struct archive_string_conv *f_sconv8, *t_sconv8; struct archive_string_conv *f_sconv16be, *f_sconv16le; FILE *fp; char buff[512]; int line = 0; int locale_is_utf8, wc_is_unicode; int sconv_opt = SCONV_SET_OPT_NORMALIZATION_D; locale_is_utf8 = (NULL != setlocale(LC_ALL, "en_US.UTF-8")); wc_is_unicode = is_wc_unicode(); /* If it doesn't exist, just warn and return. */ if (!locale_is_utf8 && !wc_is_unicode) { skipping("A test of string normalization for NFD requires " "a suitable locale; en_US.UTF-8 not available on this " "system"); return; } archive_string_init(&utf8); memset(&mstr, 0, sizeof(mstr)); /* * Create string conversion objects. */ assert((a = archive_read_new()) != NULL); assertA(NULL != (f_sconv8 = archive_string_conversion_from_charset(a, "UTF-8", 0))); assertA(NULL != (f_sconv16be = archive_string_conversion_from_charset(a, "UTF-16BE", 0))); assertA(NULL != (f_sconv16le = archive_string_conversion_from_charset(a, "UTF-16LE", 0))); assert((a2 = archive_write_new()) != NULL); assertA(NULL != (t_sconv8 = archive_string_conversion_to_charset(a2, "UTF-8", 0))); if (f_sconv8 == NULL || f_sconv16be == NULL || f_sconv16le == NULL || t_sconv8 == NULL) { /* We cannot continue this test. */ assertEqualInt(ARCHIVE_OK, archive_read_free(a)); return; } archive_string_conversion_set_opt(f_sconv8, sconv_opt); archive_string_conversion_set_opt(f_sconv16be, sconv_opt); archive_string_conversion_set_opt(f_sconv16le, sconv_opt); archive_string_conversion_set_opt(t_sconv8, sconv_opt); /* Open a test pattern file. */ assert((fp = fopen(testdata, "r")) != NULL); /* * Read test data. * Test data format: * ';' '\n' * Unicode pattern format: * [0-9A-F]{4,5}([ ][0-9A-F]{4,5}){0,} */ while (fgets(buff, sizeof(buff), fp) != NULL) { char nfc[80], nfd[80]; char utf8_nfc[80], utf8_nfd[80]; char utf16be_nfc[80], utf16be_nfd[80]; char utf16le_nfc[80], utf16le_nfd[80]; wchar_t wc_nfc[40], wc_nfd[40]; char *e, *p; const wchar_t *wp; const char *mp; size_t mplen; int should_be_nfc; line++; if (buff[0] == '#') continue; p = strchr(buff, ';'); if (p == NULL) continue; *p++ = '\0'; /* Copy an NFC pattern */ strncpy(nfc, buff, sizeof(nfc)-1); nfc[sizeof(nfc)-1] = '\0'; e = p; p = strchr(p, '\n'); if (p == NULL) continue; *p = '\0'; /* Copy an NFD pattern */ strncpy(nfd, e, sizeof(nfd)-1); nfd[sizeof(nfd)-1] = '\0'; /* * Get an NFC patterns. */ should_be_nfc = scan_unicode_pattern(utf8_nfc, wc_nfc, utf16be_nfc, utf16le_nfc, nfc, 1); /* * Get an NFD patterns. */ scan_unicode_pattern(utf8_nfd, wc_nfd, utf16be_nfd, utf16le_nfd, nfd, 0); if (locale_is_utf8) { /* * Normalize an NFC string for import. */ assertEqualInt(0, archive_strcpy_l( &utf8, utf8_nfc, f_sconv8)); if (should_be_nfc) { failure("NFC(%s) should not be converted to" " NFD(%s):%d", nfc, nfd, line); assertEqualUTF8String(utf8_nfc, utf8.s); } else { failure("NFC(%s) should be converted to" " NFD(%s):%d", nfc, nfd, line); assertEqualUTF8String(utf8_nfd, utf8.s); } /* * Normalize an NFD string for import. */ assertEqualInt(0, archive_strcpy_l( &utf8, utf8_nfd, f_sconv8)); failure("NFD(%s) should not be any changed:%d", nfd, line); assertEqualUTF8String(utf8_nfd, utf8.s); /* * Copy an NFD string for export. */ assertEqualInt(0, archive_strcpy_l( &utf8, utf8_nfd, t_sconv8)); failure("NFD(%s) should not be any changed:%d", nfd, line); assertEqualUTF8String(utf8_nfd, utf8.s); /* * Normalize an NFC string in UTF-16BE for import. */ assertEqualInt(0, archive_strncpy_l( &utf8, utf16be_nfc, 100000, f_sconv16be)); if (should_be_nfc) { failure("NFC(%s) should not be converted to" " NFD(%s):%d", nfc, nfd, line); assertEqualUTF8String(utf8_nfc, utf8.s); } else { failure("NFC(%s) should be converted to" " NFD(%s):%d", nfc, nfd, line); assertEqualUTF8String(utf8_nfd, utf8.s); } /* * Normalize an NFC string in UTF-16LE for import. */ assertEqualInt(0, archive_strncpy_l( &utf8, utf16le_nfc, 100000, f_sconv16le)); if (should_be_nfc) { failure("NFC(%s) should not be converted to" " NFD(%s):%d", nfc, nfd, line); assertEqualUTF8String(utf8_nfc, utf8.s); } else { failure("NFC(%s) should be converted to" " NFD(%s):%d", nfc, nfd, line); assertEqualUTF8String(utf8_nfd, utf8.s); } } /* * Test for archive_mstring interface. * In specific, Windows platform UTF-16BE is directly * converted to/from wide-character to avoid the effect of * current locale since windows platform cannot make * locale UTF-8. */ if (locale_is_utf8 || wc_is_unicode) { /* * Normalize an NFD string in UTF-8 for import. */ assertEqualInt(0, archive_mstring_copy_mbs_len_l( &mstr, utf8_nfc, 100000, f_sconv8)); assertEqualInt(0, archive_mstring_get_wcs(a, &mstr, &wp)); if (should_be_nfc) { failure("UTF-8 NFC(%s) should not be converted " "to WCS NFD(%s):%d", nfc, nfd, line); assertEqualWString(wc_nfc, wp); } else { failure("UTF-8 NFC(%s) should be converted " "to WCS NFD(%s):%d", nfc, nfd, line); assertEqualWString(wc_nfd, wp); } /* * Normalize an NFD string in UTF-16BE for import. */ assertEqualInt(0, archive_mstring_copy_mbs_len_l( &mstr, utf16be_nfc, 100000, f_sconv16be)); assertEqualInt(0, archive_mstring_get_wcs(a, &mstr, &wp)); if (should_be_nfc) { failure("UTF-16BE NFC(%s) should not be " "converted to WCS NFD(%s):%d", nfc, nfd, line); assertEqualWString(wc_nfc, wp); } else { failure("UTF-16BE NFC(%s) should be converted " "to WCS NFD(%s):%d", nfc, nfd, line); assertEqualWString(wc_nfd, wp); } /* * Normalize an NFD string in UTF-16LE for import. */ assertEqualInt(0, archive_mstring_copy_mbs_len_l( &mstr, utf16le_nfc, 100000, f_sconv16le)); assertEqualInt(0, archive_mstring_get_wcs(a, &mstr, &wp)); if (should_be_nfc) { failure("UTF-16LE NFC(%s) should not be " "converted to WCS NFD(%s):%d", nfc, nfd, line); assertEqualWString(wc_nfc, wp); } else { failure("UTF-16LE NFC(%s) should be converted " "to WCS NFD(%s):%d", nfc, nfd, line); assertEqualWString(wc_nfd, wp); } /* * Copy an NFD wide-string for export. */ assertEqualInt(0, archive_mstring_copy_wcs( &mstr, wc_nfd)); assertEqualInt(0, archive_mstring_get_mbs_l( a, &mstr, &mp, &mplen, t_sconv8)); failure("WCS NFD(%s) should be UTF-8 NFD:%d" ,nfd, line); assertEqualUTF8String(utf8_nfd, mp); } } archive_string_free(&utf8); archive_mstring_clean(&mstr); fclose(fp); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a2)); } static void test_archive_string_canonicalization(void) { struct archive *a; struct archive_string_conv *sconv; setlocale(LC_ALL, "en_US.UTF-8"); assert((a = archive_read_new()) != NULL); assertA(NULL != (sconv = archive_string_conversion_to_charset(a, "UTF-8", 1))); failure("Charset name should be UTF-8"); assertEqualString("UTF-8", archive_string_conversion_charset_name(sconv)); assertA(NULL != (sconv = archive_string_conversion_to_charset(a, "UTF8", 1))); failure("Charset name should be UTF-8"); assertEqualString("UTF-8", archive_string_conversion_charset_name(sconv)); assertA(NULL != (sconv = archive_string_conversion_to_charset(a, "utf8", 1))); failure("Charset name should be UTF-8"); assertEqualString("UTF-8", archive_string_conversion_charset_name(sconv)); assertA(NULL != (sconv = archive_string_conversion_to_charset(a, "UTF-16BE", 1))); failure("Charset name should be UTF-16BE"); assertEqualString("UTF-16BE", archive_string_conversion_charset_name(sconv)); assertA(NULL != (sconv = archive_string_conversion_to_charset(a, "UTF16BE", 1))); failure("Charset name should be UTF-16BE"); assertEqualString("UTF-16BE", archive_string_conversion_charset_name(sconv)); assertA(NULL != (sconv = archive_string_conversion_to_charset(a, "utf16be", 1))); failure("Charset name should be UTF-16BE"); assertEqualString("UTF-16BE", archive_string_conversion_charset_name(sconv)); assertA(NULL != (sconv = archive_string_conversion_to_charset(a, "UTF-16LE", 1))); failure("Charset name should be UTF-16LE"); assertEqualString("UTF-16LE", archive_string_conversion_charset_name(sconv)); assertA(NULL != (sconv = archive_string_conversion_to_charset(a, "UTF16LE", 1))); failure("Charset name should be UTF-16LE"); assertEqualString("UTF-16LE", archive_string_conversion_charset_name(sconv)); assertA(NULL != (sconv = archive_string_conversion_to_charset(a, "utf16le", 1))); failure("Charset name should be UTF-16LE"); assertEqualString("UTF-16LE", archive_string_conversion_charset_name(sconv)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } static void check_string(struct archive *a, struct archive_mstring *mstr, struct archive_string_conv *sc, const char *exp, const wchar_t *wexp) { /* Do all the tests on a copy so that we can have a clear initial state every time */ struct archive_mstring mstr2; const char *p = NULL; const wchar_t *wp = NULL; size_t len = 0; memset(&mstr2, 0, sizeof(mstr2)); archive_mstring_copy(&mstr2, mstr); assertEqualInt(0, archive_mstring_get_mbs(a, &mstr2, &p)); assertEqualString(exp, p); p = NULL; archive_mstring_copy(&mstr2, mstr); assertEqualInt(0, archive_mstring_get_utf8(a, &mstr2, &p)); assertEqualString(exp, p); p = NULL; archive_mstring_copy(&mstr2, mstr); assertEqualInt(0, archive_mstring_get_wcs(a, &mstr2, &wp)); assertEqualWString(wexp, wp); wp = NULL; archive_mstring_copy(&mstr2, mstr); assertEqualInt(0, archive_mstring_get_mbs_l(a, &mstr2, &p, &len, sc)); assertEqualString(exp, p); assertEqualInt(len, strlen(exp)); p = NULL; len = 0; archive_mstring_clean(&mstr2); } /* * Make sure no matter what the input encoding is, the string can be * converted too all the output encodings. */ static void test_archive_string_set_get(void) { struct archive *a; struct archive_mstring mstr; struct archive_string_conv *sc; setlocale(LC_ALL, "en_US.UTF-8"); assert((a = archive_read_new()) != NULL); memset(&mstr, 0, sizeof(mstr)); assertA(NULL != (sc = archive_string_conversion_to_charset(a, "UTF-8", 1))); failure("Charset name should be UTF-8"); assertEqualString("UTF-8", archive_string_conversion_charset_name(sc)); assertEqualInt(0, archive_mstring_copy_mbs(&mstr, "AAA")); check_string(a, &mstr, sc, "AAA", L"AAA"); assertEqualInt(4, archive_mstring_copy_utf8(&mstr, "BBBB")); check_string(a, &mstr, sc, "BBBB", L"BBBB"); assertEqualInt(0, archive_mstring_copy_wcs(&mstr, L"CCC12")); check_string(a, &mstr, sc, "CCC12", L"CCC12"); assertEqualInt(0, archive_mstring_copy_mbs_len_l(&mstr, "DDDD-l", 6, sc)); check_string(a, &mstr, sc, "DDDD-l", L"DDDD-l"); assertEqualInt(0, archive_mstring_update_utf8(a, &mstr, "EEEEE---H")); check_string(a, &mstr, sc, "EEEEE---H", L"EEEEE---H"); archive_mstring_clean(&mstr); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_archive_string_conversion) { static const char reffile[] = "test_archive_string_conversion.txt.Z"; static const char testdata[] = "testdata.txt"; struct archive *a; struct archive_entry *ae; char buff[512]; ssize_t size; FILE *fp; /* * Extract a test pattern file. */ extract_reference_file(reffile); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_raw(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, reffile, 512)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assert((fp = fopen(testdata, "w")) != NULL); while ((size = archive_read_data(a, buff, 512)) > 0) assertEqualInt(size, fwrite(buff, 1, size, fp)); assertEqualInt(0, fclose(fp)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); test_archive_string_normalization_nfc(testdata); test_archive_string_normalization_mac_nfd(testdata); test_archive_string_canonicalization(); test_archive_string_set_get(); } DEFINE_TEST(test_archive_string_conversion_utf16_utf8) { #if !defined(_WIN32) || defined(__CYGWIN__) skipping("This test is meant to verify unicode string handling on Windows"); #else struct archive_mstring mstr; const char* utf8_string; memset(&mstr, 0, sizeof(mstr)); assertEqualInt(ARCHIVE_OK, archive_mstring_copy_wcs(&mstr, L"\U0000043f\U00000440\U00000438")); /* Conversion from WCS to UTF-8 should always succeed */ assertEqualInt(ARCHIVE_OK, archive_mstring_get_utf8(NULL, &mstr, &utf8_string)); assertEqualString("\xD0\xBF\xD1\x80\xD0\xB8", utf8_string); archive_mstring_clean(&mstr); #endif } DEFINE_TEST(test_archive_string_conversion_utf8_utf16) { #if !defined(_WIN32) || defined(__CYGWIN__) skipping("This test is meant to verify unicode string handling on Windows"); #else struct archive_mstring mstr; const wchar_t* wcs_string; memset(&mstr, 0, sizeof(mstr)); assertEqualInt(6, archive_mstring_copy_utf8(&mstr, "\xD0\xBF\xD1\x80\xD0\xB8")); /* Conversion from UTF-8 to WCS should always succeed */ assertEqualInt(ARCHIVE_OK, archive_mstring_get_wcs(NULL, &mstr, &wcs_string)); assertEqualWString(L"\U0000043f\U00000440\U00000438", wcs_string); archive_mstring_clean(&mstr); #endif } DEFINE_TEST(test_archive_string_update_utf8_win) { #if !defined(_WIN32) || defined(__CYGWIN__) skipping("This test is meant to verify unicode string handling on Windows" " with the C locale"); #else static const char utf8_string[] = "\xD0\xBF\xD1\x80\xD0\xB8"; static const wchar_t wcs_string[] = L"\U0000043f\U00000440\U00000438"; struct archive_mstring mstr; int r; memset(&mstr, 0, sizeof(mstr)); r = archive_mstring_update_utf8(NULL, &mstr, utf8_string); /* On Windows, this should reliably fail with the C locale */ assertEqualInt(-1, r); assertEqualInt(0, mstr.aes_set & AES_SET_MBS); /* NOTE: We access the internals to validate that they were set by the * 'archive_mstring_update_utf8' function */ /* UTF-8 should always be set */ assertEqualInt(AES_SET_UTF8, mstr.aes_set & AES_SET_UTF8); assertEqualString(utf8_string, mstr.aes_utf8.s); /* WCS should always be set as well */ assertEqualInt(AES_SET_WCS, mstr.aes_set & AES_SET_WCS); assertEqualWString(wcs_string, mstr.aes_wcs.s); archive_mstring_clean(&mstr); #endif } DEFINE_TEST(test_archive_string_update_utf8_utf8) { static const char utf8_string[] = "\xD0\xBF\xD1\x80\xD0\xB8"; static const wchar_t wcs_string[] = L"\U0000043f\U00000440\U00000438"; struct archive_mstring mstr; int r; memset(&mstr, 0, sizeof(mstr)); if (setlocale(LC_ALL, "en_US.UTF-8") == NULL) { skipping("UTF-8 not supported on this system."); return; } r = archive_mstring_update_utf8(NULL, &mstr, utf8_string); /* All conversions should have succeeded */ assertEqualInt(0, r); assertEqualInt(AES_SET_MBS | AES_SET_WCS | AES_SET_UTF8, mstr.aes_set); assertEqualString(utf8_string, mstr.aes_utf8.s); assertEqualString(utf8_string, mstr.aes_mbs.s); assertEqualWString(wcs_string, mstr.aes_wcs.s); archive_mstring_clean(&mstr); } DEFINE_TEST(test_archive_string_update_utf8_koi8) { static const char utf8_string[] = "\xD0\xBF\xD1\x80\xD0\xB8"; static const char koi8_string[] = "\xD0\xD2\xC9"; static const wchar_t wcs_string[] = L"\U0000043f\U00000440\U00000438"; struct archive_mstring mstr; + struct archive *a; int r; memset(&mstr, 0, sizeof(mstr)); if (setlocale(LC_ALL, "ru_RU.KOI8-R") == NULL) { skipping("KOI8-R locale not available on this system."); return; } + a = archive_write_new(); + assertEqualInt(ARCHIVE_OK, archive_write_set_format_pax(a)); + if (archive_write_set_options(a, "hdrcharset=UTF-8") != ARCHIVE_OK) { + skipping("This system cannot convert character-set" + " from KOI8-R to UTF-8."); + archive_write_free(a); + return; + } + archive_write_free(a); r = archive_mstring_update_utf8(NULL, &mstr, utf8_string); /* All conversions should have succeeded */ assertEqualInt(0, r); assertEqualInt(AES_SET_MBS | AES_SET_WCS | AES_SET_UTF8, mstr.aes_set); assertEqualString(utf8_string, mstr.aes_utf8.s); assertEqualString(koi8_string, mstr.aes_mbs.s); #if defined(_WIN32) && !defined(__CYGWIN__) assertEqualWString(wcs_string, mstr.aes_wcs.s); #else /* No guarantee of how WCS strings behave, however this test test is * primarily meant for Windows */ (void)wcs_string; #endif archive_mstring_clean(&mstr); } diff --git a/contrib/libarchive/libarchive/test/test_entry.c b/contrib/libarchive/libarchive/test/test_entry.c index 9b21b83ecdfb..cff9c5c86efc 100644 --- a/contrib/libarchive/libarchive/test/test_entry.c +++ b/contrib/libarchive/libarchive/test/test_entry.c @@ -1,979 +1,990 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" #include #ifdef HAVE_LINUX_FS_H #include /* for Linux file flags */ #endif #ifndef HAVE_WCSCPY static wchar_t * wcscpy(wchar_t *s1, const wchar_t *s2) { wchar_t *dest = s1; while ((*s1 = *s2) != L'\0') ++s1, ++s2; return dest; } #endif /* * Most of these tests are system-independent, though a few depend on * features of the local system. Such tests are conditionalized on * the platform name. On unsupported platforms, only the * system-independent features will be tested. * * No, I don't want to use config.h in the test files because I want * the tests to also serve as a check on the correctness of config.h. * A mis-configured library build should cause tests to fail. */ DEFINE_TEST(test_entry) { char buff[128]; wchar_t wbuff[128]; struct stat st; struct archive_entry *e, *e2; const struct stat *pst; unsigned long set, clear; /* For fflag testing. */ int type, permset, tag, qual; /* For ACL testing. */ const char *name; /* For ACL testing. */ const char *xname; /* For xattr tests. */ const void *xval; /* For xattr tests. */ size_t xsize; /* For xattr tests. */ wchar_t wc; long l; int i; assert((e = archive_entry_new()) != NULL); /* * Verify that the AE_IF* defines match S_IF* defines * on this platform. See comments in archive_entry.h. */ #ifdef S_IFREG assertEqualInt(S_IFREG, AE_IFREG); #endif #ifdef S_IFLNK assertEqualInt(S_IFLNK, AE_IFLNK); #endif #ifdef S_IFSOCK assertEqualInt(S_IFSOCK, AE_IFSOCK); #endif #ifdef S_IFCHR assertEqualInt(S_IFCHR, AE_IFCHR); #endif /* Work around MinGW, which defines S_IFBLK wrong. */ /* sourceforge.net/tracker/?func=detail&atid=102435&aid=1942809&group_id=2435 */ #if defined(S_IFBLK) && !defined(_WIN32) assertEqualInt(S_IFBLK, AE_IFBLK); #endif #ifdef S_IFDIR assertEqualInt(S_IFDIR, AE_IFDIR); #endif #ifdef S_IFIFO assertEqualInt(S_IFIFO, AE_IFIFO); #endif /* * Basic set/read tests for all fields. * We should be able to set any field and read * back the same value. * * For methods that "copy" a string, we should be able * to overwrite the original passed-in string without * changing the value in the entry. * * The following tests are ordered alphabetically by the * name of the field. */ /* atime */ archive_entry_set_atime(e, 13579, 24680); assertEqualInt(archive_entry_atime(e), 13579); assertEqualInt(archive_entry_atime_nsec(e), 24680); archive_entry_set_atime(e, 13580, 1000000001L); assertEqualInt(archive_entry_atime(e), 13581); assertEqualInt(archive_entry_atime_nsec(e), 1); archive_entry_set_atime(e, 13580, -7); assertEqualInt(archive_entry_atime(e), 13579); assertEqualInt(archive_entry_atime_nsec(e), 999999993); archive_entry_unset_atime(e); assertEqualInt(archive_entry_atime(e), 0); assertEqualInt(archive_entry_atime_nsec(e), 0); assert(!archive_entry_atime_is_set(e)); /* birthtime */ archive_entry_set_birthtime(e, 17579, 24990); assertEqualInt(archive_entry_birthtime(e), 17579); assertEqualInt(archive_entry_birthtime_nsec(e), 24990); archive_entry_set_birthtime(e, 17580, 1234567890L); assertEqualInt(archive_entry_birthtime(e), 17581); assertEqualInt(archive_entry_birthtime_nsec(e), 234567890); archive_entry_set_birthtime(e, 17581, -24990); assertEqualInt(archive_entry_birthtime(e), 17580); assertEqualInt(archive_entry_birthtime_nsec(e), 999975010); archive_entry_unset_birthtime(e); assertEqualInt(archive_entry_birthtime(e), 0); assertEqualInt(archive_entry_birthtime_nsec(e), 0); assert(!archive_entry_birthtime_is_set(e)); /* ctime */ archive_entry_set_ctime(e, 13580, 24681); assertEqualInt(archive_entry_ctime(e), 13580); assertEqualInt(archive_entry_ctime_nsec(e), 24681); archive_entry_set_ctime(e, 13581, 2008182348L); assertEqualInt(archive_entry_ctime(e), 13583); assertEqualInt(archive_entry_ctime_nsec(e), 8182348); archive_entry_set_ctime(e, 13582, -24681); assertEqualInt(archive_entry_ctime(e), 13581); assertEqualInt(archive_entry_ctime_nsec(e), 999975319); archive_entry_unset_ctime(e); assertEqualInt(archive_entry_ctime(e), 0); assertEqualInt(archive_entry_ctime_nsec(e), 0); assert(!archive_entry_ctime_is_set(e)); /* dev */ assert(!archive_entry_dev_is_set(e)); archive_entry_set_dev(e, 235); assert(archive_entry_dev_is_set(e)); assertEqualInt(archive_entry_dev(e), 235); /* devmajor/devminor are tested specially below. */ /* filetype */ archive_entry_set_filetype(e, AE_IFREG); assertEqualInt(archive_entry_filetype(e), AE_IFREG); /* fflags are tested specially below */ /* gid */ archive_entry_set_gid(e, 204); assertEqualInt(archive_entry_gid(e), 204); /* gname */ archive_entry_set_gname(e, "group"); assertEqualString(archive_entry_gname(e), "group"); assertEqualString(archive_entry_gname_utf8(e), "group"); assertEqualWString(archive_entry_gname_w(e), L"group"); wcscpy(wbuff, L"wgroup"); archive_entry_copy_gname_w(e, wbuff); assertEqualWString(archive_entry_gname_w(e), L"wgroup"); memset(wbuff, 0, sizeof(wbuff)); assertEqualWString(archive_entry_gname_w(e), L"wgroup"); assertEqualString(archive_entry_gname_utf8(e), "wgroup"); assertEqualString(archive_entry_gname(e), "wgroup"); archive_entry_set_gname_utf8(e, "group"); assertEqualString(archive_entry_gname_utf8(e), "group"); assertEqualWString(archive_entry_gname_w(e), L"group"); assertEqualString(archive_entry_gname(e), "group"); archive_entry_update_gname_utf8(e, "group2"); assertEqualString(archive_entry_gname_utf8(e), "group2"); assertEqualWString(archive_entry_gname_w(e), L"group2"); assertEqualString(archive_entry_gname(e), "group2"); /* hardlink */ archive_entry_set_hardlink(e, "hardlinkname"); assertEqualString(archive_entry_hardlink(e), "hardlinkname"); assertEqualString(archive_entry_hardlink_utf8(e), "hardlinkname"); assertEqualWString(archive_entry_hardlink_w(e), L"hardlinkname"); strcpy(buff, "hardlinkname2"); archive_entry_copy_hardlink(e, buff); assertEqualString(archive_entry_hardlink(e), "hardlinkname2"); assertEqualWString(archive_entry_hardlink_w(e), L"hardlinkname2"); assertEqualString(archive_entry_hardlink_utf8(e), "hardlinkname2"); memset(buff, 0, sizeof(buff)); assertEqualString(archive_entry_hardlink(e), "hardlinkname2"); assertEqualString(archive_entry_hardlink_utf8(e), "hardlinkname2"); assertEqualWString(archive_entry_hardlink_w(e), L"hardlinkname2"); archive_entry_copy_hardlink(e, NULL); assertEqualString(archive_entry_hardlink(e), NULL); assertEqualWString(archive_entry_hardlink_w(e), NULL); assertEqualString(archive_entry_hardlink_utf8(e), NULL); wcscpy(wbuff, L"whardlink"); archive_entry_copy_hardlink_w(e, wbuff); assertEqualWString(archive_entry_hardlink_w(e), L"whardlink"); assertEqualString(archive_entry_hardlink_utf8(e), "whardlink"); assertEqualString(archive_entry_hardlink(e), "whardlink"); memset(wbuff, 0, sizeof(wbuff)); assertEqualWString(archive_entry_hardlink_w(e), L"whardlink"); archive_entry_copy_hardlink_w(e, NULL); assertEqualString(archive_entry_hardlink(e), NULL); assertEqualWString(archive_entry_hardlink_w(e), NULL); archive_entry_set_hardlink_utf8(e, "hardlinkname"); assertEqualString(archive_entry_hardlink_utf8(e), "hardlinkname"); assertEqualWString(archive_entry_hardlink_w(e), L"hardlinkname"); assertEqualString(archive_entry_hardlink(e), "hardlinkname"); archive_entry_update_hardlink_utf8(e, "hardlinkname2"); assertEqualString(archive_entry_hardlink_utf8(e), "hardlinkname2"); assertEqualWString(archive_entry_hardlink_w(e), L"hardlinkname2"); assertEqualString(archive_entry_hardlink(e), "hardlinkname2"); /* ino */ assert(!archive_entry_ino_is_set(e)); archive_entry_set_ino(e, 8593); assert(archive_entry_ino_is_set(e)); assertEqualInt(archive_entry_ino(e), 8593); assertEqualInt(archive_entry_ino64(e), 8593); archive_entry_set_ino64(e, 8594); assert(archive_entry_ino_is_set(e)); assertEqualInt(archive_entry_ino(e), 8594); assertEqualInt(archive_entry_ino64(e), 8594); /* link */ archive_entry_set_hardlink(e, "hardlinkname"); archive_entry_set_symlink(e, NULL); archive_entry_set_link(e, "link"); assertEqualString(archive_entry_hardlink(e), "link"); assertEqualString(archive_entry_symlink(e), NULL); archive_entry_copy_link(e, "link2"); assertEqualString(archive_entry_hardlink(e), "link2"); assertEqualString(archive_entry_symlink(e), NULL); archive_entry_copy_link_w(e, L"link3"); assertEqualString(archive_entry_hardlink(e), "link3"); assertEqualString(archive_entry_symlink(e), NULL); archive_entry_set_hardlink(e, NULL); archive_entry_set_symlink(e, "symlink"); archive_entry_set_link(e, "link"); assertEqualString(archive_entry_hardlink(e), NULL); assertEqualString(archive_entry_symlink(e), "link"); archive_entry_copy_link(e, "link2"); assertEqualString(archive_entry_hardlink(e), NULL); assertEqualString(archive_entry_symlink(e), "link2"); archive_entry_copy_link_w(e, L"link3"); assertEqualString(archive_entry_hardlink(e), NULL); assertEqualString(archive_entry_symlink(e), "link3"); /* Arbitrarily override symlink if both hardlink and symlink set. */ archive_entry_set_hardlink(e, "hardlink"); archive_entry_set_symlink(e, "symlink"); archive_entry_set_link(e, "link"); assertEqualString(archive_entry_hardlink(e), NULL); assertEqualString(archive_entry_symlink(e), "link"); /* mode */ archive_entry_set_mode(e, 0123456); assertEqualInt(archive_entry_mode(e), 0123456); /* mtime */ archive_entry_set_mtime(e, 13581, 24682); assertEqualInt(archive_entry_mtime(e), 13581); assertEqualInt(archive_entry_mtime_nsec(e), 24682); archive_entry_set_mtime(e, 13582, 1358297468); assertEqualInt(archive_entry_mtime(e), 13583); assertEqualInt(archive_entry_mtime_nsec(e), 358297468); archive_entry_set_mtime(e, 13583, -24682); assertEqualInt(archive_entry_mtime(e), 13582); assertEqualInt(archive_entry_mtime_nsec(e), 999975318); archive_entry_unset_mtime(e); assertEqualInt(archive_entry_mtime(e), 0); assertEqualInt(archive_entry_mtime_nsec(e), 0); assert(!archive_entry_mtime_is_set(e)); /* nlink */ archive_entry_set_nlink(e, 736); assertEqualInt(archive_entry_nlink(e), 736); /* pathname */ archive_entry_set_pathname(e, "path"); assertEqualString(archive_entry_pathname(e), "path"); assertEqualString(archive_entry_pathname_utf8(e), "path"); assertEqualWString(archive_entry_pathname_w(e), L"path"); archive_entry_set_pathname(e, "path"); assertEqualString(archive_entry_pathname(e), "path"); assertEqualWString(archive_entry_pathname_w(e), L"path"); assertEqualString(archive_entry_pathname_utf8(e), "path"); strcpy(buff, "path2"); archive_entry_copy_pathname(e, buff); assertEqualString(archive_entry_pathname(e), "path2"); assertEqualWString(archive_entry_pathname_w(e), L"path2"); assertEqualString(archive_entry_pathname_utf8(e), "path2"); memset(buff, 0, sizeof(buff)); assertEqualString(archive_entry_pathname(e), "path2"); assertEqualString(archive_entry_pathname_utf8(e), "path2"); assertEqualWString(archive_entry_pathname_w(e), L"path2"); wcscpy(wbuff, L"wpath"); archive_entry_copy_pathname_w(e, wbuff); assertEqualWString(archive_entry_pathname_w(e), L"wpath"); assertEqualString(archive_entry_pathname_utf8(e), "wpath"); assertEqualString(archive_entry_pathname(e), "wpath"); memset(wbuff, 0, sizeof(wbuff)); assertEqualWString(archive_entry_pathname_w(e), L"wpath"); assertEqualString(archive_entry_pathname(e), "wpath"); assertEqualString(archive_entry_pathname_utf8(e), "wpath"); archive_entry_set_pathname_utf8(e, "path"); assertEqualWString(archive_entry_pathname_w(e), L"path"); assertEqualString(archive_entry_pathname(e), "path"); assertEqualString(archive_entry_pathname_utf8(e), "path"); archive_entry_update_pathname_utf8(e, "path2"); assertEqualWString(archive_entry_pathname_w(e), L"path2"); assertEqualString(archive_entry_pathname(e), "path2"); assertEqualString(archive_entry_pathname_utf8(e), "path2"); /* rdev */ archive_entry_set_rdev(e, 532); assertEqualInt(archive_entry_rdev(e), 532); /* rdevmajor/rdevminor are tested specially below. */ /* size */ archive_entry_set_size(e, 987654321); assertEqualInt(archive_entry_size(e), 987654321); archive_entry_unset_size(e); assertEqualInt(archive_entry_size(e), 0); assert(!archive_entry_size_is_set(e)); /* sourcepath */ archive_entry_copy_sourcepath(e, "path1"); assertEqualString(archive_entry_sourcepath(e), "path1"); /* symlink */ archive_entry_set_symlink(e, "symlinkname"); assertEqualString(archive_entry_symlink(e), "symlinkname"); assertEqualString(archive_entry_symlink_utf8(e), "symlinkname"); assertEqualWString(archive_entry_symlink_w(e), L"symlinkname"); strcpy(buff, "symlinkname2"); archive_entry_copy_symlink(e, buff); assertEqualString(archive_entry_symlink(e), "symlinkname2"); assertEqualWString(archive_entry_symlink_w(e), L"symlinkname2"); assertEqualString(archive_entry_symlink_utf8(e), "symlinkname2"); memset(buff, 0, sizeof(buff)); assertEqualString(archive_entry_symlink(e), "symlinkname2"); assertEqualString(archive_entry_symlink_utf8(e), "symlinkname2"); assertEqualWString(archive_entry_symlink_w(e), L"symlinkname2"); archive_entry_copy_symlink_w(e, NULL); assertEqualWString(archive_entry_symlink_w(e), NULL); assertEqualString(archive_entry_symlink(e), NULL); assertEqualString(archive_entry_symlink_utf8(e), NULL); archive_entry_copy_symlink_w(e, L"wsymlink"); assertEqualWString(archive_entry_symlink_w(e), L"wsymlink"); assertEqualString(archive_entry_symlink_utf8(e), "wsymlink"); assertEqualString(archive_entry_symlink(e), "wsymlink"); archive_entry_copy_symlink(e, NULL); assertEqualWString(archive_entry_symlink_w(e), NULL); assertEqualString(archive_entry_symlink(e), NULL); assertEqualString(archive_entry_symlink_utf8(e), NULL); archive_entry_set_symlink_utf8(e, "symlinkname"); assertEqualWString(archive_entry_symlink_w(e), L"symlinkname"); assertEqualString(archive_entry_symlink(e), "symlinkname"); assertEqualString(archive_entry_symlink_utf8(e), "symlinkname"); archive_entry_update_symlink_utf8(e, "symlinkname2"); assertEqualWString(archive_entry_symlink_w(e), L"symlinkname2"); assertEqualString(archive_entry_symlink(e), "symlinkname2"); assertEqualString(archive_entry_symlink_utf8(e), "symlinkname2"); /* uid */ archive_entry_set_uid(e, 83); assertEqualInt(archive_entry_uid(e), 83); /* uname */ archive_entry_set_uname(e, "user"); assertEqualString(archive_entry_uname(e), "user"); assertEqualString(archive_entry_uname_utf8(e), "user"); assertEqualWString(archive_entry_uname_w(e), L"user"); wcscpy(wbuff, L"wuser"); archive_entry_copy_uname_w(e, wbuff); assertEqualWString(archive_entry_uname_w(e), L"wuser"); memset(wbuff, 0, sizeof(wbuff)); assertEqualWString(archive_entry_uname_w(e), L"wuser"); assertEqualString(archive_entry_uname_utf8(e), "wuser"); assertEqualString(archive_entry_uname(e), "wuser"); archive_entry_set_uname_utf8(e, "user"); assertEqualString(archive_entry_uname_utf8(e), "user"); assertEqualWString(archive_entry_uname_w(e), L"user"); assertEqualString(archive_entry_uname(e), "user"); archive_entry_set_uname_utf8(e, "user"); assertEqualWString(archive_entry_uname_w(e), L"user"); assertEqualString(archive_entry_uname(e), "user"); assertEqualString(archive_entry_uname_utf8(e), "user"); archive_entry_update_uname_utf8(e, "user2"); assertEqualWString(archive_entry_uname_w(e), L"user2"); assertEqualString(archive_entry_uname(e), "user2"); assertEqualString(archive_entry_uname_utf8(e), "user2"); /* Test fflags interface. */ archive_entry_set_fflags(e, 0x55, 0xAA); archive_entry_fflags(e, &set, &clear); failure("Testing set/get of fflags data."); assertEqualInt(set, 0x55); failure("Testing set/get of fflags data."); assertEqualInt(clear, 0xAA); #ifdef __FreeBSD__ /* Converting fflags bitmap to string is currently system-dependent. */ /* TODO: Make this system-independent. */ assertEqualString(archive_entry_fflags_text(e), "uappnd,nouchg,nodump,noopaque,uunlnk,nosystem"); #endif #if defined(__FreeBSD__) || defined(__APPLE__) /* Test archive_entry_copy_fflags_text_w() */ archive_entry_copy_fflags_text_w(e, L" ,nouappnd, nouchg, dump,hidden"); archive_entry_fflags(e, &set, &clear); assertEqualInt(UF_HIDDEN, set); assertEqualInt(UF_NODUMP | UF_IMMUTABLE | UF_APPEND, clear); /* Test archive_entry_copy_fflags_text() */ archive_entry_copy_fflags_text(e, " ,nouappnd, nouchg, dump,hidden"); archive_entry_fflags(e, &set, &clear); assertEqualInt(UF_HIDDEN, set); assertEqualInt(UF_NODUMP | UF_IMMUTABLE | UF_APPEND, clear); #elif defined(_WIN32) && !defined(__CYGWIN__) archive_entry_copy_fflags_text_w(e, L"rdonly,hidden,nosystem"); archive_entry_fflags(e, &set, &clear); assertEqualInt(FILE_ATTRIBUTE_READONLY | FILE_ATTRIBUTE_HIDDEN, set); assertEqualInt(FILE_ATTRIBUTE_SYSTEM, clear); archive_entry_copy_fflags_text(e, "rdonly,hidden,nosystem"); archive_entry_fflags(e, &set, &clear); assertEqualInt(FILE_ATTRIBUTE_READONLY | FILE_ATTRIBUTE_HIDDEN, set); assertEqualInt(FILE_ATTRIBUTE_SYSTEM, clear); #elif defined FS_IOC_GETFLAGS /* Linux */ archive_entry_copy_fflags_text_w(e, L"sappnd,schg,dump,noundel"); archive_entry_fflags(e, &set, &clear); assertEqualInt(FS_APPEND_FL | FS_IMMUTABLE_FL, set); assertEqualInt(FS_NODUMP_FL | FS_UNRM_FL, clear); archive_entry_copy_fflags_text(e, "sappnd,schg,dump,noundel"); archive_entry_fflags(e, &set, &clear); assertEqualInt(FS_APPEND_FL | FS_IMMUTABLE_FL, set); assertEqualInt(FS_NODUMP_FL | FS_UNRM_FL, clear); #endif /* See test_acl_basic.c for tests of ACL set/get consistency. */ /* Test xattrs set/get consistency. */ archive_entry_xattr_add_entry(e, "xattr1", "xattrvalue1", 12); assertEqualInt(1, archive_entry_xattr_reset(e)); assertEqualInt(0, archive_entry_xattr_next(e, &xname, &xval, &xsize)); assertEqualString(xname, "xattr1"); assertEqualString(xval, "xattrvalue1"); assertEqualInt((int)xsize, 12); assertEqualInt(1, archive_entry_xattr_count(e)); assertEqualInt(ARCHIVE_WARN, archive_entry_xattr_next(e, &xname, &xval, &xsize)); assertEqualString(xname, NULL); assertEqualString(xval, NULL); assertEqualInt((int)xsize, 0); archive_entry_xattr_clear(e); assertEqualInt(0, archive_entry_xattr_reset(e)); assertEqualInt(ARCHIVE_WARN, archive_entry_xattr_next(e, &xname, &xval, &xsize)); assertEqualString(xname, NULL); assertEqualString(xval, NULL); assertEqualInt((int)xsize, 0); archive_entry_xattr_add_entry(e, "xattr1", "xattrvalue1", 12); assertEqualInt(1, archive_entry_xattr_reset(e)); archive_entry_xattr_add_entry(e, "xattr2", "xattrvalue2", 12); assertEqualInt(2, archive_entry_xattr_reset(e)); assertEqualInt(0, archive_entry_xattr_next(e, &xname, &xval, &xsize)); assertEqualInt(0, archive_entry_xattr_next(e, &xname, &xval, &xsize)); assertEqualInt(ARCHIVE_WARN, archive_entry_xattr_next(e, &xname, &xval, &xsize)); assertEqualString(xname, NULL); assertEqualString(xval, NULL); assertEqualInt((int)xsize, 0); /* * Test clone() implementation. */ /* Set values in 'e' */ archive_entry_clear(e); archive_entry_set_atime(e, 13579, 24680); archive_entry_set_birthtime(e, 13779, 24990); archive_entry_set_ctime(e, 13580, 24681); archive_entry_set_dev(e, 235); archive_entry_set_fflags(e, 0x55, 0xAA); archive_entry_set_gid(e, 204); archive_entry_set_gname(e, "group"); archive_entry_set_hardlink(e, "hardlinkname"); archive_entry_set_ino(e, 8593); archive_entry_set_mode(e, 0123456); archive_entry_set_mtime(e, 13581, 24682); archive_entry_set_nlink(e, 736); archive_entry_set_pathname(e, "path"); archive_entry_set_rdev(e, 532); archive_entry_set_size(e, 987654321); archive_entry_copy_sourcepath(e, "source"); archive_entry_set_uid(e, 83); archive_entry_set_uname(e, "user"); /* Add an ACL entry. */ archive_entry_acl_add_entry(e, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_READ, ARCHIVE_ENTRY_ACL_USER, 77, "user77"); /* Add an extended attribute. */ archive_entry_xattr_add_entry(e, "xattr1", "xattrvalue", 11); /* Make a clone. */ e2 = archive_entry_clone(e); /* Clone should have same contents. */ assertEqualInt(archive_entry_atime(e2), 13579); assertEqualInt(archive_entry_atime_nsec(e2), 24680); assertEqualInt(archive_entry_birthtime(e2), 13779); assertEqualInt(archive_entry_birthtime_nsec(e2), 24990); assertEqualInt(archive_entry_ctime(e2), 13580); assertEqualInt(archive_entry_ctime_nsec(e2), 24681); assertEqualInt(archive_entry_dev(e2), 235); archive_entry_fflags(e, &set, &clear); assertEqualInt(clear, 0xAA); assertEqualInt(set, 0x55); assertEqualInt(archive_entry_gid(e2), 204); assertEqualString(archive_entry_gname(e2), "group"); assertEqualString(archive_entry_hardlink(e2), "hardlinkname"); assertEqualInt(archive_entry_ino(e2), 8593); assertEqualInt(archive_entry_mode(e2), 0123456); assertEqualInt(archive_entry_mtime(e2), 13581); assertEqualInt(archive_entry_mtime_nsec(e2), 24682); assertEqualInt(archive_entry_nlink(e2), 736); assertEqualString(archive_entry_pathname(e2), "path"); assertEqualInt(archive_entry_rdev(e2), 532); assertEqualInt(archive_entry_size(e2), 987654321); assertEqualString(archive_entry_sourcepath(e2), "source"); assertEqualString(archive_entry_symlink(e2), NULL); assertEqualInt(archive_entry_uid(e2), 83); assertEqualString(archive_entry_uname(e2), "user"); /* Verify ACL was copied. */ assertEqualInt(4, archive_entry_acl_reset(e2, ARCHIVE_ENTRY_ACL_TYPE_ACCESS)); /* First three are standard permission bits. */ assertEqualInt(0, archive_entry_acl_next(e2, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, &type, &permset, &tag, &qual, &name)); assertEqualInt(type, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); assertEqualInt(permset, 4); assertEqualInt(tag, ARCHIVE_ENTRY_ACL_USER_OBJ); assertEqualInt(qual, -1); assertEqualString(name, NULL); assertEqualInt(0, archive_entry_acl_next(e2, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, &type, &permset, &tag, &qual, &name)); assertEqualInt(type, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); assertEqualInt(permset, 5); assertEqualInt(tag, ARCHIVE_ENTRY_ACL_GROUP_OBJ); assertEqualInt(qual, -1); assertEqualString(name, NULL); assertEqualInt(0, archive_entry_acl_next(e2, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, &type, &permset, &tag, &qual, &name)); assertEqualInt(type, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); assertEqualInt(permset, 6); assertEqualInt(tag, ARCHIVE_ENTRY_ACL_OTHER); assertEqualInt(qual, -1); assertEqualString(name, NULL); /* Fourth is custom one. */ assertEqualInt(0, archive_entry_acl_next(e2, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, &type, &permset, &tag, &qual, &name)); assertEqualInt(type, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); assertEqualInt(permset, ARCHIVE_ENTRY_ACL_READ); assertEqualInt(tag, ARCHIVE_ENTRY_ACL_USER); assertEqualInt(qual, 77); assertEqualString(name, "user77"); /* Verify xattr was copied. */ assertEqualInt(1, archive_entry_xattr_reset(e2)); assertEqualInt(0, archive_entry_xattr_next(e2, &xname, &xval, &xsize)); assertEqualString(xname, "xattr1"); assertEqualString(xval, "xattrvalue"); assertEqualInt((int)xsize, 11); assertEqualInt(ARCHIVE_WARN, archive_entry_xattr_next(e2, &xname, &xval, &xsize)); assertEqualString(xname, NULL); assertEqualString(xval, NULL); assertEqualInt((int)xsize, 0); /* Change the original */ archive_entry_set_atime(e, 13580, 24690); archive_entry_set_birthtime(e, 13980, 24999); archive_entry_set_ctime(e, 13590, 24691); archive_entry_set_dev(e, 245); archive_entry_set_fflags(e, 0x85, 0xDA); archive_entry_set_filetype(e, AE_IFLNK); archive_entry_set_gid(e, 214); archive_entry_set_gname(e, "grouper"); archive_entry_set_hardlink(e, "hardlinkpath"); archive_entry_set_ino(e, 8763); archive_entry_set_mode(e, 0123654); archive_entry_set_mtime(e, 18351, 28642); archive_entry_set_nlink(e, 73); archive_entry_set_pathname(e, "pathest"); archive_entry_set_rdev(e, 132); archive_entry_set_size(e, 987456321); archive_entry_copy_sourcepath(e, "source2"); archive_entry_set_symlink(e, "symlinkpath"); archive_entry_set_uid(e, 93); archive_entry_set_uname(e, "username"); archive_entry_acl_clear(e); archive_entry_xattr_clear(e); /* Clone should still have same contents. */ assertEqualInt(archive_entry_atime(e2), 13579); assertEqualInt(archive_entry_atime_nsec(e2), 24680); assertEqualInt(archive_entry_birthtime(e2), 13779); assertEqualInt(archive_entry_birthtime_nsec(e2), 24990); assertEqualInt(archive_entry_ctime(e2), 13580); assertEqualInt(archive_entry_ctime_nsec(e2), 24681); assertEqualInt(archive_entry_dev(e2), 235); archive_entry_fflags(e2, &set, &clear); assertEqualInt(clear, 0xAA); assertEqualInt(set, 0x55); assertEqualInt(archive_entry_gid(e2), 204); assertEqualString(archive_entry_gname(e2), "group"); assertEqualString(archive_entry_hardlink(e2), "hardlinkname"); assertEqualInt(archive_entry_ino(e2), 8593); assertEqualInt(archive_entry_mode(e2), 0123456); assertEqualInt(archive_entry_mtime(e2), 13581); assertEqualInt(archive_entry_mtime_nsec(e2), 24682); assertEqualInt(archive_entry_nlink(e2), 736); assertEqualString(archive_entry_pathname(e2), "path"); assertEqualInt(archive_entry_rdev(e2), 532); assertEqualInt(archive_entry_size(e2), 987654321); assertEqualString(archive_entry_sourcepath(e2), "source"); assertEqualString(archive_entry_symlink(e2), NULL); assertEqualInt(archive_entry_uid(e2), 83); assertEqualString(archive_entry_uname(e2), "user"); /* Verify ACL was unchanged. */ assertEqualInt(4, archive_entry_acl_reset(e2, ARCHIVE_ENTRY_ACL_TYPE_ACCESS)); /* First three are standard permission bits. */ assertEqualInt(0, archive_entry_acl_next(e2, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, &type, &permset, &tag, &qual, &name)); assertEqualInt(type, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); assertEqualInt(permset, 4); assertEqualInt(tag, ARCHIVE_ENTRY_ACL_USER_OBJ); assertEqualInt(qual, -1); assertEqualString(name, NULL); assertEqualInt(0, archive_entry_acl_next(e2, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, &type, &permset, &tag, &qual, &name)); assertEqualInt(type, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); assertEqualInt(permset, 5); assertEqualInt(tag, ARCHIVE_ENTRY_ACL_GROUP_OBJ); assertEqualInt(qual, -1); assertEqualString(name, NULL); assertEqualInt(0, archive_entry_acl_next(e2, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, &type, &permset, &tag, &qual, &name)); assertEqualInt(type, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); assertEqualInt(permset, 6); assertEqualInt(tag, ARCHIVE_ENTRY_ACL_OTHER); assertEqualInt(qual, -1); assertEqualString(name, NULL); /* Fourth is custom one. */ assertEqualInt(0, archive_entry_acl_next(e2, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, &type, &permset, &tag, &qual, &name)); assertEqualInt(type, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); assertEqualInt(permset, ARCHIVE_ENTRY_ACL_READ); assertEqualInt(tag, ARCHIVE_ENTRY_ACL_USER); assertEqualInt(qual, 77); assertEqualString(name, "user77"); assertEqualInt(1, archive_entry_acl_next(e2, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, &type, &permset, &tag, &qual, &name)); assertEqualInt(type, 0); assertEqualInt(permset, 0); assertEqualInt(tag, 0); assertEqualInt(qual, -1); assertEqualString(name, NULL); /* Verify xattr was unchanged. */ assertEqualInt(1, archive_entry_xattr_reset(e2)); /* Release clone. */ archive_entry_free(e2); /* Verify that symlink is copied over by `clone` */ archive_entry_set_symlink(e, "symlinkpath"); e2 = archive_entry_clone(e); assertEqualString(archive_entry_hardlink(e2), NULL); assertEqualString(archive_entry_symlink(e2), "symlinkpath"); archive_entry_free(e2); /* * Test clear() implementation. */ archive_entry_clear(e); assertEqualInt(archive_entry_atime(e), 0); assertEqualInt(archive_entry_atime_nsec(e), 0); assertEqualInt(archive_entry_birthtime(e), 0); assertEqualInt(archive_entry_birthtime_nsec(e), 0); assertEqualInt(archive_entry_ctime(e), 0); assertEqualInt(archive_entry_ctime_nsec(e), 0); assertEqualInt(archive_entry_dev(e), 0); archive_entry_fflags(e, &set, &clear); assertEqualInt(clear, 0); assertEqualInt(set, 0); assertEqualInt(archive_entry_filetype(e), 0); assertEqualInt(archive_entry_gid(e), 0); assertEqualString(archive_entry_gname(e), NULL); assertEqualString(archive_entry_hardlink(e), NULL); assertEqualInt(archive_entry_ino(e), 0); assertEqualInt(archive_entry_mode(e), 0); assertEqualInt(archive_entry_mtime(e), 0); assertEqualInt(archive_entry_mtime_nsec(e), 0); assertEqualInt(archive_entry_nlink(e), 0); assertEqualString(archive_entry_pathname(e), NULL); assertEqualInt(archive_entry_rdev(e), 0); assertEqualInt(archive_entry_size(e), 0); assertEqualString(archive_entry_symlink(e), NULL); assertEqualInt(archive_entry_uid(e), 0); assertEqualString(archive_entry_uname(e), NULL); /* ACLs should be cleared. */ assertEqualInt(archive_entry_acl_count(e, ARCHIVE_ENTRY_ACL_TYPE_ACCESS), 0); assertEqualInt(archive_entry_acl_count(e, ARCHIVE_ENTRY_ACL_TYPE_DEFAULT), 0); /* Extended attributes should be cleared. */ assertEqualInt(archive_entry_xattr_count(e), 0); /* * Test archive_entry_copy_stat(). */ memset(&st, 0, sizeof(st)); /* Set all of the standard 'struct stat' fields. */ st.st_atime = 456789; st.st_ctime = 345678; st.st_dev = 123; st.st_gid = 34; st.st_ino = 234; st.st_mode = 077777; st.st_mtime = 234567; st.st_nlink = 345; st.st_size = 123456789; st.st_uid = 23; #ifdef __FreeBSD__ /* On FreeBSD, high-res timestamp data should come through. */ st.st_atimespec.tv_nsec = 6543210; st.st_ctimespec.tv_nsec = 5432109; st.st_mtimespec.tv_nsec = 3210987; st.st_birthtimespec.tv_nsec = 7459386; #endif /* Copy them into the entry. */ archive_entry_copy_stat(e, &st); /* Read each one back separately and compare. */ assertEqualInt(archive_entry_atime(e), 456789); assertEqualInt(archive_entry_ctime(e), 345678); assertEqualInt(archive_entry_dev(e), 123); assertEqualInt(archive_entry_gid(e), 34); assertEqualInt(archive_entry_ino(e), 234); assertEqualInt(archive_entry_mode(e), 077777); assertEqualInt(archive_entry_mtime(e), 234567); assertEqualInt(archive_entry_nlink(e), 345); assertEqualInt(archive_entry_size(e), 123456789); assertEqualInt(archive_entry_uid(e), 23); #if __FreeBSD__ /* On FreeBSD, high-res timestamp data should come through. */ assertEqualInt(archive_entry_atime_nsec(e), 6543210); assertEqualInt(archive_entry_ctime_nsec(e), 5432109); assertEqualInt(archive_entry_mtime_nsec(e), 3210987); assertEqualInt(archive_entry_birthtime_nsec(e), 7459386); #endif /* * Test archive_entry_stat(). */ /* First, clear out any existing stat data. */ memset(&st, 0, sizeof(st)); archive_entry_copy_stat(e, &st); /* Set a bunch of fields individually. */ archive_entry_set_atime(e, 456789, 321); archive_entry_set_ctime(e, 345678, 432); archive_entry_set_dev(e, 123); archive_entry_set_gid(e, 34); archive_entry_set_ino(e, 234); archive_entry_set_mode(e, 012345); archive_entry_set_mode(e, 012345); archive_entry_set_mtime(e, 234567, 543); archive_entry_set_nlink(e, 345); archive_entry_set_size(e, 123456789); archive_entry_set_uid(e, 23); /* Retrieve a stat structure. */ assert((pst = archive_entry_stat(e)) != NULL); if (pst == NULL) return; /* Check that the values match. */ assertEqualInt(pst->st_atime, 456789); assertEqualInt(pst->st_ctime, 345678); assertEqualInt(pst->st_dev, 123); assertEqualInt(pst->st_gid, 34); assertEqualInt(pst->st_ino, 234); assertEqualInt(pst->st_mode, 012345); assertEqualInt(pst->st_mtime, 234567); assertEqualInt(pst->st_nlink, 345); assertEqualInt(pst->st_size, 123456789); assertEqualInt(pst->st_uid, 23); #ifdef __FreeBSD__ /* On FreeBSD, high-res timestamp data should come through. */ assertEqualInt(pst->st_atimespec.tv_nsec, 321); assertEqualInt(pst->st_ctimespec.tv_nsec, 432); assertEqualInt(pst->st_mtimespec.tv_nsec, 543); #endif /* Changing any one value should update struct stat. */ archive_entry_set_atime(e, 456788, 0); assert((pst = archive_entry_stat(e)) != NULL); if (pst == NULL) return; assertEqualInt(pst->st_atime, 456788); archive_entry_set_ctime(e, 345677, 431); assert((pst = archive_entry_stat(e)) != NULL); if (pst == NULL) return; assertEqualInt(pst->st_ctime, 345677); archive_entry_set_dev(e, 122); assert((pst = archive_entry_stat(e)) != NULL); if (pst == NULL) return; assertEqualInt(pst->st_dev, 122); archive_entry_set_gid(e, 33); assert((pst = archive_entry_stat(e)) != NULL); if (pst == NULL) return; assertEqualInt(pst->st_gid, 33); archive_entry_set_ino(e, 233); assert((pst = archive_entry_stat(e)) != NULL); if (pst == NULL) return; assertEqualInt(pst->st_ino, 233); archive_entry_set_mode(e, 012344); assert((pst = archive_entry_stat(e)) != NULL); if (pst == NULL) return; assertEqualInt(pst->st_mode, 012344); archive_entry_set_mtime(e, 234566, 542); assert((pst = archive_entry_stat(e)) != NULL); if (pst == NULL) return; assertEqualInt(pst->st_mtime, 234566); archive_entry_set_nlink(e, 344); assert((pst = archive_entry_stat(e)) != NULL); if (pst == NULL) return; assertEqualInt(pst->st_nlink, 344); archive_entry_set_size(e, 123456788); assert((pst = archive_entry_stat(e)) != NULL); if (pst == NULL) return; assertEqualInt(pst->st_size, 123456788); archive_entry_set_uid(e, 22); assert((pst = archive_entry_stat(e)) != NULL); if (pst == NULL) return; assertEqualInt(pst->st_uid, 22); + + /* Check behavior with large sizes. */ + archive_entry_set_size(e, INT64_MAX - 1); + assert((pst = archive_entry_stat(e)) != NULL); + if (pst == NULL) + return; + if (sizeof(pst->st_size) < sizeof(int64_t)) + assertEqualInt(pst->st_size, 0); + else + assertEqualInt(pst->st_size, INT64_MAX - 1); + /* We don't need to check high-res fields here. */ /* * Test dev/major/minor interfaces. Setting 'dev' or 'rdev' * should change the corresponding major/minor values, and * vice versa. * * The test here is system-specific because it assumes that * makedev(), major(), and minor() are defined in sys/stat.h. * I'm not too worried about it, though, because the code is * simple. If it works on FreeBSD, it's unlikely to be broken * anywhere else. Note: The functionality is present on every * platform even if these tests only run some places; * libarchive's more extensive configuration logic should find * the necessary definitions on every platform. */ #if __FreeBSD__ archive_entry_set_dev(e, 0x12345678); assertEqualInt(archive_entry_devmajor(e), major(0x12345678)); assertEqualInt(archive_entry_devminor(e), minor(0x12345678)); assertEqualInt(archive_entry_dev(e), 0x12345678); archive_entry_set_devmajor(e, 0xfe); archive_entry_set_devminor(e, 0xdcba98); assertEqualInt(archive_entry_devmajor(e), 0xfe); assertEqualInt(archive_entry_devminor(e), 0xdcba98); assertEqualInt(archive_entry_dev(e), makedev(0xfe, 0xdcba98)); archive_entry_set_rdev(e, 0x12345678); assertEqualInt(archive_entry_rdevmajor(e), major(0x12345678)); assertEqualInt(archive_entry_rdevminor(e), minor(0x12345678)); assertEqualInt(archive_entry_rdev(e), 0x12345678); archive_entry_set_rdevmajor(e, 0xfe); archive_entry_set_rdevminor(e, 0xdcba98); assertEqualInt(archive_entry_rdevmajor(e), 0xfe); assertEqualInt(archive_entry_rdevminor(e), 0xdcba98); assertEqualInt(archive_entry_rdev(e), makedev(0xfe, 0xdcba98)); #endif /* * Exercise the character-conversion logic, if we can. */ if (NULL == setlocale(LC_ALL, "en_US.UTF-8")) { skipping("Can't exercise charset-conversion logic without" " a suitable locale."); } else { /* A filename that cannot be converted to wide characters. */ archive_entry_copy_pathname(e, "abc\314\214mno\374xyz"); failure("Converting invalid chars to Unicode should fail."); assert(NULL == archive_entry_pathname_w(e)); /* failure("Converting invalid chars to UTF-8 should fail."); assert(NULL == archive_entry_pathname_utf8(e)); */ /* A group name that cannot be converted. */ archive_entry_copy_gname(e, "abc\314\214mno\374xyz"); failure("Converting invalid chars to Unicode should fail."); assert(NULL == archive_entry_gname_w(e)); /* A user name that cannot be converted. */ archive_entry_copy_uname(e, "abc\314\214mno\374xyz"); failure("Converting invalid chars to Unicode should fail."); assert(NULL == archive_entry_uname_w(e)); /* A hardlink target that cannot be converted. */ archive_entry_copy_hardlink(e, "abc\314\214mno\374xyz"); failure("Converting invalid chars to Unicode should fail."); assert(NULL == archive_entry_hardlink_w(e)); /* A symlink target that cannot be converted. */ archive_entry_copy_symlink(e, "abc\314\214mno\374xyz"); failure("Converting invalid chars to Unicode should fail."); assert(NULL == archive_entry_symlink_w(e)); } l = 0x12345678L; wc = (wchar_t)l; /* Wide character too big for UTF-8. */ if (NULL == setlocale(LC_ALL, "C") || (long)wc != l) { skipping("Testing charset conversion failure requires 32-bit wchar_t and support for \"C\" locale."); } else { /* * Build the string L"xxx\U12345678yyy\u5678zzz" without * using wcscpy or C99 \u#### syntax. */ name = "xxxAyyyBzzz"; for (i = 0; i < (int)strlen(name); ++i) wbuff[i] = name[i]; wbuff[3] = (wchar_t)0x12345678; wbuff[7] = (wchar_t)0x5678; /* A Unicode filename that cannot be converted to UTF-8. */ archive_entry_copy_pathname_w(e, wbuff); failure("Converting wide characters from Unicode should fail."); assertEqualString(NULL, archive_entry_pathname(e)); } /* Release the experimental entry. */ archive_entry_free(e); } diff --git a/contrib/libarchive/libarchive/test/test_read_filter_gzip_recursive.c b/contrib/libarchive/libarchive/test/test_read_filter_gzip_recursive.c index 0042a0511d5a..51b614b6c023 100644 --- a/contrib/libarchive/libarchive/test/test_read_filter_gzip_recursive.c +++ b/contrib/libarchive/libarchive/test/test_read_filter_gzip_recursive.c @@ -1,50 +1,50 @@ /*- * Copyright (c) 2003-2024 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" DEFINE_TEST(test_read_filter_gzip_recursive) { const char *name = "test_read_filter_gzip_recursive.gz"; struct archive *a; - if (!canGzip()) { - skipping("gzip not available"); + if (archive_zlib_version() == NULL) { + skipping("zlib not available"); return; } assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); extract_reference_file(name); assertEqualIntA(a, ARCHIVE_FATAL, archive_read_open_filename(a, name, 200)); /* Verify that the filter detection worked. */ assertEqualInt(archive_filter_code(a, 0), ARCHIVE_FILTER_GZIP); assertEqualString(archive_filter_name(a, 0), "gzip"); assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } diff --git a/contrib/libarchive/libarchive/test/test_read_format_7zip.c b/contrib/libarchive/libarchive/test/test_read_format_7zip.c index ad10ef06bbc5..3236fee2c9d3 100644 --- a/contrib/libarchive/libarchive/test/test_read_format_7zip.c +++ b/contrib/libarchive/libarchive/test/test_read_format_7zip.c @@ -1,1608 +1,1629 @@ /*- * Copyright (c) 2011 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" #if HAVE_LZMA_H #include #endif #if defined(_WIN32) && !defined(__CYGWIN__) #define close _close #define open _open #endif #define __LIBARCHIVE_BUILD /* * Extract a non-encoded file. * The header of the 7z archive files is not encoded. */ static void test_copy(int use_open_fd) { const char *refname = "test_read_format_7zip_copy.7z"; struct archive_entry *ae; struct archive *a; char buff[128]; int fd = -1; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); if (use_open_fd) { fd = open(refname, O_RDONLY | O_BINARY); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_fd(a, fd, 10240)); } else { assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); } /* Verify regular file1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file1", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(60, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assert(archive_read_has_encrypted_entries(a) > ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED); assertEqualInt(60, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, " ", 4); assertEqualInt(1, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); if (fd != -1) close(fd); } /* * An archive file has no entry. */ static void test_empty_archive(void) { const char *refname = "test_read_format_7zip_empty_archive.7z"; struct archive_entry *ae; struct archive *a; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualInt(0, archive_file_count(a)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } /* * An archive file has one empty file. It means there is no content * in the archive file except for a header. */ static void test_empty_file(void) { const char *refname = "test_read_format_7zip_empty_file.7z"; struct archive_entry *ae; struct archive *a; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); /* Verify regular empty. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("empty", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(0, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(1, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } /* * Extract an encoded file. * The header of the 7z archive files is not encoded. */ static void test_plain_header(const char *refname) { struct archive_entry *ae; struct archive *a; char buff[128]; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); /* Verify regular file1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file1", archive_entry_pathname(ae)); assertEqualInt(1322058763, archive_entry_mtime(ae)); assertEqualInt(2844, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(sizeof(buff), archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "The libarchive distribution ", 28); assertEqualInt(1, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } /* * Extract multi files. * The header of the 7z archive files is encoded with LZMA. */ static void test_extract_all_files(const char *refname) { struct archive_entry *ae; struct archive *a; char buff[128]; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); /* Verify regular file1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("dir1/file1", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(13, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(13, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\n", 13); /* Verify regular file2. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file2", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(26, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(26, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\n", 26); /* Verify regular file3. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file3", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(39, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(39, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\ncccccccccccc\n", 39); /* Verify regular file4. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file4", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(52, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(52, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\ncccccccccccc\ndddddddddddd\n", 52); /* Verify directory dir1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFDIR | 0755), archive_entry_mode(ae)); assertEqualString("dir1/", archive_entry_pathname(ae)); assertEqualInt(2764801, archive_entry_mtime(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(5, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } /* * Extract multi files. * Like test_extract_all_files, but with zstandard compression. */ static void test_extract_all_files_zstd(const char *refname) { struct archive_entry *ae; struct archive *a; char buff[128]; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); /* Verify directory dir1. Note that this comes before the dir1/file1 entry in recent versions of 7-Zip. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFDIR | 0755), archive_entry_mode(ae)); assertEqualString("dir1/", archive_entry_pathname(ae)); assertEqualInt(2764801, archive_entry_mtime(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); /* Verify regular file1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("dir1/file1", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(13, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(13, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\n", 13); /* Verify regular file2. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file2", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(26, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(26, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\n", 26); /* Verify regular file3. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file3", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(39, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(39, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\ncccccccccccc\n", 39); /* Verify regular file4. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file4", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(52, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(52, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\ncccccccccccc\ndddddddddddd\n", 52); assertEqualInt(5, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } /* * Extract file from an archives using ZSTD compression with and without BCJ. */ static void test_extract_file_zstd_bcj_nobjc(const char *refname) { struct archive_entry *ae; struct archive *a; char buff[4096]; uint32_t computed_crc = 0; uint32_t expected_crc = 0xbd66eebc; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); /* Verify regular file: hw. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0775), archive_entry_mode(ae)); assertEqualString("hw", archive_entry_pathname(ae)); assertEqualInt(1685913368, archive_entry_mtime(ae)); assertEqualInt(15952, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); for (;;) { la_ssize_t bytes_read = archive_read_data(a, buff, sizeof(buff)); assert(bytes_read >= 0); if (bytes_read == 0) break; computed_crc = bitcrc32(computed_crc, buff, bytes_read); } assertEqualInt(computed_crc, expected_crc); assertEqualInt(1, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } /* * Extract last file. * The header of the 7z archive files is encoded with LZMA. */ static void test_extract_last_file(const char *refname) { struct archive_entry *ae; struct archive *a; char buff[128]; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); /* Verify regular file1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("dir1/file1", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(13, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); /* Verify regular file2. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file2", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(26, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); /* Verify regular file3. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file3", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(39, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); /* Verify regular file4. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file4", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(52, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(52, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\ncccccccccccc\ndddddddddddd\n", 52); /* Verify directory dir1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFDIR | 0755), archive_entry_mode(ae)); assertEqualString("dir1/", archive_entry_pathname(ae)); assertEqualInt(2764801, archive_entry_mtime(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(5, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } /* * Extract a mixed archive file which has both LZMA and LZMA2 encoded files. * LZMA: file1, file2, file3, file4 * LZMA2: zfile1, zfile2, zfile3, zfile4 */ static void test_extract_all_files2(const char *refname) { struct archive_entry *ae; struct archive *a; char buff[128]; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); /* Verify regular file1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("dir1/file1", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(13, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(13, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\n", 13); /* Verify regular file2. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file2", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(26, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(26, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\n", 26); /* Verify regular file3. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file3", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(39, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(39, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\ncccccccccccc\n", 39); /* Verify regular file4. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file4", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(52, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(52, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\ncccccccccccc\ndddddddddddd\n", 52); /* Verify regular zfile1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("dir1/zfile1", archive_entry_pathname(ae)); assertEqualInt(5184001, archive_entry_mtime(ae)); assertEqualInt(13, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(13, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\n", 13); /* Verify regular zfile2. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("zfile2", archive_entry_pathname(ae)); assertEqualInt(5184001, archive_entry_mtime(ae)); assertEqualInt(26, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(26, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\n", 26); /* Verify regular zfile3. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("zfile3", archive_entry_pathname(ae)); assertEqualInt(5184001, archive_entry_mtime(ae)); assertEqualInt(39, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(39, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\ncccccccccccc\n", 39); /* Verify regular zfile4. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("zfile4", archive_entry_pathname(ae)); assertEqualInt(5184001, archive_entry_mtime(ae)); assertEqualInt(52, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(52, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "aaaaaaaaaaaa\nbbbbbbbbbbbb\ncccccccccccc\ndddddddddddd\n", 52); /* Verify directory dir1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFDIR | 0755), archive_entry_mode(ae)); assertEqualString("dir1/", archive_entry_pathname(ae)); assertEqualInt(2764801, archive_entry_mtime(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(9, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } /* * Extract a file compressed with DELTA + LZMA[12]. */ static void test_delta_lzma(const char *refname) { struct archive_entry *ae; struct archive *a; size_t remaining; ssize_t bytes; char buff[1024]; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); /* Verify regular file1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file1", archive_entry_pathname(ae)); assertEqualInt(172802, archive_entry_mtime(ae)); assertEqualInt(27627, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); remaining = (size_t)archive_entry_size(ae); while (remaining) { if (remaining < sizeof(buff)) assertEqualInt(remaining, bytes = archive_read_data(a, buff, sizeof(buff))); else assertEqualInt(sizeof(buff), bytes = archive_read_data(a, buff, sizeof(buff))); if (bytes > 0) remaining -= bytes; else break; } assertEqualInt(0, remaining); assertEqualInt(1, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } /* * Extract a file compressed with BCJ + LZMA2. */ static void test_bcj(const char *refname) { struct archive_entry *ae; struct archive *a; size_t remaining; ssize_t bytes; char buff[1024]; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); /* Verify regular x86exe. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0444), archive_entry_mode(ae) & ~0111); assertEqualString("x86exe", archive_entry_pathname(ae)); assertEqualInt(172802, archive_entry_mtime(ae)); assertEqualInt(27328, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); remaining = (size_t)archive_entry_size(ae); while (remaining) { if (remaining < sizeof(buff)) assertEqualInt(remaining, bytes = archive_read_data(a, buff, sizeof(buff))); else assertEqualInt(sizeof(buff), bytes = archive_read_data(a, buff, sizeof(buff))); if (bytes > 0) remaining -= bytes; else break; } assertEqualInt(0, remaining); assertEqualInt(1, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } /* * Extract a file compressed with PPMd. */ static void test_ppmd(void) { const char *refname = "test_read_format_7zip_ppmd.7z"; struct archive_entry *ae; struct archive *a; size_t remaining; ssize_t bytes; char buff[1024]; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); /* Verify regular file1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("ppmd_test.txt", archive_entry_pathname(ae)); assertEqualInt(1322464589, archive_entry_mtime(ae)); assertEqualInt(102400, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); remaining = (size_t)archive_entry_size(ae); while (remaining) { if (remaining < sizeof(buff)) assertEqualInt(remaining, bytes = archive_read_data(a, buff, sizeof(buff))); else assertEqualInt(sizeof(buff), bytes = archive_read_data(a, buff, sizeof(buff))); if (bytes > 0) remaining -= bytes; else break; } assertEqualInt(0, remaining); assertEqualInt(1, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } static void test_symname(void) { const char *refname = "test_read_format_7zip_symbolic_name.7z"; struct archive_entry *ae; struct archive *a; char buff[128]; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); /* Verify regular file1. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("file1", archive_entry_pathname(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(32, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(32, archive_read_data(a, buff, sizeof(buff))); assertEqualMem(buff, "hellohellohello\nhellohellohello\n", 32); /* Verify symbolic-link symlinkfile. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFLNK | 0755), archive_entry_mode(ae)); assertEqualString("symlinkfile", archive_entry_pathname(ae)); assertEqualString("file1", archive_entry_symlink(ae)); assertEqualInt(86401, archive_entry_mtime(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualInt(2, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_7ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip) { struct archive *a; assert((a = archive_read_new()) != NULL); /* Extracting with liblzma */ if (ARCHIVE_OK != archive_read_support_filter_xz(a)) { skipping("7zip:lzma decoding is not supported on this " "platform"); } else { test_symname(); test_extract_all_files("test_read_format_7zip_copy_2.7z"); test_extract_last_file("test_read_format_7zip_copy_2.7z"); test_extract_all_files2("test_read_format_7zip_lzma1_lzma2.7z"); test_bcj("test_read_format_7zip_bcj2_copy_lzma.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_bzip2) { struct archive *a; assert((a = archive_read_new()) != NULL); /* Extracting with libbzip2 */ if (ARCHIVE_OK != archive_read_support_filter_bzip2(a)) { skipping("7zip:bzip2 decoding is not supported on this platform"); } else { test_plain_header("test_read_format_7zip_bzip2.7z"); test_bcj("test_read_format_7zip_bcj_bzip2.7z"); test_bcj("test_read_format_7zip_bcj2_bzip2.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_from_fd) { test_copy(1);/* read a 7zip file from a file descriptor. */ } DEFINE_TEST(test_read_format_7zip_copy) { test_copy(0); test_bcj("test_read_format_7zip_bcj_copy.7z"); test_bcj("test_read_format_7zip_bcj2_copy_1.7z"); test_bcj("test_read_format_7zip_bcj2_copy_2.7z"); } DEFINE_TEST(test_read_format_7zip_deflate) { struct archive *a; assert((a = archive_read_new()) != NULL); /* Extracting with libz */ if (ARCHIVE_OK != archive_read_support_filter_gzip(a)) { skipping( "7zip:deflate decoding is not supported on this platform"); } else { test_plain_header("test_read_format_7zip_deflate.7z"); test_bcj("test_read_format_7zip_bcj_deflate.7z"); test_bcj("test_read_format_7zip_bcj2_deflate.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_zstd) { struct archive *a; assert((a = archive_read_new()) != NULL); /* Extracting with libzstd */ if (ARCHIVE_OK != archive_read_support_filter_zstd(a)) { skipping( "7zip:zstd decoding is not supported on this platform"); } else if (ARCHIVE_OK != archive_read_support_filter_xz(a)) { // The directory header entries in the test file uses lzma. skipping( "7zip:lzma decoding is not supported on this platform"); } else { test_extract_all_files_zstd("test_read_format_7zip_zstd.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_zstd_solid) { struct archive *a; assert((a = archive_read_new()) != NULL); /* Extracting with libzstd */ if (ARCHIVE_OK != archive_read_support_filter_zstd(a)) { skipping( "7zip:zstd decoding is not supported on this platform"); } else if (ARCHIVE_OK != archive_read_support_filter_xz(a)) { // The directory header entries in the test file uses lzma. skipping( "7zip:lzma decoding is not supported on this platform"); } else { test_extract_all_files_zstd("test_read_format_7zip_solid_zstd.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_zstd_bcj) { struct archive *a; assert((a = archive_read_new()) != NULL); /* Extracting with libzstd */ if (ARCHIVE_OK != archive_read_support_filter_zstd(a)) { skipping( "7zip:zstd decoding is not supported on this platform"); } else { test_extract_file_zstd_bcj_nobjc("test_read_format_7zip_zstd_bcj.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_zstd_nobcj) { struct archive *a; assert((a = archive_read_new()) != NULL); /* Extracting with libzstd */ if (ARCHIVE_OK != archive_read_support_filter_zstd(a)) { skipping( "7zip:zstd decoding is not supported on this platform"); } else { test_extract_file_zstd_bcj_nobjc("test_read_format_7zip_zstd_nobcj.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_empty) { test_empty_archive(); test_empty_file(); } DEFINE_TEST(test_read_format_7zip_lzma1) { struct archive *a; assert((a = archive_read_new()) != NULL); /* Extracting with liblzma */ if (ARCHIVE_OK != archive_read_support_filter_xz(a)) { skipping("7zip:lzma decoding is not supported on this " "platform"); } else { test_plain_header("test_read_format_7zip_lzma1.7z"); test_extract_all_files("test_read_format_7zip_lzma1_2.7z"); test_extract_last_file("test_read_format_7zip_lzma1_2.7z"); test_bcj("test_read_format_7zip_bcj_lzma1.7z"); test_bcj("test_read_format_7zip_bcj2_lzma1_1.7z"); test_bcj("test_read_format_7zip_bcj2_lzma1_2.7z"); test_delta_lzma("test_read_format_7zip_delta_lzma1.7z"); test_delta_lzma("test_read_format_7zip_delta4_lzma1.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_lzma2) { struct archive *a; assert((a = archive_read_new()) != NULL); /* Extracting with liblzma */ if (ARCHIVE_OK != archive_read_support_filter_xz(a)) { skipping("7zip:lzma decoding is not supported on this " "platform"); } else { test_plain_header("test_read_format_7zip_lzma2.7z"); test_bcj("test_read_format_7zip_bcj_lzma2.7z"); test_bcj("test_read_format_7zip_bcj2_lzma2_1.7z"); test_bcj("test_read_format_7zip_bcj2_lzma2_2.7z"); test_delta_lzma("test_read_format_7zip_delta_lzma2.7z"); test_delta_lzma("test_read_format_7zip_delta4_lzma2.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } static void test_arm_filter(const char *refname) { struct archive *a; struct archive_entry *ae; char buff[7804]; uint32_t computed_crc = 0; uint32_t expected_crc = 0x355ec4e1; assert((a = archive_read_new()) != NULL); extract_reference_file(refname); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0755), archive_entry_mode(ae)); assertEqualString("hw-gnueabihf", archive_entry_pathname(ae)); assertEqualInt(sizeof(buff), archive_entry_size(ae)); assertEqualInt(sizeof(buff), archive_read_data(a, buff, sizeof(buff))); computed_crc = bitcrc32(computed_crc, buff, sizeof(buff)); assertEqualInt(computed_crc, expected_crc); assertEqualInt(1, archive_file_count(a)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_zstd_arm) { struct archive *a; assert((a = archive_read_new()) != NULL); if (ARCHIVE_OK != archive_read_support_filter_zstd(a)) { skipping( "7zip:zstd decoding is not supported on this platform"); } else { test_arm_filter("test_read_format_7zip_zstd_arm.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_lzma2_arm) { struct archive *a; assert((a = archive_read_new()) != NULL); if (ARCHIVE_OK != archive_read_support_filter_lzma(a)) { skipping( "7zip:lzma decoding is not supported on this platform"); } else { test_arm_filter("test_read_format_7zip_lzma2_arm.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_ppmd) { test_ppmd(); } static void test_arm64_filter(const char *refname) { struct archive *a; struct archive_entry *ae; char buff[70368]; uint32_t computed_crc = 0; uint32_t expected_crc = 0xde97d594; assert((a = archive_read_new()) != NULL); extract_reference_file(refname); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0775), archive_entry_mode(ae)); assertEqualString("hw-arm64", archive_entry_pathname(ae)); assertEqualInt(sizeof(buff), archive_entry_size(ae)); assertEqualInt(sizeof(buff), archive_read_data(a, buff, sizeof(buff))); computed_crc = bitcrc32(computed_crc, buff, sizeof(buff)); assertEqualInt(computed_crc, expected_crc); assertEqualInt(1, archive_file_count(a)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_lzma2_arm64) { #ifdef LZMA_FILTER_ARM64 struct archive *a; assert((a = archive_read_new()) != NULL); if (ARCHIVE_OK != archive_read_support_filter_lzma(a)) { skipping( "7zip:lzma decoding is not supported on this platform"); } else { test_arm64_filter("test_read_format_7zip_lzma2_arm64.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); #else skipping("This version of liblzma does not support LZMA_FILTER_ARM64"); #endif } DEFINE_TEST(test_read_format_7zip_deflate_arm64) { struct archive *a; assert((a = archive_read_new()) != NULL); if (ARCHIVE_OK != archive_read_support_filter_gzip(a)) { skipping( "7zip:deflate decoding is not supported on this platform"); } else { test_arm64_filter("test_read_format_7zip_deflate_arm64.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_win_attrib) { struct archive *a; assert((a = archive_read_new()) != NULL); if (ARCHIVE_OK != archive_read_support_filter_lzma(a)) { skipping( "7zip:lzma decoding is not supported on this platform"); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); return; } assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); // This archive has four files and four directories: // * hidden directory // * readonly directory // * regular directory // * system directory // * regular "archive" file // * hidden file // * readonly file // * system file const char *refname = "test_read_format_7zip_win_attrib.7z"; extract_reference_file(refname); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); struct archive_entry *ae; assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("hidden_dir/", archive_entry_pathname(ae)); assertEqualInt((AE_IFDIR | 0755), archive_entry_mode(ae)); assertEqualString("hidden", archive_entry_fflags_text(ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("readonly_dir/", archive_entry_pathname(ae)); assertEqualInt((AE_IFDIR | 0555), archive_entry_mode(ae)); assertEqualString("rdonly", archive_entry_fflags_text(ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("regular_dir/", archive_entry_pathname(ae)); assertEqualInt((AE_IFDIR | 0755), archive_entry_mode(ae)); assertEqualString(NULL, archive_entry_fflags_text(ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("system_dir/", archive_entry_pathname(ae)); assertEqualInt((AE_IFDIR | 0755), archive_entry_mode(ae)); assertEqualString("system", archive_entry_fflags_text(ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("archive_file.txt", archive_entry_pathname(ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString(NULL, archive_entry_fflags_text(ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("hidden_file.txt", archive_entry_pathname(ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("hidden", archive_entry_fflags_text(ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("readonly_file.txt", archive_entry_pathname(ae)); assertEqualInt((AE_IFREG | 0444), archive_entry_mode(ae)); assertEqualString("rdonly", archive_entry_fflags_text(ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("system_file.txt", archive_entry_pathname(ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("system", archive_entry_fflags_text(ae)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_sfx_pe) { /* * This is a regular 7z SFX PE file * created by 7z tool v22.01 on Windows 64-bit */ struct archive *a; struct archive_entry *ae; int bs = 10240; char buff[32]; const char reffile[] = "test_read_format_7zip_sfx_pe.exe"; const char test_txt[] = "123"; int size = sizeof(test_txt) - 1; - extract_reference_file(reffile); assert((a = archive_read_new()) != NULL); - assertA(0 == archive_read_support_filter_all(a)); - assertA(0 == archive_read_support_format_all(a)); - assertA(0 == archive_read_open_filename(a, reffile, bs)); - assertA(0 == archive_read_next_header(a, &ae)); - assertEqualString("test.txt.txt", archive_entry_pathname(ae)); + if (ARCHIVE_OK != archive_read_support_filter_lzma(a)) { + skipping( + "7zip:lzma decoding is not supported on this platform"); + } else { + extract_reference_file(reffile); + assertA(0 == archive_read_support_filter_all(a)); + assertA(0 == archive_read_support_format_all(a)); + assertA(0 == archive_read_open_filename(a, reffile, bs)); - assertA(size == archive_read_data(a, buff, size)); - assertEqualMem(buff, test_txt, size); + assertA(0 == archive_read_next_header(a, &ae)); + assertEqualString("test.txt.txt", archive_entry_pathname(ae)); + + assertA(size == archive_read_data(a, buff, size)); + assertEqualMem(buff, test_txt, size); + + assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); + } - assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_sfx_modified_pe) { /* * This test simulates a modified 7z SFX PE * the compressed data in the SFX file is still stored as PE overlay * but the decompressor code is replaced */ struct archive *a; struct archive_entry *ae; int bs = 10240; char buff[32]; const char reffile[] = "test_read_format_7zip_sfx_modified_pe.exe"; const char test_txt[] = "123"; int size = sizeof(test_txt) - 1; - extract_reference_file(reffile); assert((a = archive_read_new()) != NULL); - assertA(0 == archive_read_support_filter_all(a)); - assertA(0 == archive_read_support_format_all(a)); - assertA(0 == archive_read_open_filename(a, reffile, bs)); - assertA(0 == archive_read_next_header(a, &ae)); - assertEqualString("test.txt.txt", archive_entry_pathname(ae)); + if (ARCHIVE_OK != archive_read_support_filter_lzma(a)) { + skipping( + "7zip:lzma decoding is not supported on this platform"); + } else { + extract_reference_file(reffile); + assertA(0 == archive_read_support_filter_all(a)); + assertA(0 == archive_read_support_format_all(a)); + assertA(0 == archive_read_open_filename(a, reffile, bs)); + + assertA(0 == archive_read_next_header(a, &ae)); + assertEqualString("test.txt.txt", archive_entry_pathname(ae)); + + assertA(size == archive_read_data(a, buff, size)); + assertEqualMem(buff, test_txt, size); - assertA(size == archive_read_data(a, buff, size)); - assertEqualMem(buff, test_txt, size); + assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); + } - assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_sfx_elf) { /* * This is a regular 7z SFX ELF file * created by 7z tool v16.02 on Ubuntu */ struct archive *a; struct archive_entry *ae; int bs = 10240; char buff[32]; const char reffile[] = "test_read_format_7zip_sfx_elf.elf"; const char test_txt[] = "123"; int size = sizeof(test_txt) - 1; - extract_reference_file(reffile); assert((a = archive_read_new()) != NULL); - assertA(0 == archive_read_support_filter_all(a)); - assertA(0 == archive_read_support_format_all(a)); - assertA(0 == archive_read_open_filename(a, reffile, bs)); - assertA(0 == archive_read_next_header(a, &ae)); - assertEqualString("test.txt.txt", archive_entry_pathname(ae)); + if (ARCHIVE_OK != archive_read_support_filter_lzma(a)) { + skipping( + "7zip:lzma decoding is not supported on this platform"); + } else { + extract_reference_file(reffile); + assertA(0 == archive_read_support_filter_all(a)); + assertA(0 == archive_read_support_format_all(a)); + assertA(0 == archive_read_open_filename(a, reffile, bs)); + + assertA(0 == archive_read_next_header(a, &ae)); + assertEqualString("test.txt.txt", archive_entry_pathname(ae)); - assertA(size == archive_read_data(a, buff, size)); - assertEqualMem(buff, test_txt, size); + assertA(size == archive_read_data(a, buff, size)); + assertEqualMem(buff, test_txt, size); - assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); - assertEqualInt(ARCHIVE_OK, archive_read_free(a)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); + } + + assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_extract_second) { struct archive *a; char buffer[256]; assert((a = archive_read_new()) != NULL); if (ARCHIVE_OK != archive_read_support_filter_lzma(a)) { skipping( "7zip:lzma decoding is not supported on this platform"); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); return; } assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); /* * The test archive has two files: first.txt which is a 65,536 file (the * size of the uncompressed buffer), and second.txt which has contents * we will validate. This test ensures we can skip first.txt and still * be able to read the contents of second.txt */ const char *refname = "test_read_format_7zip_extract_second.7z"; extract_reference_file(refname); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); struct archive_entry *ae; assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("first.txt", archive_entry_pathname(ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("second.txt", archive_entry_pathname(ae)); assertEqualInt(23, archive_read_data(a, buffer, sizeof(buffer))); assertEqualMem("This is from second.txt", buffer, 23); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } #ifdef LZMA_FILTER_RISCV static void test_riscv_filter(const char *refname) { struct archive *a; struct archive_entry *ae; char buff[8488]; uint32_t computed_crc = 0; uint32_t expected_crc = 0xf7ed24e7; assert((a = archive_read_new()) != NULL); extract_reference_file(refname); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0775), archive_entry_mode(ae)); assertEqualString("hw-riscv64", archive_entry_pathname(ae)); assertEqualInt(sizeof(buff), archive_entry_size(ae)); assertEqualInt(sizeof(buff), archive_read_data(a, buff, sizeof(buff))); computed_crc = bitcrc32(computed_crc, buff, sizeof(buff)); assertEqualInt(computed_crc, expected_crc); assertEqualInt(1, archive_file_count(a)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } #endif DEFINE_TEST(test_read_format_7zip_lzma2_riscv) { #ifdef LZMA_FILTER_RISCV struct archive *a; assert((a = archive_read_new()) != NULL); if (ARCHIVE_OK != archive_read_support_filter_lzma(a)) { skipping("7zip:lzma decoding is not supported on this platform"); } else { test_riscv_filter("test_read_format_7zip_lzma2_riscv.7z"); } #else skipping("This version of liblzma does not support LZMA_FILTER_RISCV"); #endif } static void test_sparc_filter(const char *refname) { struct archive *a; struct archive_entry *ae; size_t expected_entry_size = 1053016; char *buff = malloc(expected_entry_size); uint32_t computed_crc = 0; uint32_t expected_crc = 0x6b5b364d; assert((a = archive_read_new()) != NULL); extract_reference_file(refname); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0775), archive_entry_mode(ae)); assertEqualString("hw-sparc64", archive_entry_pathname(ae)); assertEqualInt(expected_entry_size, archive_entry_size(ae)); assertEqualInt(expected_entry_size, archive_read_data(a, buff, expected_entry_size)); computed_crc = bitcrc32(computed_crc, buff, expected_entry_size); assertEqualInt(computed_crc, expected_crc); assertEqualInt(1, archive_file_count(a)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); free(buff); } DEFINE_TEST(test_read_format_7zip_lzma2_sparc) { struct archive *a; assert((a = archive_read_new()) != NULL); if (ARCHIVE_OK != archive_read_support_filter_lzma(a)) { skipping( "7zip:lzma decoding is not supported on this platform"); } else { test_sparc_filter("test_read_format_7zip_lzma2_sparc.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_zstd_sparc) { struct archive *a; assert((a = archive_read_new()) != NULL); if (ARCHIVE_OK != archive_read_support_filter_zstd(a)) { skipping( "7zip:zstd decoding is not supported on this platform"); } else { test_sparc_filter("test_read_format_7zip_zstd_sparc.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } static void test_powerpc_filter(const char *refname) { struct archive *a; struct archive_entry *ae; size_t expected_entry_size = 68340; char *buff = malloc(expected_entry_size); uint32_t computed_crc = 0; uint32_t expected_crc = 0x71fb03c9; assert((a = archive_read_new()) != NULL); extract_reference_file(refname); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0775), archive_entry_mode(ae)); assertEqualString("hw-powerpc", archive_entry_pathname(ae)); assertEqualInt(expected_entry_size, archive_entry_size(ae)); assertEqualInt(expected_entry_size, archive_read_data(a, buff, expected_entry_size)); computed_crc = bitcrc32(computed_crc, buff, expected_entry_size); assertEqualInt(computed_crc, expected_crc); assertEqualInt(1, archive_file_count(a)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); free(buff); } DEFINE_TEST(test_read_format_7zip_deflate_powerpc) { struct archive *a; assert((a = archive_read_new()) != NULL); if (ARCHIVE_OK != archive_read_support_filter_gzip(a)) { skipping( "7zip:deflate decoding is not supported on this platform"); } else { test_powerpc_filter("test_read_format_7zip_deflate_powerpc.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_7zip_lzma2_powerpc) { struct archive *a; assert((a = archive_read_new()) != NULL); - if (ARCHIVE_OK != archive_read_support_filter_gzip(a)) { + if (ARCHIVE_OK != archive_read_support_filter_lzma(a)) { skipping( - "7zip:deflate decoding is not supported on this platform"); + "7zip:lzma decoding is not supported on this platform"); } else { test_powerpc_filter("test_read_format_7zip_lzma2_powerpc.7z"); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } diff --git a/contrib/libarchive/libarchive/test/test_read_format_rar5.c b/contrib/libarchive/libarchive/test/test_read_format_rar5.c index fd233277bc1b..6ab0d236a1ed 100644 --- a/contrib/libarchive/libarchive/test/test_read_format_rar5.c +++ b/contrib/libarchive/libarchive/test/test_read_format_rar5.c @@ -1,1430 +1,1496 @@ /*- * Copyright (c) 2018 Grzegorz Antoniak * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" /* Some tests will want to calculate some CRC32's, and this header can * help. */ #define __LIBARCHIVE_BUILD #include #define PROLOGUE(reffile) \ struct archive_entry *ae; \ struct archive *a; \ \ (void) a; /* Make the compiler happy if we won't use this variables */ \ (void) ae; /* in the test cases. */ \ \ extract_reference_file(reffile); \ assert((a = archive_read_new()) != NULL); \ assertA(0 == archive_read_support_filter_all(a)); \ assertA(0 == archive_read_support_format_all(a)); \ assertA(0 == archive_read_open_filename(a, reffile, 10240)) #define PROLOGUE_MULTI(reffile) \ struct archive_entry *ae; \ struct archive *a; \ \ (void) a; \ (void) ae; \ \ extract_reference_files(reffile); \ assert((a = archive_read_new()) != NULL); \ assertA(0 == archive_read_support_filter_all(a)); \ assertA(0 == archive_read_support_format_all(a)); \ assertA(0 == archive_read_open_filenames(a, reffile, 10240)) #define EPILOGUE() \ assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); \ assertEqualInt(ARCHIVE_OK, archive_read_free(a)) static int verify_data(const uint8_t* data_ptr, int magic, int size) { int i = 0; /* This is how the test data inside test files was generated; * we are re-generating it here and we check if our re-generated * test data is the same as in the test file. If this test is * failing it's either because there's a bug in the test case, * or the unpacked data is corrupted. */ for(i = 0; i < size / 4; ++i) { const int k = i + 1; const signed int* lptr = (const signed int*) &data_ptr[i * 4]; signed int val = k * k - 3 * k + (1 + magic); if(val < 0) val = 0; /* *lptr is a value inside unpacked test file, val is the * value that should be in the unpacked test file. */ if(i4le(lptr) != (uint32_t) val) return 0; } return 1; } static int extract_one(struct archive* a, struct archive_entry* ae, uint32_t crc) { la_ssize_t fsize, bytes_read; uint8_t* buf; int ret = 1; uint32_t computed_crc; fsize = (la_ssize_t) archive_entry_size(ae); buf = malloc(fsize); if(buf == NULL) return 1; bytes_read = archive_read_data(a, buf, fsize); if(bytes_read != fsize) { assertEqualInt(bytes_read, fsize); goto fn_exit; } computed_crc = bitcrc32(0, buf, fsize); assertEqualInt(computed_crc, crc); ret = 0; fn_exit: free(buf); return ret; } DEFINE_TEST(test_read_format_rar5_set_format) { struct archive *a; struct archive_entry *ae; const char reffile[] = "test_read_format_rar5_stored.rar"; extract_reference_file(reffile); assert((a = archive_read_new()) != NULL); assertA(0 == archive_read_support_filter_all(a)); assertA(0 == archive_read_set_format(a, ARCHIVE_FORMAT_RAR_V5)); assertA(0 == archive_read_open_filename(a, reffile, 10240)); assertA(0 == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_stored) { const char helloworld_txt[] = "hello libarchive test suite!\n"; la_ssize_t file_size = sizeof(helloworld_txt) - 1; char buff[64]; PROLOGUE("test_read_format_rar5_stored.rar"); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("helloworld.txt", archive_entry_pathname(ae)); assertA((int) archive_entry_mtime(ae) > 0); assertA((int) archive_entry_ctime(ae) == 0); assertA((int) archive_entry_atime(ae) == 0); assertEqualInt(file_size, archive_entry_size(ae)); assertEqualInt(33188, archive_entry_mode(ae)); assertA(file_size == archive_read_data(a, buff, file_size)); assertEqualMem(buff, helloworld_txt, file_size); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_compressed) { const int DATA_SIZE = 1200; uint8_t buff[1200]; PROLOGUE("test_read_format_rar5_compressed.rar"); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA((int) archive_entry_mtime(ae) > 0); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); assertA(1 == verify_data(buff, 0, DATA_SIZE)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_multiple_files) { const int DATA_SIZE = 4096; uint8_t buff[4096]; PROLOGUE("test_read_format_rar5_multiple_files.rar"); /* There should be 4 files inside this test file. Check for their * existence, and also check the contents of those test files. */ assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(1 == verify_data(buff, 1, DATA_SIZE)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test2.bin", archive_entry_pathname(ae)); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(1 == verify_data(buff, 2, DATA_SIZE)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(1 == verify_data(buff, 3, DATA_SIZE)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(1 == verify_data(buff, 4, DATA_SIZE)); /* There should be no more files in this archive. */ assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } /* This test is really the same as the test above, but it deals with a solid * archive instead of a regular archive. The test solid archive contains the * same set of files as regular test archive, but it's size is 2x smaller, * because solid archives reuse the window buffer from previous compressed * files, so it's able to compress lots of small files more effectively. */ DEFINE_TEST(test_read_format_rar5_multiple_files_solid) { const int DATA_SIZE = 4096; uint8_t buff[4096]; PROLOGUE("test_read_format_rar5_multiple_files_solid.rar"); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(1 == verify_data(buff, 1, DATA_SIZE)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test2.bin", archive_entry_pathname(ae)); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(1 == verify_data(buff, 2, DATA_SIZE)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(1 == verify_data(buff, 3, DATA_SIZE)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(1 == verify_data(buff, 4, DATA_SIZE)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_multiarchive_skip_all) { const char* reffiles[] = { "test_read_format_rar5_multiarchive.part01.rar", "test_read_format_rar5_multiarchive.part02.rar", "test_read_format_rar5_multiarchive.part03.rar", "test_read_format_rar5_multiarchive.part04.rar", "test_read_format_rar5_multiarchive.part05.rar", "test_read_format_rar5_multiarchive.part06.rar", "test_read_format_rar5_multiarchive.part07.rar", "test_read_format_rar5_multiarchive.part08.rar", NULL }; PROLOGUE_MULTI(reffiles); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("home/antek/temp/build/unrar5/libarchive/bin/bsdcat_test", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("home/antek/temp/build/unrar5/libarchive/bin/bsdtar_test", archive_entry_pathname(ae)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_multiarchive_skip_all_but_first) { const char* reffiles[] = { "test_read_format_rar5_multiarchive.part01.rar", "test_read_format_rar5_multiarchive.part02.rar", "test_read_format_rar5_multiarchive.part03.rar", "test_read_format_rar5_multiarchive.part04.rar", "test_read_format_rar5_multiarchive.part05.rar", "test_read_format_rar5_multiarchive.part06.rar", "test_read_format_rar5_multiarchive.part07.rar", "test_read_format_rar5_multiarchive.part08.rar", NULL }; PROLOGUE_MULTI(reffiles); assertA(0 == archive_read_next_header(a, &ae)); assertA(0 == extract_one(a, ae, 0x35277473)); assertA(0 == archive_read_next_header(a, &ae)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_multiarchive_skip_all_but_second) { const char* reffiles[] = { "test_read_format_rar5_multiarchive.part01.rar", "test_read_format_rar5_multiarchive.part02.rar", "test_read_format_rar5_multiarchive.part03.rar", "test_read_format_rar5_multiarchive.part04.rar", "test_read_format_rar5_multiarchive.part05.rar", "test_read_format_rar5_multiarchive.part06.rar", "test_read_format_rar5_multiarchive.part07.rar", "test_read_format_rar5_multiarchive.part08.rar", NULL }; PROLOGUE_MULTI(reffiles); assertA(0 == archive_read_next_header(a, &ae)); assertA(0 == archive_read_next_header(a, &ae)); assertA(0 == extract_one(a, ae, 0xE59665F8)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_blake2) { const la_ssize_t proper_size = 814; uint8_t buf[814]; PROLOGUE("test_read_format_rar5_blake2.rar"); assertA(0 == archive_read_next_header(a, &ae)); assertEqualInt(proper_size, archive_entry_size(ae)); /* Should blake2 calculation fail, we'll get a failure return * value from archive_read_data(). */ assertA(proper_size == archive_read_data(a, buf, proper_size)); /* To be extra pedantic, let's also check crc32 of the poem. */ assertEqualInt(bitcrc32(0, buf, proper_size), 0x7E5EC49E); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_arm_filter) { /* This test unpacks a file that uses an ARM filter. The DELTA * and X86 filters are tested implicitly in the "multiarchive_skip" * test. */ const la_ssize_t proper_size = 90808; uint8_t buf[90808]; PROLOGUE("test_read_format_rar5_arm.rar"); assertA(0 == archive_read_next_header(a, &ae)); assertEqualInt(proper_size, archive_entry_size(ae)); assertA(proper_size == archive_read_data(a, buf, proper_size)); /* Yes, RARv5 unpacker itself should calculate the CRC, but in case * the DONT_FAIL_ON_CRC_ERROR define option is enabled during compilation, * let's still fail the test if the unpacked data is wrong. */ assertEqualInt(bitcrc32(0, buf, proper_size), 0x886F91EB); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_stored_skip_all) { const char* fname = "test_read_format_rar5_stored_manyfiles.rar"; PROLOGUE(fname); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("make_uue.tcl", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("cebula.txt", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_stored_skip_in_part) { const char* fname = "test_read_format_rar5_stored_manyfiles.rar"; char buf[6]; /* Skip first, extract in part rest. */ PROLOGUE(fname); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("make_uue.tcl", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("cebula.txt", archive_entry_pathname(ae)); assertA(6 == archive_read_data(a, buf, 6)); assertEqualInt(0, memcmp(buf, "Cebula", 6)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(4 == archive_read_data(a, buf, 4)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_stored_skip_all_but_first) { const char* fname = "test_read_format_rar5_stored_manyfiles.rar"; char buf[405]; /* Extract first, skip rest. */ PROLOGUE(fname); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("make_uue.tcl", archive_entry_pathname(ae)); assertA(405 == archive_read_data(a, buf, sizeof(buf))); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("cebula.txt", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_stored_skip_all_in_part) { const char* fname = "test_read_format_rar5_stored_manyfiles.rar"; char buf[4]; /* Extract in part all */ PROLOGUE(fname); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("make_uue.tcl", archive_entry_pathname(ae)); assertA(4 == archive_read_data(a, buf, 4)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("cebula.txt", archive_entry_pathname(ae)); assertA(4 == archive_read_data(a, buf, 4)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(4 == archive_read_data(a, buf, 4)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_multiarchive_solid_extr_all) { const char* reffiles[] = { "test_read_format_rar5_multiarchive_solid.part01.rar", "test_read_format_rar5_multiarchive_solid.part02.rar", "test_read_format_rar5_multiarchive_solid.part03.rar", "test_read_format_rar5_multiarchive_solid.part04.rar", NULL }; PROLOGUE_MULTI(reffiles); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("cebula.txt", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x7E5EC49E)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x7cca70cd)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x7e13b2c6)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test2.bin", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0xf166afcb)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x9fb123d9)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x10c43ed4)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test5.bin", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0xb9d155f2)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test6.bin", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x36a448ff)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("elf-Linux-ARMv7-ls", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x886F91EB)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_multiarchive_solid_skip_all) { const char* reffiles[] = { "test_read_format_rar5_multiarchive_solid.part01.rar", "test_read_format_rar5_multiarchive_solid.part02.rar", "test_read_format_rar5_multiarchive_solid.part03.rar", "test_read_format_rar5_multiarchive_solid.part04.rar", NULL }; PROLOGUE_MULTI(reffiles); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("cebula.txt", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test2.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test5.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test6.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("elf-Linux-ARMv7-ls", archive_entry_pathname(ae)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_multiarchive_solid_skip_all_but_first) { const char* reffiles[] = { "test_read_format_rar5_multiarchive_solid.part01.rar", "test_read_format_rar5_multiarchive_solid.part02.rar", "test_read_format_rar5_multiarchive_solid.part03.rar", "test_read_format_rar5_multiarchive_solid.part04.rar", NULL }; PROLOGUE_MULTI(reffiles); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("cebula.txt", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x7E5EC49E)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test2.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test5.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test6.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("elf-Linux-ARMv7-ls", archive_entry_pathname(ae)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } /* "skip_all_but_scnd" -> am I hitting the test name limit here after * expansion of "scnd" to "second"? */ DEFINE_TEST(test_read_format_rar5_multiarchive_solid_skip_all_but_scnd) { const char* reffiles[] = { "test_read_format_rar5_multiarchive_solid.part01.rar", "test_read_format_rar5_multiarchive_solid.part02.rar", "test_read_format_rar5_multiarchive_solid.part03.rar", "test_read_format_rar5_multiarchive_solid.part04.rar", NULL }; PROLOGUE_MULTI(reffiles); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("cebula.txt", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x7CCA70CD)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test2.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test5.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test6.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("elf-Linux-ARMv7-ls", archive_entry_pathname(ae)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_multiarchive_solid_skip_all_but_third) { const char* reffiles[] = { "test_read_format_rar5_multiarchive_solid.part01.rar", "test_read_format_rar5_multiarchive_solid.part02.rar", "test_read_format_rar5_multiarchive_solid.part03.rar", "test_read_format_rar5_multiarchive_solid.part04.rar", NULL }; PROLOGUE_MULTI(reffiles); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("cebula.txt", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x7E13B2C6)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test2.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test5.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test6.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("elf-Linux-ARMv7-ls", archive_entry_pathname(ae)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_multiarchive_solid_skip_all_but_last) { const char* reffiles[] = { "test_read_format_rar5_multiarchive_solid.part01.rar", "test_read_format_rar5_multiarchive_solid.part02.rar", "test_read_format_rar5_multiarchive_solid.part03.rar", "test_read_format_rar5_multiarchive_solid.part04.rar", NULL }; PROLOGUE_MULTI(reffiles); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("cebula.txt", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test2.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test5.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test6.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("elf-Linux-ARMv7-ls", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x886F91EB)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_solid_skip_all) { const char* reffile = "test_read_format_rar5_solid.rar"; /* Skip all */ PROLOGUE(reffile); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test2.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test5.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test6.bin", archive_entry_pathname(ae)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_solid_skip_all_but_first) { const char* reffile = "test_read_format_rar5_solid.rar"; /* Extract first, skip rest */ PROLOGUE(reffile); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x7CCA70CD)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test2.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test5.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test6.bin", archive_entry_pathname(ae)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_solid_skip_all_but_second) { const char* reffile = "test_read_format_rar5_solid.rar"; /* Skip first, extract second, skip rest */ PROLOGUE(reffile); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x7E13B2C6)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test2.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test5.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test6.bin", archive_entry_pathname(ae)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_solid_skip_all_but_last) { const char* reffile = "test_read_format_rar5_solid.rar"; /* Skip all but last, extract last */ PROLOGUE(reffile); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test2.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test5.bin", archive_entry_pathname(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test6.bin", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0x36A448FF)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_extract_win32) { PROLOGUE("test_read_format_rar5_win32.rar"); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("testdir", archive_entry_pathname(ae)); assertEqualInt(archive_entry_mode(ae), AE_IFDIR | 0755); assertA(0 == extract_one(a, ae, 0)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0644); assertA(0 == extract_one(a, ae, 0x7CCA70CD)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test1.bin", archive_entry_pathname(ae)); assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0644); assertA(0 == extract_one(a, ae, 0x7E13B2C6)); assertA(0 == archive_read_next_header(a, &ae)); /* Read only file */ assertEqualString("test2.bin", archive_entry_pathname(ae)); assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0444); assertA(0 == extract_one(a, ae, 0xF166AFCB)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test3.bin", archive_entry_pathname(ae)); assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0644); assertA(0 == extract_one(a, ae, 0x9FB123D9)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test4.bin", archive_entry_pathname(ae)); assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0644); assertA(0 == extract_one(a, ae, 0x10C43ED4)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test5.bin", archive_entry_pathname(ae)); assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0644); assertA(0 == extract_one(a, ae, 0xB9D155F2)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test6.bin", archive_entry_pathname(ae)); assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0644); assertA(0 == extract_one(a, ae, 0x36A448FF)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_unicode) { #if !defined(WIN32) || defined(__CYGWIN__) skipping("Skipping test on non-Windows"); return; #else /* Corresponds to the names: * 👋🌎.txt * 𝒮𝓎𝓂𝒷𝑜𝓁𝒾𝒸 𝐿𝒾𝓃𝓀.txt * Ⓗⓐⓡⓓ Ⓛⓘⓝⓚ.txt */ const wchar_t* emoji_name = L"\U0001f44b\U0001f30e.txt"; const wchar_t* italic_name = L"\U0001d4ae\U0001d4ce\U0001d4c2\U0001d4b7\U0001d45c\U0001d4c1\U0001d4be\U0001d4b8 \U0001d43f\U0001d4be\U0001d4c3\U0001d4c0.txt"; const wchar_t* circle_name = L"\u24bd\u24d0\u24e1\u24d3 \u24c1\u24d8\u24dd\u24da.txt"; PROLOGUE("test_read_format_rar5_unicode.rar"); assertA(0 == archive_read_next_header(a, &ae)); assertEqualWString(emoji_name, archive_entry_pathname_w(ae)); assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0644); assertA(0 == archive_read_next_header(a, &ae)); assertEqualWString(circle_name, archive_entry_pathname_w(ae)); assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0644); assertEqualWString(emoji_name, archive_entry_hardlink_w(ae)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualWString(italic_name, archive_entry_pathname_w(ae)); assertEqualInt(archive_entry_mode(ae), AE_IFLNK | 0644); assertEqualWString(emoji_name, archive_entry_symlink_w(ae)); EPILOGUE(); #endif } DEFINE_TEST(test_read_format_rar5_block_by_block) { /* This test uses strange buffer sizes intentionally. */ struct archive_entry *ae; struct archive *a; uint8_t buf[173]; ssize_t bytes_read; uint32_t computed_crc = 0; extract_reference_file("test_read_format_rar5_compressed.rar"); assert((a = archive_read_new()) != NULL); assertA(0 == archive_read_support_filter_all(a)); assertA(0 == archive_read_support_format_all(a)); assertA(0 == archive_read_open_filename(a, "test_read_format_rar5_compressed.rar", 130)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.bin", archive_entry_pathname(ae)); assertEqualInt(1200, archive_entry_size(ae)); /* File size is 1200 bytes, we're reading it using a buffer of 173 bytes. * Libarchive is configured to use a buffer of 130 bytes. */ while(1) { /* archive_read_data should return one of: * a) 0, if there is no more data to be read, * b) negative value, if there was an error, * c) positive value, meaning how many bytes were read. */ bytes_read = archive_read_data(a, buf, sizeof(buf)); assertA(bytes_read >= 0); if(bytes_read <= 0) break; computed_crc = bitcrc32(computed_crc, buf, bytes_read); } assertEqualInt(computed_crc, 0x7CCA70CD); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_owner) { const int DATA_SIZE = 5; uint8_t buff[5]; PROLOGUE("test_read_format_rar5_owner.rar"); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("root.txt", archive_entry_pathname(ae)); assertEqualString("root", archive_entry_uname(ae)); assertEqualString("wheel", archive_entry_gname(ae)); assertA((int) archive_entry_mtime(ae) > 0); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("nobody.txt", archive_entry_pathname(ae)); assertEqualString("nobody", archive_entry_uname(ae)); assertEqualString("nogroup", archive_entry_gname(ae)); assertA((int) archive_entry_mtime(ae) > 0); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("numeric.txt", archive_entry_pathname(ae)); assertEqualInt(9999, archive_entry_uid(ae)); assertEqualInt(8888, archive_entry_gid(ae)); assertA((int) archive_entry_mtime(ae) > 0); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_symlink) { const int DATA_SIZE = 5; uint8_t buff[5]; PROLOGUE("test_read_format_rar5_symlink.rar"); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("file.txt", archive_entry_pathname(ae)); assertEqualInt(AE_IFREG, archive_entry_filetype(ae)); assertA((int) archive_entry_mtime(ae) > 0); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("symlink.txt", archive_entry_pathname(ae)); assertEqualInt(AE_IFLNK, archive_entry_filetype(ae)); assertEqualString("file.txt", archive_entry_symlink(ae)); assertEqualInt(AE_SYMLINK_TYPE_FILE, archive_entry_symlink_type(ae)); assertEqualInt(0, archive_entry_size(ae)); assertA(0 == archive_read_data(a, NULL, (size_t)archive_entry_size(ae))); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("dirlink", archive_entry_pathname(ae)); assertEqualInt(AE_IFLNK, archive_entry_filetype(ae)); assertEqualString("dir", archive_entry_symlink(ae)); assertEqualInt(AE_SYMLINK_TYPE_DIRECTORY, archive_entry_symlink_type(ae)); assertEqualInt(0, archive_entry_size(ae)); assertA(0 == archive_read_data(a, NULL, (size_t)archive_entry_size(ae))); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("dir", archive_entry_pathname(ae)); assertEqualInt(AE_IFDIR, archive_entry_filetype(ae)); assertEqualInt(0, archive_entry_size(ae)); assertA(0 == archive_read_data(a, NULL, (size_t)archive_entry_size(ae))); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_hardlink) { const int DATA_SIZE = 5; uint8_t buff[5]; PROLOGUE("test_read_format_rar5_hardlink.rar"); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("file.txt", archive_entry_pathname(ae)); assertEqualInt(AE_IFREG, archive_entry_filetype(ae)); assertA((int) archive_entry_mtime(ae) > 0); assertEqualInt(DATA_SIZE, archive_entry_size(ae)); assertA(DATA_SIZE == archive_read_data(a, buff, DATA_SIZE)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("hardlink.txt", archive_entry_pathname(ae)); assertEqualInt(AE_IFREG, archive_entry_filetype(ae)); assertEqualString("file.txt", archive_entry_hardlink(ae)); assertEqualInt(0, archive_entry_size(ae)); assertA(0 == archive_read_data(a, NULL, (size_t)archive_entry_size(ae))); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_extra_field_version) { PROLOGUE("test_read_format_rar5_extra_field_version.rar"); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("bin/2to3;1", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0xF24181B7)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("bin/2to3", archive_entry_pathname(ae)); assertA(0 == extract_one(a, ae, 0xF24181B7)); assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_readtables_overflow) { uint8_t buf[16]; PROLOGUE("test_read_format_rar5_readtables_overflow.rar"); /* This archive is invalid. However, processing it shouldn't cause any * buffer overflow errors during reading rar5 tables. */ (void) archive_read_next_header(a, &ae); (void) archive_read_data(a, buf, sizeof(buf)); (void) archive_read_next_header(a, &ae); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_leftshift1) { uint8_t buf[16]; PROLOGUE("test_read_format_rar5_leftshift1.rar"); /* This archive is invalid. However, processing it shouldn't cause any * errors related to undefined operations when using -fsanitize. */ (void) archive_read_next_header(a, &ae); (void) archive_read_data(a, buf, sizeof(buf)); (void) archive_read_next_header(a, &ae); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_leftshift2) { uint8_t buf[16]; PROLOGUE("test_read_format_rar5_leftshift2.rar"); /* This archive is invalid. However, processing it shouldn't cause any * errors related to undefined operations when using -fsanitize. */ (void) archive_read_next_header(a, &ae); (void) archive_read_data(a, buf, sizeof(buf)); (void) archive_read_next_header(a, &ae); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_truncated_huff) { uint8_t buf[16]; PROLOGUE("test_read_format_rar5_truncated_huff.rar"); /* This archive is invalid. However, processing it shouldn't cause any * errors related to undefined operations when using -fsanitize. */ (void) archive_read_next_header(a, &ae); (void) archive_read_data(a, buf, sizeof(buf)); (void) archive_read_next_header(a, &ae); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_invalid_dict_reference) { uint8_t buf[16]; PROLOGUE("test_read_format_rar5_invalid_dict_reference.rar"); /* This test should fail on parsing the header. */ assertA(archive_read_next_header(a, &ae) != ARCHIVE_OK); /* This archive is invalid. However, processing it shouldn't cause any * errors related to buffer underflow when using -fsanitize. */ assertA(archive_read_data(a, buf, sizeof(buf)) <= 0); /* This test only cares about not returning success here. */ assertA(ARCHIVE_OK != archive_read_next_header(a, &ae)); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_distance_overflow) { uint8_t buf[16]; PROLOGUE("test_read_format_rar5_distance_overflow.rar"); /* This archive is invalid. However, processing it shouldn't cause any * errors related to variable overflows when using -fsanitize. */ (void) archive_read_next_header(a, &ae); (void) archive_read_data(a, buf, sizeof(buf)); (void) archive_read_next_header(a, &ae); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_nonempty_dir_stream) { uint8_t buf[16]; PROLOGUE("test_read_format_rar5_nonempty_dir_stream.rar"); /* This archive is invalid. However, processing it shouldn't cause any * errors related to buffer overflows when using -fsanitize. */ (void) archive_read_next_header(a, &ae); (void) archive_read_data(a, buf, sizeof(buf)); (void) archive_read_next_header(a, &ae); EPILOGUE(); } +DEFINE_TEST(test_read_format_rar5_nonempty_dir_data) +{ + PROLOGUE("test_read_format_rar5_dirdata.rar"); + + /* This archive is invalid. It declares a directory entry with nonzero + data size. */ + + assertA(archive_read_next_header(a, &ae) == ARCHIVE_FATAL); + + EPILOGUE(); +} + DEFINE_TEST(test_read_format_rar5_fileattr) { unsigned long set, clear, flag; flag = 0; PROLOGUE("test_read_format_rar5_fileattr.rar"); assertA(0 == archive_read_next_header(a, &ae)); assertEqualInt(archive_entry_mode(ae), 0444 | AE_IFREG); assertEqualString("readonly.txt", archive_entry_pathname(ae)); assertEqualString("rdonly", archive_entry_fflags_text(ae)); archive_entry_fflags(ae, &set, &clear); #if defined(__FreeBSD__) flag = UF_READONLY; #elif defined(_WIN32) && !defined(__CYGWIN__) flag = FILE_ATTRIBUTE_READONLY; #endif assertEqualInt(flag, set & flag); assertA(0 == archive_read_next_header(a, &ae)); assertEqualInt(archive_entry_mode(ae), 0644 | AE_IFREG); assertEqualString("hidden.txt", archive_entry_pathname(ae)); assertEqualString("hidden", archive_entry_fflags_text(ae)); archive_entry_fflags(ae, &set, &clear); #if defined(__FreeBSD__) flag = UF_HIDDEN; #elif defined(_WIN32) && !defined(__CYGWIN__) flag = FILE_ATTRIBUTE_HIDDEN; #endif assertEqualInt(flag, set & flag); assertA(0 == archive_read_next_header(a, &ae)); assertEqualInt(archive_entry_mode(ae), 0644 | AE_IFREG); assertEqualString("system.txt", archive_entry_pathname(ae)); assertEqualString("system", archive_entry_fflags_text(ae)); archive_entry_fflags(ae, &set, &clear); #if defined(__FreeBSD__) flag = UF_SYSTEM; #elif defined(_WIN32) && !defined(__CYGWIN__) flag = FILE_ATTRIBUTE_SYSTEM; #endif assertEqualInt(flag, set & flag); assertA(0 == archive_read_next_header(a, &ae)); assertEqualInt(archive_entry_mode(ae), 0444 | AE_IFREG); assertEqualString("ro_hidden.txt", archive_entry_pathname(ae)); assertEqualString("rdonly,hidden", archive_entry_fflags_text(ae)); archive_entry_fflags(ae, &set, &clear); #if defined(__FreeBSD__) flag = UF_READONLY | UF_HIDDEN; #elif defined(_WIN32) && !defined(__CYGWIN__) flag = FILE_ATTRIBUTE_READONLY | FILE_ATTRIBUTE_HIDDEN; #endif assertEqualInt(flag, set & flag); assertA(0 == archive_read_next_header(a, &ae)); assertEqualInt(archive_entry_mode(ae), 0555 | AE_IFDIR); assertEqualString("dir_readonly", archive_entry_pathname(ae)); assertEqualString("rdonly", archive_entry_fflags_text(ae)); archive_entry_fflags(ae, &set, &clear); #if defined(__FreeBSD__) flag = UF_READONLY; #elif defined(_WIN32) && !defined(__CYGWIN__) flag = FILE_ATTRIBUTE_READONLY; #endif assertEqualInt(flag, set & flag); assertA(0 == archive_read_next_header(a, &ae)); assertEqualInt(archive_entry_mode(ae), 0755 | AE_IFDIR); assertEqualString("dir_hidden", archive_entry_pathname(ae)); assertEqualString("hidden", archive_entry_fflags_text(ae)); archive_entry_fflags(ae, &set, &clear); #if defined(__FreeBSD__) flag = UF_HIDDEN; #elif defined(_WIN32) && !defined(__CYGWIN__) flag = FILE_ATTRIBUTE_HIDDEN; #endif assertEqualInt(flag, set & flag); assertA(0 == archive_read_next_header(a, &ae)); assertEqualInt(archive_entry_mode(ae), 0755 | AE_IFDIR); assertEqualString("dir_system", archive_entry_pathname(ae)); assertEqualString("system", archive_entry_fflags_text(ae)); archive_entry_fflags(ae, &set, &clear); #if defined(__FreeBSD__) flag = UF_SYSTEM; #elif defined(_WIN32) && !defined(__CYGWIN__) flag = FILE_ATTRIBUTE_SYSTEM; #endif assertEqualInt(flag, set & flag); assertA(0 == archive_read_next_header(a, &ae)); assertEqualInt(archive_entry_mode(ae), 0555 | AE_IFDIR); assertEqualString("dir_rohidden", archive_entry_pathname(ae)); assertEqualString("rdonly,hidden", archive_entry_fflags_text(ae)); archive_entry_fflags(ae, &set, &clear); #if defined(__FreeBSD__) flag = UF_READONLY | UF_HIDDEN; #elif defined(_WIN32) && !defined(__CYGWIN__) flag = FILE_ATTRIBUTE_READONLY | FILE_ATTRIBUTE_HIDDEN; #endif assertEqualInt(flag, set & flag); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_different_window_size) { char buf[4096]; PROLOGUE("test_read_format_rar5_different_window_size.rar"); /* Return codes of those calls are ignored, because this sample file * is invalid. However, the unpacker shouldn't produce any SIGSEGV * errors during processing. */ (void) archive_read_next_header(a, &ae); while(0 < archive_read_data(a, buf, sizeof(buf))) {} (void) archive_read_next_header(a, &ae); while(0 < archive_read_data(a, buf, sizeof(buf))) {} (void) archive_read_next_header(a, &ae); while(0 < archive_read_data(a, buf, sizeof(buf))) {} EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_window_buf_and_size_desync) { /* oss fuzz 30442 */ char buf[4096]; PROLOGUE("test_read_format_rar5_window_buf_and_size_desync.rar"); /* Return codes of those calls are ignored, because this sample file * is invalid. However, the unpacker shouldn't produce any SIGSEGV * errors during processing. */ (void) archive_read_next_header(a, &ae); while(0 < archive_read_data(a, buf, 46)) {} EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_arm_filter_on_window_boundary) { char buf[4096]; PROLOGUE("test_read_format_rar5_arm_filter_on_window_boundary.rar"); /* Return codes of those calls are ignored, because this sample file * is invalid. However, the unpacker shouldn't produce any SIGSEGV * errors during processing. */ (void) archive_read_next_header(a, &ae); while(0 < archive_read_data(a, buf, sizeof(buf))) {} EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_different_solid_window_size) { char buf[4096]; PROLOGUE("test_read_format_rar5_different_solid_window_size.rar"); /* Return codes of those calls are ignored, because this sample file * is invalid. However, the unpacker shouldn't produce any SIGSEGV * errors during processing. */ (void) archive_read_next_header(a, &ae); while(0 < archive_read_data(a, buf, sizeof(buf))) {} (void) archive_read_next_header(a, &ae); while(0 < archive_read_data(a, buf, sizeof(buf))) {} (void) archive_read_next_header(a, &ae); while(0 < archive_read_data(a, buf, sizeof(buf))) {} EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_different_winsize_on_merge) { char buf[4096]; PROLOGUE("test_read_format_rar5_different_winsize_on_merge.rar"); /* Return codes of those calls are ignored, because this sample file * is invalid. However, the unpacker shouldn't produce any SIGSEGV * errors during processing. */ (void) archive_read_next_header(a, &ae); while(0 < archive_read_data(a, buf, sizeof(buf))) {} EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_block_size_is_too_small) { char buf[4096]; PROLOGUE("test_read_format_rar5_block_size_is_too_small.rar"); /* This file is damaged, so those functions should return failure. * Additionally, SIGSEGV shouldn't be raised during execution * of those functions. */ assertA(archive_read_next_header(a, &ae) != ARCHIVE_OK); assertA(archive_read_data(a, buf, sizeof(buf)) <= 0); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_sfx) { struct archive *a; struct archive_entry *ae; int bs = 10240; char buff[32]; const char reffile[] = "test_read_format_rar5_sfx.exe"; const char test_txt[] = "123"; int size = sizeof(test_txt) - 1; extract_reference_file(reffile); assert((a = archive_read_new()) != NULL); assertA(0 == archive_read_support_filter_all(a)); assertA(0 == archive_read_support_format_all(a)); assertA(0 == archive_read_open_filename(a, reffile, bs)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualString("test.txt.txt", archive_entry_pathname(ae)); assertA(size == archive_read_data(a, buff, size)); assertEqualMem(buff, test_txt, size); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_decode_number_out_of_bounds_read) { /* oss fuzz 30448 */ char buf[4096]; PROLOGUE("test_read_format_rar5_decode_number_out_of_bounds_read.rar"); /* Return codes of those calls are ignored, because this sample file * is invalid. However, the unpacker shouldn't produce any SIGSEGV * errors during processing. */ (void) archive_read_next_header(a, &ae); while(0 < archive_read_data(a, buf, sizeof(buf))) {} EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_bad_window_size_in_multiarchive_file) { /* oss fuzz 30459 */ char buf[4096]; PROLOGUE("test_read_format_rar5_bad_window_sz_in_mltarc_file.rar"); /* This file is damaged, so those functions should return failure. * Additionally, SIGSEGV shouldn't be raised during execution * of those functions. */ (void) archive_read_next_header(a, &ae); while(0 < archive_read_data(a, buf, sizeof(buf))) {} (void) archive_read_next_header(a, &ae); while(0 < archive_read_data(a, buf, sizeof(buf))) {} EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_read_data_block_uninitialized_offset) { const void *buf; size_t size; la_int64_t offset; PROLOGUE("test_read_format_rar5_compressed.rar"); assertA(0 == archive_read_next_header(a, &ae)); /* A real code may pass a pointer to an uninitialized variable as an offset * output argument. Here we want to check this situation. But because * relying on a value of an uninitialized variable in a test is not a good * idea, let's pretend that 0xdeadbeef is a random value of the * uninitialized variable. */ offset = 0xdeadbeef; assertEqualInt(ARCHIVE_OK, archive_read_data_block(a, &buf, &size, &offset)); /* The test archive doesn't contain a sparse file. And because of that, here * we assume that the first returned offset should be 0. */ assertEqualInt(0, offset); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_data_ready_pointer_leak) { /* oss fuzz 70024 */ char buf[4096]; PROLOGUE("test_read_format_rar5_data_ready_pointer_leak.rar"); /* Return codes of those calls are ignored, because this sample file * is invalid. However, the unpacker shouldn't produce any SIGSEGV * errors during processing. */ (void) archive_read_next_header(a, &ae); (void) archive_read_data(a, buf, sizeof(buf)); (void) archive_read_next_header(a, &ae); (void) archive_read_data(a, buf, sizeof(buf)); (void) archive_read_data(a, buf, sizeof(buf)); (void) archive_read_next_header(a, &ae); /* This call shouldn't produce SIGSEGV. */ (void) archive_read_data(a, buf, sizeof(buf)); EPILOGUE(); } + +DEFINE_TEST(test_read_format_rar5_only_crypt_exfld) +{ + /* GH #2711 */ + + char buf[4096]; + PROLOGUE("test_read_format_rar5_only_crypt_exfld.rar"); + + /* The reader should allow iteration through files, but should fail + during data extraction. */ + + assertA(archive_read_next_header(a, &ae) == ARCHIVE_OK); + assertA(archive_read_data(a, buf, sizeof(buf)) == ARCHIVE_FATAL); + + /* The reader should also provide a valid error message. */ + assertA(archive_error_string(a) != NULL); + + EPILOGUE(); +} + +DEFINE_TEST(test_read_format_rar5_only_unsupported_exfld) +{ + /* GH #2711 */ + + char buf[4096]; + PROLOGUE("test_read_format_rar5_unsupported_exfld.rar"); + + /* The reader should allow iteration through files, and it should + succeed with data extraction. */ + + assertA(archive_read_next_header(a, &ae) == ARCHIVE_OK); + + /* 48 is the expected number of bytes that should be extracted */ + assertA(archive_read_data(a, buf, sizeof(buf)) == 48); + + EPILOGUE(); +} + +DEFINE_TEST(test_read_format_rar5_invalidhash_and_validhtime_exfld) +{ + /* GH #2711 */ + + char buf[4096]; + PROLOGUE("test_read_format_rar5_invalid_hash_valid_htime_exfld.rar"); + + /* The reader should report an error when trying to process this data. + Returning EOF here means that the reader has failed to identify + malformed structure. */ + + assertA(archive_read_next_header(a, &ae) < 0); + assertA(archive_read_data(a, buf, sizeof(buf)) < 0); + + EPILOGUE(); +} diff --git a/contrib/libarchive/libarchive/test/test_read_format_rar5_dirdata.rar.uu b/contrib/libarchive/libarchive/test/test_read_format_rar5_dirdata.rar.uu new file mode 100644 index 000000000000..c7928f344a80 --- /dev/null +++ b/contrib/libarchive/libarchive/test/test_read_format_rar5_dirdata.rar.uu @@ -0,0 +1,6 @@ +begin 644 - +M4F%R(1H'`0`BD'[;,`$%,#8P`0&`@("``B?GD;$U`@(+@X``"_C5%:2#``(` +M`#"``S`P,#`P,#`P,#!);S#6KA',@]:N$?*IN;YV[8"1S>?4^`,#`R,#`P,# +-`P,#1)'C@XX*4`O.^P`` +` +end diff --git a/contrib/libarchive/libarchive/test/test_read_format_rar5_invalid_hash_valid_htime_exfld.rar.uu b/contrib/libarchive/libarchive/test/test_read_format_rar5_invalid_hash_valid_htime_exfld.rar.uu new file mode 100644 index 000000000000..399acd814ae4 --- /dev/null +++ b/contrib/libarchive/libarchive/test/test_read_format_rar5_invalid_hash_valid_htime_exfld.rar.uu @@ -0,0 +1,6 @@ +begin 644 - +M4F%R(1H'`0`SDK7E"@$%!@`%`0&`@`#^T/5L)`(###$$,>V#`D840I@``0AF +M:6QE+G1X=`@"OX0]``$"`P(#`&EN=F%L:60@2$%32"!E>'1R82P@86YD(&QA +>=&5R(&$@=F%L:60@2%1)344@97AT)!,``0AF +M:6QE+G1X="0!```&``````````````````````````````````````````!R +M87(U('-T;W)E9"!F:6QE('=I=&@@;VYL>2!A($-265!4(&5X=')A(&9I96QD +'(#W?A@(%!``` +` +end diff --git a/contrib/libarchive/libarchive/test/test_read_format_rar5_unsupported_exfld.rar.uu b/contrib/libarchive/libarchive/test/test_read_format_rar5_unsupported_exfld.rar.uu new file mode 100644 index 000000000000..16b456bf4873 --- /dev/null +++ b/contrib/libarchive/libarchive/test/test_read_format_rar5_unsupported_exfld.rar.uu @@ -0,0 +1,6 @@ +begin 644 - +M4F%R(1H'`0`SDK7E"@$%!@`%`0&`@`#>[JDS)@(##C`$,.V#`BX6Z[0``0AF +M:6QE+G1X=`W_____#WA6-!(`````^E#_1.OY96VEI*Z used1); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_memory(a, buff, used2)); for (i = 0; i < 999; i++) { snprintf(path, sizeof(path), "file%03d", i); if (!assertEqualInt(0, archive_read_next_header(a, &ae))) break; assertEqualString(path, archive_entry_pathname(ae)); assertEqualInt((int)datasize, archive_entry_size(ae)); } assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); /* * Test various premature shutdown scenarios to make sure we * don't crash or leak memory. */ assert((a = archive_write_new()) != NULL); if (use_prog) assertEqualIntA(a, ARCHIVE_WARN, archive_write_add_filter_bzip2(a)); else assertEqualIntA(a, ARCHIVE_OK, archive_write_add_filter_bzip2(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); assert((a = archive_write_new()) != NULL); if (use_prog) assertEqualIntA(a, ARCHIVE_WARN, archive_write_add_filter_bzip2(a)); else assertEqualIntA(a, ARCHIVE_OK, archive_write_add_filter_bzip2(a)); assertEqualInt(ARCHIVE_OK, archive_write_close(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_ustar(a)); if (use_prog) assertEqualIntA(a, ARCHIVE_WARN, archive_write_add_filter_bzip2(a)); else assertEqualIntA(a, ARCHIVE_OK, archive_write_add_filter_bzip2(a)); assertEqualInt(ARCHIVE_OK, archive_write_close(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_ustar(a)); if (use_prog) assertEqualIntA(a, ARCHIVE_WARN, archive_write_add_filter_bzip2(a)); else assertEqualIntA(a, ARCHIVE_OK, archive_write_add_filter_bzip2(a)); assertEqualIntA(a, ARCHIVE_OK, archive_write_open_memory(a, buff, buffsize, &used2)); assertEqualInt(ARCHIVE_OK, archive_write_close(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); + /* + * Test behavior after a fatal error (triggered by giving + * archive_write_open_memory() a very small buffer). + */ + if (!use_prog) { + used1 = 0; + assert((a = archive_write_new()) != NULL); + assertEqualIntA(a, ARCHIVE_OK, + archive_write_set_format_ustar(a)); + assertEqualIntA(a, ARCHIVE_OK, + archive_write_add_filter_bzip2(a)); + assertEqualIntA(a, ARCHIVE_OK, + archive_write_open_memory(a, buff, 100, &used1)); + assert((ae = archive_entry_new()) != NULL); + archive_entry_set_filetype(ae, AE_IFREG); + archive_entry_set_size(ae, 4000000); + archive_entry_copy_pathname(ae, "file"); + assertEqualIntA(a, ARCHIVE_OK, + archive_write_header(a, ae)); + for (i = 0; i < 1000000; i++) { + r = archive_write_data(a, &i, 4); + if (r == ARCHIVE_FATAL) + break; + } + assertEqualIntA(a, ARCHIVE_FATAL, r); + archive_entry_free(ae); + assertEqualInt(ARCHIVE_OK, archive_write_free(a)); + } + /* * Clean up. */ free(data); free(buff); } diff --git a/contrib/libarchive/libarchive/test/test_write_filter_gzip.c b/contrib/libarchive/libarchive/test/test_write_filter_gzip.c index 8fbdbed09744..a6681d7618b1 100644 --- a/contrib/libarchive/libarchive/test/test_write_filter_gzip.c +++ b/contrib/libarchive/libarchive/test/test_write_filter_gzip.c @@ -1,300 +1,306 @@ /*- * Copyright (c) 2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" /* * A basic exercise of gzip reading and writing. * * TODO: Add a reference file and make sure we can decompress that. */ DEFINE_TEST(test_write_filter_gzip) { struct archive_entry *ae; struct archive* a; char *buff, *data; size_t buffsize, datasize; unsigned char *rbuff; char path[16]; size_t used1, used2; int i, r, use_prog = 0; buffsize = 2000000; assert(NULL != (buff = malloc(buffsize))); if (buff == NULL) return; datasize = 10000; assert(NULL != (data = malloc(datasize))); if (data == NULL) { free(buff); return; } memset(data, 0, datasize); /* * Write a 100 files and read them all back. * Use default compression level (6). */ assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_ustar(a)); r = archive_write_add_filter_gzip(a); if (r != ARCHIVE_OK) { if (canGzip() && r == ARCHIVE_WARN) use_prog = 1; else { skipping("gzip writing not supported on this platform"); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); free(buff); free(data); return; } } assertEqualIntA(a, ARCHIVE_OK, archive_write_set_bytes_per_block(a, 10)); assertEqualInt(ARCHIVE_FILTER_GZIP, archive_filter_code(a, 0)); assertEqualString("gzip", archive_filter_name(a, 0)); assertEqualIntA(a, ARCHIVE_OK, archive_write_open_memory(a, buff, buffsize, &used1)); assert((ae = archive_entry_new()) != NULL); archive_entry_set_filetype(ae, AE_IFREG); archive_entry_set_size(ae, datasize); for (i = 0; i < 100; i++) { snprintf(path, sizeof(path), "file%03d", i); archive_entry_copy_pathname(ae, path); assertEqualIntA(a, ARCHIVE_OK, archive_write_header(a, ae)); assertA(datasize == (size_t)archive_write_data(a, data, datasize)); } archive_entry_free(ae); assertEqualIntA(a, ARCHIVE_OK, archive_write_close(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); /* Basic gzip header tests */ rbuff = (unsigned char *)buff; assertEqualInt(rbuff[0], 0x1f); assertEqualInt(rbuff[1], 0x8b); assertEqualInt(rbuff[2], 0x08); assertEqualInt(rbuff[3], 0x00); assertEqualInt(rbuff[8], 0); /* RFC 1952 flag for compression level 6 */ assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); r = archive_read_support_filter_gzip(a); if (r == ARCHIVE_WARN) { skipping("Can't verify gzip writing by reading back;" " gzip reading not fully supported on this platform"); } else { assertEqualIntA(a, ARCHIVE_OK, archive_read_open_memory(a, buff, used1)); for (i = 0; i < 100; i++) { snprintf(path, sizeof(path), "file%03d", i); if (!assertEqualInt(ARCHIVE_OK, archive_read_next_header(a, &ae))) break; assertEqualString(path, archive_entry_pathname(ae)); assertEqualInt((int)datasize, archive_entry_size(ae)); } assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); /* * Repeat the cycle again, this time setting some compression * options. Compression level is 9. */ assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_ustar(a)); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_bytes_per_block(a, 10)); assertEqualIntA(a, (use_prog)?ARCHIVE_WARN:ARCHIVE_OK, archive_write_add_filter_gzip(a)); assertEqualIntA(a, ARCHIVE_FAILED, archive_write_set_options(a, "gzip:nonexistent-option=0")); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_options(a, "gzip:compression-level=1")); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_filter_option(a, NULL, "compression-level", "9")); assertEqualIntA(a, ARCHIVE_FAILED, archive_write_set_filter_option(a, NULL, "compression-level", "abc")); assertEqualIntA(a, ARCHIVE_FAILED, archive_write_set_filter_option(a, NULL, "compression-level", "99")); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_options(a, "gzip:compression-level=9")); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_options(a, "gzip:original-filename=testorgfilename")); assertEqualIntA(a, ARCHIVE_OK, archive_write_open_memory(a, buff, buffsize, &used2)); for (i = 0; i < 100; i++) { snprintf(path, sizeof(path), "file%03d", i); assert((ae = archive_entry_new()) != NULL); archive_entry_copy_pathname(ae, path); archive_entry_set_size(ae, datasize); archive_entry_set_filetype(ae, AE_IFREG); assertEqualIntA(a, ARCHIVE_OK, archive_write_header(a, ae)); assertA(datasize == (size_t)archive_write_data(a, data, datasize)); archive_entry_free(ae); } assertEqualIntA(a, ARCHIVE_OK, archive_write_close(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); /* Basic gzip header tests */ rbuff = (unsigned char *)buff; assertEqualInt(rbuff[0], 0x1f); assertEqualInt(rbuff[1], 0x8b); assertEqualInt(rbuff[2], 0x08); - assertEqualInt(rbuff[3], 0x08); - assertEqualInt(rbuff[8], 2); /* RFC 1952 flag for compression level 9 */ - assertEqualString((const char*)rbuff+10, "testorgfilename"); + /* RFC 1952 flag for compression level 9 */ + assertEqualInt(rbuff[8], 2); + /* External gzip program might not save filename */ + if (!use_prog || rbuff[3] == 0x08) { + assertEqualInt(rbuff[3], 0x08); + assertEqualString((const char*)rbuff+10, "testorgfilename"); + } else { + assertEqualInt(rbuff[3], 0x00); + } /* Curiously, this test fails; the test data above compresses * better at default compression than at level 9. */ /* failure("compression-level=9 wrote %d bytes, default wrote %d bytes", (int)used2, (int)used1); assert(used2 < used1); */ assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); r = archive_read_support_filter_gzip(a); if (r != ARCHIVE_OK && !use_prog) { skipping("gzip reading not fully supported on this platform"); } else { assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_memory(a, buff, used2)); for (i = 0; i < 100; i++) { snprintf(path, sizeof(path), "file%03d", i); if (!assertEqualInt(ARCHIVE_OK, archive_read_next_header(a, &ae))) break; assertEqualString(path, archive_entry_pathname(ae)); assertEqualInt((int)datasize, archive_entry_size(ae)); } assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); /* * Repeat again, with compression level 1 */ assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_ustar(a)); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_bytes_per_block(a, 10)); assertEqualIntA(a, (use_prog)?ARCHIVE_WARN:ARCHIVE_OK, archive_write_add_filter_gzip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_filter_option(a, NULL, "compression-level", "1")); assertEqualIntA(a, ARCHIVE_OK, archive_write_open_memory(a, buff, buffsize, &used2)); for (i = 0; i < 100; i++) { snprintf(path, sizeof(path), "file%03d", i); assert((ae = archive_entry_new()) != NULL); archive_entry_copy_pathname(ae, path); archive_entry_set_size(ae, datasize); archive_entry_set_filetype(ae, AE_IFREG); assertEqualIntA(a, ARCHIVE_OK, archive_write_header(a, ae)); failure("Writing file %s", path); assertEqualIntA(a, datasize, (size_t)archive_write_data(a, data, datasize)); archive_entry_free(ae); } assertEqualIntA(a, ARCHIVE_OK, archive_write_close(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); /* Level 1 really does result in larger data. */ failure("Compression-level=1 wrote %d bytes; default wrote %d bytes", (int)used2, (int)used1); assert(used2 > used1); /* Basic gzip header tests */ rbuff = (unsigned char *)buff; assertEqualInt(rbuff[0], 0x1f); assertEqualInt(rbuff[1], 0x8b); assertEqualInt(rbuff[2], 0x08); assertEqualInt(rbuff[3], 0x00); assertEqualInt(rbuff[8], 4); /* RFC 1952 flag for compression level 1 */ assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); r = archive_read_support_filter_gzip(a); if (r == ARCHIVE_WARN) { skipping("gzip reading not fully supported on this platform"); } else { assertEqualIntA(a, ARCHIVE_OK, archive_read_open_memory(a, buff, used2)); for (i = 0; i < 100; i++) { snprintf(path, sizeof(path), "file%03d", i); if (!assertEqualInt(ARCHIVE_OK, archive_read_next_header(a, &ae))) break; assertEqualString(path, archive_entry_pathname(ae)); assertEqualInt((int)datasize, archive_entry_size(ae)); } assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); } assertEqualInt(ARCHIVE_OK, archive_read_free(a)); /* * Test various premature shutdown scenarios to make sure we * don't crash or leak memory. */ assert((a = archive_write_new()) != NULL); assertEqualIntA(a, (use_prog)?ARCHIVE_WARN:ARCHIVE_OK, archive_write_add_filter_gzip(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); assert((a = archive_write_new()) != NULL); assertEqualIntA(a, (use_prog)?ARCHIVE_WARN:ARCHIVE_OK, archive_write_add_filter_gzip(a)); assertEqualInt(ARCHIVE_OK, archive_write_close(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_ustar(a)); assertEqualIntA(a, (use_prog)?ARCHIVE_WARN:ARCHIVE_OK, archive_write_add_filter_gzip(a)); assertEqualInt(ARCHIVE_OK, archive_write_close(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_ustar(a)); assertEqualIntA(a, (use_prog)?ARCHIVE_WARN:ARCHIVE_OK, archive_write_add_filter_gzip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_write_open_memory(a, buff, buffsize, &used2)); assertEqualInt(ARCHIVE_OK, archive_write_close(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); /* * Clean up. */ free(data); free(buff); } diff --git a/contrib/libarchive/libarchive/test/test_write_filter_gzip_timestamp.c b/contrib/libarchive/libarchive/test/test_write_filter_gzip_timestamp.c index a148f818dcec..d0496b025b64 100644 --- a/contrib/libarchive/libarchive/test/test_write_filter_gzip_timestamp.c +++ b/contrib/libarchive/libarchive/test/test_write_filter_gzip_timestamp.c @@ -1,117 +1,120 @@ /*- * Copyright (c) 2007 Tim Kientzle * Copyright (c) 2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" DEFINE_TEST(test_write_filter_gzip_timestamp) { struct archive_entry *ae; struct archive* a; char *buff, *data; size_t buffsize, datasize; size_t used1; int r, use_prog = 0; buffsize = 10000; assert(NULL != (buff = malloc(buffsize))); if (buff == NULL) return; datasize = 10000; assert(NULL != (data = malloc(datasize))); if (data == NULL) { free(buff); return; } memset(data, 0, datasize); /* Test1: set "gzip:timestamp" option. */ assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_ustar(a)); r = archive_write_add_filter_gzip(a); if (r != ARCHIVE_OK) { if (canGzip() && r == ARCHIVE_WARN) use_prog = 1; else { skipping("gzip writing not supported on this platform"); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); free(buff); free(data); return; } } assertEqualIntA(a, ARCHIVE_OK, archive_write_set_options(a, "gzip:timestamp")); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_bytes_per_block(a, 10)); assertEqualInt(ARCHIVE_FILTER_GZIP, archive_filter_code(a, 0)); assertEqualString("gzip", archive_filter_name(a, 0)); assertEqualIntA(a, ARCHIVE_OK, archive_write_open_memory(a, buff, buffsize, &used1)); assert((ae = archive_entry_new()) != NULL); archive_entry_set_filetype(ae, AE_IFREG); archive_entry_set_size(ae, datasize); archive_entry_copy_pathname(ae, "file"); assertEqualIntA(a, ARCHIVE_OK, archive_write_header(a, ae)); assertEqualIntA(a, datasize, archive_write_data(a, data, datasize)); archive_entry_free(ae); assertEqualIntA(a, ARCHIVE_OK, archive_write_close(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); - failure("Timestamp should be recorded"); - assert(memcmp(buff + 4, "\x00\x00\x00\x00", 4) != 0); + /* External gzip program might not save timestamp */ + if (!use_prog) { + failure("Timestamp should be recorded"); + assert(memcmp(buff + 4, "\x00\x00\x00\x00", 4) != 0); + } /* Test2: set "gzip:!timestamp" option. */ assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_ustar(a)); assertEqualIntA(a, (use_prog)?ARCHIVE_WARN:ARCHIVE_OK, archive_write_add_filter_gzip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_options(a, "gzip:!timestamp")); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_bytes_per_block(a, 10)); assertEqualInt(ARCHIVE_FILTER_GZIP, archive_filter_code(a, 0)); assertEqualString("gzip", archive_filter_name(a, 0)); assertEqualIntA(a, ARCHIVE_OK, archive_write_open_memory(a, buff, buffsize, &used1)); assert((ae = archive_entry_new()) != NULL); archive_entry_set_filetype(ae, AE_IFREG); archive_entry_set_size(ae, datasize); archive_entry_copy_pathname(ae, "file"); assertEqualIntA(a, ARCHIVE_OK, archive_write_header(a, ae)); assertEqualIntA(a, datasize, archive_write_data(a, data, datasize)); archive_entry_free(ae); assertEqualIntA(a, ARCHIVE_OK, archive_write_close(a)); assertEqualInt(ARCHIVE_OK, archive_write_free(a)); failure("Timestamp should not be recorded"); assertEqualMem(buff + 4, "\x00\x00\x00\x00", 4); /* * Clean up. */ free(data); free(buff); } diff --git a/contrib/libarchive/libarchive_fe/err.c b/contrib/libarchive/libarchive_fe/lafe_err.c similarity index 99% rename from contrib/libarchive/libarchive_fe/err.c rename to contrib/libarchive/libarchive_fe/lafe_err.c index f6dcf44af347..b108b9b1fea7 100644 --- a/contrib/libarchive/libarchive_fe/err.c +++ b/contrib/libarchive/libarchive_fe/lafe_err.c @@ -1,98 +1,98 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "lafe_platform.h" #ifdef HAVE_STDARG_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif -#include "err.h" +#include "lafe_err.h" static void lafe_vwarnc(int, const char *, va_list) __LA_PRINTFLIKE(2, 0); static const char *lafe_progname; const char * lafe_getprogname(void) { return lafe_progname; } void lafe_setprogname(const char *name, const char *defaultname) { if (name == NULL) name = defaultname; #if defined(_WIN32) && !defined(__CYGWIN__) lafe_progname = strrchr(name, '\\'); if (strrchr(name, '/') > lafe_progname) #endif lafe_progname = strrchr(name, '/'); if (lafe_progname != NULL) lafe_progname++; else lafe_progname = name; } static void lafe_vwarnc(int code, const char *fmt, va_list ap) { fprintf(stderr, "%s: ", lafe_progname); vfprintf(stderr, fmt, ap); if (code != 0) fprintf(stderr, ": %s", strerror(code)); fprintf(stderr, "\n"); } void lafe_warnc(int code, const char *fmt, ...) { va_list ap; va_start(ap, fmt); lafe_vwarnc(code, fmt, ap); va_end(ap); } void lafe_errc(int eval, int code, const char *fmt, ...) { va_list ap; va_start(ap, fmt); lafe_vwarnc(code, fmt, ap); va_end(ap); exit(eval); } diff --git a/contrib/libarchive/libarchive_fe/err.h b/contrib/libarchive/libarchive_fe/lafe_err.h similarity index 100% rename from contrib/libarchive/libarchive_fe/err.h rename to contrib/libarchive/libarchive_fe/lafe_err.h diff --git a/contrib/libarchive/libarchive_fe/line_reader.c b/contrib/libarchive/libarchive_fe/line_reader.c index 7f0429ece3cf..0af9db53c0a2 100644 --- a/contrib/libarchive/libarchive_fe/line_reader.c +++ b/contrib/libarchive/libarchive_fe/line_reader.c @@ -1,166 +1,166 @@ /*- * Copyright (c) 2008 Tim Kientzle * Copyright (c) 2010 Joerg Sonnenberger * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "lafe_platform.h" #include #include #include #include -#include "err.h" +#include "lafe_err.h" #include "line_reader.h" #if defined(_WIN32) && !defined(__CYGWIN__) && !defined(__BORLANDC__) #define strdup _strdup #endif /* * Read lines from file and do something with each one. If option_null * is set, lines are terminated with zero bytes; otherwise, they're * terminated with newlines. * * This uses a self-sizing buffer to handle arbitrarily-long lines. */ struct lafe_line_reader { FILE *f; char *buff, *buff_end, *line_start, *line_end; char *pathname; size_t buff_length; int nullSeparator; /* Lines separated by null, not CR/CRLF/etc. */ }; struct lafe_line_reader * lafe_line_reader(const char *pathname, int nullSeparator) { struct lafe_line_reader *lr; lr = calloc(1, sizeof(*lr)); if (lr == NULL) lafe_errc(1, ENOMEM, "Can't open %s", pathname); lr->nullSeparator = nullSeparator; lr->pathname = strdup(pathname); if (strcmp(pathname, "-") == 0) lr->f = stdin; else lr->f = fopen(pathname, "r"); if (lr->f == NULL) lafe_errc(1, errno, "Couldn't open %s", pathname); lr->buff_length = 8192; lr->line_start = lr->line_end = lr->buff_end = lr->buff = NULL; return (lr); } static void lafe_line_reader_find_eol(struct lafe_line_reader *lr) { lr->line_end += strcspn(lr->line_end, lr->nullSeparator ? "" : "\x0d\x0a"); *lr->line_end = '\0'; /* Noop if line_end == buff_end */ } const char * lafe_line_reader_next(struct lafe_line_reader *lr) { size_t bytes_wanted, bytes_read, new_buff_size; char *line_start, *p; for (;;) { /* If there's a line in the buffer, return it immediately. */ while (lr->line_end < lr->buff_end) { line_start = lr->line_start; lr->line_start = ++lr->line_end; lafe_line_reader_find_eol(lr); if (lr->nullSeparator || line_start[0] != '\0') return (line_start); } /* If we're at end-of-file, process the final data. */ if (lr->f == NULL) { if (lr->line_start == lr->buff_end) return (NULL); /* No more text */ line_start = lr->line_start; lr->line_start = lr->buff_end; return (line_start); } /* Buffer only has part of a line. */ if (lr->line_start > lr->buff) { /* Move a leftover fractional line to the beginning. */ memmove(lr->buff, lr->line_start, lr->buff_end - lr->line_start); lr->buff_end -= lr->line_start - lr->buff; lr->line_end -= lr->line_start - lr->buff; lr->line_start = lr->buff; } else { /* Line is too big; enlarge the buffer. */ new_buff_size = lr->buff_length * 2; if (new_buff_size <= lr->buff_length) lafe_errc(1, ENOMEM, "Line too long in %s", lr->pathname); lr->buff_length = new_buff_size; /* * Allocate one extra byte to allow terminating * the buffer. */ p = realloc(lr->buff, new_buff_size + 1); if (p == NULL) lafe_errc(1, ENOMEM, "Line too long in %s", lr->pathname); lr->buff_end = p + (lr->buff_end - lr->buff); lr->line_end = p + (lr->line_end - lr->buff); lr->line_start = lr->buff = p; } /* Get some more data into the buffer. */ bytes_wanted = lr->buff + lr->buff_length - lr->buff_end; bytes_read = fread(lr->buff_end, 1, bytes_wanted, lr->f); lr->buff_end += bytes_read; *lr->buff_end = '\0'; /* Always terminate buffer */ lafe_line_reader_find_eol(lr); if (ferror(lr->f)) lafe_errc(1, errno, "Can't read %s", lr->pathname); if (feof(lr->f)) { if (lr->f != stdin) fclose(lr->f); lr->f = NULL; } } } void lafe_line_reader_free(struct lafe_line_reader *lr) { free(lr->buff); free(lr->pathname); free(lr); } diff --git a/contrib/libarchive/libarchive_fe/passphrase.c b/contrib/libarchive/libarchive_fe/passphrase.c index 527ad2d63654..1b7a50453664 100644 --- a/contrib/libarchive/libarchive_fe/passphrase.c +++ b/contrib/libarchive/libarchive_fe/passphrase.c @@ -1,343 +1,343 @@ /*- * Copyright (c) 2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /* $OpenBSD: readpassphrase.c,v 1.27 2019/01/25 00:19:25 millert Exp $ */ /* * Copyright (c) 2000-2002, 2007, 2010 * Todd C. Miller * * Permission to use, copy, modify, and distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. * * Sponsored in part by the Defense Advanced Research Projects * Agency (DARPA) and Air Force Research Laboratory, Air Force * Materiel Command, USAF, under agreement number F39502-99-1-0512. */ /* OPENBSD ORIGINAL: lib/libc/gen/readpassphrase.c */ #include "lafe_platform.h" #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_READPASSPHRASE_H #include #endif -#include "err.h" +#include "lafe_err.h" #include "passphrase.h" #ifndef HAVE_READPASSPHRASE #define RPP_ECHO_OFF 0x00 /* Turn off echo (default). */ #define RPP_ECHO_ON 0x01 /* Leave echo on. */ #define RPP_REQUIRE_TTY 0x02 /* Fail if there is no tty. */ #define RPP_FORCELOWER 0x04 /* Force input to lower case. */ #define RPP_FORCEUPPER 0x08 /* Force input to upper case. */ #define RPP_SEVENBIT 0x10 /* Strip the high bit from input. */ #define RPP_STDIN 0x20 /* Read from stdin, not /dev/tty */ #if defined(_WIN32) && !defined(__CYGWIN__) #include #include static char * readpassphrase(const char *prompt, char *buf, size_t bufsiz, int flags) { HANDLE hStdin, hStdout; DWORD mode, rbytes; BOOL success; (void)flags; hStdin = GetStdHandle(STD_INPUT_HANDLE); if (hStdin == INVALID_HANDLE_VALUE) return (NULL); hStdout = GetStdHandle(STD_OUTPUT_HANDLE); if (hStdout == INVALID_HANDLE_VALUE) return (NULL); success = GetConsoleMode(hStdin, &mode); if (!success) return (NULL); mode &= ~ENABLE_ECHO_INPUT; mode |= ENABLE_LINE_INPUT | ENABLE_PROCESSED_INPUT; success = SetConsoleMode(hStdin, mode); if (!success) return (NULL); success = WriteFile(hStdout, prompt, (DWORD)strlen(prompt), NULL, NULL); if (!success) return (NULL); success = ReadFile(hStdin, buf, (DWORD)bufsiz - 1, &rbytes, NULL); if (!success) return (NULL); WriteFile(hStdout, "\r\n", 2, NULL, NULL); buf[rbytes] = '\0'; /* Remove trailing carriage return(s). */ buf[strcspn(buf, "\r\n")] = '\0'; return (buf); } #elif defined(HAVE_TCGETATTR) && defined(HAVE_TCSETATTR) #include #include #include #ifdef HAVE_PATHS_H #include #endif #include #include #include #include #ifndef _PATH_TTY #define _PATH_TTY "/dev/tty" #endif #ifdef TCSASOFT # define _T_FLUSH (TCSAFLUSH|TCSASOFT) #else # define _T_FLUSH (TCSAFLUSH) #endif /* SunOS 4.x which lacks _POSIX_VDISABLE, but has VDISABLE */ #if !defined(_POSIX_VDISABLE) && defined(VDISABLE) # define _POSIX_VDISABLE VDISABLE #endif #define M(a,b) (a > b ? a : b) #define MAX_SIGNO M(M(M(SIGALRM, SIGHUP), \ M(SIGINT, SIGPIPE)), \ M(M(SIGQUIT, SIGTERM), \ M(M(SIGTSTP, SIGTTIN), SIGTTOU))) static volatile sig_atomic_t signo[MAX_SIGNO + 1]; static void handler(int s) { assert(s <= MAX_SIGNO); signo[s] = 1; } static char * readpassphrase(const char *prompt, char *buf, size_t bufsiz, int flags) { ssize_t nr; int input, output, save_errno, i, need_restart; char ch, *p, *end; struct termios term, oterm; #ifdef HAVE_SIGACTION struct sigaction sa, savealrm, saveint, savehup, savequit, saveterm; struct sigaction savetstp, savettin, savettou, savepipe; #endif /* I suppose we could alloc on demand in this case (XXX). */ if (bufsiz == 0) { errno = EINVAL; return(NULL); } restart: for (i = 0; i <= MAX_SIGNO; i++) signo[i] = 0; nr = -1; save_errno = 0; need_restart = 0; /* * Read and write to /dev/tty if available. If not, read from * stdin and write to stderr unless a tty is required. */ if ((flags & RPP_STDIN) || (input = output = open(_PATH_TTY, O_RDWR)) == -1) { if (flags & RPP_REQUIRE_TTY) { errno = ENOTTY; return(NULL); } input = STDIN_FILENO; output = STDERR_FILENO; } /* * Turn off echo if possible. * If we are using a tty but are not the foreground pgrp this will * generate SIGTTOU, so do it *before* installing the signal handlers. */ if (input != STDIN_FILENO && tcgetattr(input, &oterm) == 0) { memcpy(&term, &oterm, sizeof(term)); if (!(flags & RPP_ECHO_ON)) term.c_lflag &= ~(ECHO | ECHONL); #ifdef VSTATUS if (term.c_cc[VSTATUS] != _POSIX_VDISABLE) term.c_cc[VSTATUS] = _POSIX_VDISABLE; #endif (void)tcsetattr(input, _T_FLUSH, &term); } else { memset(&term, 0, sizeof(term)); term.c_lflag |= ECHO; memset(&oterm, 0, sizeof(oterm)); oterm.c_lflag |= ECHO; } #ifdef HAVE_SIGACTION /* * Catch signals that would otherwise cause the user to end * up with echo turned off in the shell. Don't worry about * things like SIGXCPU and SIGVTALRM for now. */ sigemptyset(&sa.sa_mask); sa.sa_flags = 0; /* don't restart system calls */ sa.sa_handler = handler; /* Keep this list in sync with MAX_SIGNO! */ (void)sigaction(SIGALRM, &sa, &savealrm); (void)sigaction(SIGHUP, &sa, &savehup); (void)sigaction(SIGINT, &sa, &saveint); (void)sigaction(SIGPIPE, &sa, &savepipe); (void)sigaction(SIGQUIT, &sa, &savequit); (void)sigaction(SIGTERM, &sa, &saveterm); (void)sigaction(SIGTSTP, &sa, &savetstp); (void)sigaction(SIGTTIN, &sa, &savettin); (void)sigaction(SIGTTOU, &sa, &savettou); #endif if (!(flags & RPP_STDIN)) { int r = write(output, prompt, strlen(prompt)); (void)r; } end = buf + bufsiz - 1; p = buf; while ((nr = read(input, &ch, 1)) == 1 && ch != '\n' && ch != '\r') { if (p < end) { if ((flags & RPP_SEVENBIT)) ch &= 0x7f; if (isalpha((unsigned char)ch)) { if ((flags & RPP_FORCELOWER)) ch = (char)tolower((unsigned char)ch); if ((flags & RPP_FORCEUPPER)) ch = (char)toupper((unsigned char)ch); } *p++ = ch; } } *p = '\0'; save_errno = errno; if (!(term.c_lflag & ECHO)) { int r = write(output, "\n", 1); (void)r; } /* Restore old terminal settings and signals. */ if (memcmp(&term, &oterm, sizeof(term)) != 0) { const int sigttou = signo[SIGTTOU]; /* Ignore SIGTTOU generated when we are not the fg pgrp. */ while (tcsetattr(input, _T_FLUSH, &oterm) == -1 && errno == EINTR && !signo[SIGTTOU]) continue; signo[SIGTTOU] = sigttou; } #ifdef HAVE_SIGACTION (void)sigaction(SIGALRM, &savealrm, NULL); (void)sigaction(SIGHUP, &savehup, NULL); (void)sigaction(SIGINT, &saveint, NULL); (void)sigaction(SIGQUIT, &savequit, NULL); (void)sigaction(SIGPIPE, &savepipe, NULL); (void)sigaction(SIGTERM, &saveterm, NULL); (void)sigaction(SIGTSTP, &savetstp, NULL); (void)sigaction(SIGTTIN, &savettin, NULL); (void)sigaction(SIGTTOU, &savettou, NULL); #endif if (input != STDIN_FILENO) (void)close(input); /* * If we were interrupted by a signal, resend it to ourselves * now that we have restored the signal handlers. */ for (i = 0; i <= MAX_SIGNO; i++) { if (signo[i]) { kill(getpid(), i); switch (i) { case SIGTSTP: case SIGTTIN: case SIGTTOU: need_restart = 1; } } } if (need_restart) goto restart; if (save_errno) errno = save_errno; return(nr == -1 ? NULL : buf); } #else static char * readpassphrase(const char *prompt, char *buf, size_t bufsiz, int flags) { return (NULL); } #endif #endif /* HAVE_READPASSPHRASE */ char * lafe_readpassphrase(const char *prompt, char *buf, size_t bufsiz) { char *p; p = readpassphrase(prompt, buf, bufsiz, RPP_ECHO_OFF); if (p == NULL) { switch (errno) { case EINTR: break; default: lafe_errc(1, errno, "Couldn't read passphrase"); /* NOTREACHED */ } } return (p); } diff --git a/contrib/libarchive/tar/bsdtar.c b/contrib/libarchive/tar/bsdtar.c index 53ac135f0129..92e86fd6bd94 100644 --- a/contrib/libarchive/tar/bsdtar.c +++ b/contrib/libarchive/tar/bsdtar.c @@ -1,1135 +1,1140 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2003-2008 Tim Kientzle * All rights reserved. */ #include "bsdtar_platform.h" #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_SYS_PARAM_H #include #endif #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_COPYFILE_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_LANGINFO_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_LOCALE_H #include #endif #ifdef HAVE_PATHS_H #include #endif #ifdef HAVE_SIGNAL_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_TIME_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #include "bsdtar.h" -#include "err.h" +#include "lafe_err.h" #if ARCHIVE_VERSION_NUMBER < 4000000 && !defined(_PATH_DEFTAPE) // Libarchive 4.0 and later will NOT define _PATH_DEFTAPE // but will honor it if it's set in the build. // Until then, we'll continue to set it by default on certain platforms: #if defined(__linux) #define _PATH_DEFTAPE "/dev/st0" #elif defined(_WIN32) && !defined(__CYGWIN__) #define _PATH_DEFTAPE "\\\\.\\tape0" #elif !defined(__APPLE__) #define _PATH_DEFTAPE "/dev/tape" #endif #endif #define _PATH_STDIO "-" #ifdef __MINGW32__ int _CRT_glob = 0; /* Disable broken CRT globbing. */ #endif #if defined(HAVE_SIGACTION) && (defined(SIGINFO) || defined(SIGUSR1)) static volatile int siginfo_occurred; static void siginfo_handler(int sig) { (void)sig; /* UNUSED */ siginfo_occurred = 1; } int need_report(void) { int r = siginfo_occurred; siginfo_occurred = 0; return (r); } #else int need_report(void) { return (0); } #endif static __LA_NORETURN void long_help(void); static void only_mode(struct bsdtar *, const char *opt, const char *valid); static void set_mode(struct bsdtar *, int opt); static __LA_NORETURN void version(void); /* A basic set of security flags to request from libarchive. */ #define SECURITY \ (ARCHIVE_EXTRACT_SECURE_SYMLINKS \ | ARCHIVE_EXTRACT_SECURE_NODOTDOT) static char const * const vcs_files[] = { /* CVS */ "CVS", ".cvsignore", /* RCS */ "RCS", /* SCCS */ "SCCS", /* SVN */ ".svn", /* git */ ".git", ".gitignore", ".gitattributes", ".gitmodules", /* Arch */ ".arch-ids", "{arch}", "=RELEASE-ID", "=meta-update", "=update", /* Bazaar */ ".bzr", ".bzrignore", ".bzrtags", /* Mercurial */ ".hg", ".hgignore", ".hgtags", /* darcs */ "_darcs", NULL }; int main(int argc, char **argv) { struct bsdtar *bsdtar, bsdtar_storage; int opt, t; int compression, compression2; const char *compression_name, *compression2_name; const char *compress_program; char *tptr, *uptr; char possible_help_request; char buff[16]; long l; time_t now; /* * Use a pointer for consistency, but stack-allocated storage * for ease of cleanup. */ bsdtar = &bsdtar_storage; memset(bsdtar, 0, sizeof(*bsdtar)); bsdtar->fd = -1; /* Mark as "unused" */ bsdtar->gid = -1; bsdtar->uid = -1; bsdtar->flags = 0; compression = compression2 = '\0'; compression_name = compression2_name = NULL; compress_program = NULL; time(&now); #if defined(HAVE_SIGACTION) { /* Set up signal handling. */ struct sigaction sa; sa.sa_handler = siginfo_handler; sigemptyset(&sa.sa_mask); sa.sa_flags = 0; #ifdef SIGINFO if (sigaction(SIGINFO, &sa, NULL)) lafe_errc(1, errno, "sigaction(SIGINFO) failed"); #endif #ifdef SIGUSR1 /* ... and treat SIGUSR1 the same way as SIGINFO. */ if (sigaction(SIGUSR1, &sa, NULL)) lafe_errc(1, errno, "sigaction(SIGUSR1) failed"); #endif #ifdef SIGPIPE /* Ignore SIGPIPE signals. */ sa.sa_handler = SIG_IGN; sigaction(SIGPIPE, &sa, NULL); +#endif +#ifdef SIGCHLD + /* Do not ignore SIGCHLD. */ + sa.sa_handler = SIG_DFL; + sigaction(SIGCHLD, &sa, NULL); #endif } #endif /* Set lafe_progname before calling lafe_warnc. */ lafe_setprogname(*argv, "bsdtar"); #if HAVE_SETLOCALE if (setlocale(LC_ALL, "") == NULL) lafe_warnc(0, "Failed to set default locale"); #endif #if defined(HAVE_NL_LANGINFO) && defined(HAVE_D_MD_ORDER) bsdtar->day_first = (*nl_langinfo(D_MD_ORDER) == 'd'); #endif possible_help_request = 0; /* Look up uid of current user for future reference */ bsdtar->user_uid = geteuid(); /* Default: open tape drive. */ bsdtar->filename = getenv("TAPE"); #if defined(_PATH_DEFTAPE) if (bsdtar->filename == NULL) { #if defined(_WIN32) && !defined(__CYGWIN__) int tapeExists = !_access(_PATH_DEFTAPE, 0); #else int tapeExists = !access(_PATH_DEFTAPE, F_OK); #endif if (tapeExists) { bsdtar->filename = _PATH_DEFTAPE; } } #endif if (bsdtar->filename == NULL) { bsdtar->filename = _PATH_STDIO; } /* Default block size settings. */ bsdtar->bytes_per_block = DEFAULT_BYTES_PER_BLOCK; /* Allow library to default this unless user specifies -b. */ bsdtar->bytes_in_last_block = -1; /* Default: preserve mod time on extract */ bsdtar->extract_flags = ARCHIVE_EXTRACT_TIME; /* Default: Perform basic security checks. */ bsdtar->extract_flags |= SECURITY; #ifndef _WIN32 /* On POSIX systems, assume --same-owner and -p when run by * the root user. This doesn't make any sense on Windows. */ if (bsdtar->user_uid == 0) { /* --same-owner */ bsdtar->extract_flags |= ARCHIVE_EXTRACT_OWNER; /* -p */ bsdtar->extract_flags |= ARCHIVE_EXTRACT_PERM; bsdtar->extract_flags |= ARCHIVE_EXTRACT_ACL; bsdtar->extract_flags |= ARCHIVE_EXTRACT_XATTR; bsdtar->extract_flags |= ARCHIVE_EXTRACT_FFLAGS; bsdtar->extract_flags |= ARCHIVE_EXTRACT_MAC_METADATA; } #endif /* * Enable Mac OS "copyfile()" extension by default. * This has no effect on other platforms. */ bsdtar->readdisk_flags |= ARCHIVE_READDISK_MAC_COPYFILE; #ifdef COPYFILE_DISABLE_VAR if (getenv(COPYFILE_DISABLE_VAR)) bsdtar->readdisk_flags &= ~ARCHIVE_READDISK_MAC_COPYFILE; #endif #if defined(__APPLE__) /* * On Mac OS ACLs are archived with copyfile() (--mac-metadata) * Translation to NFSv4 ACLs has to be requested explicitly with --acls */ bsdtar->readdisk_flags |= ARCHIVE_READDISK_NO_ACL; #endif bsdtar->matching = archive_match_new(); if (bsdtar->matching == NULL) lafe_errc(1, errno, "Out of memory"); bsdtar->cset = cset_new(); if (bsdtar->cset == NULL) lafe_errc(1, errno, "Out of memory"); bsdtar->argv = argv; bsdtar->argc = argc; /* * Comments following each option indicate where that option * originated: SUSv2, POSIX, GNU tar, star, etc. If there's * no such comment, then I don't know of anyone else who * implements that option. */ while ((opt = bsdtar_getopt(bsdtar)) != -1) { switch (opt) { case 'a': /* GNU tar */ bsdtar->flags |= OPTFLAG_AUTO_COMPRESS; break; case OPTION_ACLS: /* GNU tar */ bsdtar->extract_flags |= ARCHIVE_EXTRACT_ACL; bsdtar->readdisk_flags &= ~ARCHIVE_READDISK_NO_ACL; bsdtar->flags |= OPTFLAG_ACLS; break; case 'B': /* GNU tar */ /* libarchive doesn't need this; just ignore it. */ break; case 'b': /* SUSv2 */ tptr = NULL; l = strtol(bsdtar->argument, &tptr, 10); if (l <= 0 || l > 8192L || *(bsdtar->argument) == '\0' || tptr == NULL || *tptr != '\0') { lafe_errc(1, 0, "Invalid or out of range " "(1..8192) argument to -b"); } bsdtar->bytes_per_block = 512 * (int)l; /* Explicit -b forces last block size. */ bsdtar->bytes_in_last_block = bsdtar->bytes_per_block; break; case OPTION_B64ENCODE: if (compression2 != '\0') lafe_errc(1, 0, "Can't specify both --uuencode and " "--b64encode"); compression2 = opt; compression2_name = "b64encode"; break; case 'C': /* GNU tar */ if (strlen(bsdtar->argument) == 0) lafe_errc(1, 0, "Meaningless option: -C ''"); set_chdir(bsdtar, bsdtar->argument); break; case 'c': /* SUSv2 */ set_mode(bsdtar, opt); break; case OPTION_CHECK_LINKS: /* GNU tar */ bsdtar->flags |= OPTFLAG_WARN_LINKS; break; case OPTION_CHROOT: /* NetBSD */ bsdtar->flags |= OPTFLAG_CHROOT; break; case OPTION_CLEAR_NOCHANGE_FFLAGS: bsdtar->extract_flags |= ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS; break; case OPTION_EXCLUDE: /* GNU tar */ if (archive_match_exclude_pattern( bsdtar->matching, bsdtar->argument) != ARCHIVE_OK) lafe_errc(1, 0, "Couldn't exclude %s", bsdtar->argument); break; case OPTION_EXCLUDE_VCS: /* GNU tar */ for(t=0; vcs_files[t]; t++) { if (archive_match_exclude_pattern( bsdtar->matching, vcs_files[t]) != ARCHIVE_OK) lafe_errc(1, 0, "Couldn't " "exclude %s", vcs_files[t]); } break; case OPTION_FFLAGS: bsdtar->extract_flags |= ARCHIVE_EXTRACT_FFLAGS; bsdtar->readdisk_flags &= ~ARCHIVE_READDISK_NO_FFLAGS; bsdtar->flags |= OPTFLAG_FFLAGS; break; case OPTION_FORMAT: /* GNU tar, others */ cset_set_format(bsdtar->cset, bsdtar->argument); break; case 'f': /* SUSv2 */ bsdtar->filename = bsdtar->argument; break; case OPTION_GID: /* cpio */ tptr = NULL; l = strtol(bsdtar->argument, &tptr, 10); if (l < 0 || l >= INT_MAX || *(bsdtar->argument) == '\0' || tptr == NULL || *tptr != '\0') { lafe_errc(1, 0, "Invalid argument to --gid"); } bsdtar->gid = (int)l; break; case OPTION_GNAME: /* cpio */ bsdtar->gname = bsdtar->argument; break; case OPTION_GROUP: /* GNU tar */ tptr = NULL; uptr = strchr(bsdtar->argument, ':'); if (uptr != NULL) { if (uptr[1] == '\0') { lafe_errc(1, 0, "Invalid argument to --group (missing id after :)"); } uptr[0] = 0; uptr++; l = strtol(uptr, &tptr, 10); if (l < 0 || l >= INT_MAX || *uptr == '\0' || tptr == NULL || *tptr != '\0') { lafe_errc(1, 0, "Invalid argument to --group (%s is not a number)", uptr); } else { bsdtar->gid = (int)l; } bsdtar->gname = bsdtar->argument; } else { l = strtol(bsdtar->argument, &tptr, 10); if (l < 0 || l >= INT_MAX || *(bsdtar->argument) == '\0' || tptr == NULL || *tptr != '\0') { bsdtar->gname = bsdtar->argument; } else { bsdtar->gid = (int)l; bsdtar->gname = ""; } } break; case OPTION_GRZIP: if (compression != '\0') lafe_errc(1, 0, "Can't specify both -%c and -%c", opt, compression); compression = opt; compression_name = "grzip"; break; case 'H': /* BSD convention */ bsdtar->symlink_mode = 'H'; break; case 'h': /* Linux Standards Base, gtar; synonym for -L */ bsdtar->symlink_mode = 'L'; /* Hack: -h by itself is the "help" command. */ possible_help_request = 1; break; case OPTION_HELP: /* GNU tar, others */ long_help(); /* NOTREACHED*/ case OPTION_HFS_COMPRESSION: /* Mac OS X v10.6 or later */ bsdtar->extract_flags |= ARCHIVE_EXTRACT_HFS_COMPRESSION_FORCED; break; case OPTION_IGNORE_ZEROS: bsdtar->flags |= OPTFLAG_IGNORE_ZEROS; break; case 'I': /* GNU tar */ /* * TODO: Allow 'names' to come from an archive, * not just a text file. Design a good UI for * allowing names and mode/owner to be read * from an archive, with contents coming from * disk. This can be used to "refresh" an * archive or to design archives with special * permissions without having to create those * permissions on disk. */ bsdtar->names_from_file = bsdtar->argument; break; case OPTION_INCLUDE: /* * No one else has the @archive extension, so * no one else needs this to filter entries * when transforming archives. */ if (archive_match_include_pattern(bsdtar->matching, bsdtar->argument) != ARCHIVE_OK) lafe_errc(1, 0, "Failed to add %s to inclusion list", bsdtar->argument); break; case 'j': /* GNU tar */ if (compression != '\0') lafe_errc(1, 0, "Can't specify both -%c and -%c", opt, compression); compression = opt; compression_name = "bzip2"; break; case 'J': /* GNU tar 1.21 and later */ if (compression != '\0') lafe_errc(1, 0, "Can't specify both -%c and -%c", opt, compression); compression = opt; compression_name = "xz"; break; case 'k': /* GNU tar */ bsdtar->extract_flags |= ARCHIVE_EXTRACT_NO_OVERWRITE; break; case OPTION_KEEP_NEWER_FILES: /* GNU tar */ bsdtar->extract_flags |= ARCHIVE_EXTRACT_NO_OVERWRITE_NEWER; break; case 'L': /* BSD convention */ bsdtar->symlink_mode = 'L'; break; case 'l': /* SUSv2 and GNU tar beginning with 1.16 */ /* GNU tar 1.13 used -l for --one-file-system */ bsdtar->flags |= OPTFLAG_WARN_LINKS; break; case OPTION_LRZIP: case OPTION_LZ4: case OPTION_LZIP: /* GNU tar beginning with 1.23 */ case OPTION_LZMA: /* GNU tar beginning with 1.20 */ case OPTION_LZOP: /* GNU tar beginning with 1.21 */ case OPTION_ZSTD: if (compression != '\0') lafe_errc(1, 0, "Can't specify both -%c and -%c", opt, compression); compression = opt; switch (opt) { case OPTION_LRZIP: compression_name = "lrzip"; break; case OPTION_LZ4: compression_name = "lz4"; break; case OPTION_LZIP: compression_name = "lzip"; break; case OPTION_LZMA: compression_name = "lzma"; break; case OPTION_LZOP: compression_name = "lzop"; break; case OPTION_ZSTD: compression_name = "zstd"; break; } break; case 'm': /* SUSv2 */ bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_TIME; break; case OPTION_MAC_METADATA: /* Mac OS X */ bsdtar->readdisk_flags |= ARCHIVE_READDISK_MAC_COPYFILE; bsdtar->extract_flags |= ARCHIVE_EXTRACT_MAC_METADATA; bsdtar->flags |= OPTFLAG_MAC_METADATA; break; case 'n': /* GNU tar */ bsdtar->flags |= OPTFLAG_NO_SUBDIRS; break; /* * Selecting files by time: * --newer-?time='date' Only files newer than 'date' * --newer-?time-than='file' Only files newer than time * on specified file (useful for incremental backups) */ case OPTION_NEWER_CTIME: /* GNU tar */ if (archive_match_include_date(bsdtar->matching, ARCHIVE_MATCH_CTIME | ARCHIVE_MATCH_NEWER, bsdtar->argument) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(bsdtar->matching)); break; case OPTION_NEWER_CTIME_THAN: if (archive_match_include_file_time(bsdtar->matching, ARCHIVE_MATCH_CTIME | ARCHIVE_MATCH_NEWER, bsdtar->argument) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(bsdtar->matching)); break; case OPTION_NEWER_MTIME: /* GNU tar */ if (archive_match_include_date(bsdtar->matching, ARCHIVE_MATCH_MTIME | ARCHIVE_MATCH_NEWER, bsdtar->argument) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(bsdtar->matching)); break; case OPTION_NEWER_MTIME_THAN: if (archive_match_include_file_time(bsdtar->matching, ARCHIVE_MATCH_MTIME | ARCHIVE_MATCH_NEWER, bsdtar->argument) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(bsdtar->matching)); break; case OPTION_NODUMP: /* star */ bsdtar->readdisk_flags |= ARCHIVE_READDISK_HONOR_NODUMP; break; case OPTION_NOPRESERVE_HFS_COMPRESSION: /* Mac OS X v10.6 or later */ bsdtar->extract_flags |= ARCHIVE_EXTRACT_NO_HFS_COMPRESSION; break; case OPTION_NO_ACLS: /* GNU tar */ bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_ACL; bsdtar->readdisk_flags |= ARCHIVE_READDISK_NO_ACL; bsdtar->flags |= OPTFLAG_NO_ACLS; break; case OPTION_NO_FFLAGS: bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_FFLAGS; bsdtar->readdisk_flags |= ARCHIVE_READDISK_NO_FFLAGS; bsdtar->flags |= OPTFLAG_NO_FFLAGS; break; case OPTION_NO_MAC_METADATA: /* Mac OS X */ bsdtar->readdisk_flags &= ~ARCHIVE_READDISK_MAC_COPYFILE; bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_MAC_METADATA; bsdtar->flags |= OPTFLAG_NO_MAC_METADATA; break; case OPTION_NO_READ_SPARSE: bsdtar->readdisk_flags |= ARCHIVE_READDISK_NO_SPARSE; bsdtar->flags |= OPTFLAG_NO_READ_SPARSE; break; case OPTION_NO_SAFE_WRITES: bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_SAFE_WRITES; break; case OPTION_NO_SAME_OWNER: /* GNU tar */ bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_OWNER; break; case OPTION_NO_SAME_PERMISSIONS: /* GNU tar */ bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_PERM; bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_ACL; bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_XATTR; bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_FFLAGS; bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_MAC_METADATA; break; case OPTION_NO_XATTRS: /* GNU tar */ bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_XATTR; bsdtar->readdisk_flags |= ARCHIVE_READDISK_NO_XATTR; bsdtar->flags |= OPTFLAG_NO_XATTRS; break; case OPTION_NULL: /* GNU tar */ bsdtar->flags |= OPTFLAG_NULL; break; case OPTION_NUMERIC_OWNER: /* GNU tar */ bsdtar->uname = ""; bsdtar->gname = ""; bsdtar->flags |= OPTFLAG_NUMERIC_OWNER; break; case 'O': /* GNU tar */ bsdtar->flags |= OPTFLAG_STDOUT; break; case 'o': /* SUSv2 and GNU conflict here, but not fatally */ bsdtar->flags |= OPTFLAG_O; break; /* * Selecting files by time: * --older-?time='date' Only files older than 'date' * --older-?time-than='file' Only files older than time * on specified file */ case OPTION_OLDER_CTIME: if (archive_match_include_date(bsdtar->matching, ARCHIVE_MATCH_CTIME | ARCHIVE_MATCH_OLDER, bsdtar->argument) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(bsdtar->matching)); break; case OPTION_OLDER_CTIME_THAN: if (archive_match_include_file_time(bsdtar->matching, ARCHIVE_MATCH_CTIME | ARCHIVE_MATCH_OLDER, bsdtar->argument) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(bsdtar->matching)); break; case OPTION_OLDER_MTIME: if (archive_match_include_date(bsdtar->matching, ARCHIVE_MATCH_MTIME | ARCHIVE_MATCH_OLDER, bsdtar->argument) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(bsdtar->matching)); break; case OPTION_OLDER_MTIME_THAN: if (archive_match_include_file_time(bsdtar->matching, ARCHIVE_MATCH_MTIME | ARCHIVE_MATCH_OLDER, bsdtar->argument) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(bsdtar->matching)); break; case OPTION_ONE_FILE_SYSTEM: /* GNU tar */ bsdtar->readdisk_flags |= ARCHIVE_READDISK_NO_TRAVERSE_MOUNTS; break; case OPTION_OPTIONS: if (bsdtar->option_options != NULL) { lafe_warnc(0, "Ignoring previous option '%s', separate multiple options with commas", bsdtar->option_options); } bsdtar->option_options = bsdtar->argument; break; case OPTION_OWNER: /* GNU tar */ tptr = NULL; uptr = strchr(bsdtar->argument, ':'); if (uptr != NULL) { if (uptr[1] == 0) { lafe_errc(1, 0, "Invalid argument to --owner (missing id after :)"); } uptr[0] = 0; uptr++; l = strtol(uptr, &tptr, 10); if (l < 0 || l >= INT_MAX || *uptr == '\0' || tptr == NULL || *tptr != '\0') { lafe_errc(1, 0, "Invalid argument to --owner (%s is not a number)", uptr); } else { bsdtar->uid = (int)l; } bsdtar->uname = bsdtar->argument; } else { l = strtol(bsdtar->argument, &tptr, 10); if (l < 0 || l >= INT_MAX || *(bsdtar->argument) == '\0' || tptr == NULL || *tptr != '\0') { bsdtar->uname = bsdtar->argument; } else { bsdtar->uid = (int)l; bsdtar->uname = ""; } } break; case OPTION_MTIME: /* GNU tar */ bsdtar->has_mtime = 1; bsdtar->mtime = archive_parse_date(now, bsdtar->argument); if (bsdtar->mtime == (time_t)-1) { lafe_errc(1, 0, "Invalid argument to --mtime (bad date string)"); } break; case OPTION_CLAMP_MTIME: /* GNU tar */ bsdtar->clamp_mtime = 1; break; #if 0 /* * The common BSD -P option is not necessary, since * our default is to archive symlinks, not follow * them. This is convenient, as -P conflicts with GNU * tar anyway. */ case 'P': /* BSD convention */ /* Default behavior, no option necessary. */ break; #endif case 'P': /* GNU tar */ bsdtar->extract_flags &= ~SECURITY; bsdtar->flags |= OPTFLAG_ABSOLUTE_PATHS; break; case 'p': /* GNU tar, star */ bsdtar->extract_flags |= ARCHIVE_EXTRACT_PERM; bsdtar->extract_flags |= ARCHIVE_EXTRACT_ACL; bsdtar->extract_flags |= ARCHIVE_EXTRACT_XATTR; bsdtar->extract_flags |= ARCHIVE_EXTRACT_FFLAGS; bsdtar->extract_flags |= ARCHIVE_EXTRACT_MAC_METADATA; break; case OPTION_PASSPHRASE: bsdtar->passphrase = bsdtar->argument; break; case OPTION_POSIX: /* GNU tar */ cset_set_format(bsdtar->cset, "pax"); break; case 'q': /* FreeBSD GNU tar --fast-read, NetBSD -q */ bsdtar->flags |= OPTFLAG_FAST_READ; break; case 'r': /* SUSv2 */ set_mode(bsdtar, opt); break; case OPTION_READ_SPARSE: bsdtar->readdisk_flags &= ~ARCHIVE_READDISK_NO_SPARSE; bsdtar->flags |= OPTFLAG_READ_SPARSE; break; case 'S': /* NetBSD pax-as-tar */ bsdtar->extract_flags |= ARCHIVE_EXTRACT_SPARSE; break; case 's': /* NetBSD pax-as-tar */ #if defined(HAVE_REGEX_H) || defined(HAVE_PCREPOSIX_H) || defined(HAVE_PCRE2POSIX_H) add_substitution(bsdtar, bsdtar->argument); #else lafe_warnc(0, "-s is not supported by this version of bsdtar"); usage(); #endif break; case OPTION_SAFE_WRITES: bsdtar->extract_flags |= ARCHIVE_EXTRACT_SAFE_WRITES; break; case OPTION_SAME_OWNER: /* GNU tar */ bsdtar->extract_flags |= ARCHIVE_EXTRACT_OWNER; break; case OPTION_STRIP_COMPONENTS: /* GNU tar 1.15 */ tptr = NULL; l = strtol(bsdtar->argument, &tptr, 10); if (l < 0 || l > 100000L || *(bsdtar->argument) == '\0' || tptr == NULL || *tptr != '\0') { lafe_errc(1, 0, "Invalid argument to " "--strip-components"); } bsdtar->strip_components = (int)l; break; case 'T': /* GNU tar */ if (bsdtar->names_from_file) lafe_errc(1, 0, "Multiple --files-from/-T options are not supported"); bsdtar->names_from_file = bsdtar->argument; break; case 't': /* SUSv2 */ set_mode(bsdtar, opt); bsdtar->verbose++; break; case OPTION_TOTALS: /* GNU tar */ bsdtar->flags |= OPTFLAG_TOTALS; break; case 'U': /* GNU tar */ bsdtar->extract_flags |= ARCHIVE_EXTRACT_UNLINK; bsdtar->flags |= OPTFLAG_UNLINK_FIRST; break; case 'u': /* SUSv2 */ set_mode(bsdtar, opt); break; case OPTION_UID: /* cpio */ tptr = NULL; l = strtol(bsdtar->argument, &tptr, 10); if (l < 0 || l >= INT_MAX || *(bsdtar->argument) == '\0' || tptr == NULL || *tptr != '\0') { lafe_errc(1, 0, "Invalid argument to --uid"); } bsdtar->uid = (int)l; break; case OPTION_UNAME: /* cpio */ bsdtar->uname = bsdtar->argument; break; case OPTION_UUENCODE: if (compression2 != '\0') lafe_errc(1, 0, "Can't specify both --uuencode and " "--b64encode"); compression2 = opt; compression2_name = "uuencode"; break; case 'v': /* SUSv2 */ bsdtar->verbose++; break; case OPTION_VERSION: /* GNU convention */ version(); /* NOTREACHED */ #if 0 /* * The -W longopt feature is handled inside of * bsdtar_getopt(), so -W is not available here. */ case 'W': /* Obscure GNU convention. */ break; #endif case 'w': /* SUSv2 */ bsdtar->flags |= OPTFLAG_INTERACTIVE; break; case 'X': /* GNU tar */ if (archive_match_exclude_pattern_from_file( bsdtar->matching, bsdtar->argument, 0) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(bsdtar->matching)); break; case 'x': /* SUSv2 */ set_mode(bsdtar, opt); break; case OPTION_XATTRS: /* GNU tar */ bsdtar->extract_flags |= ARCHIVE_EXTRACT_XATTR; bsdtar->readdisk_flags &= ~ARCHIVE_READDISK_NO_XATTR; bsdtar->flags |= OPTFLAG_XATTRS; break; case 'y': /* FreeBSD version of GNU tar */ if (compression != '\0') lafe_errc(1, 0, "Can't specify both -%c and -%c", opt, compression); compression = opt; compression_name = "bzip2"; break; case 'Z': /* GNU tar */ if (compression != '\0') lafe_errc(1, 0, "Can't specify both -%c and -%c", opt, compression); compression = opt; compression_name = "compress"; break; case 'z': /* GNU tar, star, many others */ if (compression != '\0') lafe_errc(1, 0, "Can't specify both -%c and -%c", opt, compression); compression = opt; compression_name = "gzip"; break; case OPTION_USE_COMPRESS_PROGRAM: compress_program = bsdtar->argument; break; default: usage(); } } /* * Sanity-check options. */ /* If no "real" mode was specified, treat -h as --help. */ if ((bsdtar->mode == '\0') && possible_help_request) { long_help(); } /* Otherwise, a mode is required. */ if (bsdtar->mode == '\0') lafe_errc(1, 0, "Must specify one of -c, -r, -t, -u, -x"); /* Check boolean options only permitted in certain modes. */ if (bsdtar->flags & OPTFLAG_AUTO_COMPRESS) { only_mode(bsdtar, "-a", "cx"); if (bsdtar->mode == 'x') { bsdtar->flags &= ~OPTFLAG_AUTO_COMPRESS; lafe_warnc(0, "Ignoring option -a in mode -x"); } } if (bsdtar->readdisk_flags & ARCHIVE_READDISK_NO_TRAVERSE_MOUNTS) only_mode(bsdtar, "--one-file-system", "cru"); if (bsdtar->flags & OPTFLAG_FAST_READ) only_mode(bsdtar, "--fast-read", "xt"); if (bsdtar->extract_flags & ARCHIVE_EXTRACT_HFS_COMPRESSION_FORCED) only_mode(bsdtar, "--hfsCompression", "x"); if (bsdtar->extract_flags & ARCHIVE_EXTRACT_NO_HFS_COMPRESSION) only_mode(bsdtar, "--nopreserveHFSCompression", "x"); if (bsdtar->readdisk_flags & ARCHIVE_READDISK_HONOR_NODUMP) only_mode(bsdtar, "--nodump", "cru"); if (bsdtar->flags & OPTFLAG_ACLS) only_mode(bsdtar, "--acls", "crux"); if (bsdtar->flags & OPTFLAG_NO_ACLS) only_mode(bsdtar, "--no-acls", "crux"); if (bsdtar->flags & OPTFLAG_XATTRS) only_mode(bsdtar, "--xattrs", "crux"); if (bsdtar->flags & OPTFLAG_NO_XATTRS) only_mode(bsdtar, "--no-xattrs", "crux"); if (bsdtar->flags & OPTFLAG_FFLAGS) only_mode(bsdtar, "--fflags", "crux"); if (bsdtar->flags & OPTFLAG_NO_FFLAGS) only_mode(bsdtar, "--no-fflags", "crux"); if (bsdtar->flags & OPTFLAG_MAC_METADATA) only_mode(bsdtar, "--mac-metadata", "crux"); if (bsdtar->flags & OPTFLAG_NO_MAC_METADATA) only_mode(bsdtar, "--no-mac-metadata", "crux"); if (bsdtar->flags & OPTFLAG_O) { switch (bsdtar->mode) { case 'c': /* * In GNU tar, -o means "old format." The * "ustar" format is the closest thing * supported by libarchive. */ cset_set_format(bsdtar->cset, "ustar"); /* TODO: bsdtar->create_format = "v7"; */ break; case 'x': /* POSIX-compatible behavior. */ bsdtar->flags |= OPTFLAG_NO_OWNER; bsdtar->extract_flags &= ~ARCHIVE_EXTRACT_OWNER; break; default: only_mode(bsdtar, "-o", "xc"); break; } } if (bsdtar->flags & OPTFLAG_STDOUT) only_mode(bsdtar, "-O", "xt"); if (bsdtar->flags & OPTFLAG_UNLINK_FIRST) only_mode(bsdtar, "-U", "x"); if (bsdtar->flags & OPTFLAG_WARN_LINKS) only_mode(bsdtar, "--check-links", "cr"); if ((bsdtar->flags & OPTFLAG_AUTO_COMPRESS) && cset_auto_compress(bsdtar->cset, bsdtar->filename)) { /* Ignore specified compressions if auto-compress works. */ compression = '\0'; compression2 = '\0'; } /* Check other parameters only permitted in certain modes. */ if (compress_program != NULL) { only_mode(bsdtar, "--use-compress-program", "cxt"); cset_add_filter_program(bsdtar->cset, compress_program); /* Ignore specified compressions. */ compression = '\0'; compression2 = '\0'; } if (compression != '\0') { switch (compression) { case 'J': case 'j': case 'y': case 'Z': case 'z': strcpy(buff, "-?"); buff[1] = (char)compression; break; default: strcpy(buff, "--"); strcat(buff, compression_name); break; } only_mode(bsdtar, buff, "cxt"); cset_add_filter(bsdtar->cset, compression_name); } if (compression2 != '\0') { strcpy(buff, "--"); strcat(buff, compression2_name); only_mode(bsdtar, buff, "cxt"); cset_add_filter(bsdtar->cset, compression2_name); } if (cset_get_format(bsdtar->cset) != NULL) only_mode(bsdtar, "--format", "cru"); if (bsdtar->symlink_mode != '\0') { strcpy(buff, "-?"); buff[1] = bsdtar->symlink_mode; only_mode(bsdtar, buff, "cru"); } if (!bsdtar->has_mtime && bsdtar->clamp_mtime) lafe_errc(1, 0, "--clamp-mtime is not valid without --mtime "); /* * When creating an archive from a directory tree, the directory * walking code will already avoid entering directories when * recursive inclusion of directory content is disabled, therefore * changing the matching behavior has no effect for creation modes. * It is relevant for extraction or listing. */ archive_match_set_inclusion_recursion(bsdtar->matching, !(bsdtar->flags & OPTFLAG_NO_SUBDIRS)); /* Filename "-" implies stdio. */ if (strcmp(bsdtar->filename, "-") == 0) bsdtar->filename = NULL; switch(bsdtar->mode) { case 'c': tar_mode_c(bsdtar); break; case 'r': tar_mode_r(bsdtar); break; case 't': tar_mode_t(bsdtar); break; case 'u': tar_mode_u(bsdtar); break; case 'x': tar_mode_x(bsdtar); break; } archive_match_free(bsdtar->matching); #if defined(HAVE_REGEX_H) || defined(HAVE_PCREPOSIX_H) || defined(HAVE_PCRE2POSIX_H) cleanup_substitution(bsdtar); #endif cset_free(bsdtar->cset); passphrase_free(bsdtar->ppbuff); if (bsdtar->return_value != 0) lafe_warnc(0, "Error exit delayed from previous errors."); return (bsdtar->return_value); } static void set_mode(struct bsdtar *bsdtar, int opt) { if (bsdtar->mode != '\0' && bsdtar->mode != opt) lafe_errc(1, 0, "Can't specify both -%c and -%c", opt, bsdtar->mode); bsdtar->mode = opt; } /* * Verify that the mode is correct. */ static void only_mode(struct bsdtar *bsdtar, const char *opt, const char *valid_modes) { if (strchr(valid_modes, bsdtar->mode) == NULL) lafe_errc(1, 0, "Option %s is not permitted in mode -%c", opt, bsdtar->mode); } void usage(void) { const char *p; p = lafe_getprogname(); fprintf(stderr, "Usage:\n"); fprintf(stderr, " List: %s -tf \n", p); fprintf(stderr, " Extract: %s -xf \n", p); fprintf(stderr, " Create: %s -cf [filenames...]\n", p); fprintf(stderr, " Help: %s --help\n", p); exit(1); } static void version(void) { printf("bsdtar %s - %s \n", BSDTAR_VERSION_STRING, archive_version_details()); exit(0); } static const char *long_help_msg = "First option must be a mode specifier:\n" " -c Create -r Add/Replace -t List -u Update -x Extract\n" "Common Options:\n" " -b # Use # 512-byte records per I/O block\n" " -f Location of archive (default " _PATH_DEFTAPE ")\n" " -v Verbose\n" " -w Interactive\n" "Create: %p -c [options] [ | | @ | -C ]\n" " , add these items to archive\n" " -z, -j, -J, --lzma Compress archive with gzip/bzip2/xz/lzma\n" " --format {ustar|pax|cpio|shar} Select archive format\n" " --exclude Skip files that match pattern\n" " --mtime Set modification times for added files\n" " --clamp-mtime Only set modification times for files newer than --mtime\n" " -C Change to before processing remaining files\n" " @ Add entries from to output\n" "List: %p -t [options] []\n" " If specified, list only entries that match\n" "Extract: %p -x [options] []\n" " If specified, extract only entries that match\n" " -k Keep (don't overwrite) existing files\n" " -m Don't restore modification times\n" " -O Write entries to stdout, don't restore to disk\n" " -p Restore permissions (including ACLs, owner, file flags)\n"; /* * Note that the word 'bsdtar' will always appear in the first line * of output. * * In particular, /bin/sh scripts that need to test for the presence * of bsdtar can use the following template: * * if (tar --help 2>&1 | grep bsdtar >/dev/null 2>&1 ) then \ * echo bsdtar; else echo not bsdtar; fi */ static void long_help(void) { const char *prog; const char *p; prog = lafe_getprogname(); fflush(stderr); p = (strcmp(prog,"bsdtar") != 0) ? "(bsdtar)" : ""; printf("%s%s: manipulate archive files\n", prog, p); for (p = long_help_msg; *p != '\0'; p++) { if (*p == '%') { if (p[1] == 'p') { fputs(prog, stdout); p++; } else putchar('%'); } else putchar(*p); } version(); } diff --git a/contrib/libarchive/tar/bsdtar.h b/contrib/libarchive/tar/bsdtar.h index 45dfeed7dce3..782d36d6f756 100644 --- a/contrib/libarchive/tar/bsdtar.h +++ b/contrib/libarchive/tar/bsdtar.h @@ -1,225 +1,231 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. */ #ifndef BSDTAR_H_INCLUDED #define BSDTAR_H_INCLUDED #include "bsdtar_platform.h" #include #define DEFAULT_BYTES_PER_BLOCK (20*512) #define ENV_READER_OPTIONS "TAR_READER_OPTIONS" #define ENV_WRITER_OPTIONS "TAR_WRITER_OPTIONS" #define IGNORE_WRONG_MODULE_NAME "__ignore_wrong_module_name__," +#if defined(_MSC_VER ) && (_MSC_VER < 1927 ) /* Check if compiler pre-dated Visual Studio 2019 Release 16.8 */ +#define ARCHIVE_RESTRICT +#else +#define ARCHIVE_RESTRICT restrict +#endif + struct creation_set; /* * The internal state for the "bsdtar" program. * * Keeping all of the state in a structure like this simplifies memory * leak testing (at exit, anything left on the heap is suspect). A * pointer to this structure is passed to most bsdtar internal * functions. */ struct bsdtar { /* Options */ const char *filename; /* -f filename */ char *pending_chdir; /* -C dir */ const char *names_from_file; /* -T file */ int bytes_per_block; /* -b block_size */ int bytes_in_last_block; /* See -b handling. */ int verbose; /* -v */ unsigned int flags; /* Bitfield of boolean options */ int extract_flags; /* Flags for extract operation */ int readdisk_flags; /* Flags for read disk operation */ int strip_components; /* Remove this many leading dirs */ int gid; /* --gid */ const char *gname; /* --gname */ int uid; /* --uid */ const char *uname; /* --uname */ const char *passphrase; /* --passphrase */ int mode; /* Program mode: 'c', 't', 'r', 'u', 'x' */ char symlink_mode; /* H or L, per BSD conventions */ const char *option_options; /* --options */ char day_first; /* show day before month in -tv output */ char has_mtime; /* --mtime exists (0 or 1) */ char clamp_mtime; /* --clamp-mtime (0 or 1)*/ time_t mtime; /* --mtime */ struct creation_set *cset; /* Option parser state */ int getopt_state; char *getopt_word; /* If >= 0, then close this when done. */ int fd; /* Miscellaneous state information */ int argc; char **argv; const char *argument; size_t gs_width; /* For 'list_item' in read.c */ size_t u_width; /* for 'list_item' in read.c */ uid_t user_uid; /* UID running this program */ int return_value; /* Value returned by main() */ char warned_lead_slash; /* Already displayed warning */ char next_line_is_dir; /* Used for -C parsing in -cT */ /* * Data for various subsystems. Full definitions are located in * the file where they are used. */ struct archive *diskreader; /* for write.c */ struct archive_entry_linkresolver *resolver; /* for write.c */ struct archive_dir *archive_dir; /* for write.c */ struct name_cache *gname_cache; /* for write.c */ char *buff; /* for write.c */ size_t buff_size; /* for write.c */ int first_fs; /* for write.c */ struct archive *matching; /* for matching.c */ struct security *security; /* for read.c */ struct name_cache *uname_cache; /* for write.c */ struct siginfo_data *siginfo; /* for siginfo.c */ struct substitution *substitution; /* for subst.c */ char *ppbuff; /* for util.c */ }; /* Options for flags bitfield */ #define OPTFLAG_AUTO_COMPRESS (0x00000001) /* -a */ #define OPTFLAG_ABSOLUTE_PATHS (0x00000002) /* -P */ #define OPTFLAG_CHROOT (0x00000004) /* --chroot */ #define OPTFLAG_FAST_READ (0x00000008) /* --fast-read */ #define OPTFLAG_IGNORE_ZEROS (0x00000010) /* --ignore-zeros */ #define OPTFLAG_INTERACTIVE (0x00000020) /* -w */ #define OPTFLAG_NO_OWNER (0x00000040) /* -o */ #define OPTFLAG_NO_SUBDIRS (0x00000080) /* -n */ #define OPTFLAG_NULL (0x00000100) /* --null */ #define OPTFLAG_NUMERIC_OWNER (0x00000200) /* --numeric-owner */ #define OPTFLAG_O (0x00000400) /* -o */ #define OPTFLAG_STDOUT (0x00000800) /* -O */ #define OPTFLAG_TOTALS (0x00001000) /* --totals */ #define OPTFLAG_UNLINK_FIRST (0x00002000) /* -U */ #define OPTFLAG_WARN_LINKS (0x00004000) /* --check-links */ #define OPTFLAG_NO_XATTRS (0x00008000) /* --no-xattrs */ #define OPTFLAG_XATTRS (0x00010000) /* --xattrs */ #define OPTFLAG_NO_ACLS (0x00020000) /* --no-acls */ #define OPTFLAG_ACLS (0x00040000) /* --acls */ #define OPTFLAG_NO_FFLAGS (0x00080000) /* --no-fflags */ #define OPTFLAG_FFLAGS (0x00100000) /* --fflags */ #define OPTFLAG_NO_MAC_METADATA (0x00200000) /* --no-mac-metadata */ #define OPTFLAG_MAC_METADATA (0x00400000) /* --mac-metadata */ #define OPTFLAG_NO_READ_SPARSE (0x00800000) /* --no-read-sparse */ #define OPTFLAG_READ_SPARSE (0x01000000) /* --read-sparse */ /* Fake short equivalents for long options that otherwise lack them. */ enum { OPTION_ACLS = 256, OPTION_B64ENCODE, OPTION_CHECK_LINKS, OPTION_CHROOT, OPTION_CLEAR_NOCHANGE_FFLAGS, OPTION_EXCLUDE, OPTION_EXCLUDE_VCS, OPTION_FFLAGS, OPTION_FORMAT, OPTION_GID, OPTION_GNAME, OPTION_GROUP, OPTION_GRZIP, OPTION_HELP, OPTION_HFS_COMPRESSION, OPTION_IGNORE_ZEROS, OPTION_INCLUDE, OPTION_KEEP_NEWER_FILES, OPTION_LRZIP, OPTION_LZ4, OPTION_LZIP, OPTION_LZMA, OPTION_LZOP, OPTION_MAC_METADATA, OPTION_NEWER_CTIME, OPTION_NEWER_CTIME_THAN, OPTION_NEWER_MTIME, OPTION_NEWER_MTIME_THAN, OPTION_NODUMP, OPTION_NOPRESERVE_HFS_COMPRESSION, OPTION_NO_ACLS, OPTION_NO_FFLAGS, OPTION_NO_MAC_METADATA, OPTION_NO_READ_SPARSE, OPTION_NO_SAFE_WRITES, OPTION_NO_SAME_OWNER, OPTION_NO_SAME_PERMISSIONS, OPTION_NO_XATTRS, OPTION_NULL, OPTION_NUMERIC_OWNER, OPTION_OLDER_CTIME, OPTION_OLDER_CTIME_THAN, OPTION_OLDER_MTIME, OPTION_OLDER_MTIME_THAN, OPTION_ONE_FILE_SYSTEM, OPTION_OPTIONS, OPTION_OWNER, OPTION_PASSPHRASE, OPTION_POSIX, OPTION_READ_SPARSE, OPTION_SAFE_WRITES, OPTION_SAME_OWNER, OPTION_STRIP_COMPONENTS, OPTION_TOTALS, OPTION_UID, OPTION_UNAME, OPTION_USE_COMPRESS_PROGRAM, OPTION_UUENCODE, OPTION_VERSION, OPTION_XATTRS, OPTION_ZSTD, OPTION_MTIME, OPTION_CLAMP_MTIME, }; int bsdtar_getopt(struct bsdtar *); void do_chdir(struct bsdtar *); int edit_pathname(struct bsdtar *, struct archive_entry *); void edit_mtime(struct bsdtar *, struct archive_entry *); int need_report(void); int pathcmp(const char *a, const char *b); -void safe_fprintf(FILE * restrict, const char * restrict fmt, ...) __LA_PRINTF(2, 3); +void safe_fprintf(FILE * ARCHIVE_RESTRICT, const char * ARCHIVE_RESTRICT fmt, ...) __LA_PRINTF(2, 3); void set_chdir(struct bsdtar *, const char *newdir); const char *tar_i64toa(int64_t); void tar_mode_c(struct bsdtar *bsdtar); void tar_mode_r(struct bsdtar *bsdtar); void tar_mode_t(struct bsdtar *bsdtar); void tar_mode_u(struct bsdtar *bsdtar); void tar_mode_x(struct bsdtar *bsdtar); __LA_NORETURN void usage(void); int yes(const char *fmt, ...) __LA_PRINTF(1, 2); #if defined(HAVE_REGEX_H) || defined(HAVE_PCREPOSIX_H) || defined(HAVE_PCRE2POSIX_H) void add_substitution(struct bsdtar *, const char *); int apply_substitution(struct bsdtar *, const char *, char **, int, int); void cleanup_substitution(struct bsdtar *); #endif void cset_add_filter(struct creation_set *, const char *); void cset_add_filter_program(struct creation_set *, const char *); int cset_auto_compress(struct creation_set *, const char *); void cset_free(struct creation_set *); const char * cset_get_format(struct creation_set *); struct creation_set *cset_new(void); int cset_read_support_filter_program(struct creation_set *, struct archive *); void cset_set_format(struct creation_set *, const char *); int cset_write_add_filters(struct creation_set *, struct archive *, const void **); const char * passphrase_callback(struct archive *, void *); void passphrase_free(char *); void list_item_verbose(struct bsdtar *, FILE *, struct archive_entry *); #endif diff --git a/contrib/libarchive/tar/cmdline.c b/contrib/libarchive/tar/cmdline.c index c766c1a52dbb..309be312c948 100644 --- a/contrib/libarchive/tar/cmdline.c +++ b/contrib/libarchive/tar/cmdline.c @@ -1,410 +1,410 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2003-2008 Tim Kientzle * All rights reserved. */ /* * Command line parser for tar. */ #include "bsdtar_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "bsdtar.h" -#include "err.h" +#include "lafe_err.h" /* * Short options for tar. Please keep this sorted. */ static const char *short_options = "aBb:C:cf:HhI:JjkLlmnOoPpqrSs:T:tUuvW:wX:xyZz"; /* * Long options for tar. Please keep this list sorted. * * The symbolic names for options that lack a short equivalent are * defined in bsdtar.h. Also note that so far I've found no need * to support optional arguments to long options. That would be * a small change to the code below. */ static const struct bsdtar_option { const char *name; int required; /* 1 if this option requires an argument. */ int equivalent; /* Equivalent short option. */ } tar_longopts[] = { { "absolute-paths", 0, 'P' }, { "append", 0, 'r' }, { "acls", 0, OPTION_ACLS }, { "auto-compress", 0, 'a' }, { "b64encode", 0, OPTION_B64ENCODE }, { "block-size", 1, 'b' }, { "blocking-factor", 1, 'b' }, { "bunzip2", 0, 'j' }, { "bzip", 0, 'j' }, { "bzip2", 0, 'j' }, { "cd", 1, 'C' }, { "check-links", 0, OPTION_CHECK_LINKS }, { "chroot", 0, OPTION_CHROOT }, { "clamp-mtime", 0, OPTION_CLAMP_MTIME }, { "clear-nochange-fflags", 0, OPTION_CLEAR_NOCHANGE_FFLAGS }, { "compress", 0, 'Z' }, { "confirmation", 0, 'w' }, { "create", 0, 'c' }, { "dereference", 0, 'L' }, { "directory", 1, 'C' }, { "disable-copyfile", 0, OPTION_NO_MAC_METADATA }, { "exclude", 1, OPTION_EXCLUDE }, { "exclude-from", 1, 'X' }, { "exclude-vcs", 0, OPTION_EXCLUDE_VCS }, { "extract", 0, 'x' }, { "fast-read", 0, 'q' }, { "fflags", 0, OPTION_FFLAGS }, { "file", 1, 'f' }, { "files-from", 1, 'T' }, { "format", 1, OPTION_FORMAT }, { "gid", 1, OPTION_GID }, { "gname", 1, OPTION_GNAME }, { "group", 1, OPTION_GROUP }, { "grzip", 0, OPTION_GRZIP }, { "gunzip", 0, 'z' }, { "gzip", 0, 'z' }, { "help", 0, OPTION_HELP }, { "hfsCompression", 0, OPTION_HFS_COMPRESSION }, { "ignore-zeros", 0, OPTION_IGNORE_ZEROS }, { "include", 1, OPTION_INCLUDE }, { "insecure", 0, 'P' }, { "interactive", 0, 'w' }, { "keep-newer-files", 0, OPTION_KEEP_NEWER_FILES }, { "keep-old-files", 0, 'k' }, { "list", 0, 't' }, { "lrzip", 0, OPTION_LRZIP }, { "lz4", 0, OPTION_LZ4 }, { "lzip", 0, OPTION_LZIP }, { "lzma", 0, OPTION_LZMA }, { "lzop", 0, OPTION_LZOP }, { "mac-metadata", 0, OPTION_MAC_METADATA }, { "modification-time", 0, 'm' }, { "mtime", 1, OPTION_MTIME }, { "newer", 1, OPTION_NEWER_CTIME }, { "newer-ctime", 1, OPTION_NEWER_CTIME }, { "newer-ctime-than", 1, OPTION_NEWER_CTIME_THAN }, { "newer-mtime", 1, OPTION_NEWER_MTIME }, { "newer-mtime-than", 1, OPTION_NEWER_MTIME_THAN }, { "newer-than", 1, OPTION_NEWER_CTIME_THAN }, { "no-acls", 0, OPTION_NO_ACLS }, { "no-fflags", 0, OPTION_NO_FFLAGS }, { "no-mac-metadata", 0, OPTION_NO_MAC_METADATA }, { "no-read-sparse", 0, OPTION_NO_READ_SPARSE }, { "no-recursion", 0, 'n' }, { "no-safe-writes", 0, OPTION_NO_SAFE_WRITES }, { "no-same-owner", 0, OPTION_NO_SAME_OWNER }, { "no-same-permissions", 0, OPTION_NO_SAME_PERMISSIONS }, { "no-xattr", 0, OPTION_NO_XATTRS }, { "no-xattrs", 0, OPTION_NO_XATTRS }, { "nodump", 0, OPTION_NODUMP }, { "nopreserveHFSCompression",0, OPTION_NOPRESERVE_HFS_COMPRESSION }, { "norecurse", 0, 'n' }, { "null", 0, OPTION_NULL }, { "numeric-owner", 0, OPTION_NUMERIC_OWNER }, { "older", 1, OPTION_OLDER_CTIME }, { "older-ctime", 1, OPTION_OLDER_CTIME }, { "older-ctime-than", 1, OPTION_OLDER_CTIME_THAN }, { "older-mtime", 1, OPTION_OLDER_MTIME }, { "older-mtime-than", 1, OPTION_OLDER_MTIME_THAN }, { "older-than", 1, OPTION_OLDER_CTIME_THAN }, { "one-file-system", 0, OPTION_ONE_FILE_SYSTEM }, { "options", 1, OPTION_OPTIONS }, { "owner", 1, OPTION_OWNER }, { "passphrase", 1, OPTION_PASSPHRASE }, { "posix", 0, OPTION_POSIX }, { "preserve-permissions", 0, 'p' }, { "read-full-blocks", 0, 'B' }, { "read-sparse", 0, OPTION_READ_SPARSE }, { "safe-writes", 0, OPTION_SAFE_WRITES }, { "same-owner", 0, OPTION_SAME_OWNER }, { "same-permissions", 0, 'p' }, { "strip-components", 1, OPTION_STRIP_COMPONENTS }, { "to-stdout", 0, 'O' }, { "totals", 0, OPTION_TOTALS }, { "uid", 1, OPTION_UID }, { "uname", 1, OPTION_UNAME }, { "uncompress", 0, 'Z' }, { "unlink", 0, 'U' }, { "unlink-first", 0, 'U' }, { "update", 0, 'u' }, { "use-compress-program", 1, OPTION_USE_COMPRESS_PROGRAM }, { "uuencode", 0, OPTION_UUENCODE }, { "verbose", 0, 'v' }, { "version", 0, OPTION_VERSION }, { "xattrs", 0, OPTION_XATTRS }, { "xz", 0, 'J' }, { "zstd", 0, OPTION_ZSTD }, { NULL, 0, 0 } }; /* * This getopt implementation has two key features that common * getopt_long() implementations lack. Apart from those, it's a * straightforward option parser, considerably simplified by not * needing to support the wealth of exotic getopt_long() features. It * has, of course, been shamelessly tailored for bsdtar. (If you're * looking for a generic getopt_long() implementation for your * project, I recommend Gregory Pietsch's public domain getopt_long() * implementation.) The two additional features are: * * Old-style tar arguments: The original tar implementation treated * the first argument word as a list of single-character option * letters. All arguments follow as separate words. For example, * tar xbf 32 /dev/tape * Here, the "xbf" is three option letters, "32" is the argument for * "b" and "/dev/tape" is the argument for "f". We support this usage * if the first command-line argument does not begin with '-'. We * also allow regular short and long options to follow, e.g., * tar xbf 32 /dev/tape -P --format=pax * * -W long options: There's an obscure GNU convention (only rarely * supported even there) that allows "-W option=argument" as an * alternative way to support long options. This was supported in * early bsdtar as a way to access long options on platforms that did * not support getopt_long() and is preserved here for backwards * compatibility. (Of course, if I'd started with a custom * command-line parser from the beginning, I would have had normal * long option support on every platform so that hack wouldn't have * been necessary. Oh, well. Some mistakes you just have to live * with.) * * TODO: We should be able to use this to pull files and intermingled * options (such as -C) from the command line in write mode. That * will require a little rethinking of the argument handling in * bsdtar.c. * * TODO: If we want to support arbitrary command-line options from -T * input (as GNU tar does), we may need to extend this to handle option * words from sources other than argv/argc. I'm not really sure if I * like that feature of GNU tar, so it's certainly not a priority. */ int bsdtar_getopt(struct bsdtar *bsdtar) { enum { state_start = 0, state_old_tar, state_next_word, state_short, state_long }; const struct bsdtar_option *popt, *match, *match2; const char *p, *long_prefix; size_t optlength; int opt; int required; again: match = NULL; match2 = NULL; long_prefix = "--"; opt = '?'; required = 0; bsdtar->argument = NULL; /* First time through, initialize everything. */ if (bsdtar->getopt_state == state_start) { /* Skip program name. */ ++bsdtar->argv; --bsdtar->argc; if (*bsdtar->argv == NULL) return (-1); /* Decide between "new style" and "old style" arguments. */ if (bsdtar->argv[0][0] == '-') { bsdtar->getopt_state = state_next_word; } else { bsdtar->getopt_state = state_old_tar; bsdtar->getopt_word = *bsdtar->argv++; --bsdtar->argc; } } /* * We're parsing old-style tar arguments */ if (bsdtar->getopt_state == state_old_tar) { /* Get the next option character. */ opt = *bsdtar->getopt_word++; if (opt == '\0') { /* New-style args can follow old-style. */ bsdtar->getopt_state = state_next_word; } else { /* See if it takes an argument. */ p = strchr(short_options, opt); if (p == NULL) return ('?'); if (p[1] == ':') { bsdtar->argument = *bsdtar->argv; if (bsdtar->argument == NULL) { lafe_warnc(0, "Option %c requires an argument", opt); return ('?'); } ++bsdtar->argv; --bsdtar->argc; } } } /* * We're ready to look at the next word in argv. */ if (bsdtar->getopt_state == state_next_word) { /* No more arguments, so no more options. */ if (bsdtar->argv[0] == NULL) return (-1); /* Doesn't start with '-', so no more options. */ if (bsdtar->argv[0][0] != '-') return (-1); /* "--" marks end of options; consume it and return. */ if (strcmp(bsdtar->argv[0], "--") == 0) { ++bsdtar->argv; --bsdtar->argc; return (-1); } /* Get next word for parsing. */ bsdtar->getopt_word = *bsdtar->argv++; --bsdtar->argc; if (bsdtar->getopt_word[1] == '-') { /* Set up long option parser. */ bsdtar->getopt_state = state_long; bsdtar->getopt_word += 2; /* Skip leading '--' */ } else { /* Set up short option parser. */ bsdtar->getopt_state = state_short; ++bsdtar->getopt_word; /* Skip leading '-' */ } } /* * We're parsing a group of POSIX-style single-character options. */ if (bsdtar->getopt_state == state_short) { /* Peel next option off of a group of short options. */ opt = *bsdtar->getopt_word++; if (opt == '\0') { /* End of this group; recurse to get next option. */ bsdtar->getopt_state = state_next_word; goto again; } /* Does this option take an argument? */ p = strchr(short_options, opt); if (p == NULL) return ('?'); if (p[1] == ':') required = 1; /* If it takes an argument, parse that. */ if (required) { /* If arg is run-in, bsdtar->getopt_word already points to it. */ if (bsdtar->getopt_word[0] == '\0') { /* Otherwise, pick up the next word. */ bsdtar->getopt_word = *bsdtar->argv; if (bsdtar->getopt_word == NULL) { lafe_warnc(0, "Option -%c requires an argument", opt); return ('?'); } ++bsdtar->argv; --bsdtar->argc; } if (opt == 'W') { bsdtar->getopt_state = state_long; long_prefix = "-W "; /* For clearer errors. */ } else { bsdtar->getopt_state = state_next_word; bsdtar->argument = bsdtar->getopt_word; } } } /* We're reading a long option, including -W long=arg convention. */ if (bsdtar->getopt_state == state_long) { /* After this long option, we'll be starting a new word. */ bsdtar->getopt_state = state_next_word; /* Option name ends at '=' if there is one. */ p = strchr(bsdtar->getopt_word, '='); if (p != NULL) { optlength = (size_t)(p - bsdtar->getopt_word); bsdtar->argument = (char *)(uintptr_t)(p + 1); } else { optlength = strlen(bsdtar->getopt_word); } /* Search the table for an unambiguous match. */ for (popt = tar_longopts; popt->name != NULL; popt++) { /* Short-circuit if first chars don't match. */ if (popt->name[0] != bsdtar->getopt_word[0]) continue; /* If option is a prefix of name in table, record it.*/ if (strncmp(bsdtar->getopt_word, popt->name, optlength) == 0) { match2 = match; /* Record up to two matches. */ match = popt; /* If it's an exact match, we're done. */ if (strlen(popt->name) == optlength) { match2 = NULL; /* Forget the others. */ break; } } } /* Fail if there wasn't a unique match. */ if (match == NULL) { lafe_warnc(0, "Option %s%s is not supported", long_prefix, bsdtar->getopt_word); return ('?'); } if (match2 != NULL) { lafe_warnc(0, "Ambiguous option %s%s (matches --%s and --%s)", long_prefix, bsdtar->getopt_word, match->name, match2->name); return ('?'); } /* We've found a unique match; does it need an argument? */ if (match->required) { /* Argument required: get next word if necessary. */ if (bsdtar->argument == NULL) { bsdtar->argument = *bsdtar->argv; if (bsdtar->argument == NULL) { lafe_warnc(0, "Option %s%s requires an argument", long_prefix, match->name); return ('?'); } ++bsdtar->argv; --bsdtar->argc; } } else { /* Argument forbidden: fail if there is one. */ if (bsdtar->argument != NULL) { lafe_warnc(0, "Option %s%s does not allow an argument", long_prefix, match->name); return ('?'); } } return (match->equivalent); } return (opt); } diff --git a/contrib/libarchive/tar/creation_set.c b/contrib/libarchive/tar/creation_set.c index 6883090418b0..51f803354a5d 100644 --- a/contrib/libarchive/tar/creation_set.c +++ b/contrib/libarchive/tar/creation_set.c @@ -1,299 +1,299 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2012 Michihiro NAKAJIMA * All rights reserved. */ #include "bsdtar_platform.h" #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "bsdtar.h" -#include "err.h" +#include "lafe_err.h" struct creation_set { char *create_format; struct filter_set { int program; /* Set 1 if filter is a program name */ char *filter_name; } *filters; int filter_count; }; struct suffix_code_t { const char *suffix; const char *form; }; static const char * get_suffix_code(const struct suffix_code_t *tbl, const char *suffix) { int i; if (suffix == NULL) return (NULL); for (i = 0; tbl[i].suffix != NULL; i++) { if (strcmp(tbl[i].suffix, suffix) == 0) return (tbl[i].form); } return (NULL); } static const char * get_filter_code(const char *suffix) { /* A pair of suffix and compression/filter. */ static const struct suffix_code_t filters[] = { { ".Z", "compress" }, { ".bz2", "bzip2" }, { ".gz", "gzip" }, { ".grz", "grzip" }, { ".lrz", "lrzip" }, { ".lz", "lzip" }, { ".lz4", "lz4" }, { ".lzo", "lzop" }, { ".lzma", "lzma" }, { ".uu", "uuencode" }, { ".xz", "xz" }, { ".zst", "zstd"}, { NULL, NULL } }; return get_suffix_code(filters, suffix); } static const char * get_format_code(const char *suffix) { /* A pair of suffix and format. */ static const struct suffix_code_t formats[] = { { ".7z", "7zip" }, { ".ar", "arbsd" }, { ".cpio", "cpio" }, { ".iso", "iso9660" }, { ".mtree", "mtree" }, { ".shar", "shar" }, { ".tar", "paxr" }, { ".warc", "warc" }, { ".xar", "xar" }, { ".zip", "zip" }, { NULL, NULL } }; return get_suffix_code(formats, suffix); } static const char * decompose_alias(const char *suffix) { static const struct suffix_code_t alias[] = { { ".taz", ".tar.gz" }, { ".tgz", ".tar.gz" }, { ".tbz", ".tar.bz2" }, { ".tbz2", ".tar.bz2" }, { ".tz2", ".tar.bz2" }, { ".tlz", ".tar.lzma" }, { ".txz", ".tar.xz" }, { ".tzo", ".tar.lzo" }, { ".taZ", ".tar.Z" }, { ".tZ", ".tar.Z" }, { ".tzst", ".tar.zst" }, { NULL, NULL } }; return get_suffix_code(alias, suffix); } static void _cset_add_filter(struct creation_set *cset, int program, const char *filter) { struct filter_set *new_ptr; char *new_filter; new_ptr = realloc(cset->filters, sizeof(*cset->filters) * (cset->filter_count + 1)); if (new_ptr == NULL) lafe_errc(1, 0, "No memory"); new_filter = strdup(filter); if (new_filter == NULL) lafe_errc(1, 0, "No memory"); cset->filters = new_ptr; cset->filters[cset->filter_count].program = program; cset->filters[cset->filter_count].filter_name = new_filter; cset->filter_count++; } void cset_add_filter(struct creation_set *cset, const char *filter) { _cset_add_filter(cset, 0, filter); } void cset_add_filter_program(struct creation_set *cset, const char *filter) { _cset_add_filter(cset, 1, filter); } int cset_read_support_filter_program(struct creation_set *cset, struct archive *a) { int cnt = 0, i; for (i = 0; i < cset->filter_count; i++) { if (cset->filters[i].program) { archive_read_support_filter_program(a, cset->filters[i].filter_name); ++cnt; } } return (cnt); } int cset_write_add_filters(struct creation_set *cset, struct archive *a, const void **filter_name) { int cnt = 0, i, r; for (i = 0; i < cset->filter_count; i++) { if (cset->filters[i].program) r = archive_write_add_filter_program(a, cset->filters[i].filter_name); else r = archive_write_add_filter_by_name(a, cset->filters[i].filter_name); if (r < ARCHIVE_WARN) { *filter_name = cset->filters[i].filter_name; return (r); } ++cnt; } return (cnt); } void cset_set_format(struct creation_set *cset, const char *format) { char *f; f = strdup(format); if (f == NULL) lafe_errc(1, 0, "No memory"); free(cset->create_format); cset->create_format = f; } const char * cset_get_format(struct creation_set *cset) { return (cset->create_format); } static void _cleanup_filters(struct filter_set *filters, int count) { int i; for (i = 0; i < count; i++) free(filters[i].filter_name); free(filters); } /* * Clean up a creation set. */ void cset_free(struct creation_set *cset) { _cleanup_filters(cset->filters, cset->filter_count); free(cset->create_format); free(cset); } struct creation_set * cset_new(void) { return calloc(1, sizeof(struct creation_set)); } /* * Build a creation set by a file name suffix. */ int cset_auto_compress(struct creation_set *cset, const char *filename) { struct filter_set *old_filters; char *name, *p; const char *code; int old_filter_count; name = strdup(filename); if (name == NULL) lafe_errc(1, 0, "No memory"); /* Save previous filters. */ old_filters = cset->filters; old_filter_count = cset->filter_count; cset->filters = NULL; cset->filter_count = 0; for (;;) { /* Get the suffix. */ p = strrchr(name, '.'); if (p == NULL) break; /* Suppose it indicates compression/filter type * such as ".gz". */ code = get_filter_code(p); if (code != NULL) { cset_add_filter(cset, code); *p = '\0'; continue; } /* Suppose it indicates format type such as ".tar". */ code = get_format_code(p); if (code != NULL) { cset_set_format(cset, code); break; } /* Suppose it indicates alias such as ".tgz". */ code = decompose_alias(p); if (code == NULL) break; /* Replace the suffix. */ *p = '\0'; name = realloc(name, strlen(name) + strlen(code) + 1); if (name == NULL) lafe_errc(1, 0, "No memory"); strcat(name, code); } free(name); if (cset->filters) { struct filter_set *v; int i, r; /* Release previous filters. */ _cleanup_filters(old_filters, old_filter_count); v = malloc(sizeof(*v) * cset->filter_count); if (v == NULL) lafe_errc(1, 0, "No memory"); /* Reverse filter sequence. */ for (i = 0, r = cset->filter_count; r > 0; ) v[i++] = cset->filters[--r]; free(cset->filters); cset->filters = v; return (1); } else { /* Put previous filters back. */ cset->filters = old_filters; cset->filter_count = old_filter_count; return (0); } } diff --git a/contrib/libarchive/tar/read.c b/contrib/libarchive/tar/read.c index 8563fe714f9a..7cbcfb19ff0a 100644 --- a/contrib/libarchive/tar/read.c +++ b/contrib/libarchive/tar/read.c @@ -1,401 +1,401 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. */ #include "bsdtar_platform.h" #ifdef HAVE_SYS_TYPES_H #include #endif #ifdef HAVE_SYS_PARAM_H #include #endif #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_GRP_H #include #endif #ifdef HAVE_IO_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_PWD_H #include #endif #ifdef HAVE_STDINT_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_TIME_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #include "bsdtar.h" -#include "err.h" +#include "lafe_err.h" struct progress_data { struct bsdtar *bsdtar; struct archive *archive; struct archive_entry *entry; }; static void read_archive(struct bsdtar *bsdtar, char mode, struct archive *); static int unmatched_inclusions_warn(struct archive *matching, const char *); void tar_mode_t(struct bsdtar *bsdtar) { read_archive(bsdtar, 't', NULL); if (unmatched_inclusions_warn(bsdtar->matching, "Not found in archive") != 0) bsdtar->return_value = 1; } void tar_mode_x(struct bsdtar *bsdtar) { struct archive *writer; writer = archive_write_disk_new(); if (writer == NULL) lafe_errc(1, ENOMEM, "Cannot allocate disk writer object"); if ((bsdtar->flags & OPTFLAG_NUMERIC_OWNER) == 0) archive_write_disk_set_standard_lookup(writer); archive_write_disk_set_options(writer, bsdtar->extract_flags); read_archive(bsdtar, 'x', writer); if (unmatched_inclusions_warn(bsdtar->matching, "Not found in archive") != 0) bsdtar->return_value = 1; archive_write_free(writer); } static void progress_func(void *cookie) { struct progress_data *progress_data = (struct progress_data *)cookie; struct bsdtar *bsdtar = progress_data->bsdtar; struct archive *a = progress_data->archive; struct archive_entry *entry = progress_data->entry; uint64_t comp, uncomp; int compression; if (!need_report()) return; if (bsdtar->verbose) fprintf(stderr, "\n"); if (a != NULL) { comp = archive_filter_bytes(a, -1); uncomp = archive_filter_bytes(a, 0); if (comp > uncomp) compression = 0; else compression = (int)((uncomp - comp) * 100 / uncomp); fprintf(stderr, "In: %s bytes, compression %d%%;", tar_i64toa(comp), compression); fprintf(stderr, " Out: %d files, %s bytes\n", archive_file_count(a), tar_i64toa(uncomp)); } if (entry != NULL) { safe_fprintf(stderr, "Current: %s", archive_entry_pathname(entry)); fprintf(stderr, " (%s bytes)\n", tar_i64toa(archive_entry_size(entry))); } } /* * Handle 'x' and 't' modes. */ static void read_archive(struct bsdtar *bsdtar, char mode, struct archive *writer) { struct progress_data progress_data; FILE *out; struct archive *a; struct archive_entry *entry; const char *reader_options; int r; while (*bsdtar->argv) { if (archive_match_include_pattern(bsdtar->matching, *bsdtar->argv) != ARCHIVE_OK) lafe_errc(1, 0, "Error inclusion pattern: %s", archive_error_string(bsdtar->matching)); bsdtar->argv++; } if (bsdtar->names_from_file != NULL) if (archive_match_include_pattern_from_file( bsdtar->matching, bsdtar->names_from_file, (bsdtar->flags & OPTFLAG_NULL)) != ARCHIVE_OK) lafe_errc(1, 0, "Error inclusion pattern: %s", archive_error_string(bsdtar->matching)); a = archive_read_new(); if (cset_read_support_filter_program(bsdtar->cset, a) == 0) archive_read_support_filter_all(a); archive_read_support_format_all(a); reader_options = getenv(ENV_READER_OPTIONS); if (reader_options != NULL) { size_t module_len = sizeof(IGNORE_WRONG_MODULE_NAME) - 1; size_t opt_len = strlen(reader_options) + 1; char *p; /* Set default read options. */ if ((p = malloc(module_len + opt_len)) == NULL) lafe_errc(1, errno, "Out of memory"); /* Prepend magic code to ignore options for * a format or modules which are not added to * the archive read object. */ memcpy(p, IGNORE_WRONG_MODULE_NAME, module_len); memcpy(p + module_len, reader_options, opt_len); r = archive_read_set_options(a, p); free(p); if (r == ARCHIVE_FATAL) lafe_errc(1, 0, "%s", archive_error_string(a)); else archive_clear_error(a); } if (ARCHIVE_OK != archive_read_set_options(a, bsdtar->option_options)) lafe_errc(1, 0, "%s", archive_error_string(a)); if (bsdtar->flags & OPTFLAG_IGNORE_ZEROS) if (archive_read_set_options(a, "read_concatenated_archives") != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); if (bsdtar->passphrase != NULL) r = archive_read_add_passphrase(a, bsdtar->passphrase); else r = archive_read_set_passphrase_callback(a, bsdtar, &passphrase_callback); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); if (archive_read_open_filename(a, bsdtar->filename, bsdtar->bytes_per_block)) lafe_errc(1, 0, "Error opening archive: %s", archive_error_string(a)); do_chdir(bsdtar); if (mode == 'x') { /* Set an extract callback so that we can handle SIGINFO. */ progress_data.bsdtar = bsdtar; progress_data.archive = a; archive_read_extract_set_progress_callback(a, progress_func, &progress_data); } if (mode == 'x' && (bsdtar->flags & OPTFLAG_CHROOT)) { #if HAVE_CHROOT if (chroot(".") != 0) lafe_errc(1, errno, "Can't chroot to \".\""); #else lafe_errc(1, 0, "chroot isn't supported on this platform"); #endif } #if defined(_WIN32) && !defined(__CYGWIN__) if (mode == 'x' && (bsdtar->flags & OPTFLAG_STDOUT)) { _setmode(1, _O_BINARY); } #endif for (;;) { /* Support --fast-read option */ const char *p; if ((bsdtar->flags & OPTFLAG_FAST_READ) && archive_match_path_unmatched_inclusions(bsdtar->matching) == 0) break; r = archive_read_next_header(a, &entry); progress_data.entry = entry; if (r == ARCHIVE_EOF) break; if (r < ARCHIVE_OK) lafe_warnc(0, "%s", archive_error_string(a)); if (r <= ARCHIVE_WARN) bsdtar->return_value = 1; if (r == ARCHIVE_RETRY) { /* Retryable error: try again */ lafe_warnc(0, "Retrying..."); continue; } if (r == ARCHIVE_FATAL) break; p = archive_entry_pathname(entry); if (p == NULL || p[0] == '\0') { lafe_warnc(0, "Archive entry has empty or unreadable filename ... skipping."); bsdtar->return_value = 1; continue; } if (bsdtar->uid >= 0) { archive_entry_set_uid(entry, bsdtar->uid); archive_entry_set_uname(entry, NULL); } if (bsdtar->gid >= 0) { archive_entry_set_gid(entry, bsdtar->gid); archive_entry_set_gname(entry, NULL); } if (bsdtar->uname) archive_entry_set_uname(entry, bsdtar->uname); if (bsdtar->gname) archive_entry_set_gname(entry, bsdtar->gname); /* * Note that pattern exclusions are checked before * pathname rewrites are handled. This gives more * control over exclusions, since rewrites always lose * information. (For example, consider a rewrite * s/foo[0-9]/foo/. If we check exclusions after the * rewrite, there would be no way to exclude foo1/bar * while allowing foo2/bar.) */ if (archive_match_excluded(bsdtar->matching, entry)) continue; /* Excluded by a pattern test. */ if (mode == 't') { /* Perversely, gtar uses -O to mean "send to stderr" * when used with -t. */ out = (bsdtar->flags & OPTFLAG_STDOUT) ? stderr : stdout; /* * TODO: Provide some reasonable way to * preview rewrites. gtar always displays * the unedited path in -t output, which means * you cannot easily preview rewrites. */ if (bsdtar->verbose < 2) safe_fprintf(out, "%s", archive_entry_pathname(entry)); else list_item_verbose(bsdtar, out, entry); fflush(out); r = archive_read_data_skip(a); if (r == ARCHIVE_WARN) { fprintf(out, "\n"); lafe_warnc(0, "%s", archive_error_string(a)); } if (r == ARCHIVE_RETRY) { fprintf(out, "\n"); lafe_warnc(0, "%s", archive_error_string(a)); } if (r == ARCHIVE_FATAL) { fprintf(out, "\n"); lafe_warnc(0, "%s", archive_error_string(a)); bsdtar->return_value = 1; break; } fprintf(out, "\n"); } else { /* Note: some rewrite failures prevent extraction. */ if (edit_pathname(bsdtar, entry)) continue; /* Excluded by a rewrite failure. */ if ((bsdtar->flags & OPTFLAG_INTERACTIVE) && !yes("extract '%s'", archive_entry_pathname(entry))) continue; if (bsdtar->verbose > 1) { /* GNU tar uses -tv format with -xvv */ safe_fprintf(stderr, "x "); list_item_verbose(bsdtar, stderr, entry); fflush(stderr); } else if (bsdtar->verbose > 0) { /* Format follows SUSv2, including the * deferred '\n'. */ safe_fprintf(stderr, "x %s", archive_entry_pathname(entry)); fflush(stderr); } /* TODO siginfo_printinfo(bsdtar, 0); */ if (bsdtar->flags & OPTFLAG_STDOUT) r = archive_read_data_into_fd(a, 1); else r = archive_read_extract2(a, entry, writer); if (r != ARCHIVE_OK) { if (!bsdtar->verbose) safe_fprintf(stderr, "%s", archive_entry_pathname(entry)); safe_fprintf(stderr, ": %s: %s", archive_error_string(a), strerror(archive_errno(a))); if (!bsdtar->verbose) fprintf(stderr, "\n"); bsdtar->return_value = 1; } if (bsdtar->verbose) fprintf(stderr, "\n"); if (r == ARCHIVE_FATAL) break; } } r = archive_read_close(a); if (r != ARCHIVE_OK) lafe_warnc(0, "%s", archive_error_string(a)); if (r <= ARCHIVE_WARN) bsdtar->return_value = 1; if (bsdtar->verbose > 2) fprintf(stdout, "Archive Format: %s, Compression: %s\n", archive_format_name(a), archive_filter_name(a, 0)); archive_read_free(a); } static int unmatched_inclusions_warn(struct archive *matching, const char *msg) { const char *p; int r; if (matching == NULL) return (0); while ((r = archive_match_path_unmatched_inclusions_next( matching, &p)) == ARCHIVE_OK) lafe_warnc(0, "%s: %s", p, msg); if (r == ARCHIVE_FATAL) lafe_errc(1, errno, "Out of memory"); return (archive_match_path_unmatched_inclusions(matching)); } diff --git a/contrib/libarchive/tar/subst.c b/contrib/libarchive/tar/subst.c index 5546b5f93b12..a5d644dc5a70 100644 --- a/contrib/libarchive/tar/subst.c +++ b/contrib/libarchive/tar/subst.c @@ -1,329 +1,329 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2008 Joerg Sonnenberger * All rights reserved. */ #include "bsdtar_platform.h" #if defined(HAVE_REGEX_H) || defined(HAVE_PCREPOSIX_H) || defined(HAVE_PCRE2POSIX_H) #include "bsdtar.h" #include #if defined(HAVE_PCREPOSIX_H) #include #elif defined(HAVE_PCRE2POSIX_H) #include #else #include #endif #include #include #ifndef REG_BASIC #define REG_BASIC 0 #endif -#include "err.h" +#include "lafe_err.h" struct subst_rule { struct subst_rule *next; regex_t re; char *result; unsigned int global:1, print:1, regular:1, symlink:1, hardlink:1, from_begin:1; }; struct substitution { struct subst_rule *first_rule, *last_rule; }; static void init_substitution(struct bsdtar *bsdtar) { struct substitution *subst; bsdtar->substitution = subst = malloc(sizeof(*subst)); if (subst == NULL) lafe_errc(1, errno, "Out of memory"); subst->first_rule = subst->last_rule = NULL; } void add_substitution(struct bsdtar *bsdtar, const char *rule_text) { struct subst_rule *rule; struct substitution *subst; const char *end_pattern, *start_subst; char *pattern; int r; if ((subst = bsdtar->substitution) == NULL) { init_substitution(bsdtar); subst = bsdtar->substitution; } rule = malloc(sizeof(*rule)); if (rule == NULL) lafe_errc(1, errno, "Out of memory"); rule->next = NULL; rule->result = NULL; if (subst->last_rule == NULL) subst->first_rule = rule; else subst->last_rule->next = rule; subst->last_rule = rule; if (*rule_text == '\0') lafe_errc(1, 0, "Empty replacement string"); end_pattern = strchr(rule_text + 1, *rule_text); if (end_pattern == NULL) lafe_errc(1, 0, "Invalid replacement string"); pattern = malloc(end_pattern - rule_text); if (pattern == NULL) lafe_errc(1, errno, "Out of memory"); memcpy(pattern, rule_text + 1, end_pattern - rule_text - 1); pattern[end_pattern - rule_text - 1] = '\0'; if ((r = regcomp(&rule->re, pattern, REG_BASIC)) != 0) { char buf[80]; regerror(r, &rule->re, buf, sizeof(buf)); lafe_errc(1, 0, "Invalid regular expression: %s", buf); } free(pattern); start_subst = end_pattern + 1; end_pattern = strchr(start_subst, *rule_text); if (end_pattern == NULL) lafe_errc(1, 0, "Invalid replacement string"); rule->result = malloc(end_pattern - start_subst + 1); if (rule->result == NULL) lafe_errc(1, errno, "Out of memory"); memcpy(rule->result, start_subst, end_pattern - start_subst); rule->result[end_pattern - start_subst] = '\0'; /* Defaults */ rule->global = 0; /* Don't do multiple replacements. */ rule->print = 0; /* Don't print. */ rule->regular = 1; /* Rewrite regular filenames. */ rule->symlink = 1; /* Rewrite symlink targets. */ rule->hardlink = 1; /* Rewrite hardlink targets. */ rule->from_begin = 0; /* Don't match from start. */ while (*++end_pattern) { switch (*end_pattern) { case 'b': case 'B': rule->from_begin = 1; break; case 'g': case 'G': rule->global = 1; break; case 'h': rule->hardlink = 1; break; case 'H': rule->hardlink = 0; break; case 'p': case 'P': rule->print = 1; break; case 'r': rule->regular = 1; break; case 'R': rule->regular = 0; break; case 's': rule->symlink = 1; break; case 'S': rule->symlink = 0; break; default: lafe_errc(1, 0, "Invalid replacement flag %c", *end_pattern); /* NOTREACHED */ } } } static void realloc_strncat(char **str, const char *append, size_t len) { char *new_str; size_t old_len; if (*str == NULL) old_len = 0; else old_len = strlen(*str); new_str = malloc(old_len + len + 1); if (new_str == NULL) lafe_errc(1, errno, "Out of memory"); if (*str != NULL) memcpy(new_str, *str, old_len); memcpy(new_str + old_len, append, len); new_str[old_len + len] = '\0'; free(*str); *str = new_str; } static void realloc_strcat(char **str, const char *append) { char *new_str; size_t old_len; if (*str == NULL) old_len = 0; else old_len = strlen(*str); new_str = malloc(old_len + strlen(append) + 1); if (new_str == NULL) lafe_errc(1, errno, "Out of memory"); if (*str != NULL) memcpy(new_str, *str, old_len); strcpy(new_str + old_len, append); free(*str); *str = new_str; } int apply_substitution(struct bsdtar *bsdtar, const char *name, char **result, int symlink_target, int hardlink_target) { const char *path = name; regmatch_t matches[10]; char* buffer = NULL; size_t i, j; struct subst_rule *rule; struct substitution *subst; int c, got_match, print_match; *result = NULL; if ((subst = bsdtar->substitution) == NULL) return 0; got_match = 0; print_match = 0; for (rule = subst->first_rule; rule != NULL; rule = rule->next) { if (symlink_target) { if (!rule->symlink) continue; } else if (hardlink_target) { if (!rule->hardlink) continue; } else { /* Regular filename. */ if (!rule->regular) continue; } if (rule->from_begin && *result) { realloc_strcat(result, name); if (buffer) buffer[0] = 0; realloc_strcat(&buffer, *result); name = buffer; (*result)[0] = 0; } while (1) { if (regexec(&rule->re, name, 10, matches, 0)) break; got_match = 1; print_match |= rule->print; realloc_strncat(result, name, matches[0].rm_so); for (i = 0, j = 0; rule->result[i] != '\0'; ++i) { if (rule->result[i] == '~') { realloc_strncat(result, rule->result + j, i - j); realloc_strncat(result, name + matches[0].rm_so, matches[0].rm_eo - matches[0].rm_so); j = i + 1; continue; } if (rule->result[i] != '\\') continue; ++i; c = rule->result[i]; switch (c) { case '~': case '\\': realloc_strncat(result, rule->result + j, i - j - 1); j = i; break; case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': realloc_strncat(result, rule->result + j, i - j - 1); if ((size_t)(c - '0') > (size_t)(rule->re.re_nsub)) { free(buffer); free(*result); *result = NULL; return -1; } realloc_strncat(result, name + matches[c - '0'].rm_so, matches[c - '0'].rm_eo - matches[c - '0'].rm_so); j = i + 1; break; default: /* Just continue; */ break; } } realloc_strcat(result, rule->result + j); name += matches[0].rm_eo; if (!rule->global) break; } } if (got_match) realloc_strcat(result, name); free(buffer); if (print_match) fprintf(stderr, "%s >> %s\n", path, *result); return got_match; } void cleanup_substitution(struct bsdtar *bsdtar) { struct subst_rule *rule; struct substitution *subst; if ((subst = bsdtar->substitution) == NULL) return; while ((rule = subst->first_rule) != NULL) { subst->first_rule = rule->next; free(rule->result); regfree(&rule->re); free(rule); } free(subst); } #endif /* defined(HAVE_REGEX_H) || defined(HAVE_PCREPOSIX_H) || defined(HAVE_PCRE2POSIX_H) */ diff --git a/contrib/libarchive/tar/test/test_crlf_mtree.c b/contrib/libarchive/tar/test/test_crlf_mtree.c new file mode 100644 index 000000000000..5ef8811369c1 --- /dev/null +++ b/contrib/libarchive/tar/test/test_crlf_mtree.c @@ -0,0 +1,74 @@ +/*- + * SPDX-License-Identifier: BSD-2-Clause + * + * Copyright (c) 2018 The FreeBSD Foundation + * All rights reserved. + * + * This software was developed by Arshan Khanifar + * under sponsorship from the FreeBSD Foundation. + */ +#include "test.h" + +DEFINE_TEST(test_crlf_mtree) +{ + char *p0; + size_t s; + int r; + p0 = NULL; + char *content = "#mtree\r\n" + "f type=file uname=\\\r\n" + "root gname=root mode=0755 content=bar/foo\r\n" + "g type=file uname=root gname=root mode=0755 content=bar/goo\r\n"; + char *filename = "output.tar"; +#if defined(_WIN32) && !defined(__CYGWIN__) + char *p; +#endif + + /* an absolute path to mtree file */ + char *mtree_file = "/METALOG.mtree"; + char *absolute_path = malloc(strlen(testworkdir) + strlen(mtree_file) + 1); + strcpy(absolute_path, testworkdir); + strcat(absolute_path, mtree_file ); + + /* Create an archive using an mtree file. */ + assertMakeFile(absolute_path, 0777, content); + assertMakeDir("bar", 0775); + assertMakeFile("bar/foo", 0777, "abc"); + assertMakeFile("bar/goo", 0777, "abc"); + +#if defined(_WIN32) && !defined(__CYGWIN__) + p = absolute_path; + while(*p != '\0') { + if (*p == '/') + *p = '\\'; + p++; + } + + r = systemf("%s -cf %s @%s >step1.out 2>step1.err", testprog, filename, absolute_path); + failure("Error invoking %s -cf %s -C bar @%s", testprog, filename, absolute_path); +#else + r = systemf("%s -cf %s \"@%s\" >step1.out 2>step1.err", testprog, filename, absolute_path); + failure("Error invoking %s -cf %s -C bar \"@%s\"", testprog, filename, absolute_path); +#endif + + assertEqualInt(r, 0); + assertEmptyFile("step1.out"); + assertEmptyFile("step1.err"); + + /* Do validation of the constructed archive. */ + + p0 = slurpfile(&s, "output.tar"); + if (!assert(p0 != NULL)) + goto done; + if (!assert(s >= 2048)) + goto done; + assertEqualMem(p0 + 0, "f", 2); + assertEqualMem(p0 + 512, "abc", 4); + assertEqualMem(p0 + 1024, "g", 2); + assertEqualMem(p0 + 1536, "abc", 4); +done: + free(p0); + free(absolute_path); +} + + diff --git a/contrib/libarchive/tar/test/test_option_safe_writes.c b/contrib/libarchive/tar/test/test_option_safe_writes.c index b88479bc5f35..d30b9a745927 100644 --- a/contrib/libarchive/tar/test/test_option_safe_writes.c +++ b/contrib/libarchive/tar/test/test_option_safe_writes.c @@ -1,58 +1,61 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2020 Martin Matuska * All rights reserved. */ #include "test.h" DEFINE_TEST(test_option_safe_writes) { /* Create files */ assertMakeDir("in", 0755); assertEqualInt(0, chdir("in")); assertMakeFile("f", 0644, "a"); assertMakeFile("fh", 0644, "b"); assertMakeFile("d", 0644, "c"); assertMakeFile("fs", 0644, "d"); assertMakeFile("ds", 0644, "e"); + assertMakeDir("fd", 0755); assertEqualInt(0, chdir("..")); /* Tar files up */ assertEqualInt(0, - systemf("%s -c -C in -f t.tar f fh d fs ds " + systemf("%s -c -C in -f t.tar f fh d fs ds fd " ">pack.out 2>pack.err", testprog)); /* Verify that nothing went to stdout or stderr. */ assertEmptyFile("pack.err"); assertEmptyFile("pack.out"); /* Create various objects */ assertMakeDir("out", 0755); assertEqualInt(0, chdir("out")); assertMakeFile("f", 0644, "a"); assertMakeHardlink("fh", "f"); + assertMakeFile("fd", 0644, "b"); assertMakeDir("d", 0755); if (canSymlink()) { assertMakeSymlink("fs", "f", 0); assertMakeSymlink("ds", "d", 1); } assertEqualInt(0, chdir("..")); /* Extract created archive with safe writes */ assertEqualInt(0, systemf("%s -x -C out --safe-writes -f t.tar " ">unpack.out 2>unpack.err", testprog)); /* Verify that nothing went to stdout or stderr. */ assertEmptyFile("unpack.err"); assertEmptyFile("unpack.out"); /* Verify that files were overwritten properly */ assertEqualInt(0, chdir("out")); assertTextFileContents("a","f"); assertTextFileContents("b","fh"); assertTextFileContents("c","d"); assertTextFileContents("d","fs"); assertTextFileContents("e","ds"); + assertIsDir("fd", 0755); } diff --git a/contrib/libarchive/tar/util.c b/contrib/libarchive/tar/util.c index c99f67797562..fc5e15cb039f 100644 --- a/contrib/libarchive/tar/util.c +++ b/contrib/libarchive/tar/util.c @@ -1,784 +1,785 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. */ #include "bsdtar_platform.h" #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_SYS_TYPES_H #include /* Linux doesn't define mode_t, etc. in sys/stat.h. */ #endif #include #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_IO_H #include #endif #ifdef HAVE_STDARG_H #include #endif #ifdef HAVE_STDINT_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_WCTYPE_H #include #else /* If we don't have wctype, we need to hack up some version of iswprint(). */ #define iswprint isprint #endif #include "bsdtar.h" -#include "err.h" +#include "lafe_err.h" #include "passphrase.h" static size_t bsdtar_expand_char(char *, size_t, size_t, char); static const char *strip_components(const char *path, int elements); #if defined(_WIN32) && !defined(__CYGWIN__) #define read _read #endif /* TODO: Hack up a version of mbtowc for platforms with no wide * character support at all. I think the following might suffice, * but it needs careful testing. * #if !HAVE_MBTOWC * #define mbtowc(wcp, p, n) ((*wcp = *p), 1) * #endif */ /* * Print a string, taking care with any non-printable characters. * * Note that we use a stack-allocated buffer to receive the formatted * string if we can. This is partly performance (avoiding a call to * malloc()), partly out of expedience (we have to call vsnprintf() * before malloc() anyway to find out how big a buffer we need; we may * as well point that first call at a small local buffer in case it * works). */ void safe_fprintf(FILE * restrict f, const char * restrict fmt, ...) { char fmtbuff_stack[256]; /* Place to format the printf() string. */ char outbuff[256]; /* Buffer for outgoing characters. */ char *fmtbuff_heap; /* If fmtbuff_stack is too small, we use malloc */ char *fmtbuff; /* Pointer to fmtbuff_stack or fmtbuff_heap. */ size_t fmtbuff_length; int length, n; va_list ap; const char *p; size_t i; wchar_t wc; char try_wc; /* Use a stack-allocated buffer if we can, for speed and safety. */ memset(fmtbuff_stack, '\0', sizeof(fmtbuff_stack)); fmtbuff_heap = NULL; fmtbuff_length = sizeof(fmtbuff_stack); fmtbuff = fmtbuff_stack; /* Try formatting into the stack buffer. */ va_start(ap, fmt); length = vsnprintf(fmtbuff, fmtbuff_length, fmt, ap); va_end(ap); /* If vsnprintf will always fail, stop early. */ if (length < 0 && errno == EOVERFLOW) return; /* If the result was too large, allocate a buffer on the heap. */ while (length < 0 || (size_t)length >= fmtbuff_length) { if (length >= 0 && (size_t)length >= fmtbuff_length) fmtbuff_length = (size_t)length + 1; else if (fmtbuff_length < 8192) fmtbuff_length *= 2; else if (fmtbuff_length < 1000000) fmtbuff_length += fmtbuff_length / 4; else { fmtbuff[fmtbuff_length - 1] = '\0'; length = (int)strlen(fmtbuff); break; } free(fmtbuff_heap); fmtbuff_heap = malloc(fmtbuff_length); /* Reformat the result into the heap buffer if we can. */ if (fmtbuff_heap != NULL) { fmtbuff = fmtbuff_heap; va_start(ap, fmt); length = vsnprintf(fmtbuff, fmtbuff_length, fmt, ap); va_end(ap); } else { /* Leave fmtbuff pointing to the truncated * string in fmtbuff_stack. */ fmtbuff_stack[sizeof(fmtbuff_stack) - 1] = '\0'; fmtbuff = fmtbuff_stack; length = (int)strlen(fmtbuff); break; } } /* Note: mbrtowc() has a cleaner API, but mbtowc() seems a bit * more portable, so we use that here instead. */ if (mbtowc(NULL, NULL, 1) == -1) { /* Reset the shift state. */ /* mbtowc() should never fail in practice, but * handle the theoretical error anyway. */ free(fmtbuff_heap); return; } /* Write data, expanding unprintable characters. */ p = fmtbuff; i = 0; try_wc = 1; while (*p != '\0') { /* Convert to wide char, test if the wide * char is printable in the current locale. */ if (try_wc && (n = mbtowc(&wc, p, length)) != -1) { length -= n; if (iswprint(wc) && wc != L'\\') { /* Printable, copy the bytes through. */ while (n-- > 0) outbuff[i++] = *p++; } else { /* Not printable, format the bytes. */ while (n-- > 0) i += bsdtar_expand_char( outbuff, sizeof(outbuff), i, *p++); } } else { /* After any conversion failure, don't bother * trying to convert the rest. */ i += bsdtar_expand_char(outbuff, sizeof(outbuff), i, *p++); try_wc = 0; } /* If our output buffer is full, dump it and keep going. */ if (i > (sizeof(outbuff) - 128)) { outbuff[i] = '\0'; fprintf(f, "%s", outbuff); i = 0; } } outbuff[i] = '\0'; fprintf(f, "%s", outbuff); /* If we allocated a heap-based formatting buffer, free it now. */ free(fmtbuff_heap); } /* * Render an arbitrary sequence of bytes into printable ASCII characters. */ static size_t bsdtar_expand_char(char *buff, size_t buffsize, size_t offset, char c) { size_t i = offset; if (isprint((unsigned char)c) && c != '\\') buff[i++] = c; else { buff[i++] = '\\'; switch (c) { case '\a': buff[i++] = 'a'; break; case '\b': buff[i++] = 'b'; break; case '\f': buff[i++] = 'f'; break; case '\n': buff[i++] = 'n'; break; #if '\r' != '\n' /* On some platforms, \n and \r are the same. */ case '\r': buff[i++] = 'r'; break; #endif case '\t': buff[i++] = 't'; break; case '\v': buff[i++] = 'v'; break; case '\\': buff[i++] = '\\'; break; default: snprintf(buff + i, buffsize - i, "%03o", 0xFF & (unsigned int)c); i += 3; } } return (i - offset); } int yes(const char *fmt, ...) { char buff[32]; char *p; ssize_t l; int read_fd = 2; /* stderr */ va_list ap; va_start(ap, fmt); vfprintf(stderr, fmt, ap); va_end(ap); fprintf(stderr, " (y/N)? "); fflush(stderr); #if defined(_WIN32) && !defined(__CYGWIN__) /* To be resilient when stdin is a pipe, bsdtar prefers to read from * stderr. On Windows, stderr cannot be read. The nearest "piping * resilient" equivalent is reopening the console input handle. */ read_fd = _open("CONIN$", O_RDONLY); if (read_fd < 0) { fprintf(stderr, "Keyboard read failed\n"); exit(1); } #endif l = read(read_fd, buff, sizeof(buff) - 1); #if defined(_WIN32) && !defined(__CYGWIN__) _close(read_fd); #endif if (l < 0) { fprintf(stderr, "Keyboard read failed\n"); exit(1); } if (l == 0) return (0); buff[l] = 0; for (p = buff; *p != '\0'; p++) { if (isspace((unsigned char)*p)) continue; switch(*p) { case 'y': case 'Y': return (1); case 'n': case 'N': return (0); default: return (0); } } return (0); } /*- * The logic here for -C attempts to avoid * chdir() as long as possible. For example: * "-C /foo -C /bar file" needs chdir("/bar") but not chdir("/foo") * "-C /foo -C bar file" needs chdir("/foo/bar") * "-C /foo -C bar /file1" does not need chdir() * "-C /foo -C bar /file1 file2" needs chdir("/foo/bar") before file2 * * The only correct way to handle this is to record a "pending" chdir * request and combine multiple requests intelligently until we * need to process a non-absolute file. set_chdir() adds the new dir * to the pending list; do_chdir() actually executes any pending chdir. * * This way, programs that build tar command lines don't have to worry * about -C with non-existent directories; such requests will only * fail if the directory must be accessed. * */ void set_chdir(struct bsdtar *bsdtar, const char *newdir) { #if defined(_WIN32) && !defined(__CYGWIN__) if (newdir[0] == '/' || newdir[0] == '\\' || /* Detect this type, for example, "C:\" or "C:/" */ (((newdir[0] >= 'a' && newdir[0] <= 'z') || (newdir[0] >= 'A' && newdir[0] <= 'Z')) && newdir[1] == ':' && (newdir[2] == '/' || newdir[2] == '\\'))) { #else if (newdir[0] == '/') { #endif /* The -C /foo -C /bar case; dump first one. */ free(bsdtar->pending_chdir); bsdtar->pending_chdir = NULL; } if (bsdtar->pending_chdir == NULL) /* Easy case: no previously-saved dir. */ bsdtar->pending_chdir = strdup(newdir); else { /* The -C /foo -C bar case; concatenate */ char *old_pending = bsdtar->pending_chdir; size_t old_len = strlen(old_pending); size_t new_len = old_len + strlen(newdir) + 2; bsdtar->pending_chdir = malloc(new_len); if (old_pending[old_len - 1] == '/') old_pending[old_len - 1] = '\0'; if (bsdtar->pending_chdir != NULL) snprintf(bsdtar->pending_chdir, new_len, "%s/%s", old_pending, newdir); free(old_pending); } if (bsdtar->pending_chdir == NULL) lafe_errc(1, errno, "No memory"); } void do_chdir(struct bsdtar *bsdtar) { if (bsdtar->pending_chdir == NULL) return; if (chdir(bsdtar->pending_chdir) != 0) { lafe_errc(1, 0, "could not chdir to '%s'", bsdtar->pending_chdir); } free(bsdtar->pending_chdir); bsdtar->pending_chdir = NULL; } static const char * strip_components(const char *p, int elements) { /* Skip as many elements as necessary. */ while (elements > 0) { switch (*p++) { case '/': #if defined(_WIN32) && !defined(__CYGWIN__) case '\\': /* Support \ path sep on Windows ONLY. */ #endif elements--; break; case '\0': /* Path is too short, skip it. */ return (NULL); } } /* Skip any / characters. This handles short paths that have * additional / termination. This also handles the case where * the logic above stops in the middle of a duplicate // * sequence (which would otherwise get converted to an * absolute path). */ for (;;) { switch (*p) { case '/': #if defined(_WIN32) && !defined(__CYGWIN__) case '\\': /* Support \ path sep on Windows ONLY. */ #endif ++p; break; case '\0': return (NULL); default: return (p); } } } static void warn_strip_leading_char(struct bsdtar *bsdtar, const char *c) { if (!bsdtar->warned_lead_slash) { lafe_warnc(0, "Removing leading '%c' from member names", c[0]); bsdtar->warned_lead_slash = 1; } } static void warn_strip_drive_letter(struct bsdtar *bsdtar) { if (!bsdtar->warned_lead_slash) { lafe_warnc(0, "Removing leading drive letter from " "member names"); bsdtar->warned_lead_slash = 1; } } /* * Convert absolute path to non-absolute path by skipping leading * absolute path prefixes. */ static const char* strip_absolute_path(struct bsdtar *bsdtar, const char *p) { const char *rp; /* Remove leading "//./" or "//?/" or "//?/UNC/" * (absolute path prefixes used by Windows API) */ if ((p[0] == '/' || p[0] == '\\') && (p[1] == '/' || p[1] == '\\') && (p[2] == '.' || p[2] == '?') && (p[3] == '/' || p[3] == '\\')) { if (p[2] == '?' && (p[4] == 'U' || p[4] == 'u') && (p[5] == 'N' || p[5] == 'n') && (p[6] == 'C' || p[6] == 'c') && (p[7] == '/' || p[7] == '\\')) p += 8; else p += 4; warn_strip_drive_letter(bsdtar); } /* Remove multiple leading slashes and Windows drive letters. */ do { rp = p; if (((p[0] >= 'a' && p[0] <= 'z') || (p[0] >= 'A' && p[0] <= 'Z')) && p[1] == ':') { p += 2; warn_strip_drive_letter(bsdtar); } /* Remove leading "/../", "/./", "//", etc. */ while (p[0] == '/' || p[0] == '\\') { if (p[1] == '.' && p[2] == '.' && (p[3] == '/' || p[3] == '\\')) { p += 3; /* Remove "/..", leave "/" for next pass. */ } else if (p[1] == '.' && (p[2] == '/' || p[2] == '\\')) { p += 2; /* Remove "/.", leave "/" for next pass. */ } else p += 1; /* Remove "/". */ warn_strip_leading_char(bsdtar, rp); } } while (rp != p); return (p); } /* * Handle --strip-components and any future path-rewriting options. * Returns non-zero if the pathname should not be extracted. * * Note: The rewrites are applied uniformly to pathnames and hardlink * names but not to symlink bodies. This is deliberate: Symlink * bodies are not necessarily filenames. Even when they are, they * need to be interpreted relative to the directory containing them, * so simple rewrites like this are rarely appropriate. * * TODO: Support pax-style regex path rewrites. */ int edit_pathname(struct bsdtar *bsdtar, struct archive_entry *entry) { const char *name = archive_entry_pathname(entry); const char *original_name = name; const char *hardlinkname = archive_entry_hardlink(entry); const char *original_hardlinkname = hardlinkname; #if defined(HAVE_REGEX_H) || defined(HAVE_PCREPOSIX_H) || defined(HAVE_PCRE2POSIX_H) char *subst_name; int r; /* Apply user-specified substitution to pathname. */ r = apply_substitution(bsdtar, name, &subst_name, 0, 0); if (r == -1) { lafe_warnc(0, "Invalid substitution, skipping entry"); return 1; } if (r == 1) { archive_entry_copy_pathname(entry, subst_name); if (*subst_name == '\0') { free(subst_name); return -1; } else free(subst_name); name = archive_entry_pathname(entry); original_name = name; } /* Apply user-specified substitution to hardlink target. */ if (hardlinkname != NULL) { r = apply_substitution(bsdtar, hardlinkname, &subst_name, 0, 1); if (r == -1) { lafe_warnc(0, "Invalid substitution, skipping entry"); return 1; } if (r == 1) { archive_entry_copy_hardlink(entry, subst_name); free(subst_name); } hardlinkname = archive_entry_hardlink(entry); original_hardlinkname = hardlinkname; } /* Apply user-specified substitution to symlink body. */ if (archive_entry_symlink(entry) != NULL) { r = apply_substitution(bsdtar, archive_entry_symlink(entry), &subst_name, 1, 0); if (r == -1) { lafe_warnc(0, "Invalid substitution, skipping entry"); return 1; } if (r == 1) { archive_entry_copy_symlink(entry, subst_name); free(subst_name); } } #endif /* Strip leading dir names as per --strip-components option. */ if (bsdtar->strip_components > 0) { name = strip_components(name, bsdtar->strip_components); if (name == NULL) return (1); if (hardlinkname != NULL) { hardlinkname = strip_components(hardlinkname, bsdtar->strip_components); if (hardlinkname == NULL) return (1); } } if ((bsdtar->flags & OPTFLAG_ABSOLUTE_PATHS) == 0) { /* By default, don't write or restore absolute pathnames. */ name = strip_absolute_path(bsdtar, name); if (*name == '\0') name = "."; if (hardlinkname != NULL) { hardlinkname = strip_absolute_path(bsdtar, hardlinkname); if (*hardlinkname == '\0') return (1); } } else { /* Strip redundant leading '/' characters. */ while (name[0] == '/' && name[1] == '/') name++; } /* Replace name in archive_entry. */ if (name != original_name) { archive_entry_copy_pathname(entry, name); } if (hardlinkname != original_hardlinkname) { archive_entry_copy_hardlink(entry, hardlinkname); } return (0); } /* * Apply --mtime and --clamp-mtime options. */ void edit_mtime(struct bsdtar *bsdtar, struct archive_entry *entry) { if (!bsdtar->has_mtime) return; __LA_TIME_T entry_mtime = archive_entry_mtime(entry); if (!bsdtar->clamp_mtime || entry_mtime > bsdtar->mtime) archive_entry_set_mtime(entry, bsdtar->mtime, 0); } /* * It would be nice to just use printf() for formatting large numbers, * but the compatibility problems are quite a headache. Hence the * following simple utility function. */ const char * tar_i64toa(int64_t n0) { static char buff[24]; uint64_t n = n0 < 0 ? -n0 : n0; char *p = buff + sizeof(buff); *--p = '\0'; do { *--p = '0' + (int)(n % 10); } while (n /= 10); if (n0 < 0) *--p = '-'; return p; } /* * Like strcmp(), but try to be a little more aware of the fact that * we're comparing two paths. Right now, it just handles leading * "./" and trailing '/' specially, so that "a/b/" == "./a/b" * * TODO: Make this better, so that "./a//b/./c/" == "a/b/c" * TODO: After this works, push it down into libarchive. * TODO: Publish the path normalization routines in libarchive so * that bsdtar can normalize paths and use fast strcmp() instead * of this. * * Note: This is currently only used within write.c, so should * not handle \ path separators. */ int pathcmp(const char *a, const char *b) { /* Skip leading './' */ if (a[0] == '.' && a[1] == '/' && a[2] != '\0') a += 2; if (b[0] == '.' && b[1] == '/' && b[2] != '\0') b += 2; /* Find the first difference, or return (0) if none. */ while (*a == *b) { if (*a == '\0') return (0); a++; b++; } /* * If one ends in '/' and the other one doesn't, * they're the same. */ if (a[0] == '/' && a[1] == '\0' && b[0] == '\0') return (0); if (a[0] == '\0' && b[0] == '/' && b[1] == '\0') return (0); /* They're really different, return the correct sign. */ return (*(const unsigned char *)a - *(const unsigned char *)b); } #define PPBUFF_SIZE 1024 const char * passphrase_callback(struct archive *a, void *_client_data) { struct bsdtar *bsdtar = (struct bsdtar *)_client_data; (void)a; /* UNUSED */ if (bsdtar->ppbuff == NULL) { bsdtar->ppbuff = malloc(PPBUFF_SIZE); if (bsdtar->ppbuff == NULL) lafe_errc(1, errno, "Out of memory"); } return lafe_readpassphrase("Enter passphrase:", bsdtar->ppbuff, PPBUFF_SIZE); } void passphrase_free(char *ppbuff) { if (ppbuff != NULL) { memset(ppbuff, 0, PPBUFF_SIZE); free(ppbuff); } } /* * Display information about the current file. * * The format here roughly duplicates the output of 'ls -l'. * This is based on SUSv2, where 'tar tv' is documented as * listing additional information in an "unspecified format," * and 'pax -l' is documented as using the same format as 'ls -l'. */ void list_item_verbose(struct bsdtar *bsdtar, FILE *out, struct archive_entry *entry) { char tmp[100]; size_t w; + size_t sw; const char *p; const char *fmt; time_t tim; static time_t now; struct tm *ltime; #if defined(HAVE_LOCALTIME_R) || defined(HAVE_LOCALTIME_S) struct tm tmbuf; #endif /* * We avoid collecting the entire list in memory at once by * listing things as we see them. However, that also means we can't * just pre-compute the field widths. Instead, we start with guesses * and just widen them as necessary. These numbers are completely * arbitrary. */ if (!bsdtar->u_width) { bsdtar->u_width = 6; bsdtar->gs_width = 13; } if (!now) time(&now); fprintf(out, "%s %u ", archive_entry_strmode(entry), archive_entry_nlink(entry)); /* Use uname if it's present, else uid. */ p = archive_entry_uname(entry); if ((p == NULL) || (*p == '\0')) { snprintf(tmp, sizeof(tmp), "%lu ", (unsigned long)archive_entry_uid(entry)); p = tmp; } w = strlen(p); if (w > bsdtar->u_width) bsdtar->u_width = w; fprintf(out, "%-*s ", (int)bsdtar->u_width, p); /* Use gname if it's present, else gid. */ p = archive_entry_gname(entry); if (p != NULL && p[0] != '\0') { fprintf(out, "%s", p); w = strlen(p); } else { snprintf(tmp, sizeof(tmp), "%lu", (unsigned long)archive_entry_gid(entry)); w = strlen(tmp); fprintf(out, "%s", tmp); } /* * Print device number or file size, right-aligned so as to make * total width of group and devnum/filesize fields be gs_width. * If gs_width is too small, grow it. */ if (archive_entry_filetype(entry) == AE_IFCHR || archive_entry_filetype(entry) == AE_IFBLK) { snprintf(tmp, sizeof(tmp), "%lu,%lu", (unsigned long)archive_entry_rdevmajor(entry), (unsigned long)archive_entry_rdevminor(entry)); } else { strcpy(tmp, tar_i64toa(archive_entry_size(entry))); } if (w + strlen(tmp) >= bsdtar->gs_width) bsdtar->gs_width = w+strlen(tmp)+1; fprintf(out, "%*s", (int)(bsdtar->gs_width - w), tmp); /* Format the time using 'ls -l' conventions. */ tim = archive_entry_mtime(entry); #define HALF_YEAR (time_t)365 * 86400 / 2 #if defined(_WIN32) && !defined(__CYGWIN__) #define DAY_FMT "%d" /* Windows' strftime function does not support %e format. */ #else #define DAY_FMT "%e" /* Day number without leading zeros */ #endif if (tim < now - HALF_YEAR || tim > now + HALF_YEAR) fmt = bsdtar->day_first ? DAY_FMT " %b %Y" : "%b " DAY_FMT " %Y"; else fmt = bsdtar->day_first ? DAY_FMT " %b %H:%M" : "%b " DAY_FMT " %H:%M"; #if defined(HAVE_LOCALTIME_S) ltime = localtime_s(&tmbuf, &tim) ? NULL : &tmbuf; #elif defined(HAVE_LOCALTIME_R) ltime = localtime_r(&tim, &tmbuf); #else ltime = localtime(&tim); #endif if (ltime) - strftime(tmp, sizeof(tmp), fmt, ltime); - else + sw = strftime(tmp, sizeof(tmp), fmt, ltime); + if (!ltime || !sw) sprintf(tmp, "-- -- ----"); fprintf(out, " %s ", tmp); safe_fprintf(out, "%s", archive_entry_pathname(entry)); /* Extra information for links. */ if (archive_entry_hardlink(entry)) /* Hard link */ safe_fprintf(out, " link to %s", archive_entry_hardlink(entry)); else if (archive_entry_symlink(entry)) /* Symbolic link */ safe_fprintf(out, " -> %s", archive_entry_symlink(entry)); } diff --git a/contrib/libarchive/tar/write.c b/contrib/libarchive/tar/write.c index 21984e980ebd..9e6c97b580b7 100644 --- a/contrib/libarchive/tar/write.c +++ b/contrib/libarchive/tar/write.c @@ -1,1047 +1,1072 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2003-2007 Tim Kientzle * Copyright (c) 2012 Michihiro NAKAJIMA * All rights reserved. */ #include "bsdtar_platform.h" #ifdef HAVE_SYS_TYPES_H #include #endif #ifdef HAVE_SYS_STAT_H #include #endif #if HAVE_SYS_XATTR_H #include #elif HAVE_ATTR_XATTR_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_GRP_H #include #endif #ifdef HAVE_IO_H #include #endif #ifdef HAVE_LIBGEN_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_PATHS_H #include #endif #ifdef HAVE_PWD_H #include #endif #ifdef HAVE_STDINT_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #include "bsdtar.h" -#include "err.h" +#include "lafe_err.h" #include "line_reader.h" #ifndef O_BINARY #define O_BINARY 0 #endif struct archive_dir_entry { struct archive_dir_entry *next; time_t mtime_sec; int mtime_nsec; char *name; }; struct archive_dir { struct archive_dir_entry *head, *tail; }; static int append_archive(struct bsdtar *, struct archive *, struct archive *ina); static int append_archive_filename(struct bsdtar *, struct archive *, const char *fname); static void archive_names_from_file(struct bsdtar *bsdtar, struct archive *a); static int copy_file_data_block(struct bsdtar *, struct archive *a, struct archive *, struct archive_entry *); static void excluded_callback(struct archive *, void *, struct archive_entry *); static void report_write(struct bsdtar *, struct archive *, struct archive_entry *, int64_t progress); static void test_for_append(struct bsdtar *); static int metadata_filter(struct archive *, void *, struct archive_entry *); static void write_archive(struct archive *, struct bsdtar *); static void write_entry(struct bsdtar *, struct archive *, struct archive_entry *); static void write_file(struct bsdtar *, struct archive *, struct archive_entry *); static void write_hierarchy(struct bsdtar *, struct archive *, const char *); #if defined(_WIN32) && !defined(__CYGWIN__) /* Not a full lseek() emulation, but enough for our needs here. */ static int seek_file(int fd, int64_t offset, int whence) { LARGE_INTEGER distance; (void)whence; /* UNUSED */ distance.QuadPart = offset; return (SetFilePointerEx((HANDLE)_get_osfhandle(fd), distance, NULL, FILE_BEGIN) ? 1 : -1); } -#define open _open + +static int +_open_wrap_sopen(char const *const path, int const oflag, ...) +{ + va_list ap; + int r, pmode; + + pmode = 0; + if (oflag & _O_CREAT) + { + va_start(ap, oflag); + pmode = va_arg(ap, int); + va_end(ap); + } + + _sopen_s(&r, path, oflag, _SH_DENYNO, pmode & 0600); + if (r < 0) + { + /* _sopen_s populates errno */ + return -1; + } + + return r; +} + +#define open _open_wrap_sopen #define close _close #define read _read #ifdef lseek #undef lseek #endif #define lseek seek_file #endif static void set_writer_options(struct bsdtar *bsdtar, struct archive *a) { const char *writer_options; int r; writer_options = getenv(ENV_WRITER_OPTIONS); if (writer_options != NULL) { size_t module_len = sizeof(IGNORE_WRONG_MODULE_NAME) - 1; size_t opt_len = strlen(writer_options) + 1; char *p; /* Set default write options. */ if ((p = malloc(module_len + opt_len)) == NULL) lafe_errc(1, errno, "Out of memory"); /* Prepend magic code to ignore options for * a format or filters which are not added to * the archive write object. */ memcpy(p, IGNORE_WRONG_MODULE_NAME, module_len); memcpy(p, writer_options, opt_len); r = archive_write_set_options(a, p); free(p); if (r < ARCHIVE_WARN) lafe_errc(1, 0, "%s", archive_error_string(a)); else archive_clear_error(a); } if (ARCHIVE_OK != archive_write_set_options(a, bsdtar->option_options)) lafe_errc(1, 0, "%s", archive_error_string(a)); } static void set_reader_options(struct bsdtar *bsdtar, struct archive *a) { const char *reader_options; int r; (void)bsdtar; /* UNUSED */ reader_options = getenv(ENV_READER_OPTIONS); if (reader_options != NULL) { size_t module_len = sizeof(IGNORE_WRONG_MODULE_NAME) - 1; size_t opt_len = strlen(reader_options) + 1; char *p; /* Set default write options. */ if ((p = malloc(module_len + opt_len)) == NULL) if (p == NULL) lafe_errc(1, errno, "Out of memory"); /* Prepend magic code to ignore options for * a format or filters which are not added to * the archive write object. */ memcpy(p, IGNORE_WRONG_MODULE_NAME, module_len); memcpy(p, reader_options, opt_len); r = archive_read_set_options(a, p); free(p); if (r < ARCHIVE_WARN) lafe_errc(1, 0, "%s", archive_error_string(a)); else archive_clear_error(a); } if (bsdtar->flags & OPTFLAG_IGNORE_ZEROS) if (archive_read_set_options(a, "read_concatenated_archives") != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); } void tar_mode_c(struct bsdtar *bsdtar) { struct archive *a; const void *filter_name; int r; if (*bsdtar->argv == NULL && bsdtar->names_from_file == NULL) lafe_errc(1, 0, "no files or directories specified"); a = archive_write_new(); /* Support any format that the library supports. */ if (cset_get_format(bsdtar->cset) == NULL) { r = archive_write_set_format_pax_restricted(a); cset_set_format(bsdtar->cset, "pax restricted"); } else { r = archive_write_set_format_by_name(a, cset_get_format(bsdtar->cset)); } if (r != ARCHIVE_OK) { fprintf(stderr, "Can't use format %s: %s\n", cset_get_format(bsdtar->cset), archive_error_string(a)); usage(); } archive_write_set_bytes_per_block(a, bsdtar->bytes_per_block); archive_write_set_bytes_in_last_block(a, bsdtar->bytes_in_last_block); r = cset_write_add_filters(bsdtar->cset, a, &filter_name); if (r < ARCHIVE_WARN) { lafe_errc(1, 0, "Unsupported compression option --%s", (const char *)filter_name); } set_writer_options(bsdtar, a); if (bsdtar->passphrase != NULL) r = archive_write_set_passphrase(a, bsdtar->passphrase); else r = archive_write_set_passphrase_callback(a, bsdtar, &passphrase_callback); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); if (ARCHIVE_OK != archive_write_open_filename(a, bsdtar->filename)) lafe_errc(1, 0, "%s", archive_error_string(a)); write_archive(a, bsdtar); } /* * Same as 'c', except we only support tar or empty formats in * uncompressed files on disk. */ void tar_mode_r(struct bsdtar *bsdtar) { int64_t end_offset; int format; struct archive *a; struct archive_entry *entry; int r; /* Sanity-test some arguments and the file. */ test_for_append(bsdtar); format = ARCHIVE_FORMAT_TAR_PAX_RESTRICTED; #if defined(__BORLANDC__) bsdtar->fd = open(bsdtar->filename, O_RDWR | O_CREAT | O_BINARY); #else bsdtar->fd = open(bsdtar->filename, O_RDWR | O_CREAT | O_BINARY, 0666); #endif if (bsdtar->fd < 0) lafe_errc(1, errno, "Cannot open %s", bsdtar->filename); a = archive_read_new(); archive_read_support_filter_all(a); archive_read_support_format_empty(a); archive_read_support_format_tar(a); archive_read_support_format_gnutar(a); set_reader_options(bsdtar, a); r = archive_read_open_fd(a, bsdtar->fd, 10240); if (r != ARCHIVE_OK) lafe_errc(1, archive_errno(a), "Can't read archive %s: %s", bsdtar->filename, archive_error_string(a)); while (0 == archive_read_next_header(a, &entry)) { if (archive_filter_code(a, 0) != ARCHIVE_FILTER_NONE) { archive_read_free(a); close(bsdtar->fd); lafe_errc(1, 0, "Cannot append to compressed archive."); } /* Keep going until we hit end-of-archive */ format = archive_format(a); } end_offset = archive_read_header_position(a); archive_read_free(a); /* Re-open archive for writing */ a = archive_write_new(); /* * Set the format to be used for writing. To allow people to * extend empty files, we need to allow them to specify the format, * which opens the possibility that they will specify a format that * doesn't match the existing format. Hence, the following bit * of arcane ugliness. */ if (cset_get_format(bsdtar->cset) != NULL) { /* If the user requested a format, use that, but ... */ archive_write_set_format_by_name(a, cset_get_format(bsdtar->cset)); /* ... complain if it's not compatible. */ format &= ARCHIVE_FORMAT_BASE_MASK; if (format != (int)(archive_format(a) & ARCHIVE_FORMAT_BASE_MASK) && format != ARCHIVE_FORMAT_EMPTY) { lafe_errc(1, 0, "Format %s is incompatible with the archive %s.", cset_get_format(bsdtar->cset), bsdtar->filename); } } else { /* * Just preserve the current format, with a little care * for formats that libarchive can't write. */ if (format == ARCHIVE_FORMAT_EMPTY) format = ARCHIVE_FORMAT_TAR_PAX_RESTRICTED; archive_write_set_format(a, format); } if (lseek(bsdtar->fd, end_offset, SEEK_SET) < 0) lafe_errc(1, errno, "Could not seek to archive end"); set_writer_options(bsdtar, a); if (ARCHIVE_OK != archive_write_open_fd(a, bsdtar->fd)) lafe_errc(1, 0, "%s", archive_error_string(a)); write_archive(a, bsdtar); /* XXX check return val XXX */ close(bsdtar->fd); bsdtar->fd = -1; } void tar_mode_u(struct bsdtar *bsdtar) { int64_t end_offset; struct archive *a; struct archive_entry *entry; int format; struct archive_dir_entry *p; struct archive_dir archive_dir; bsdtar->archive_dir = &archive_dir; memset(&archive_dir, 0, sizeof(archive_dir)); format = ARCHIVE_FORMAT_TAR_PAX_RESTRICTED; /* Sanity-test some arguments and the file. */ test_for_append(bsdtar); bsdtar->fd = open(bsdtar->filename, O_RDWR | O_BINARY); if (bsdtar->fd < 0) lafe_errc(1, errno, "Cannot open %s", bsdtar->filename); a = archive_read_new(); archive_read_support_filter_all(a); archive_read_support_format_tar(a); archive_read_support_format_gnutar(a); set_reader_options(bsdtar, a); if (archive_read_open_fd(a, bsdtar->fd, bsdtar->bytes_per_block) != ARCHIVE_OK) { lafe_errc(1, 0, "Can't open %s: %s", bsdtar->filename, archive_error_string(a)); } /* Build a list of all entries and their recorded mod times. */ while (0 == archive_read_next_header(a, &entry)) { if (archive_filter_code(a, 0) != ARCHIVE_FILTER_NONE) { archive_read_free(a); close(bsdtar->fd); lafe_errc(1, 0, "Cannot append to compressed archive."); } if (archive_match_exclude_entry(bsdtar->matching, ARCHIVE_MATCH_MTIME | ARCHIVE_MATCH_OLDER | ARCHIVE_MATCH_EQUAL, entry) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(bsdtar->matching)); /* Record the last format determination we see */ format = archive_format(a); /* Keep going until we hit end-of-archive */ } end_offset = archive_read_header_position(a); archive_read_free(a); /* Re-open archive for writing. */ a = archive_write_new(); /* * Set format to same one auto-detected above. */ archive_write_set_format(a, format); archive_write_set_bytes_per_block(a, bsdtar->bytes_per_block); archive_write_set_bytes_in_last_block(a, bsdtar->bytes_in_last_block); if (lseek(bsdtar->fd, end_offset, SEEK_SET) < 0) lafe_errc(1, errno, "Could not seek to archive end"); set_writer_options(bsdtar, a); if (ARCHIVE_OK != archive_write_open_fd(a, bsdtar->fd)) lafe_errc(1, 0, "%s", archive_error_string(a)); write_archive(a, bsdtar); close(bsdtar->fd); bsdtar->fd = -1; while (bsdtar->archive_dir->head != NULL) { p = bsdtar->archive_dir->head->next; free(bsdtar->archive_dir->head->name); free(bsdtar->archive_dir->head); bsdtar->archive_dir->head = p; } bsdtar->archive_dir->tail = NULL; } /* * Write user-specified files/dirs to opened archive. */ static void write_archive(struct archive *a, struct bsdtar *bsdtar) { const char *arg; struct archive_entry *entry, *sparse_entry; /* Choose a suitable copy buffer size */ bsdtar->buff_size = 64 * 1024; while (bsdtar->buff_size < (size_t)bsdtar->bytes_per_block) bsdtar->buff_size *= 2; /* Try to compensate for space we'll lose to alignment. */ bsdtar->buff_size += 16 * 1024; /* Allocate a buffer for file data. */ if ((bsdtar->buff = malloc(bsdtar->buff_size)) == NULL) lafe_errc(1, 0, "cannot allocate memory"); if ((bsdtar->resolver = archive_entry_linkresolver_new()) == NULL) lafe_errc(1, 0, "cannot create link resolver"); archive_entry_linkresolver_set_strategy(bsdtar->resolver, archive_format(a)); /* Create a read_disk object. */ if ((bsdtar->diskreader = archive_read_disk_new()) == NULL) lafe_errc(1, 0, "Cannot create read_disk object"); /* Tell the read_disk how handle symlink. */ switch (bsdtar->symlink_mode) { case 'H': archive_read_disk_set_symlink_hybrid(bsdtar->diskreader); break; case 'L': archive_read_disk_set_symlink_logical(bsdtar->diskreader); break; default: archive_read_disk_set_symlink_physical(bsdtar->diskreader); break; } /* Register entry filters. */ archive_read_disk_set_matching(bsdtar->diskreader, bsdtar->matching, excluded_callback, bsdtar); archive_read_disk_set_metadata_filter_callback( bsdtar->diskreader, metadata_filter, bsdtar); /* Set the behavior of archive_read_disk. */ archive_read_disk_set_behavior(bsdtar->diskreader, bsdtar->readdisk_flags); archive_read_disk_set_standard_lookup(bsdtar->diskreader); if (bsdtar->names_from_file != NULL) archive_names_from_file(bsdtar, a); while (*bsdtar->argv) { arg = *bsdtar->argv; if (arg[0] == '-' && arg[1] == 'C') { arg += 2; if (*arg == '\0') { bsdtar->argv++; arg = *bsdtar->argv; if (arg == NULL) { lafe_warnc(0, "%s", "Missing argument for -C"); bsdtar->return_value = 1; goto cleanup; } if (*arg == '\0') { lafe_warnc(0, "Meaningless argument for -C: ''"); bsdtar->return_value = 1; goto cleanup; } } set_chdir(bsdtar, arg); } else { if (*arg != '/') do_chdir(bsdtar); /* Handle a deferred -C */ if (*arg == '@') { if (append_archive_filename(bsdtar, a, arg + 1) != 0) break; } else write_hierarchy(bsdtar, a, arg); } bsdtar->argv++; } archive_read_disk_set_matching(bsdtar->diskreader, NULL, NULL, NULL); archive_read_disk_set_metadata_filter_callback( bsdtar->diskreader, NULL, NULL); entry = NULL; archive_entry_linkify(bsdtar->resolver, &entry, &sparse_entry); while (entry != NULL) { int r; struct archive_entry *entry2; struct archive *disk = bsdtar->diskreader; /* * This tricky code here is to correctly read the contents * of the entry because the disk reader bsdtar->diskreader * is pointing at does not have any information about the * entry by this time and using archive_read_data_block() * with the disk reader consequently must fail. And we * have to re-open the entry to read the contents. */ /* TODO: Work with -C option as well. */ r = archive_read_disk_open(disk, archive_entry_sourcepath(entry)); if (r != ARCHIVE_OK) { lafe_warnc(archive_errno(disk), "%s", archive_error_string(disk)); bsdtar->return_value = 1; goto next_entry; } /* * Invoke archive_read_next_header2() to work * archive_read_data_block(), which is called via write_file(), * without failure. */ entry2 = archive_entry_new(); r = archive_read_next_header2(disk, entry2); archive_entry_free(entry2); if (r != ARCHIVE_OK) { lafe_warnc(archive_errno(disk), "%s", archive_error_string(disk)); if (r == ARCHIVE_FATAL) bsdtar->return_value = 1; archive_read_close(disk); goto next_entry; } write_file(bsdtar, a, entry); archive_read_close(disk); next_entry: archive_entry_free(entry); entry = NULL; archive_entry_linkify(bsdtar->resolver, &entry, &sparse_entry); } if (archive_write_close(a)) { lafe_warnc(0, "%s", archive_error_string(a)); bsdtar->return_value = 1; } cleanup: /* Free file data buffer. */ free(bsdtar->buff); archive_entry_linkresolver_free(bsdtar->resolver); bsdtar->resolver = NULL; archive_read_free(bsdtar->diskreader); bsdtar->diskreader = NULL; if (bsdtar->flags & OPTFLAG_TOTALS) { fprintf(stderr, "Total bytes written: %s\n", tar_i64toa(archive_filter_bytes(a, -1))); } archive_write_free(a); } /* * Archive names specified in file. * * Unless --null was specified, a line containing exactly "-C" will * cause the next line to be a directory to pass to chdir(). If * --null is specified, then a line "-C" is just another filename. */ static void archive_names_from_file(struct bsdtar *bsdtar, struct archive *a) { struct lafe_line_reader *lr; const char *line; bsdtar->next_line_is_dir = 0; lr = lafe_line_reader(bsdtar->names_from_file, (bsdtar->flags & OPTFLAG_NULL)); while ((line = lafe_line_reader_next(lr)) != NULL) { if (bsdtar->next_line_is_dir) { if (*line != '\0') set_chdir(bsdtar, line); else { lafe_warnc(0, "Meaningless argument for -C: ''"); bsdtar->return_value = 1; } bsdtar->next_line_is_dir = 0; } else if (((bsdtar->flags & OPTFLAG_NULL) == 0) && strcmp(line, "-C") == 0) bsdtar->next_line_is_dir = 1; else { if (*line != '/') do_chdir(bsdtar); /* Handle a deferred -C */ write_hierarchy(bsdtar, a, line); } } lafe_line_reader_free(lr); if (bsdtar->next_line_is_dir) lafe_errc(1, errno, "Unexpected end of filename list; " "directory expected after -C"); } /* * Copy from specified archive to current archive. Returns non-zero * for write errors (which force us to terminate the entire archiving * operation). If there are errors reading the input archive, we set * bsdtar->return_value but return zero, so the overall archiving * operation will complete and return non-zero. */ static int append_archive_filename(struct bsdtar *bsdtar, struct archive *a, const char *raw_filename) { struct archive *ina; const char *filename = raw_filename; int rc; if (strcmp(filename, "-") == 0) filename = NULL; /* Library uses NULL for stdio. */ ina = archive_read_new(); archive_read_support_format_all(ina); archive_read_support_filter_all(ina); set_reader_options(bsdtar, ina); archive_read_set_options(ina, "mtree:checkfs"); if (bsdtar->passphrase != NULL) rc = archive_read_add_passphrase(a, bsdtar->passphrase); else rc = archive_read_set_passphrase_callback(ina, bsdtar, &passphrase_callback); if (rc != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); if (archive_read_open_filename(ina, filename, bsdtar->bytes_per_block)) { lafe_warnc(0, "%s", archive_error_string(ina)); bsdtar->return_value = 1; return (0); } rc = append_archive(bsdtar, a, ina); if (rc != ARCHIVE_OK) { lafe_warnc(0, "Error reading archive %s: %s", raw_filename, archive_error_string(ina)); bsdtar->return_value = 1; } archive_read_free(ina); return (rc); } static int append_archive(struct bsdtar *bsdtar, struct archive *a, struct archive *ina) { struct archive_entry *in_entry; int e; while (ARCHIVE_OK == (e = archive_read_next_header(ina, &in_entry))) { if (archive_match_excluded(bsdtar->matching, in_entry)) continue; if(edit_pathname(bsdtar, in_entry)) continue; if ((bsdtar->flags & OPTFLAG_INTERACTIVE) && !yes("copy '%s'", archive_entry_pathname(in_entry))) continue; edit_mtime(bsdtar, in_entry); if (bsdtar->verbose > 1) { safe_fprintf(stderr, "a "); list_item_verbose(bsdtar, stderr, in_entry); } else if (bsdtar->verbose > 0) safe_fprintf(stderr, "a %s", archive_entry_pathname(in_entry)); if (need_report()) report_write(bsdtar, a, in_entry, 0); e = archive_write_header(a, in_entry); if (e != ARCHIVE_OK) { if (!bsdtar->verbose) lafe_warnc(0, "%s: %s", archive_entry_pathname(in_entry), archive_error_string(a)); else fprintf(stderr, ": %s", archive_error_string(a)); } if (e == ARCHIVE_FATAL) exit(1); if (e >= ARCHIVE_WARN) { if (archive_entry_size(in_entry) == 0) archive_read_data_skip(ina); else if (copy_file_data_block(bsdtar, a, ina, in_entry)) exit(1); } if (bsdtar->verbose) fprintf(stderr, "\n"); } return (e == ARCHIVE_EOF ? ARCHIVE_OK : e); } /* Helper function to copy file to archive. */ static int copy_file_data_block(struct bsdtar *bsdtar, struct archive *a, struct archive *in_a, struct archive_entry *entry) { size_t bytes_read; ssize_t bytes_written; int64_t offset, progress = 0; char *null_buff = NULL; const void *buff; int r; while ((r = archive_read_data_block(in_a, &buff, &bytes_read, &offset)) == ARCHIVE_OK) { if (need_report()) report_write(bsdtar, a, entry, progress); if (offset > progress) { int64_t sparse = offset - progress; size_t ns; if (null_buff == NULL) { null_buff = bsdtar->buff; memset(null_buff, 0, bsdtar->buff_size); } while (sparse > 0) { if (sparse > (int64_t)bsdtar->buff_size) ns = bsdtar->buff_size; else ns = (size_t)sparse; bytes_written = archive_write_data(a, null_buff, ns); if (bytes_written < 0) { /* Write failed; this is bad */ lafe_warnc(0, "%s", archive_error_string(a)); return (-1); } if ((size_t)bytes_written < ns) { /* Write was truncated; warn but * continue. */ lafe_warnc(0, "%s: Truncated write; file may " "have grown while being archived.", archive_entry_pathname(entry)); return (0); } progress += bytes_written; sparse -= bytes_written; } } bytes_written = archive_write_data(a, buff, bytes_read); if (bytes_written < 0) { /* Write failed; this is bad */ lafe_warnc(0, "%s", archive_error_string(a)); return (-1); } if ((size_t)bytes_written < bytes_read) { /* Write was truncated; warn but continue. */ lafe_warnc(0, "%s: Truncated write; file may have grown " "while being archived.", archive_entry_pathname(entry)); return (0); } progress += bytes_written; } if (r < ARCHIVE_WARN) { lafe_warnc(archive_errno(a), "%s", archive_error_string(a)); return (-1); } return (0); } static void excluded_callback(struct archive *a, void *_data, struct archive_entry *entry) { struct bsdtar *bsdtar = (struct bsdtar *)_data; if (bsdtar->flags & OPTFLAG_NO_SUBDIRS) return; if (!archive_read_disk_can_descend(a)) return; if ((bsdtar->flags & OPTFLAG_INTERACTIVE) && !yes("add '%s'", archive_entry_pathname(entry))) return; archive_read_disk_descend(a); } static int metadata_filter(struct archive *a, void *_data, struct archive_entry *entry) { struct bsdtar *bsdtar = (struct bsdtar *)_data; /* XXX TODO: check whether this filesystem is * synthetic and/or local. Add a new * --local-only option to skip non-local * filesystems. Skip synthetic filesystems * regardless. * * The results should be cached, since * tree.c doesn't usually visit a directory * and the directory contents together. A simple * move-to-front list should perform quite well. * * Use archive_read_disk_current_filesystem_is_remote(). */ /* * If the user vetoes this file/directory, skip it. * We want this to be fairly late; if some other * check would veto this file, we shouldn't bother * the user with it. */ if ((bsdtar->flags & OPTFLAG_INTERACTIVE) && !yes("add '%s'", archive_entry_pathname(entry))) return (0); /* Note: if user vetoes, we won't descend. */ if (((bsdtar->flags & OPTFLAG_NO_SUBDIRS) == 0) && archive_read_disk_can_descend(a)) archive_read_disk_descend(a); return (1); } /* * Add the file or dir hierarchy named by 'path' to the archive */ static void write_hierarchy(struct bsdtar *bsdtar, struct archive *a, const char *path) { struct archive *disk = bsdtar->diskreader; struct archive_entry *entry = NULL, *spare_entry = NULL; int r; r = archive_read_disk_open(disk, path); if (r != ARCHIVE_OK) { lafe_warnc(archive_errno(disk), "%s", archive_error_string(disk)); bsdtar->return_value = 1; return; } bsdtar->first_fs = -1; for (;;) { archive_entry_free(entry); entry = archive_entry_new(); r = archive_read_next_header2(disk, entry); if (r == ARCHIVE_EOF) break; else if (r != ARCHIVE_OK) { lafe_warnc(archive_errno(disk), "%s", archive_error_string(disk)); if (r == ARCHIVE_FATAL || r == ARCHIVE_FAILED) { bsdtar->return_value = 1; archive_entry_free(entry); archive_read_close(disk); return; } else if (r < ARCHIVE_WARN) continue; } if (bsdtar->uid >= 0) { archive_entry_set_uid(entry, bsdtar->uid); if (!bsdtar->uname) archive_entry_set_uname(entry, archive_read_disk_uname(bsdtar->diskreader, bsdtar->uid)); } if (bsdtar->gid >= 0) { archive_entry_set_gid(entry, bsdtar->gid); if (!bsdtar->gname) archive_entry_set_gname(entry, archive_read_disk_gname(bsdtar->diskreader, bsdtar->gid)); } if (bsdtar->uname) archive_entry_set_uname(entry, bsdtar->uname); if (bsdtar->gname) archive_entry_set_gname(entry, bsdtar->gname); /* * Rewrite the pathname to be archived. If rewrite * fails, skip the entry. */ if (edit_pathname(bsdtar, entry)) continue; /* Rewrite the mtime. */ edit_mtime(bsdtar, entry); /* Display entry as we process it. */ if (bsdtar->verbose > 1) { safe_fprintf(stderr, "a "); list_item_verbose(bsdtar, stderr, entry); } else if (bsdtar->verbose > 0) { /* This format is required by SUSv2. */ safe_fprintf(stderr, "a %s", archive_entry_pathname(entry)); } /* Non-regular files get archived with zero size. */ if (archive_entry_filetype(entry) != AE_IFREG) archive_entry_set_size(entry, 0); archive_entry_linkify(bsdtar->resolver, &entry, &spare_entry); while (entry != NULL) { write_file(bsdtar, a, entry); if (entry != spare_entry) { archive_entry_free(entry); } entry = spare_entry; spare_entry = NULL; } if (bsdtar->verbose) fprintf(stderr, "\n"); } archive_entry_free(entry); archive_read_close(disk); } /* * Write a single file (or directory or other filesystem object) to * the archive. */ static void write_file(struct bsdtar *bsdtar, struct archive *a, struct archive_entry *entry) { write_entry(bsdtar, a, entry); } /* * Write a single entry to the archive. */ static void write_entry(struct bsdtar *bsdtar, struct archive *a, struct archive_entry *entry) { int e; e = archive_write_header(a, entry); if (e != ARCHIVE_OK) { if (bsdtar->verbose > 1) { safe_fprintf(stderr, "a "); list_item_verbose(bsdtar, stderr, entry); lafe_warnc(0, ": %s", archive_error_string(a)); } else { lafe_warnc(0, "%s: %s", archive_entry_pathname(entry), archive_error_string(a)); } } if (e == ARCHIVE_FATAL) exit(1); /* * If we opened a file earlier, write it out now. Note that * the format handler might have reset the size field to zero * to inform us that the archive body won't get stored. In * that case, just skip the write. */ if (e >= ARCHIVE_WARN && archive_entry_size(entry) > 0) { if (copy_file_data_block(bsdtar, a, bsdtar->diskreader, entry)) exit(1); } } static void report_write(struct bsdtar *bsdtar, struct archive *a, struct archive_entry *entry, int64_t progress) { uint64_t comp, uncomp; int compression; if (bsdtar->verbose) fprintf(stderr, "\n"); comp = archive_filter_bytes(a, -1); uncomp = archive_filter_bytes(a, 0); fprintf(stderr, "In: %d files, %s bytes;", archive_file_count(a), tar_i64toa(uncomp)); if (comp >= uncomp) compression = 0; else compression = (int)((uncomp - comp) * 100 / uncomp); fprintf(stderr, " Out: %s bytes, compression %d%%\n", tar_i64toa(comp), compression); /* Can't have two calls to tar_i64toa() pending, so split the output. */ safe_fprintf(stderr, "Current: %s (%s", archive_entry_pathname(entry), tar_i64toa(progress)); fprintf(stderr, "/%s bytes)\n", tar_i64toa(archive_entry_size(entry))); } static void test_for_append(struct bsdtar *bsdtar) { struct stat s; if (*bsdtar->argv == NULL && bsdtar->names_from_file == NULL) lafe_errc(1, 0, "no files or directories specified"); if (bsdtar->filename == NULL) lafe_errc(1, 0, "Cannot append to stdout."); if (stat(bsdtar->filename, &s) != 0) return; if (!S_ISREG(s.st_mode) && !S_ISBLK(s.st_mode)) lafe_errc(1, 0, "Cannot append to %s: not a regular file.", bsdtar->filename); /* Is this an appropriate check here on Windows? */ /* if (GetFileType(handle) != FILE_TYPE_DISK) lafe_errc(1, 0, "Cannot append"); */ } diff --git a/contrib/libarchive/test_utils/test_common.h b/contrib/libarchive/test_utils/test_common.h index 064f0a6ec241..a9df300e3be8 100644 --- a/contrib/libarchive/test_utils/test_common.h +++ b/contrib/libarchive/test_utils/test_common.h @@ -1,470 +1,476 @@ /* * Copyright (c) 2003-2017 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #ifndef TEST_COMMON_H #define TEST_COMMON_H /* * The goal of this file (and the matching test.c) is to * simplify the very repetitive test-*.c test programs. */ #if defined(HAVE_CONFIG_H) /* Most POSIX platforms use the 'configure' script to build config.h */ #include "config.h" #elif defined(__FreeBSD__) /* Building as part of FreeBSD system requires a pre-built config.h. */ #include "config_freebsd.h" #elif defined(__NetBSD__) /* Building as part of NetBSD system requires a pre-built config.h. */ #include "config_netbsd.h" #elif defined(_WIN32) && !defined(__CYGWIN__) /* Win32 can't run the 'configure' script. */ #include "config_windows.h" #else /* Warn if the library hasn't been (automatically or manually) configured. */ #error Oops: No config.h and no pre-built configuration in test.h. #endif #include /* Windows requires this before sys/stat.h */ #if !HAVE_SUSECONDS_T #define suseconds_t long #endif #include #if HAVE_DIRENT_H #include #endif #ifdef HAVE_DIRECT_H #include #define dirent direct #endif #include #include #ifdef HAVE_IO_H #include #endif #ifdef HAVE_STDINT_H #include #endif #include #include #include #include #include #ifdef HAVE_UNISTD_H #include #endif #include #ifdef HAVE_ACL_LIBACL_H #include #endif #ifdef HAVE_SYS_ACL_H #include #endif #ifdef HAVE_SYS_RICHACL_H #include #endif #ifdef HAVE_WINDOWS_H #define NOCRYPT #include #include #endif /* * System-specific tweaks. We really want to minimize these * as much as possible, since they make it harder to understand * the mainline code. */ /* Windows (including Visual Studio and MinGW but not Cygwin) */ #if defined(_WIN32) && !defined(__CYGWIN__) #if !defined(__BORLANDC__) #undef chdir #define chdir _chdir #define strdup _strdup #endif #endif /* Visual Studio */ #if defined(_MSC_VER) && _MSC_VER < 1900 #define snprintf sprintf_s #endif #if defined(__BORLANDC__) #pragma warn -8068 /* Constant out of range in comparison. */ #endif #if defined(__GNUC__) && (__GNUC__ > 2 || \ (__GNUC__ == 2 && __GNUC_MINOR__ >= 7)) # ifdef __MINGW_PRINTF_FORMAT # define __LA_PRINTF_FORMAT __MINGW_PRINTF_FORMAT # else # define __LA_PRINTF_FORMAT __printf__ # endif # define __LA_PRINTFLIKE(f,a) __attribute__((__format__(__LA_PRINTF_FORMAT, f, a))) #else # define __LA_PRINTFLIKE(f,a) #endif /* Haiku OS and QNX */ #if defined(__HAIKU__) || defined(__QNXNTO__) /* Haiku and QNX have typedefs in stdint.h (needed for int64_t) */ #include #endif #ifndef O_BINARY #define O_BINARY 0 #endif #ifndef __LIBARCHIVE_TEST_COMMON #define __LIBARCHIVE_TEST_COMMON #endif #include "archive_platform_acl.h" #define ARCHIVE_TEST_ACL_TYPE_POSIX1E 1 #define ARCHIVE_TEST_ACL_TYPE_NFS4 2 #include "archive_platform_xattr.h" /* * Redefine DEFINE_TEST for use in defining the test functions. */ #undef DEFINE_TEST #define DEFINE_TEST(name) void name(void); void name(void) /* An implementation of the standard assert() macro */ #define assert(e) assertion_assert(__FILE__, __LINE__, (e), #e, NULL) /* chdir() and error if it fails */ #define assertChdir(path) \ assertion_chdir(__FILE__, __LINE__, path) /* change file/directory permissions and errors if it fails */ #define assertChmod(pathname, mode) \ assertion_chmod(__FILE__, __LINE__, pathname, mode) /* Assert two files have the same file flags */ #define assertEqualFflags(patha, pathb) \ assertion_compare_fflags(__FILE__, __LINE__, patha, pathb, 0) /* Assert two integers are the same. Reports value of each one if not. */ #define assertEqualInt(v1,v2) \ assertion_equal_int(__FILE__, __LINE__, (v1), #v1, (v2), #v2, NULL) #define assertEqualAddress(v1,v2) \ assertion_equal_address(__FILE__, __LINE__, (v1), #v1, (v2), #v2, NULL) /* Assert two strings are the same. Reports value of each one if not. */ #define assertEqualString(v1,v2) \ assertion_equal_string(__FILE__, __LINE__, (v1), #v1, (v2), #v2, NULL, 0) #define assertEqualUTF8String(v1,v2) \ assertion_equal_string(__FILE__, __LINE__, (v1), #v1, (v2), #v2, NULL, 1) /* As above, but v1 and v2 are wchar_t * */ #define assertEqualWString(v1,v2) \ assertion_equal_wstring(__FILE__, __LINE__, (v1), #v1, (v2), #v2, NULL) /* As above, but raw blocks of bytes. */ #define assertEqualMem(v1, v2, l) \ assertion_equal_mem(__FILE__, __LINE__, (v1), #v1, (v2), #v2, (l), #l, NULL) /* Assert that memory is full of a specified byte */ #define assertMemoryFilledWith(v1, l, b) \ assertion_memory_filled_with(__FILE__, __LINE__, (v1), #v1, (l), #l, (b), #b, NULL) /* Assert two files are the same. */ #define assertEqualFile(f1, f2) \ assertion_equal_file(__FILE__, __LINE__, (f1), (f2)) /* Assert that a file is empty. */ #define assertEmptyFile(pathname) \ assertion_empty_file(__FILE__, __LINE__, (pathname)) /* Assert that a file is not empty. */ #define assertNonEmptyFile(pathname) \ assertion_non_empty_file(__FILE__, __LINE__, (pathname)) #define assertFileAtime(pathname, sec, nsec) \ assertion_file_atime(__FILE__, __LINE__, pathname, sec, nsec) #define assertFileAtimeRecent(pathname) \ assertion_file_atime_recent(__FILE__, __LINE__, pathname) #define assertFileBirthtime(pathname, sec, nsec) \ assertion_file_birthtime(__FILE__, __LINE__, pathname, sec, nsec) #define assertFileBirthtimeRecent(pathname) \ assertion_file_birthtime_recent(__FILE__, __LINE__, pathname) /* Assert that a file exists; supports printf-style arguments. */ #define assertFileExists(pathname) \ assertion_file_exists(__FILE__, __LINE__, pathname) /* Assert that a file exists. */ #define assertFileNotExists(pathname) \ assertion_file_not_exists(__FILE__, __LINE__, pathname) /* Assert that file contents match a string. */ #define assertFileContents(data, data_size, pathname) \ assertion_file_contents(__FILE__, __LINE__, data, data_size, pathname) /* Verify that a file does not contain invalid strings */ #define assertFileContainsNoInvalidStrings(pathname, strings) \ assertion_file_contains_no_invalid_strings(__FILE__, __LINE__, pathname, strings) #define assertFileMtime(pathname, sec, nsec) \ assertion_file_mtime(__FILE__, __LINE__, pathname, sec, nsec) #define assertFileMtimeRecent(pathname) \ assertion_file_mtime_recent(__FILE__, __LINE__, pathname) #define assertFileNLinks(pathname, nlinks) \ assertion_file_nlinks(__FILE__, __LINE__, pathname, nlinks) #define assertFileSize(pathname, size) \ assertion_file_size(__FILE__, __LINE__, pathname, size) #define assertFileMode(pathname, mode) \ assertion_file_mode(__FILE__, __LINE__, pathname, mode) #define assertTextFileContents(text, pathname) \ assertion_text_file_contents(__FILE__, __LINE__, text, pathname) #define assertFileContainsLinesAnyOrder(pathname, lines) \ assertion_file_contains_lines_any_order(__FILE__, __LINE__, pathname, lines) #define assertIsDir(pathname, mode) \ assertion_is_dir(__FILE__, __LINE__, pathname, mode) #define assertIsHardlink(path1, path2) \ assertion_is_hardlink(__FILE__, __LINE__, path1, path2) #define assertIsNotHardlink(path1, path2) \ assertion_is_not_hardlink(__FILE__, __LINE__, path1, path2) #define assertIsReg(pathname, mode) \ assertion_is_reg(__FILE__, __LINE__, pathname, mode) #define assertIsSymlink(pathname, contents, isdir) \ assertion_is_symlink(__FILE__, __LINE__, pathname, contents, isdir) /* Create a directory, report error if it fails. */ #define assertMakeDir(dirname, mode) \ assertion_make_dir(__FILE__, __LINE__, dirname, mode) #define assertMakeFile(path, mode, contents) \ assertion_make_file(__FILE__, __LINE__, path, mode, -1, contents) #define assertMakeBinFile(path, mode, csize, contents) \ assertion_make_file(__FILE__, __LINE__, path, mode, csize, contents) #define assertMakeHardlink(newfile, oldfile) \ assertion_make_hardlink(__FILE__, __LINE__, newfile, oldfile) #define assertMakeSymlink(newfile, linkto, targetIsDir) \ assertion_make_symlink(__FILE__, __LINE__, newfile, linkto, targetIsDir) #define assertSetNodump(path) \ assertion_set_nodump(__FILE__, __LINE__, path) #define assertUmask(mask) \ assertion_umask(__FILE__, __LINE__, mask) /* Assert that two files have unequal file flags */ #define assertUnequalFflags(patha, pathb) \ assertion_compare_fflags(__FILE__, __LINE__, patha, pathb, 1) #define assertUtimes(pathname, atime, atime_nsec, mtime, mtime_nsec) \ assertion_utimes(__FILE__, __LINE__, pathname, atime, atime_nsec, mtime, mtime_nsec) #ifndef PROGRAM #define assertEntrySetAcls(entry, acls, count) \ assertion_entry_set_acls(__FILE__, __LINE__, entry, acls, count) #define assertEntryCompareAcls(entry, acls, count, type, mode) \ assertion_entry_compare_acls(__FILE__, __LINE__, entry, acls, count, type, mode) #endif /* * This would be simple with C99 variadic macros, but I don't want to * require that. Instead, I insert a function call before each * skipping() call to pass the file and line information down. Crude, * but effective. */ #define skipping \ skipping_setup(__FILE__, __LINE__);test_skipping /* Function declarations. These are defined in test_utility.c. */ void failure(const char *fmt, ...) __LA_PRINTFLIKE(1, 2); int assertion_assert(const char *, int, int, const char *, void *); int assertion_chdir(const char *, int, const char *); int assertion_chmod(const char *, int, const char *, int); int assertion_compare_fflags(const char *, int, const char *, const char *, int); int assertion_empty_file(const char *, int, const char *); int assertion_equal_file(const char *, int, const char *, const char *); int assertion_equal_int(const char *, int, long long, const char *, long long, const char *, void *); int assertion_equal_address(const char *, int, const void *, const char *, const void *, const char *, void *); int assertion_equal_mem(const char *, int, const void *, const char *, const void *, const char *, size_t, const char *, void *); int assertion_memory_filled_with(const char *, int, const void *, const char *, size_t, const char *, char, const char *, void *); int assertion_equal_string(const char *, int, const char *v1, const char *, const char *v2, const char *, void *, int); int assertion_equal_wstring(const char *, int, const wchar_t *v1, const char *, const wchar_t *v2, const char *, void *); int assertion_file_atime(const char *, int, const char *, long, long); int assertion_file_atime_recent(const char *, int, const char *); int assertion_file_birthtime(const char *, int, const char *, long, long); int assertion_file_birthtime_recent(const char *, int, const char *); int assertion_file_contains_lines_any_order(const char *, int, const char *, const char **); int assertion_file_contains_no_invalid_strings(const char *, int, const char *, const char **); int assertion_file_contents(const char *, int, const void *, int, const char *); int assertion_file_exists(const char *, int, const char *); int assertion_file_mode(const char *, int, const char *, int); int assertion_file_mtime(const char *, int, const char *, long, long); int assertion_file_mtime_recent(const char *, int, const char *); int assertion_file_nlinks(const char *, int, const char *, int); int assertion_file_not_exists(const char *, int, const char *); int assertion_file_size(const char *, int, const char *, long); int assertion_is_dir(const char *, int, const char *, int); int assertion_is_hardlink(const char *, int, const char *, const char *); int assertion_is_not_hardlink(const char *, int, const char *, const char *); int assertion_is_reg(const char *, int, const char *, int); int assertion_is_symlink(const char *, int, const char *, const char *, int); int assertion_make_dir(const char *, int, const char *, int); int assertion_make_file(const char *, int, const char *, int, int, const void *); int assertion_make_hardlink(const char *, int, const char *newpath, const char *); int assertion_make_symlink(const char *, int, const char *newpath, const char *, int); int assertion_non_empty_file(const char *, int, const char *); int assertion_set_nodump(const char *, int, const char *); int assertion_text_file_contents(const char *, int, const char *buff, const char *f); int assertion_umask(const char *, int, int); int assertion_utimes(const char *, int, const char *, time_t, suseconds_t, time_t, suseconds_t); int assertion_version(const char*, int, const char *, const char *); void skipping_setup(const char *, int); void test_skipping(const char *fmt, ...) __LA_PRINTFLIKE(1, 2); /* Like sprintf, then system() */ int systemf(const char *fmt, ...) __LA_PRINTFLIKE(1, 2); /* Delay until time() returns a value after this. */ void sleepUntilAfter(time_t); /* Return true if this platform can create symlinks. */ int canSymlink(void); /* Return true if this platform can run the "bzip2" program. */ int canBzip2(void); /* Return true if this platform can run the "grzip" program. */ int canGrzip(void); /* Return true if this platform can run the "gzip" program. */ int canGzip(void); -/* Return true if this platform can run the specified command. */ -int canRunCommand(const char *); +/* Return true if this platform can run the specified command. + * + * Result can be optionally cached with `*tested`: + * - 0 if not tested yet + * - <0 if already tested negative + * - >0 if already tested positive + */ +int canRunCommand(const char *cmd, int *tested); /* Return true if this platform can run the "lrzip" program. */ int canLrzip(void); /* Return true if this platform can run the "lz4" program. */ int canLz4(void); /* Return true if this platform can run the "zstd" program. */ int canZstd(void); /* Return true if this platform can run the "lzip" program. */ int canLzip(void); /* Return true if this platform can run the "lzma" program. */ int canLzma(void); /* Return true if this platform can run the "lzop" program. */ int canLzop(void); /* Return true if this platform can run the "xz" program. */ int canXz(void); /* Return true if this filesystem can handle nodump flags. */ int canNodump(void); /* Set test ACLs */ int setTestAcl(const char *path); /* Get extended attribute */ void *getXattr(const char *, const char *, size_t *); /* Set extended attribute */ int setXattr(const char *, const char *, const void *, size_t); /* Return true if the file has large i-node number(>0xffffffff). */ int is_LargeInode(const char *); #if ARCHIVE_ACL_SUNOS /* Fetch ACLs on Solaris using acl() or facl() */ void *sunacl_get(int cmd, int *aclcnt, int fd, const char *path); #endif /* Suck file into string allocated via malloc(). Call free() when done. */ /* Supports printf-style args: slurpfile(NULL, "%s/myfile", refdir); */ char *slurpfile(size_t *, const char *fmt, ...) __LA_PRINTFLIKE(2, 3); /* Dump block of bytes to a file. */ void dumpfile(const char *filename, void *, size_t); /* Extracts named reference file to the current directory. */ void extract_reference_file(const char *); /* Copies named reference file to the current directory. */ void copy_reference_file(const char *); /* Extracts a list of files to the current directory. * List must be NULL terminated. */ void extract_reference_files(const char **); /* Subtract umask from mode */ mode_t umasked(mode_t expected_mode); /* Path to working directory for current test */ extern const char *testworkdir; #ifndef PROGRAM /* * Special interfaces for libarchive test harness. */ #include "archive.h" #include "archive_entry.h" /* ACL structure */ struct archive_test_acl_t { int type; /* Type of ACL */ int permset; /* Permissions for this class of users. */ int tag; /* Owner, User, Owning group, group, other, etc. */ int qual; /* GID or UID of user/group, depending on tag. */ const char *name; /* Name of user/group, depending on tag. */ }; /* Set ACLs */ int assertion_entry_set_acls(const char *, int, struct archive_entry *, struct archive_test_acl_t *, int); /* Compare ACLs */ int assertion_entry_compare_acls(const char *, int, struct archive_entry *, struct archive_test_acl_t *, int, int, int); /* Special customized read-from-memory interface. */ int read_open_memory(struct archive *, const void *, size_t, size_t); /* _minimal version exercises a slightly different set of libarchive APIs. */ int read_open_memory_minimal(struct archive *, const void *, size_t, size_t); /* _seek version produces a seekable file. */ int read_open_memory_seek(struct archive *, const void *, size_t, size_t); /* Versions of above that accept an archive argument for additional info. */ #define assertA(e) assertion_assert(__FILE__, __LINE__, (e), #e, (a)) #define assertEqualIntA(a,v1,v2) \ assertion_equal_int(__FILE__, __LINE__, (v1), #v1, (v2), #v2, (a)) #define assertEqualStringA(a,v1,v2) \ assertion_equal_string(__FILE__, __LINE__, (v1), #v1, (v2), #v2, (a), 0) #else /* defined(PROGRAM) */ /* * Special interfaces for program test harness. */ /* Pathname of exe to be tested. */ extern const char *testprogfile; /* Name of exe to use in printf-formatted command strings. */ /* On Windows, this includes leading/trailing quotes. */ extern const char *testprog; void assertVersion(const char *prog, const char *base); #endif /* defined(PROGRAM) */ #ifdef USE_DMALLOC #include #endif #include "test_utils.h" #endif /* !TEST_COMMON_H */ diff --git a/contrib/libarchive/test_utils/test_main.c b/contrib/libarchive/test_utils/test_main.c index fe330e5a052e..f4d443060d88 100644 --- a/contrib/libarchive/test_utils/test_main.c +++ b/contrib/libarchive/test_utils/test_main.c @@ -1,4381 +1,4297 @@ /* * Copyright (c) 2003-2009 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" #include "test_utils.h" #ifdef HAVE_SYS_IOCTL_H #include #endif #ifdef HAVE_SYS_TIME_H #include #endif #include #ifdef HAVE_ICONV_H #include #endif /* * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. * As the include guards don't agree, the order of include is important. */ #ifdef HAVE_LINUX_EXT2_FS_H #include /* for Linux file flags */ #endif #if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) #include /* Linux file flags, broken on Cygwin */ #endif #ifdef HAVE_LINUX_FS_H #include #endif #include #include #ifdef HAVE_SIGNAL_H #include #endif #include #include #ifdef HAVE_SIGNAL_H #endif #ifdef HAVE_ACL_LIBACL_H #include #endif #ifdef HAVE_SYS_TYPES_H #include #endif #ifdef HAVE_SYS_ACL_H #include #endif #ifdef HAVE_SYS_EA_H #include #endif #ifdef HAVE_SYS_EXTATTR_H #include #endif #if HAVE_SYS_XATTR_H #include #elif HAVE_ATTR_XATTR_H #include #endif #ifdef HAVE_SYS_RICHACL_H #include #endif #if HAVE_MEMBERSHIP_H #include #endif #ifndef nitems #define nitems(arr) (sizeof(arr) / sizeof((arr)[0])) #endif /* * * Windows support routines * * Note: Configuration is a tricky issue. Using HAVE_* feature macros * in the test harness is dangerous because they cover up * configuration errors. The classic example of this is omitting a * configure check. If libarchive and libarchive_test both look for * the same feature macro, such errors are hard to detect. Platform * macros (e.g., _WIN32 or __GNUC__) are a little better, but can * easily lead to very messy code. It's best to limit yourself * to only the most generic programming techniques in the test harness * and thus avoid conditionals altogether. Where that's not possible, * try to minimize conditionals by grouping platform-specific tests in * one place (e.g., test_acl_freebsd) or by adding new assert() * functions (e.g., assertMakeHardlink()) to cover up platform * differences. Platform-specific coding in libarchive_test is often * a symptom that some capability is missing from libarchive itself. */ #if defined(_WIN32) && !defined(__CYGWIN__) #include #include #include #ifndef F_OK #define F_OK (0) #endif #ifndef S_ISDIR #define S_ISDIR(m) ((m) & _S_IFDIR) #endif #ifndef S_ISREG #define S_ISREG(m) ((m) & _S_IFREG) #endif #if !defined(__BORLANDC__) #define access _access #undef chdir #define chdir _chdir #undef chmod #define chmod _chmod #endif #ifndef fileno #define fileno _fileno #endif /*#define fstat _fstat64*/ #if !defined(__BORLANDC__) #define getcwd _getcwd #endif #define lstat stat /*#define lstat _stat64*/ /*#define stat _stat64*/ #define rmdir _rmdir #if !defined(__BORLANDC__) #define strdup _strdup #define umask _umask #endif #define int64_t __int64 #endif #if defined(HAVE__CrtSetReportMode) # include #endif mode_t umasked(mode_t expected_mode) { mode_t mode = umask(0); umask(mode); return expected_mode & ~mode; } /* Path to working directory for current test */ const char *testworkdir; #ifdef PROGRAM /* Pathname of exe to be tested. */ const char *testprogfile; /* Name of exe to use in printf-formatted command strings. */ /* On Windows, this includes leading/trailing quotes. */ const char *testprog; #endif #if defined(_WIN32) && !defined(__CYGWIN__) static void *GetFunctionKernel32(const char *); static int my_CreateSymbolicLinkA(const char *, const char *, int); static int my_CreateHardLinkA(const char *, const char *); static int my_GetFileInformationByName(const char *, BY_HANDLE_FILE_INFORMATION *); typedef struct _REPARSE_DATA_BUFFER { ULONG ReparseTag; USHORT ReparseDataLength; USHORT Reserved; union { struct { USHORT SubstituteNameOffset; USHORT SubstituteNameLength; USHORT PrintNameOffset; USHORT PrintNameLength; ULONG Flags; WCHAR PathBuffer[1]; } SymbolicLinkReparseBuffer; struct { USHORT SubstituteNameOffset; USHORT SubstituteNameLength; USHORT PrintNameOffset; USHORT PrintNameLength; WCHAR PathBuffer[1]; } MountPointReparseBuffer; struct { UCHAR DataBuffer[1]; } GenericReparseBuffer; } DUMMYUNIONNAME; } REPARSE_DATA_BUFFER, *PREPARSE_DATA_BUFFER; static void * GetFunctionKernel32(const char *name) { static HINSTANCE lib; static int set; if (!set) { set = 1; lib = LoadLibrary("kernel32.dll"); } if (lib == NULL) { fprintf(stderr, "Can't load kernel32.dll?!\n"); exit(1); } return (void *)GetProcAddress(lib, name); } static int my_CreateSymbolicLinkA(const char *linkname, const char *target, int targetIsDir) { static BOOLEAN (WINAPI *f)(LPCSTR, LPCSTR, DWORD); DWORD attrs; static int set; int ret, tmpflags; size_t llen, tlen; int flags = 0; char *src, *tgt, *p; if (!set) { set = 1; f = GetFunctionKernel32("CreateSymbolicLinkA"); } if (f == NULL) return (0); tlen = strlen(target); llen = strlen(linkname); if (tlen == 0 || llen == 0) return (0); tgt = malloc(tlen + 1); if (tgt == NULL) return (0); src = malloc(llen + 1); if (src == NULL) { free(tgt); return (0); } /* * Translate slashes to backslashes */ p = src; while(*linkname != '\0') { if (*linkname == '/') *p = '\\'; else *p = *linkname; linkname++; p++; } *p = '\0'; p = tgt; while(*target != '\0') { if (*target == '/') *p = '\\'; else *p = *target; target++; p++; } *p = '\0'; /* * Each test has to specify if a file or a directory symlink * should be created. */ if (targetIsDir) { #if defined(SYMBOLIC_LINK_FLAG_DIRECTORY) flags |= SYMBOLIC_LINK_FLAG_DIRECTORY; #else flags |= 0x1; #endif } #if defined(SYMBOLIC_LINK_FLAG_ALLOW_UNPRIVILEGED_CREATE) tmpflags = flags | SYMBOLIC_LINK_FLAG_ALLOW_UNPRIVILEGED_CREATE; #else tmpflags = flags | 0x2; #endif /* * Windows won't overwrite existing links */ attrs = GetFileAttributesA(linkname); if (attrs != INVALID_FILE_ATTRIBUTES) { if (attrs & FILE_ATTRIBUTE_DIRECTORY) RemoveDirectoryA(linkname); else DeleteFileA(linkname); } ret = (*f)(src, tgt, tmpflags); /* * Prior to Windows 10 the SYMBOLIC_LINK_FLAG_ALLOW_UNPRIVILEGED_CREATE * is not understood */ if (!ret) ret = (*f)(src, tgt, flags); free(src); free(tgt); return (ret); } static int my_CreateHardLinkA(const char *linkname, const char *target) { static BOOLEAN (WINAPI *f)(LPCSTR, LPCSTR, LPSECURITY_ATTRIBUTES); static int set; if (!set) { set = 1; f = GetFunctionKernel32("CreateHardLinkA"); } return f == NULL ? 0 : (*f)(linkname, target, NULL); } static int my_GetFileInformationByName(const char *path, BY_HANDLE_FILE_INFORMATION *bhfi) { HANDLE h; int r; memset(bhfi, 0, sizeof(*bhfi)); h = CreateFileA(path, FILE_READ_ATTRIBUTES, 0, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); if (h == INVALID_HANDLE_VALUE) return (0); r = GetFileInformationByHandle(h, bhfi); CloseHandle(h); return (r); } #endif #if defined(HAVE__CrtSetReportMode) && !defined(__WATCOMC__) static void invalid_parameter_handler(const wchar_t * expression, const wchar_t * function, const wchar_t * file, unsigned int line, uintptr_t pReserved) { /* nop */ // Silence unused-parameter compiler warnings. (void)expression; (void)function; (void)file; (void)line; (void)pReserved; } #endif /* * * OPTIONS FLAGS * */ /* Enable core dump on failure. */ static int dump_on_failure = 0; /* Default is to remove temp dirs and log data for successful tests. */ static int keep_temp_files = 0; /* Default is to only return a failure code (1) if there were test failures. If enabled, exit with code 2 if there were no failures, but some tests were skipped. */ static int fail_if_tests_skipped = 0; /* Default is to run the specified tests once and report errors. */ static int until_failure = 0; /* Default is to just report pass/fail for each test. */ static int verbosity = 0; #define VERBOSITY_SUMMARY_ONLY -1 /* -q */ #define VERBOSITY_PASSFAIL 0 /* Default */ #define VERBOSITY_LIGHT_REPORT 1 /* -v */ #define VERBOSITY_FULL 2 /* -vv */ /* A few places generate even more output for verbosity > VERBOSITY_FULL, * mostly for debugging the test harness itself. */ /* Cumulative count of assertion failures. */ static int failures = 0; /* Cumulative count of reported skips. */ static int skips = 0; /* Cumulative count of assertions checked. */ static int assertions = 0; /* Directory where uuencoded reference files can be found. */ static const char *refdir; /* * Report log information selectively to console and/or disk log. */ static int log_console = 0; static FILE *logfile; static void __LA_PRINTFLIKE(1, 0) vlogprintf(const char *fmt, va_list ap) { #ifdef va_copy va_list lfap; va_copy(lfap, ap); #endif if (log_console) vfprintf(stdout, fmt, ap); if (logfile != NULL) #ifdef va_copy vfprintf(logfile, fmt, lfap); va_end(lfap); #else vfprintf(logfile, fmt, ap); #endif } static void __LA_PRINTFLIKE(1, 2) logprintf(const char *fmt, ...) { va_list ap; va_start(ap, fmt); vlogprintf(fmt, ap); va_end(ap); } /* Set up a message to display only if next assertion fails. */ static char msgbuff[4096]; static const char *msg, *nextmsg; void failure(const char *fmt, ...) { va_list ap; if (fmt == NULL) { nextmsg = NULL; } else { va_start(ap, fmt); vsnprintf(msgbuff, sizeof(msgbuff), fmt, ap); va_end(ap); nextmsg = msgbuff; } } /* * Copy arguments into file-local variables. * This was added to permit vararg assert() functions without needing * variadic wrapper macros. Turns out that the vararg capability is almost * never used, so almost all of the vararg assertions can be simplified * by removing the vararg capability and reworking the wrapper macro to * pass __FILE__, __LINE__ directly into the function instead of using * this hook. I suspect this machinery is used so rarely that we * would be better off just removing it entirely. That would simplify * the code here noticeably. */ static const char *skipping_filename; static int skipping_line; void skipping_setup(const char *filename, int line) { skipping_filename = filename; skipping_line = line; } /* Called at the beginning of each assert() function. */ static void assertion_count(const char *file, int line) { (void)file; /* UNUSED */ (void)line; /* UNUSED */ ++assertions; /* Proper handling of "failure()" message. */ msg = nextmsg; nextmsg = NULL; /* Uncomment to print file:line after every assertion. * Verbose, but occasionally useful in tracking down crashes. */ /* printf("Checked %s:%d\n", file, line); */ } /* * For each test source file, we remember how many times each * assertion was reported. Cleared before each new test, * used by test_summarize(). */ static struct line { int count; int skip; } failed_lines[10000]; static const char *failed_filename; /* Count this failure, setup up log destination and handle initial report. */ static void __LA_PRINTFLIKE(3, 4) failure_start(const char *filename, int line, const char *fmt, ...) { va_list ap; /* Record another failure for this line. */ ++failures; failed_filename = filename; failed_lines[line].count++; /* Determine whether to log header to console. */ switch (verbosity) { case VERBOSITY_LIGHT_REPORT: log_console = (failed_lines[line].count < 2); break; default: log_console = (verbosity >= VERBOSITY_FULL); } /* Log file:line header for this failure */ va_start(ap, fmt); #if _MSC_VER logprintf("%s(%d): ", filename, line); #else logprintf("%s:%d: ", filename, line); #endif vlogprintf(fmt, ap); va_end(ap); logprintf("\n"); if (msg != NULL && msg[0] != '\0') { logprintf(" Description: %s\n", msg); msg = NULL; } /* Determine whether to log details to console. */ if (verbosity == VERBOSITY_LIGHT_REPORT) log_console = 0; } /* Complete reporting of failed tests. */ /* * The 'extra' hook here is used by libarchive to include libarchive * error messages with assertion failures. It could also be used * to add strerror() output, for example. Just define the EXTRA_DUMP() * macro appropriately. */ static void failure_finish(void *extra) { (void)extra; /* UNUSED (maybe) */ #ifdef EXTRA_DUMP if (extra != NULL) { logprintf(" errno: %d\n", EXTRA_ERRNO(extra)); logprintf(" detail: %s\n", EXTRA_DUMP(extra)); } #endif if (dump_on_failure) { fprintf(stderr, " *** forcing core dump so failure can be debugged ***\n"); abort(); } } /* Inform user that we're skipping some checks. */ void test_skipping(const char *fmt, ...) { char buff[1024]; va_list ap; va_start(ap, fmt); vsnprintf(buff, sizeof(buff), fmt, ap); va_end(ap); /* Use failure() message if set. */ msg = nextmsg; nextmsg = NULL; /* failure_start() isn't quite right, but is awfully convenient. */ failure_start(skipping_filename, skipping_line, "SKIPPING: %s", buff); --failures; /* Undo failures++ in failure_start() */ /* Don't failure_finish() here. */ /* Mark as skip, so doesn't count as failed test. */ failed_lines[skipping_line].skip = 1; ++skips; } /* * * ASSERTIONS * */ /* Generic assert() just displays the failed condition. */ int assertion_assert(const char *file, int line, int value, const char *condition, void *extra) { assertion_count(file, line); if (!value) { failure_start(file, line, "Assertion failed: %s", condition); failure_finish(extra); } return (value); } /* chdir() and report any errors */ int assertion_chdir(const char *file, int line, const char *pathname) { assertion_count(file, line); if (chdir(pathname) == 0) return (1); failure_start(file, line, "chdir(\"%s\")", pathname); failure_finish(NULL); return (0); } /* change file/directory permissions and errors if it fails */ int assertion_chmod(const char *file, int line, const char *pathname, int mode) { assertion_count(file, line); if (chmod(pathname, (mode_t)mode) == 0) return (1); failure_start(file, line, "chmod(\"%s\", %4.o)", pathname, (unsigned int)mode); failure_finish(NULL); return (0); } /* Verify two integers are equal. */ int assertion_equal_int(const char *file, int line, long long v1, const char *e1, long long v2, const char *e2, void *extra) { assertion_count(file, line); if (v1 == v2) return (1); failure_start(file, line, "%s != %s", e1, e2); logprintf(" %s=%lld (0x%llx, 0%llo)\n", e1, v1, (unsigned long long)v1, (unsigned long long)v1); logprintf(" %s=%lld (0x%llx, 0%llo)\n", e2, v2, (unsigned long long)v2, (unsigned long long)v2); failure_finish(extra); return (0); } /* Verify two pointers are equal. */ int assertion_equal_address(const char *file, int line, const void *v1, const char *e1, const void *v2, const char *e2, void *extra) { assertion_count(file, line); if (v1 == v2) return (1); failure_start(file, line, "%s != %s", e1, e2); logprintf(" %s=0x%llx\n", e1, (unsigned long long)(uintptr_t)v1); logprintf(" %s=0x%llx\n", e2, (unsigned long long)(uintptr_t)v2); failure_finish(extra); return (0); } /* * Utility to convert a single UTF-8 sequence. */ static int _utf8_to_unicode(uint32_t *pwc, const char *s, size_t n) { static const char utf8_count[256] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 00 - 0F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 10 - 1F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 20 - 2F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 30 - 3F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 40 - 4F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 50 - 5F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 60 - 6F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 70 - 7F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 80 - 8F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 90 - 9F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* A0 - AF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* B0 - BF */ 0, 0, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* C0 - CF */ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* D0 - DF */ 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,/* E0 - EF */ 4, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 /* F0 - FF */ }; int ch; int cnt; uint32_t wc; *pwc = 0; /* Sanity check. */ if (n == 0) return (0); /* * Decode 1-4 bytes depending on the value of the first byte. */ ch = (unsigned char)*s; if (ch == 0) return (0); /* Standard: return 0 for end-of-string. */ cnt = utf8_count[ch]; /* Invalid sequence or there are not plenty bytes. */ if (n < (size_t)cnt) return (-1); /* Make a Unicode code point from a single UTF-8 sequence. */ switch (cnt) { case 1: /* 1 byte sequence. */ *pwc = ch & 0x7f; return (cnt); case 2: /* 2 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) return (-1); *pwc = ((ch & 0x1f) << 6) | (s[1] & 0x3f); return (cnt); case 3: /* 3 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) return (-1); if ((s[2] & 0xc0) != 0x80) return (-1); wc = ((ch & 0x0f) << 12) | ((s[1] & 0x3f) << 6) | (s[2] & 0x3f); if (wc < 0x800) return (-1);/* Overlong sequence. */ break; case 4: /* 4 bytes sequence. */ if (n < 4) return (-1); if ((s[1] & 0xc0) != 0x80) return (-1); if ((s[2] & 0xc0) != 0x80) return (-1); if ((s[3] & 0xc0) != 0x80) return (-1); wc = ((ch & 0x07) << 18) | ((s[1] & 0x3f) << 12) | ((s[2] & 0x3f) << 6) | (s[3] & 0x3f); if (wc < 0x10000) return (-1);/* Overlong sequence. */ break; default: return (-1); } /* The code point larger than 0x10FFFF is not legal * Unicode values. */ if (wc > 0x10FFFF) return (-1); /* Correctly gets a Unicode, returns used bytes. */ *pwc = wc; return (cnt); } static void strdump(const char *e, const char *p, int ewidth, int utf8) { const char *q = p; logprintf(" %*s = ", ewidth, e); if (p == NULL) { logprintf("NULL\n"); return; } logprintf("\""); while (*p != '\0') { unsigned int c = 0xff & *p++; switch (c) { case '\a': logprintf("\\a"); break; case '\b': logprintf("\\b"); break; case '\n': logprintf("\\n"); break; case '\r': logprintf("\\r"); break; default: if (c >= 32 && c < 127) logprintf("%c", (int)c); else logprintf("\\x%02X", c); } } logprintf("\""); logprintf(" (length %d)", q == NULL ? -1 : (int)strlen(q)); /* * If the current string is UTF-8, dump its code points. */ if (utf8) { size_t len; uint32_t uc; int n; int cnt = 0; p = q; len = strlen(p); logprintf(" ["); while ((n = _utf8_to_unicode(&uc, p, len)) > 0) { if (p != q) logprintf(" "); logprintf("%04X", uc); p += n; len -= n; cnt++; } logprintf("]"); logprintf(" (count %d", cnt); if (n < 0) { logprintf(",unknown %zu bytes", len); } logprintf(")"); } logprintf("\n"); } /* Verify two strings are equal, dump them if not. */ int assertion_equal_string(const char *file, int line, const char *v1, const char *e1, const char *v2, const char *e2, void *extra, int utf8) { int l1, l2; assertion_count(file, line); if (v1 == v2 || (v1 != NULL && v2 != NULL && strcmp(v1, v2) == 0)) return (1); failure_start(file, line, "%s != %s", e1, e2); l1 = (int)strlen(e1); l2 = (int)strlen(e2); if (l1 < l2) l1 = l2; strdump(e1, v1, l1, utf8); strdump(e2, v2, l1, utf8); failure_finish(extra); return (0); } static void wcsdump(const char *e, const wchar_t *w) { logprintf(" %s = ", e); if (w == NULL) { logprintf("(null)"); return; } logprintf("\""); while (*w != L'\0') { unsigned int c = *w++; if (c >= 32 && c < 127) logprintf("%c", (int)c); else if (c < 256) logprintf("\\x%02X", c); else if (c < 0x10000) logprintf("\\u%04X", c); else logprintf("\\U%08X", c); } logprintf("\"\n"); } #ifndef HAVE_WCSCMP static int wcscmp(const wchar_t *s1, const wchar_t *s2) { while (*s1 == *s2++) { if (*s1++ == L'\0') return 0; } if (*s1 > *--s2) return 1; else return -1; } #endif /* Verify that two wide strings are equal, dump them if not. */ int assertion_equal_wstring(const char *file, int line, const wchar_t *v1, const char *e1, const wchar_t *v2, const char *e2, void *extra) { assertion_count(file, line); if (v1 == v2) return (1); if (v1 != NULL && v2 != NULL && wcscmp(v1, v2) == 0) return (1); failure_start(file, line, "%s != %s", e1, e2); wcsdump(e1, v1); wcsdump(e2, v2); failure_finish(extra); return (0); } /* * Pretty standard hexdump routine. As a bonus, if ref != NULL, then * any bytes in p that differ from ref will be highlighted with '_' * before and after the hex value. */ static void hexdump(const char *p, const char *ref, size_t l, size_t offset) { size_t i, j; char sep; if (p == NULL) { logprintf("(null)\n"); return; } for(i=0; i < l; i+=16) { logprintf("%04x", (unsigned)(i + offset)); sep = ' '; for (j = 0; j < 16 && i + j < l; j++) { if (ref != NULL && p[i + j] != ref[i + j]) sep = '_'; logprintf("%c%02x", sep, 0xff & (unsigned int)p[i+j]); if (ref != NULL && p[i + j] == ref[i + j]) sep = ' '; } for (; j < 16; j++) { logprintf("%c ", sep); sep = ' '; } logprintf("%c", sep); for (j=0; j < 16 && i + j < l; j++) { int c = p[i + j]; if (c >= ' ' && c <= 126) logprintf("%c", c); else logprintf("."); } logprintf("\n"); } } /* Verify that two blocks of memory are the same, display the first * block of differences if they're not. */ int assertion_equal_mem(const char *file, int line, const void *_v1, const char *e1, const void *_v2, const char *e2, size_t l, const char *ld, void *extra) { const char *v1 = (const char *)_v1; const char *v2 = (const char *)_v2; size_t offset; assertion_count(file, line); if (v1 == v2 || (v1 != NULL && v2 != NULL && memcmp(v1, v2, l) == 0)) return (1); if (v1 == NULL || v2 == NULL) return (0); failure_start(file, line, "%s != %s", e1, e2); logprintf(" size %s = %d\n", ld, (int)l); /* Dump 48 bytes (3 lines) so that the first difference is * in the second line. */ offset = 0; while (l > 64 && memcmp(v1, v2, 32) == 0) { /* Two lines agree, so step forward one line. */ v1 += 16; v2 += 16; l -= 16; offset += 16; } logprintf(" Dump of %s\n", e1); hexdump(v1, v2, l < 128 ? l : 128, offset); logprintf(" Dump of %s\n", e2); hexdump(v2, v1, l < 128 ? l : 128, offset); logprintf("\n"); failure_finish(extra); return (0); } /* Verify that a block of memory is filled with the specified byte. */ int assertion_memory_filled_with(const char *file, int line, const void *_v1, const char *vd, size_t l, const char *ld, char b, const char *bd, void *extra) { const char *v1 = (const char *)_v1; size_t c = 0; size_t i; (void)ld; /* UNUSED */ assertion_count(file, line); for (i = 0; i < l; ++i) { if (v1[i] == b) { ++c; } } if (c == l) return (1); failure_start(file, line, "%s (size %d) not filled with %s", vd, (int)l, bd); logprintf(" Only %d bytes were correct\n", (int)c); failure_finish(extra); return (0); } /* Verify that the named file exists and is empty. */ int assertion_empty_file(const char *filename, int line, const char *f1) { char buff[1024]; struct stat st; ssize_t s; FILE *f; assertion_count(filename, line); if (stat(f1, &st) != 0) { failure_start(filename, line, "Stat failed: %s", f1); failure_finish(NULL); return (0); } if (st.st_size == 0) return (1); failure_start(filename, line, "File should be empty: %s", f1); logprintf(" File size: %d\n", (int)st.st_size); logprintf(" Contents:\n"); f = fopen(f1, "rb"); if (f == NULL) { logprintf(" Unable to open %s\n", f1); } else { s = ((off_t)sizeof(buff) < st.st_size) ? (ssize_t)sizeof(buff) : (ssize_t)st.st_size; s = fread(buff, 1, s, f); hexdump(buff, NULL, s, 0); fclose(f); } failure_finish(NULL); return (0); } /* Verify that the named file exists and is not empty. */ int assertion_non_empty_file(const char *filename, int line, const char *f1) { struct stat st; assertion_count(filename, line); if (stat(f1, &st) != 0) { failure_start(filename, line, "Stat failed: %s", f1); failure_finish(NULL); return (0); } if (st.st_size == 0) { failure_start(filename, line, "File empty: %s", f1); failure_finish(NULL); return (0); } return (1); } /* Verify that two files have the same contents. */ /* TODO: hexdump the first bytes that actually differ. */ int assertion_equal_file(const char *filename, int line, const char *fn1, const char *fn2) { char buff1[1024]; char buff2[1024]; FILE *f1, *f2; int n1, n2; assertion_count(filename, line); f1 = fopen(fn1, "rb"); f2 = fopen(fn2, "rb"); if (f1 == NULL || f2 == NULL) { if (f1) fclose(f1); if (f2) fclose(f2); return (0); } for (;;) { n1 = (int)fread(buff1, 1, sizeof(buff1), f1); n2 = (int)fread(buff2, 1, sizeof(buff2), f2); if (n1 != n2) break; if (n1 == 0 && n2 == 0) { fclose(f1); fclose(f2); return (1); } if (memcmp(buff1, buff2, n1) != 0) break; } fclose(f1); fclose(f2); failure_start(filename, line, "Files not identical"); logprintf(" file1=\"%s\"\n", fn1); logprintf(" file2=\"%s\"\n", fn2); failure_finish(NULL); return (0); } /* Verify that the named file does exist. */ int assertion_file_exists(const char *filename, int line, const char *f) { assertion_count(filename, line); #if defined(_WIN32) && !defined(__CYGWIN__) if (!_access(f, 0)) return (1); #else if (!access(f, F_OK)) return (1); #endif failure_start(filename, line, "File should exist: %s", f); failure_finish(NULL); return (0); } /* Verify that the named file doesn't exist. */ int assertion_file_not_exists(const char *filename, int line, const char *f) { assertion_count(filename, line); #if defined(_WIN32) && !defined(__CYGWIN__) if (_access(f, 0)) return (1); #else if (access(f, F_OK)) return (1); #endif failure_start(filename, line, "File should not exist: %s", f); failure_finish(NULL); return (0); } /* Compare the contents of a file to a block of memory. */ int assertion_file_contents(const char *filename, int line, const void *buff, int s, const char *fn) { char *contents; FILE *f; int n; assertion_count(filename, line); f = fopen(fn, "rb"); if (f == NULL) { failure_start(filename, line, "File should exist: %s", fn); failure_finish(NULL); return (0); } contents = malloc(s * 2); n = (int)fread(contents, 1, s * 2, f); fclose(f); if (n == s && memcmp(buff, contents, s) == 0) { free(contents); return (1); } failure_start(filename, line, "File contents don't match"); logprintf(" file=\"%s\"\n", fn); if (n > 0) hexdump(contents, buff, n > 512 ? 512 : n, 0); else { logprintf(" File empty, contents should be:\n"); hexdump(buff, NULL, s > 512 ? 512 : s, 0); } failure_finish(NULL); free(contents); return (0); } /* Check the contents of a text file, being tolerant of line endings. */ int assertion_text_file_contents(const char *filename, int line, const char *buff, const char *fn) { char *contents; const char *btxt, *ftxt; FILE *f; int n, s; assertion_count(filename, line); f = fopen(fn, "r"); if (f == NULL) { failure_start(filename, line, "File doesn't exist: %s", fn); failure_finish(NULL); return (0); } s = (int)strlen(buff); contents = malloc(s * 2 + 128); n = (int)fread(contents, 1, s * 2 + 128 - 1, f); if (n >= 0) contents[n] = '\0'; fclose(f); /* Compare texts. */ btxt = buff; ftxt = (const char *)contents; while (*btxt != '\0' && *ftxt != '\0') { if (*btxt == *ftxt) { ++btxt; ++ftxt; continue; } if (btxt[0] == '\n' && ftxt[0] == '\r' && ftxt[1] == '\n') { /* Pass over different new line characters. */ ++btxt; ftxt += 2; continue; } break; } if (*btxt == '\0' && *ftxt == '\0') { free(contents); return (1); } failure_start(filename, line, "Contents don't match"); logprintf(" file=\"%s\"\n", fn); if (n > 0) { hexdump(contents, buff, n, 0); logprintf(" expected\n"); hexdump(buff, contents, s, 0); } else { logprintf(" File empty, contents should be:\n"); hexdump(buff, NULL, s, 0); } failure_finish(NULL); free(contents); return (0); } /* Verify that a text file contains the specified lines, regardless of order */ /* This could be more efficient if we sorted both sets of lines, etc, but * since this is used only for testing and only ever deals with a dozen or so * lines at a time, this relatively crude approach is just fine. */ int assertion_file_contains_lines_any_order(const char *file, int line, const char *pathname, const char *lines[]) { char *buff; size_t buff_size; size_t expected_count, actual_count, i, j; char **expected = NULL; char *p, **actual = NULL; char c; int expected_failure = 0, actual_failure = 0; assertion_count(file, line); buff = slurpfile(&buff_size, "%s", pathname); if (buff == NULL) { failure_start(pathname, line, "Can't read file: %s", pathname); failure_finish(NULL); return (0); } /* Make a copy of the provided lines and count up the expected * file size. */ for (i = 0; lines[i] != NULL; ++i) { } expected_count = i; if (expected_count) { expected = calloc(expected_count, sizeof(*expected)); if (expected == NULL) { failure_start(pathname, line, "Can't allocate memory"); failure_finish(NULL); goto cleanup; } for (i = 0; lines[i] != NULL; ++i) { expected[i] = strdup(lines[i]); if (expected[i] == NULL) { failure_start(pathname, line, "Can't allocate memory"); failure_finish(NULL); goto cleanup; } } } /* Break the file into lines */ actual_count = 0; for (c = '\0', p = buff; p < buff + buff_size; ++p) { if (*p == '\x0d' || *p == '\x0a') *p = '\0'; if (c == '\0' && *p != '\0') ++actual_count; c = *p; } if (actual_count) { actual = calloc(actual_count, sizeof(char *)); if (actual == NULL) { failure_start(pathname, line, "Can't allocate memory"); failure_finish(NULL); goto cleanup; } for (j = 0, p = buff; p < buff + buff_size; p += 1 + strlen(p)) { if (*p != '\0') { actual[j] = p; ++j; } } } /* Erase matching lines from both lists */ for (i = 0; i < expected_count; ++i) { for (j = 0; j < actual_count; ++j) { if (actual[j] == NULL) continue; if (strcmp(expected[i], actual[j]) == 0) { free(expected[i]); expected[i] = NULL; actual[j] = NULL; break; } } } /* If there's anything left, it's a failure */ for (i = 0; i < expected_count; ++i) { if (expected[i] != NULL) ++expected_failure; } for (j = 0; j < actual_count; ++j) { if (actual[j] != NULL) ++actual_failure; } if (expected_failure == 0 && actual_failure == 0) { free(actual); free(expected); free(buff); return (1); } failure_start(file, line, "File doesn't match: %s", pathname); for (i = 0; i < expected_count; ++i) { if (expected[i] != NULL) { logprintf(" Expected but not present: %s\n", expected[i]); free(expected[i]); expected[i] = NULL; } } for (j = 0; j < actual_count; ++j) { if (actual[j] != NULL) logprintf(" Present but not expected: %s\n", actual[j]); } failure_finish(NULL); cleanup: free(actual); if (expected != NULL) { for (i = 0; i < expected_count; ++i) if (expected[i] != NULL) free(expected[i]); free(expected); } free(buff); return (0); } /* Verify that a text file does not contains the specified strings */ int assertion_file_contains_no_invalid_strings(const char *file, int line, const char *pathname, const char *strings[]) { char *buff; int i; buff = slurpfile(NULL, "%s", pathname); if (buff == NULL) { failure_start(file, line, "Can't read file: %s", pathname); failure_finish(NULL); return (0); } for (i = 0; strings[i] != NULL; ++i) { if (strstr(buff, strings[i]) != NULL) { failure_start(file, line, "Invalid string in %s: %s", pathname, strings[i]); failure_finish(NULL); free(buff); return(0); } } free(buff); return (0); } /* Test that two paths point to the same file. */ /* As a side-effect, asserts that both files exist. */ static int is_hardlink(const char *file, int line, const char *path1, const char *path2) { #if defined(_WIN32) && !defined(__CYGWIN__) BY_HANDLE_FILE_INFORMATION bhfi1, bhfi2; int r; assertion_count(file, line); r = my_GetFileInformationByName(path1, &bhfi1); if (r == 0) { failure_start(file, line, "File %s can't be inspected?", path1); failure_finish(NULL); return (0); } r = my_GetFileInformationByName(path2, &bhfi2); if (r == 0) { failure_start(file, line, "File %s can't be inspected?", path2); failure_finish(NULL); return (0); } return (bhfi1.dwVolumeSerialNumber == bhfi2.dwVolumeSerialNumber && bhfi1.nFileIndexHigh == bhfi2.nFileIndexHigh && bhfi1.nFileIndexLow == bhfi2.nFileIndexLow); #else struct stat st1, st2; int r; assertion_count(file, line); r = lstat(path1, &st1); if (r != 0) { failure_start(file, line, "File should exist: %s", path1); failure_finish(NULL); return (0); } r = lstat(path2, &st2); if (r != 0) { failure_start(file, line, "File should exist: %s", path2); failure_finish(NULL); return (0); } return (st1.st_ino == st2.st_ino && st1.st_dev == st2.st_dev); #endif } int assertion_is_hardlink(const char *file, int line, const char *path1, const char *path2) { if (is_hardlink(file, line, path1, path2)) return (1); failure_start(file, line, "Files %s and %s are not hardlinked", path1, path2); failure_finish(NULL); return (0); } int assertion_is_not_hardlink(const char *file, int line, const char *path1, const char *path2) { if (!is_hardlink(file, line, path1, path2)) return (1); failure_start(file, line, "Files %s and %s should not be hardlinked", path1, path2); failure_finish(NULL); return (0); } /* Verify a/b/mtime of 'pathname'. */ /* If 'recent', verify that it's within last 10 seconds. */ static int assertion_file_time(const char *file, int line, const char *pathname, long t, long nsec, char type, int recent) { long long filet, filet_nsec; int r; #if defined(_WIN32) && !defined(__CYGWIN__) #define EPOC_TIME (116444736000000000ULL) FILETIME fxtime, fbirthtime, fatime, fmtime; ULARGE_INTEGER wintm; HANDLE h; fxtime.dwLowDateTime = 0; fxtime.dwHighDateTime = 0; assertion_count(file, line); /* Note: FILE_FLAG_BACKUP_SEMANTICS applies to open * a directory file. If not, CreateFile() will fail when * the pathname is a directory. */ h = CreateFileA(pathname, FILE_READ_ATTRIBUTES, 0, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); if (h == INVALID_HANDLE_VALUE) { failure_start(file, line, "Can't access %s\n", pathname); failure_finish(NULL); return (0); } r = GetFileTime(h, &fbirthtime, &fatime, &fmtime); switch (type) { case 'a': fxtime = fatime; break; case 'b': fxtime = fbirthtime; break; case 'm': fxtime = fmtime; break; } CloseHandle(h); if (r == 0) { failure_start(file, line, "Can't GetFileTime %s\n", pathname); failure_finish(NULL); return (0); } wintm.LowPart = fxtime.dwLowDateTime; wintm.HighPart = fxtime.dwHighDateTime; filet = (wintm.QuadPart - EPOC_TIME) / 10000000; filet_nsec = ((wintm.QuadPart - EPOC_TIME) % 10000000) * 100; nsec = (nsec / 100) * 100; /* Round the request */ #else struct stat st; assertion_count(file, line); r = lstat(pathname, &st); if (r != 0) { failure_start(file, line, "Can't stat %s\n", pathname); failure_finish(NULL); return (0); } switch (type) { case 'a': filet = st.st_atime; break; case 'm': filet = st.st_mtime; break; case 'b': filet = 0; break; default: fprintf(stderr, "INTERNAL: Bad type %c for file time", type); exit(1); } #if defined(__FreeBSD__) switch (type) { case 'a': filet_nsec = st.st_atimespec.tv_nsec; break; case 'b': filet = st.st_birthtime; /* FreeBSD filesystems that don't support birthtime * (e.g., UFS1) always return -1 here. */ if (filet == -1) { return (1); } filet_nsec = st.st_birthtimespec.tv_nsec; break; case 'm': filet_nsec = st.st_mtimespec.tv_nsec; break; default: fprintf(stderr, "INTERNAL: Bad type %c for file time", type); exit(1); } /* FreeBSD generally only stores to microsecond res, so round. */ filet_nsec = (filet_nsec / 1000) * 1000; nsec = (nsec / 1000) * 1000; #else filet_nsec = nsec = 0; /* Generic POSIX only has whole seconds. */ if (type == 'b') return (1); /* Generic POSIX doesn't have birthtime */ #if defined(__HAIKU__) if (type == 'a') return (1); /* Haiku doesn't have atime. */ #endif #endif #endif if (recent) { /* Check that requested time is up-to-date. */ time_t now = time(NULL); if (filet < now - 10 || filet > now + 1) { failure_start(file, line, "File %s has %ctime %lld, %lld seconds ago\n", pathname, type, filet, now - filet); failure_finish(NULL); return (0); } } else if (filet != t || filet_nsec != nsec) { failure_start(file, line, "File %s has %ctime %lld.%09lld, expected %ld.%09ld", pathname, type, filet, filet_nsec, t, nsec); failure_finish(NULL); return (0); } return (1); } /* Verify atime of 'pathname'. */ int assertion_file_atime(const char *file, int line, const char *pathname, long t, long nsec) { return assertion_file_time(file, line, pathname, t, nsec, 'a', 0); } /* Verify atime of 'pathname' is up-to-date. */ int assertion_file_atime_recent(const char *file, int line, const char *pathname) { return assertion_file_time(file, line, pathname, 0, 0, 'a', 1); } /* Verify birthtime of 'pathname'. */ int assertion_file_birthtime(const char *file, int line, const char *pathname, long t, long nsec) { return assertion_file_time(file, line, pathname, t, nsec, 'b', 0); } /* Verify birthtime of 'pathname' is up-to-date. */ int assertion_file_birthtime_recent(const char *file, int line, const char *pathname) { return assertion_file_time(file, line, pathname, 0, 0, 'b', 1); } /* Verify mode of 'pathname'. */ int assertion_file_mode(const char *file, int line, const char *pathname, int expected_mode) { int mode; int r; assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) failure_start(file, line, "assertFileMode not yet implemented for Windows"); (void)mode; /* UNUSED */ (void)r; /* UNUSED */ (void)pathname; /* UNUSED */ (void)expected_mode; /* UNUSED */ #else { struct stat st; r = lstat(pathname, &st); mode = (int)(st.st_mode & 0777); } if (r == 0 && mode == expected_mode) return (1); failure_start(file, line, "File %s has mode %o, expected %o", pathname, (unsigned int)mode, (unsigned int)expected_mode); #endif failure_finish(NULL); return (0); } /* Verify mtime of 'pathname'. */ int assertion_file_mtime(const char *file, int line, const char *pathname, long t, long nsec) { return assertion_file_time(file, line, pathname, t, nsec, 'm', 0); } /* Verify mtime of 'pathname' is up-to-date. */ int assertion_file_mtime_recent(const char *file, int line, const char *pathname) { return assertion_file_time(file, line, pathname, 0, 0, 'm', 1); } /* Verify number of links to 'pathname'. */ int assertion_file_nlinks(const char *file, int line, const char *pathname, int nlinks) { #if defined(_WIN32) && !defined(__CYGWIN__) BY_HANDLE_FILE_INFORMATION bhfi; int r; assertion_count(file, line); r = my_GetFileInformationByName(pathname, &bhfi); if (r != 0 && bhfi.nNumberOfLinks == (DWORD)nlinks) return (1); failure_start(file, line, "File %s has %jd links, expected %d", pathname, (intmax_t)bhfi.nNumberOfLinks, nlinks); failure_finish(NULL); return (0); #else struct stat st; int r; assertion_count(file, line); r = lstat(pathname, &st); if (r == 0 && (int)st.st_nlink == nlinks) return (1); failure_start(file, line, "File %s has %jd links, expected %d", pathname, (intmax_t)st.st_nlink, nlinks); failure_finish(NULL); return (0); #endif } /* Verify size of 'pathname'. */ int assertion_file_size(const char *file, int line, const char *pathname, long size) { int64_t filesize; int r; assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) { BY_HANDLE_FILE_INFORMATION bhfi; r = !my_GetFileInformationByName(pathname, &bhfi); filesize = ((int64_t)bhfi.nFileSizeHigh << 32) + bhfi.nFileSizeLow; } #else { struct stat st; r = lstat(pathname, &st); filesize = st.st_size; } #endif if (r == 0 && filesize == size) return (1); failure_start(file, line, "File %s has size %ld, expected %ld", pathname, (long)filesize, (long)size); failure_finish(NULL); return (0); } /* Assert that 'pathname' is a dir. If mode >= 0, verify that too. */ int assertion_is_dir(const char *file, int line, const char *pathname, int mode) { struct stat st; int r; #if defined(_WIN32) && !defined(__CYGWIN__) (void)mode; /* UNUSED */ #endif assertion_count(file, line); r = lstat(pathname, &st); if (r != 0) { failure_start(file, line, "Dir should exist: %s", pathname); failure_finish(NULL); return (0); } if (!S_ISDIR(st.st_mode)) { failure_start(file, line, "%s is not a dir", pathname); failure_finish(NULL); return (0); } #if !defined(_WIN32) || defined(__CYGWIN__) /* Windows doesn't handle permissions the same way as POSIX, * so just ignore the mode tests. */ /* TODO: Can we do better here? */ if (mode >= 0 && (mode_t)mode != (st.st_mode & 07777)) { failure_start(file, line, "Dir %s has wrong mode", pathname); logprintf(" Expected: 0%3o\n", (unsigned int)mode); logprintf(" Found: 0%3o\n", (unsigned int)st.st_mode & 07777); failure_finish(NULL); return (0); } #endif return (1); } /* Verify that 'pathname' is a regular file. If 'mode' is >= 0, * verify that too. */ int assertion_is_reg(const char *file, int line, const char *pathname, int mode) { struct stat st; int r; #if defined(_WIN32) && !defined(__CYGWIN__) (void)mode; /* UNUSED */ #endif assertion_count(file, line); r = lstat(pathname, &st); if (r != 0 || !S_ISREG(st.st_mode)) { failure_start(file, line, "File should exist: %s", pathname); failure_finish(NULL); return (0); } #if !defined(_WIN32) || defined(__CYGWIN__) /* Windows doesn't handle permissions the same way as POSIX, * so just ignore the mode tests. */ /* TODO: Can we do better here? */ if (mode >= 0 && (mode_t)mode != (st.st_mode & 07777)) { failure_start(file, line, "File %s has wrong mode", pathname); logprintf(" Expected: 0%3o\n", (unsigned int)mode); logprintf(" Found: 0%3o\n", (unsigned int)st.st_mode & 07777); failure_finish(NULL); return (0); } #endif return (1); } /* * Check whether 'pathname' is a symbolic link. If 'contents' is * non-NULL, verify that the symlink has those contents. * * On platforms with directory symlinks, set isdir to 0 to test for a file * symlink and to 1 to test for a directory symlink. On other platforms * the variable is ignored. */ static int is_symlink(const char *file, int line, const char *pathname, const char *contents, int isdir) { #if defined(_WIN32) && !defined(__CYGWIN__) HANDLE h; DWORD inbytes; REPARSE_DATA_BUFFER *buf; BY_HANDLE_FILE_INFORMATION st; size_t len, len2; wchar_t *linknamew, *contentsw; const char *p; char *s, *pn; int ret = 0; BYTE *indata; const DWORD flag = FILE_FLAG_BACKUP_SEMANTICS | FILE_FLAG_OPEN_REPARSE_POINT; /* Replace slashes with backslashes in pathname */ pn = malloc(strlen(pathname) + 1); if (pn == NULL) { failure_start(file, line, "Can't allocate memory"); failure_finish(NULL); return (0); } for (p = pathname, s = pn; *p != '\0'; p++, s++) { if (*p == '/') *s = '\\'; else *s = *p; } *s = '\0'; h = CreateFileA(pn, 0, FILE_SHARE_READ, NULL, OPEN_EXISTING, flag, NULL); free(pn); if (h == INVALID_HANDLE_VALUE) { failure_start(file, line, "Can't access %s\n", pathname); failure_finish(NULL); return (0); } ret = GetFileInformationByHandle(h, &st); if (ret == 0) { failure_start(file, line, "Can't stat: %s", pathname); failure_finish(NULL); } else if ((st.dwFileAttributes & FILE_ATTRIBUTE_REPARSE_POINT) == 0) { failure_start(file, line, "Not a symlink: %s", pathname); failure_finish(NULL); ret = 0; } if (isdir && ((st.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) == 0)) { failure_start(file, line, "Not a directory symlink: %s", pathname); failure_finish(NULL); ret = 0; } if (!isdir && ((st.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) != 0)) { failure_start(file, line, "Not a file symlink: %s", pathname); failure_finish(NULL); ret = 0; } if (ret == 0) { CloseHandle(h); return (0); } indata = malloc(MAXIMUM_REPARSE_DATA_BUFFER_SIZE); ret = DeviceIoControl(h, FSCTL_GET_REPARSE_POINT, NULL, 0, indata, 1024, &inbytes, NULL); CloseHandle(h); if (ret == 0) { free(indata); failure_start(file, line, "Could not retrieve symlink target: %s", pathname); failure_finish(NULL); return (0); } buf = (REPARSE_DATA_BUFFER *) indata; if (buf->ReparseTag != IO_REPARSE_TAG_SYMLINK) { free(indata); /* File is not a symbolic link */ failure_start(file, line, "Not a symlink: %s", pathname); failure_finish(NULL); return (0); } if (contents == NULL) { free(indata); return (1); } len = buf->SymbolicLinkReparseBuffer.SubstituteNameLength; linknamew = malloc(len + sizeof(wchar_t)); if (linknamew == NULL) { free(indata); return (0); } memcpy(linknamew, &((BYTE *)buf->SymbolicLinkReparseBuffer.PathBuffer) [buf->SymbolicLinkReparseBuffer.SubstituteNameOffset], len); free(indata); linknamew[len / sizeof(wchar_t)] = L'\0'; contentsw = malloc(len + sizeof(wchar_t)); if (contentsw == NULL) { free(linknamew); return (0); } len2 = mbsrtowcs(contentsw, &contents, (len + sizeof(wchar_t) / sizeof(wchar_t)), NULL); if (len2 > 0 && wcscmp(linknamew, contentsw) != 0) ret = 1; free(linknamew); free(contentsw); return (ret); #else char buff[300]; struct stat st; ssize_t linklen; int r; (void)isdir; /* UNUSED */ assertion_count(file, line); r = lstat(pathname, &st); if (r != 0) { failure_start(file, line, "Symlink should exist: %s", pathname); failure_finish(NULL); return (0); } if (!S_ISLNK(st.st_mode)) return (0); if (contents == NULL) return (1); linklen = readlink(pathname, buff, sizeof(buff) - 1); if (linklen < 0) { failure_start(file, line, "Can't read symlink %s", pathname); failure_finish(NULL); return (0); } buff[linklen] = '\0'; if (strcmp(buff, contents) != 0) return (0); return (1); #endif } /* Assert that path is a symlink that (optionally) contains contents. */ int assertion_is_symlink(const char *file, int line, const char *path, const char *contents, int isdir) { if (is_symlink(file, line, path, contents, isdir)) return (1); if (contents) failure_start(file, line, "File %s is not a symlink to %s", path, contents); else failure_start(file, line, "File %s is not a symlink", path); failure_finish(NULL); return (0); } /* Create a directory and report any errors. */ int assertion_make_dir(const char *file, int line, const char *dirname, int mode) { assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) (void)mode; /* UNUSED */ if (0 == _mkdir(dirname)) return (1); #else if (0 == mkdir(dirname, (mode_t)mode)) { if (0 == chmod(dirname, (mode_t)mode)) { assertion_file_mode(file, line, dirname, mode); return (1); } } #endif failure_start(file, line, "Could not create directory %s", dirname); failure_finish(NULL); return(0); } /* Create a file with the specified contents and report any failures. */ int assertion_make_file(const char *file, int line, const char *path, int mode, int csize, const void *contents) { #if defined(_WIN32) && !defined(__CYGWIN__) /* TODO: Rework this to set file mode as well. */ FILE *f; (void)mode; /* UNUSED */ assertion_count(file, line); f = fopen(path, "wb"); if (f == NULL) { failure_start(file, line, "Could not create file %s", path); failure_finish(NULL); return (0); } if (contents != NULL) { size_t wsize; if (csize < 0) wsize = strlen(contents); else wsize = (size_t)csize; if (wsize != fwrite(contents, 1, wsize, f)) { fclose(f); failure_start(file, line, "Could not write file %s", path); failure_finish(NULL); return (0); } } fclose(f); return (1); #else int fd; assertion_count(file, line); fd = open(path, O_CREAT | O_WRONLY, mode >= 0 ? mode : 0644); if (fd < 0) { failure_start(file, line, "Could not create %s", path); failure_finish(NULL); return (0); } #ifdef HAVE_FCHMOD if (0 != fchmod(fd, (mode_t)mode)) #else if (0 != chmod(path, (mode_t)mode)) #endif { failure_start(file, line, "Could not chmod %s", path); failure_finish(NULL); close(fd); return (0); } if (contents != NULL) { ssize_t wsize; if (csize < 0) wsize = (ssize_t)strlen(contents); else wsize = (ssize_t)csize; if (wsize != write(fd, contents, wsize)) { close(fd); failure_start(file, line, "Could not write to %s", path); failure_finish(NULL); close(fd); return (0); } } close(fd); assertion_file_mode(file, line, path, mode); return (1); #endif } /* Create a hardlink and report any failures. */ int assertion_make_hardlink(const char *file, int line, const char *newpath, const char *linkto) { int succeeded; assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) succeeded = my_CreateHardLinkA(newpath, linkto); #elif HAVE_LINK succeeded = !link(linkto, newpath); #else succeeded = 0; #endif if (succeeded) return (1); failure_start(file, line, "Could not create hardlink"); logprintf(" New link: %s\n", newpath); logprintf(" Old name: %s\n", linkto); failure_finish(NULL); return(0); } /* * Create a symlink and report any failures. * * Windows symlinks need to know if the target is a directory. */ int assertion_make_symlink(const char *file, int line, const char *newpath, const char *linkto, int targetIsDir) { #if defined(_WIN32) && !defined(__CYGWIN__) assertion_count(file, line); if (my_CreateSymbolicLinkA(newpath, linkto, targetIsDir)) return (1); #elif HAVE_SYMLINK (void)targetIsDir; /* UNUSED */ assertion_count(file, line); if (0 == symlink(linkto, newpath)) return (1); #else (void)targetIsDir; /* UNUSED */ #endif failure_start(file, line, "Could not create symlink"); logprintf(" New link: %s\n", newpath); logprintf(" Old name: %s\n", linkto); failure_finish(NULL); return(0); } /* Set umask, report failures. */ int assertion_umask(const char *file, int line, int mask) { assertion_count(file, line); (void)file; /* UNUSED */ (void)line; /* UNUSED */ umask((mode_t)mask); return (1); } /* Set times, report failures. */ int assertion_utimes(const char *file, int line, const char *pathname, time_t at, suseconds_t at_nsec, time_t mt, suseconds_t mt_nsec) { int r; #if defined(_WIN32) && !defined(__CYGWIN__) #define WINTIME(sec, nsec) (((sec * 10000000LL) + EPOC_TIME)\ + (((nsec)/1000)*10)) HANDLE h; ULARGE_INTEGER wintm; FILETIME fatime, fmtime; FILETIME *pat, *pmt; assertion_count(file, line); h = CreateFileA(pathname,GENERIC_READ | GENERIC_WRITE, FILE_SHARE_READ | FILE_SHARE_WRITE, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); if (h == INVALID_HANDLE_VALUE) { failure_start(file, line, "Can't access %s\n", pathname); failure_finish(NULL); return (0); } if (at > 0 || at_nsec > 0) { wintm.QuadPart = WINTIME(at, at_nsec); fatime.dwLowDateTime = wintm.LowPart; fatime.dwHighDateTime = wintm.HighPart; pat = &fatime; } else pat = NULL; if (mt > 0 || mt_nsec > 0) { wintm.QuadPart = WINTIME(mt, mt_nsec); fmtime.dwLowDateTime = wintm.LowPart; fmtime.dwHighDateTime = wintm.HighPart; pmt = &fmtime; } else pmt = NULL; if (pat != NULL || pmt != NULL) r = SetFileTime(h, NULL, pat, pmt); else r = 1; CloseHandle(h); if (r == 0) { failure_start(file, line, "Can't SetFileTime %s\n", pathname); failure_finish(NULL); return (0); } return (1); #else /* defined(_WIN32) && !defined(__CYGWIN__) */ struct stat st; struct timeval times[2]; #if !defined(__FreeBSD__) mt_nsec = at_nsec = 0; /* Generic POSIX only has whole seconds. */ #endif if (mt == 0 && mt_nsec == 0 && at == 0 && at_nsec == 0) return (1); r = lstat(pathname, &st); if (r < 0) { failure_start(file, line, "Can't stat %s\n", pathname); failure_finish(NULL); return (0); } if (mt == 0 && mt_nsec == 0) { mt = st.st_mtime; #if defined(__FreeBSD__) mt_nsec = st.st_mtimespec.tv_nsec; /* FreeBSD generally only stores to microsecond res, so round. */ mt_nsec = (mt_nsec / 1000) * 1000; #endif } if (at == 0 && at_nsec == 0) { at = st.st_atime; #if defined(__FreeBSD__) at_nsec = st.st_atimespec.tv_nsec; /* FreeBSD generally only stores to microsecond res, so round. */ at_nsec = (at_nsec / 1000) * 1000; #endif } times[1].tv_sec = mt; times[1].tv_usec = mt_nsec / 1000; times[0].tv_sec = at; times[0].tv_usec = at_nsec / 1000; #ifdef HAVE_LUTIMES r = lutimes(pathname, times); #else r = utimes(pathname, times); #endif if (r < 0) { failure_start(file, line, "Can't utimes %s\n", pathname); failure_finish(NULL); return (0); } return (1); #endif /* defined(_WIN32) && !defined(__CYGWIN__) */ } /* Compare file flags */ int assertion_compare_fflags(const char *file, int line, const char *patha, const char *pathb, int nomatch) { #if defined(HAVE_STRUCT_STAT_ST_FLAGS) && defined(UF_NODUMP) struct stat sa, sb; assertion_count(file, line); if (stat(patha, &sa) < 0) return (0); if (stat(pathb, &sb) < 0) return (0); if (!nomatch && sa.st_flags != sb.st_flags) { failure_start(file, line, "File flags should be identical: " "%s=%#010x %s=%#010x", patha, sa.st_flags, pathb, sb.st_flags); failure_finish(NULL); return (0); } if (nomatch && sa.st_flags == sb.st_flags) { failure_start(file, line, "File flags should be different: " "%s=%#010x %s=%#010x", patha, sa.st_flags, pathb, sb.st_flags); failure_finish(NULL); return (0); } #elif (defined(FS_IOC_GETFLAGS) && defined(HAVE_WORKING_FS_IOC_GETFLAGS) && \ defined(FS_NODUMP_FL)) || \ (defined(EXT2_IOC_GETFLAGS) && defined(HAVE_WORKING_EXT2_IOC_GETFLAGS) \ && defined(EXT2_NODUMP_FL)) int fd, r, flagsa, flagsb; assertion_count(file, line); fd = open(patha, O_RDONLY | O_NONBLOCK); if (fd < 0) { failure_start(file, line, "Can't open %s\n", patha); failure_finish(NULL); return (0); } r = ioctl(fd, #ifdef FS_IOC_GETFLAGS FS_IOC_GETFLAGS, #else EXT2_IOC_GETFLAGS, #endif &flagsa); close(fd); if (r < 0) { failure_start(file, line, "Can't get flags %s\n", patha); failure_finish(NULL); return (0); } fd = open(pathb, O_RDONLY | O_NONBLOCK); if (fd < 0) { failure_start(file, line, "Can't open %s\n", pathb); failure_finish(NULL); return (0); } r = ioctl(fd, #ifdef FS_IOC_GETFLAGS FS_IOC_GETFLAGS, #else EXT2_IOC_GETFLAGS, #endif &flagsb); close(fd); if (r < 0) { failure_start(file, line, "Can't get flags %s\n", pathb); failure_finish(NULL); return (0); } if (!nomatch && flagsa != flagsb) { failure_start(file, line, "File flags should be identical: " "%s=%#010x %s=%#010x", patha, flagsa, pathb, flagsb); failure_finish(NULL); return (0); } if (nomatch && flagsa == flagsb) { failure_start(file, line, "File flags should be different: " "%s=%#010x %s=%#010x", patha, flagsa, pathb, flagsb); failure_finish(NULL); return (0); } #else (void)patha; /* UNUSED */ (void)pathb; /* UNUSED */ (void)nomatch; /* UNUSED */ assertion_count(file, line); #endif return (1); } /* Set nodump, report failures. */ int assertion_set_nodump(const char *file, int line, const char *pathname) { #if defined(HAVE_STRUCT_STAT_ST_FLAGS) && defined(UF_NODUMP) int r; assertion_count(file, line); r = chflags(pathname, UF_NODUMP); if (r < 0) { failure_start(file, line, "Can't set nodump %s\n", pathname); failure_finish(NULL); return (0); } #elif (defined(FS_IOC_GETFLAGS) && defined(HAVE_WORKING_FS_IOC_GETFLAGS) && \ defined(FS_NODUMP_FL)) || \ (defined(EXT2_IOC_GETFLAGS) && defined(HAVE_WORKING_EXT2_IOC_GETFLAGS) \ && defined(EXT2_NODUMP_FL)) int fd, r, flags; assertion_count(file, line); fd = open(pathname, O_RDONLY | O_NONBLOCK); if (fd < 0) { failure_start(file, line, "Can't open %s\n", pathname); failure_finish(NULL); return (0); } r = ioctl(fd, #ifdef FS_IOC_GETFLAGS FS_IOC_GETFLAGS, #else EXT2_IOC_GETFLAGS, #endif &flags); if (r < 0) { failure_start(file, line, "Can't get flags %s\n", pathname); failure_finish(NULL); return (0); } #ifdef FS_NODUMP_FL flags |= FS_NODUMP_FL; #else flags |= EXT2_NODUMP_FL; #endif r = ioctl(fd, #ifdef FS_IOC_SETFLAGS FS_IOC_SETFLAGS, #else EXT2_IOC_SETFLAGS, #endif &flags); if (r < 0) { failure_start(file, line, "Can't set nodump %s\n", pathname); failure_finish(NULL); return (0); } close(fd); #else (void)pathname; /* UNUSED */ assertion_count(file, line); #endif return (1); } #ifdef PROGRAM static void assert_version_id(char **qq, size_t *ss) { char *q = *qq; size_t s = *ss; /* Version number is a series of digits and periods. */ while (s > 0 && (*q == '.' || (*q >= '0' && *q <= '9'))) { ++q; --s; } if (q[0] == 'd' && q[1] == 'e' && q[2] == 'v') { q += 3; s -= 3; } /* Skip a single trailing a,b,c, or d. */ if (*q == 'a' || *q == 'b' || *q == 'c' || *q == 'd') ++q; /* Version number terminated by space. */ failure("No space after version: ``%s''", q); assert(s > 1); failure("No space after version: ``%s''", q); assert(*q == ' '); ++q; --s; *qq = q; *ss = s; } /* * Check program version */ void assertVersion(const char *prog, const char *base) { int r; char *p, *q; size_t s; size_t prog_len = strlen(base); r = systemf("%s --version >version.stdout 2>version.stderr", prog); if (r != 0) r = systemf("%s -W version >version.stdout 2>version.stderr", prog); failure("Unable to run either %s --version or %s -W version", prog, prog); if (!assert(r == 0)) return; /* --version should generate nothing to stdout. */ assertEmptyFile("version.stderr"); /* Verify format of version message. */ q = p = slurpfile(&s, "version.stdout"); /* Version message should start with name of program, then space. */ assert(s > prog_len + 1); failure("Version must start with '%s': ``%s''", base, p); if (!assertEqualMem(q, base, prog_len)) { free(p); return; } q += prog_len; s -= prog_len; assert(*q == ' '); q++; s--; assert_version_id(&q, &s); /* Separator. */ failure("No `-' between program name and versions: ``%s''", p); assertEqualMem(q, "- ", 2); q += 2; s -= 2; failure("Not long enough for libarchive version: ``%s''", p); assert(s > 11); failure("Libarchive version must start with `libarchive': ``%s''", p); assertEqualMem(q, "libarchive ", 11); q += 11; s -= 11; assert_version_id(&q, &s); /* Skip arbitrary third-party version numbers. */ while (s > 0 && (*q == ' ' || *q == '-' || *q == '/' || *q == '.' || *q == '_' || isalnum((unsigned char)*q))) { ++q; --s; } /* All terminated by end-of-line. */ assert(s >= 1); /* Skip an optional CR character (e.g., Windows) */ failure("Version output must end with \\n or \\r\\n"); if (*q == '\r') { ++q; --s; } assertEqualMem(q, "\n", 1); free(p); } #endif /* PROGRAM */ /* * * UTILITIES for use by tests. * */ /* * Check whether platform supports symlinks. This is intended * for tests to use in deciding whether to bother testing symlink * support; if the platform doesn't support symlinks, there's no point * in checking whether the program being tested can create them. * * Note that the first time this test is called, we actually go out to * disk to create and verify a symlink. This is necessary because * symlink support is actually a property of a particular filesystem * and can thus vary between directories on a single system. After * the first call, this returns the cached result from memory, so it's * safe to call it as often as you wish. */ int canSymlink(void) { /* Remember the test result */ static int value = 0, tested = 0; if (tested) return (value); ++tested; assertion_make_file(__FILE__, __LINE__, "canSymlink.0", 0644, 1, "a"); /* Note: Cygwin has its own symlink() emulation that does not * use the Win32 CreateSymbolicLink() function. */ #if defined(_WIN32) && !defined(__CYGWIN__) value = my_CreateSymbolicLinkA("canSymlink.1", "canSymlink.0", 0) && is_symlink(__FILE__, __LINE__, "canSymlink.1", "canSymlink.0", 0); #elif HAVE_SYMLINK value = (0 == symlink("canSymlink.0", "canSymlink.1")) && is_symlink(__FILE__, __LINE__, "canSymlink.1","canSymlink.0", 0); #endif return (value); } /* Platform-dependent options for hiding the output of a subcommand. */ #if defined(_WIN32) && !defined(__CYGWIN__) static const char *redirectArgs = ">NUL 2>NUL"; /* Win32 cmd.exe */ #else static const char *redirectArgs = ">/dev/null 2>/dev/null"; /* POSIX 'sh' */ #endif + /* - * Can this platform run the bzip2 program? + * Can this platform run the specified command? */ int -canBzip2(void) +canRunCommand(const char *cmd, int *tested) { - static int tested = 0, value = 0; - if (!tested) { - tested = 1; - if (systemf("bzip2 --help %s", redirectArgs) == 0) - value = 1; - } - return (value); + int value = tested ? *tested : 0; + if (!value) { + value = systemf("%s %s", cmd, redirectArgs) ? -1 : +1; + if (tested) + *tested = value; + } + return (value > 0); } +#define CAN_RUN_FUNC(Program, Command) \ + int can##Program(void) { \ + static int tested = 0; \ + return canRunCommand((Command), &tested); \ + } + +/* + * Can this platform run the bzip2 program? + */ +CAN_RUN_FUNC(Bzip2, "bzip2 --help") + /* * Can this platform run the grzip program? */ -int -canGrzip(void) -{ - static int tested = 0, value = 0; - if (!tested) { - tested = 1; - if (systemf("grzip -V %s", redirectArgs) == 0) - value = 1; - } - return (value); -} +CAN_RUN_FUNC(Grzip, "grzip -V") /* * Can this platform run the gzip program? */ -int -canGzip(void) -{ - static int tested = 0, value = 0; - if (!tested) { - tested = 1; - if (systemf("gzip --help %s", redirectArgs) == 0) - value = 1; - } - return (value); -} +CAN_RUN_FUNC(Gzip, "gzip --help") /* * Can this platform run the lrzip program? */ -int -canRunCommand(const char *cmd) -{ - static int tested = 0, value = 0; - if (!tested) { - tested = 1; - if (systemf("%s %s", cmd, redirectArgs) == 0) - value = 1; - } - return (value); -} - -int -canLrzip(void) -{ - static int tested = 0, value = 0; - if (!tested) { - tested = 1; - if (systemf("lrzip -V %s", redirectArgs) == 0) - value = 1; - } - return (value); -} +CAN_RUN_FUNC(Lrzip, "lrzip -V") /* * Can this platform run the lz4 program? */ -int -canLz4(void) -{ - static int tested = 0, value = 0; - if (!tested) { - tested = 1; - if (systemf("lz4 --help %s", redirectArgs) == 0) - value = 1; - } - return (value); -} +CAN_RUN_FUNC(Lz4, "lz4 --help") /* * Can this platform run the zstd program? */ -int -canZstd(void) -{ - static int tested = 0, value = 0; - if (!tested) { - tested = 1; - if (systemf("zstd --help %s", redirectArgs) == 0) - value = 1; - } - return (value); -} +CAN_RUN_FUNC(Zstd, "zstd --help") /* * Can this platform run the lzip program? */ -int -canLzip(void) -{ - static int tested = 0, value = 0; - if (!tested) { - tested = 1; - if (systemf("lzip --help %s", redirectArgs) == 0) - value = 1; - } - return (value); -} +CAN_RUN_FUNC(Lzip, "lzip --help") /* * Can this platform run the lzma program? */ -int -canLzma(void) -{ - static int tested = 0, value = 0; - if (!tested) { - tested = 1; - if (systemf("lzma --help %s", redirectArgs) == 0) - value = 1; - } - return (value); -} +CAN_RUN_FUNC(Lzma, "lzma --help") /* * Can this platform run the lzop program? */ -int -canLzop(void) -{ - static int tested = 0, value = 0; - if (!tested) { - tested = 1; - if (systemf("lzop --help %s", redirectArgs) == 0) - value = 1; - } - return (value); -} +CAN_RUN_FUNC(Lzop, "lzop --help") /* * Can this platform run the xz program? */ -int -canXz(void) -{ - static int tested = 0, value = 0; - if (!tested) { - tested = 1; - if (systemf("xz --help %s", redirectArgs) == 0) - value = 1; - } - return (value); -} +CAN_RUN_FUNC(Xz, "xz --help") /* * Can this filesystem handle nodump flags. */ int canNodump(void) { #if defined(HAVE_STRUCT_STAT_ST_FLAGS) && defined(UF_NODUMP) const char *path = "cannodumptest"; struct stat sb; assertion_make_file(__FILE__, __LINE__, path, 0644, 0, NULL); if (chflags(path, UF_NODUMP) < 0) return (0); if (stat(path, &sb) < 0) return (0); if (sb.st_flags & UF_NODUMP) return (1); #elif (defined(FS_IOC_GETFLAGS) && defined(HAVE_WORKING_FS_IOC_GETFLAGS) \ && defined(FS_NODUMP_FL)) || \ (defined(EXT2_IOC_GETFLAGS) && defined(HAVE_WORKING_EXT2_IOC_GETFLAGS) \ && defined(EXT2_NODUMP_FL)) const char *path = "cannodumptest"; int fd, r, flags; assertion_make_file(__FILE__, __LINE__, path, 0644, 0, NULL); fd = open(path, O_RDONLY | O_NONBLOCK); if (fd < 0) return (0); r = ioctl(fd, #ifdef FS_IOC_GETFLAGS FS_IOC_GETFLAGS, #else EXT2_IOC_GETFLAGS, #endif &flags); if (r < 0) return (0); #ifdef FS_NODUMP_FL flags |= FS_NODUMP_FL; #else flags |= EXT2_NODUMP_FL; #endif r = ioctl(fd, #ifdef FS_IOC_SETFLAGS FS_IOC_SETFLAGS, #else EXT2_IOC_SETFLAGS, #endif &flags); if (r < 0) return (0); close(fd); fd = open(path, O_RDONLY | O_NONBLOCK); if (fd < 0) return (0); r = ioctl(fd, #ifdef FS_IOC_GETFLAGS FS_IOC_GETFLAGS, #else EXT2_IOC_GETFLAGS, #endif &flags); if (r < 0) return (0); close(fd); #ifdef FS_NODUMP_FL if (flags & FS_NODUMP_FL) #else if (flags & EXT2_NODUMP_FL) #endif return (1); #endif return (0); } /* Get extended attribute value from a path */ void * getXattr(const char *path, const char *name, size_t *sizep) { void *value = NULL; #if ARCHIVE_XATTR_SUPPORT ssize_t size; #if ARCHIVE_XATTR_LINUX size = lgetxattr(path, name, NULL, 0); #elif ARCHIVE_XATTR_DARWIN size = getxattr(path, name, NULL, 0, 0, XATTR_NOFOLLOW); #elif ARCHIVE_XATTR_AIX size = lgetea(path, name, NULL, 0); #elif ARCHIVE_XATTR_FREEBSD size = extattr_get_link(path, EXTATTR_NAMESPACE_USER, name + 5, NULL, 0); #endif if (size >= 0) { value = malloc(size); #if ARCHIVE_XATTR_LINUX size = lgetxattr(path, name, value, size); #elif ARCHIVE_XATTR_DARWIN size = getxattr(path, name, value, size, 0, XATTR_NOFOLLOW); #elif ARCHIVE_XATTR_AIX size = lgetea(path, name, value, size); #elif ARCHIVE_XATTR_FREEBSD size = extattr_get_link(path, EXTATTR_NAMESPACE_USER, name + 5, value, size); #endif if (size < 0) { free(value); value = NULL; } } if (size < 0) *sizep = 0; else *sizep = (size_t)size; #else /* !ARCHIVE_XATTR_SUPPORT */ (void)path; /* UNUSED */ (void)name; /* UNUSED */ *sizep = 0; #endif /* !ARCHIVE_XATTR_SUPPORT */ return (value); } /* * Set extended attribute on a path * Returns 0 on error, 1 on success */ int setXattr(const char *path, const char *name, const void *value, size_t size) { #if ARCHIVE_XATTR_SUPPORT #if ARCHIVE_XATTR_LINUX if (lsetxattr(path, name, value, size, 0) == 0) #elif ARCHIVE_XATTR_DARWIN if (setxattr(path, name, value, size, 0, XATTR_NOFOLLOW) == 0) #elif ARCHIVE_XATTR_AIX if (lsetea(path, name, value, size, 0) == 0) #elif ARCHIVE_XATTR_FREEBSD if (extattr_set_link(path, EXTATTR_NAMESPACE_USER, name + 5, value, size) > -1) #else if (0) #endif return (1); #else /* !ARCHIVE_XATTR_SUPPORT */ (void)path; /* UNUSED */ (void)name; /* UNUSED */ (void)value; /* UNUSED */ (void)size; /* UNUSED */ #endif /* !ARCHIVE_XATTR_SUPPORT */ return (0); } #if ARCHIVE_ACL_SUNOS /* Fetch ACLs on Solaris using acl() or facl() */ void * sunacl_get(int cmd, int *aclcnt, int fd, const char *path) { int cnt, cntcmd; size_t size; void *aclp; if (cmd == GETACL) { cntcmd = GETACLCNT; size = sizeof(aclent_t); } #if ARCHIVE_ACL_SUNOS_NFS4 else if (cmd == ACE_GETACL) { cntcmd = ACE_GETACLCNT; size = sizeof(ace_t); } #endif else { errno = EINVAL; *aclcnt = -1; return (NULL); } aclp = NULL; cnt = -2; while (cnt == -2 || (cnt == -1 && errno == ENOSPC)) { if (path != NULL) cnt = acl(path, cntcmd, 0, NULL); else cnt = facl(fd, cntcmd, 0, NULL); if (cnt > 0) { if (aclp == NULL) aclp = malloc(cnt * size); else aclp = realloc(NULL, cnt * size); if (aclp != NULL) { if (path != NULL) cnt = acl(path, cmd, cnt, aclp); else cnt = facl(fd, cmd, cnt, aclp); } } else { free(aclp); aclp = NULL; break; } } *aclcnt = cnt; return (aclp); } #endif /* ARCHIVE_ACL_SUNOS */ /* * Set test ACLs on a path * Return values: * 0: error setting ACLs * ARCHIVE_TEST_ACL_TYPE_POSIX1E: POSIX.1E ACLs have been set * ARCHIVE_TEST_ACL_TYPE_NFS4: NFSv4 or extended ACLs have been set */ int setTestAcl(const char *path) { #if ARCHIVE_ACL_SUPPORT int r = 1; #if ARCHIVE_ACL_LIBACL || ARCHIVE_ACL_FREEBSD || ARCHIVE_ACL_DARWIN acl_t acl; #endif #if ARCHIVE_ACL_LIBRICHACL struct richacl *richacl; #endif #if ARCHIVE_ACL_LIBACL || ARCHIVE_ACL_FREEBSD const char *acltext_posix1e = "user:1:rw-," "group:15:r-x," "user::rwx," "group::rwx," "other::r-x," "mask::rwx"; #elif ARCHIVE_ACL_SUNOS /* Solaris POSIX.1e */ aclent_t aclp_posix1e[] = { { USER_OBJ, -1, 4 | 2 | 1 }, { USER, 1, 4 | 2 }, { GROUP_OBJ, -1, 4 | 2 | 1 }, { GROUP, 15, 4 | 1 }, { CLASS_OBJ, -1, 4 | 2 | 1 }, { OTHER_OBJ, -1, 4 | 2 | 1 } }; #endif #if ARCHIVE_ACL_FREEBSD /* FreeBSD NFS4 */ const char *acltext_nfs4 = "user:1:rwpaRcs::allow:1," "group:15:rxaRcs::allow:15," "owner@:rwpxaARWcCos::allow," "group@:rwpxaRcs::allow," "everyone@:rxaRcs::allow"; #elif ARCHIVE_ACL_LIBRICHACL const char *acltext_nfs4 = "owner:rwpxaARWcCoS::mask," "group:rwpxaRcS::mask," "other:rxaRcS::mask," "user:1:rwpaRcS::allow," "group:15:rxaRcS::allow," "owner@:rwpxaARWcCoS::allow," "group@:rwpxaRcS::allow," "everyone@:rxaRcS::allow"; #elif ARCHIVE_ACL_SUNOS_NFS4 /* Solaris NFS4 */ ace_t aclp_nfs4[] = { { 1, ACE_READ_DATA | ACE_WRITE_DATA | ACE_APPEND_DATA | ACE_READ_ATTRIBUTES | ACE_READ_NAMED_ATTRS | ACE_READ_ACL | ACE_SYNCHRONIZE, 0, ACE_ACCESS_ALLOWED_ACE_TYPE }, { 15, ACE_READ_DATA | ACE_EXECUTE | ACE_READ_ATTRIBUTES | ACE_READ_NAMED_ATTRS | ACE_READ_ACL | ACE_SYNCHRONIZE, ACE_IDENTIFIER_GROUP, ACE_ACCESS_ALLOWED_ACE_TYPE }, { -1, ACE_READ_DATA | ACE_WRITE_DATA | ACE_APPEND_DATA | ACE_EXECUTE | ACE_READ_ATTRIBUTES | ACE_WRITE_ATTRIBUTES | ACE_READ_NAMED_ATTRS | ACE_WRITE_NAMED_ATTRS | ACE_READ_ACL | ACE_WRITE_ACL | ACE_WRITE_OWNER | ACE_SYNCHRONIZE, ACE_OWNER, ACE_ACCESS_ALLOWED_ACE_TYPE }, { -1, ACE_READ_DATA | ACE_WRITE_DATA | ACE_APPEND_DATA | ACE_EXECUTE | ACE_READ_ATTRIBUTES | ACE_READ_NAMED_ATTRS | ACE_READ_ACL | ACE_SYNCHRONIZE, ACE_GROUP | ACE_IDENTIFIER_GROUP, ACE_ACCESS_ALLOWED_ACE_TYPE }, { -1, ACE_READ_DATA | ACE_EXECUTE | ACE_READ_ATTRIBUTES | ACE_READ_NAMED_ATTRS | ACE_READ_ACL | ACE_SYNCHRONIZE, ACE_EVERYONE, ACE_ACCESS_ALLOWED_ACE_TYPE } }; #elif ARCHIVE_ACL_DARWIN /* Mac OS X */ acl_entry_t aclent; acl_permset_t permset; const uid_t uid = 1; uuid_t uuid; const acl_perm_t acl_perms[] = { ACL_READ_DATA, ACL_WRITE_DATA, ACL_APPEND_DATA, ACL_EXECUTE, ACL_READ_ATTRIBUTES, ACL_READ_EXTATTRIBUTES, ACL_READ_SECURITY, #if HAVE_DECL_ACL_SYNCHRONIZE ACL_SYNCHRONIZE #endif }; #endif /* ARCHIVE_ACL_DARWIN */ #if ARCHIVE_ACL_FREEBSD acl = acl_from_text(acltext_nfs4); failure("acl_from_text() error: %s", strerror(errno)); if (assert(acl != NULL) == 0) return (0); #elif ARCHIVE_ACL_LIBRICHACL richacl = richacl_from_text(acltext_nfs4, NULL, NULL); failure("richacl_from_text() error: %s", strerror(errno)); if (assert(richacl != NULL) == 0) return (0); #elif ARCHIVE_ACL_DARWIN acl = acl_init(1); failure("acl_init() error: %s", strerror(errno)); if (assert(acl != NULL) == 0) return (0); r = acl_create_entry(&acl, &aclent); failure("acl_create_entry() error: %s", strerror(errno)); if (assertEqualInt(r, 0) == 0) goto testacl_free; r = acl_set_tag_type(aclent, ACL_EXTENDED_ALLOW); failure("acl_set_tag_type() error: %s", strerror(errno)); if (assertEqualInt(r, 0) == 0) goto testacl_free; r = acl_get_permset(aclent, &permset); failure("acl_get_permset() error: %s", strerror(errno)); if (assertEqualInt(r, 0) == 0) goto testacl_free; for (size_t i = 0; i < nitems(acl_perms); i++) { r = acl_add_perm(permset, acl_perms[i]); failure("acl_add_perm() error: %s", strerror(errno)); if (assertEqualInt(r, 0) == 0) goto testacl_free; } r = acl_set_permset(aclent, permset); failure("acl_set_permset() error: %s", strerror(errno)); if (assertEqualInt(r, 0) == 0) goto testacl_free; r = mbr_uid_to_uuid(uid, uuid); failure("mbr_uid_to_uuid() error: %s", strerror(errno)); if (assertEqualInt(r, 0) == 0) goto testacl_free; r = acl_set_qualifier(aclent, uuid); failure("acl_set_qualifier() error: %s", strerror(errno)); if (assertEqualInt(r, 0) == 0) goto testacl_free; #endif /* ARCHIVE_ACL_DARWIN */ #if ARCHIVE_ACL_NFS4 #if ARCHIVE_ACL_FREEBSD r = acl_set_file(path, ACL_TYPE_NFS4, acl); acl_free(acl); #elif ARCHIVE_ACL_LIBRICHACL r = richacl_set_file(path, richacl); richacl_free(richacl); #elif ARCHIVE_ACL_SUNOS_NFS4 r = acl(path, ACE_SETACL, (int)(sizeof(aclp_nfs4)/sizeof(aclp_nfs4[0])), aclp_nfs4); #elif ARCHIVE_ACL_DARWIN r = acl_set_file(path, ACL_TYPE_EXTENDED, acl); acl_free(acl); #endif if (r == 0) return (ARCHIVE_TEST_ACL_TYPE_NFS4); #endif /* ARCHIVE_ACL_NFS4 */ #if ARCHIVE_ACL_POSIX1E #if ARCHIVE_ACL_FREEBSD || ARCHIVE_ACL_LIBACL acl = acl_from_text(acltext_posix1e); failure("acl_from_text() error: %s", strerror(errno)); if (assert(acl != NULL) == 0) return (0); r = acl_set_file(path, ACL_TYPE_ACCESS, acl); acl_free(acl); #elif ARCHIVE_ACL_SUNOS r = acl(path, SETACL, (int)(sizeof(aclp_posix1e)/sizeof(aclp_posix1e[0])), aclp_posix1e); #endif if (r == 0) return (ARCHIVE_TEST_ACL_TYPE_POSIX1E); else return (0); #endif /* ARCHIVE_ACL_POSIX1E */ #if ARCHIVE_ACL_DARWIN testacl_free: acl_free(acl); #endif #endif /* ARCHIVE_ACL_SUPPORT */ (void)path; /* UNUSED */ return (0); } /* * Sleep as needed; useful for verifying disk timestamp changes by * ensuring that the wall-clock time has actually changed before we * go back to re-read something from disk. */ void sleepUntilAfter(time_t t) { while (t >= time(NULL)) #if defined(_WIN32) && !defined(__CYGWIN__) Sleep(500); #else sleep(1); #endif } /* * Call standard system() call, but build up the command line using * sprintf() conventions. */ int systemf(const char *fmt, ...) { char buff[8192]; va_list ap; int r; va_start(ap, fmt); vsnprintf(buff, sizeof(buff), fmt, ap); if (verbosity > VERBOSITY_FULL) logprintf("Cmd: %s\n", buff); r = system(buff); va_end(ap); return (r); } /* * Slurp a file into memory for ease of comparison and testing. * Returns size of file in 'sizep' if non-NULL, null-terminates * data in memory for ease of use. */ char * slurpfile(size_t * sizep, const char *fmt, ...) { char filename[8192]; struct stat st; va_list ap; char *p; ssize_t bytes_read; FILE *f; int r; va_start(ap, fmt); vsnprintf(filename, sizeof(filename), fmt, ap); va_end(ap); f = fopen(filename, "rb"); if (f == NULL) { /* Note: No error; non-existent file is okay here. */ return (NULL); } r = fstat(fileno(f), &st); if (r != 0) { logprintf("Can't stat file %s\n", filename); fclose(f); return (NULL); } p = malloc((size_t)st.st_size + 1); if (p == NULL) { logprintf("Can't allocate %ld bytes of memory to read file %s\n", (long int)st.st_size, filename); fclose(f); return (NULL); } bytes_read = fread(p, 1, (size_t)st.st_size, f); if (bytes_read < st.st_size) { logprintf("Can't read file %s\n", filename); fclose(f); free(p); return (NULL); } p[st.st_size] = '\0'; if (sizep != NULL) *sizep = (size_t)st.st_size; fclose(f); return (p); } /* * Slurp a file into memory for ease of comparison and testing. * Returns size of file in 'sizep' if non-NULL, null-terminates * data in memory for ease of use. */ void dumpfile(const char *filename, void *data, size_t len) { ssize_t bytes_written; FILE *f; f = fopen(filename, "wb"); if (f == NULL) { logprintf("Can't open file %s for writing\n", filename); return; } bytes_written = fwrite(data, 1, len, f); if (bytes_written < (ssize_t)len) logprintf("Can't write file %s\n", filename); fclose(f); } /* Read a uuencoded file from the reference directory, decode, and * write the result into the current directory. */ #define VALID_UUDECODE(c) (c >= 32 && c <= 96) #define UUDECODE(c) (((c) - 0x20) & 0x3f) void extract_reference_file(const char *name) { char buff[1024]; FILE *in, *out; snprintf(buff, sizeof(buff), "%s/%s.uu", refdir, name); in = fopen(buff, "r"); failure("Couldn't open reference file %s", buff); assert(in != NULL); if (in == NULL) return; /* Read up to and including the 'begin' line. */ for (;;) { if (fgets(buff, sizeof(buff), in) == NULL) { /* TODO: This is a failure. */ return; } if (memcmp(buff, "begin ", 6) == 0) break; } /* Now, decode the rest and write it. */ out = fopen(name, "wb"); while (fgets(buff, sizeof(buff), in) != NULL) { char *p = buff; int bytes; if (memcmp(buff, "end", 3) == 0) break; bytes = UUDECODE(*p++); while (bytes > 0) { int n = 0; /* Write out 1-3 bytes from that. */ assert(VALID_UUDECODE(p[0])); assert(VALID_UUDECODE(p[1])); n = UUDECODE(*p++) << 18; n |= UUDECODE(*p++) << 12; fputc(n >> 16, out); --bytes; if (bytes > 0) { assert(VALID_UUDECODE(p[0])); n |= UUDECODE(*p++) << 6; fputc((n >> 8) & 0xFF, out); --bytes; } if (bytes > 0) { assert(VALID_UUDECODE(p[0])); n |= UUDECODE(*p++); fputc(n & 0xFF, out); --bytes; } } } fclose(out); fclose(in); } void copy_reference_file(const char *name) { char buff[1024]; FILE *in, *out; size_t rbytes; snprintf(buff, sizeof(buff), "%s/%s", refdir, name); in = fopen(buff, "rb"); failure("Couldn't open reference file %s", buff); assert(in != NULL); if (in == NULL) return; /* Now, decode the rest and write it. */ /* Not a lot of error checking here; the input better be right. */ out = fopen(name, "wb"); while ((rbytes = fread(buff, 1, sizeof(buff), in)) > 0) { if (fwrite(buff, 1, rbytes, out) != rbytes) { logprintf("Error: fwrite\n"); break; } } fclose(out); fclose(in); } int is_LargeInode(const char *file) { #if defined(_WIN32) && !defined(__CYGWIN__) BY_HANDLE_FILE_INFORMATION bhfi; int r; r = my_GetFileInformationByName(file, &bhfi); if (r != 0) return (0); return (bhfi.nFileIndexHigh & 0x0000FFFFUL); #else struct stat st; int64_t ino; if (stat(file, &st) < 0) return (0); ino = (int64_t)st.st_ino; return (ino > 0xffffffff); #endif } void extract_reference_files(const char **names) { while (names && *names) extract_reference_file(*names++); } #ifndef PROGRAM /* Set ACLs */ int assertion_entry_set_acls(const char *file, int line, struct archive_entry *ae, struct archive_test_acl_t *acls, int n) { int i, r, ret; assertion_count(file, line); ret = 0; archive_entry_acl_clear(ae); for (i = 0; i < n; i++) { r = archive_entry_acl_add_entry(ae, acls[i].type, acls[i].permset, acls[i].tag, acls[i].qual, acls[i].name); if (r != 0) { ret = 1; failure_start(file, line, "type=%#010x, " "permset=%#010x, tag=%d, qual=%d name=%s", (unsigned int)acls[i].type, (unsigned int)acls[i].permset, acls[i].tag, acls[i].qual, acls[i].name); failure_finish(NULL); } } return (ret); } static int archive_test_acl_match(struct archive_test_acl_t *acl, int type, int permset, int tag, int qual, const char *name) { if (type != acl->type) return (0); if (permset != acl->permset) return (0); if (tag != acl->tag) return (0); if (tag == ARCHIVE_ENTRY_ACL_USER_OBJ) return (1); if (tag == ARCHIVE_ENTRY_ACL_GROUP_OBJ) return (1); if (tag == ARCHIVE_ENTRY_ACL_EVERYONE) return (1); if (tag == ARCHIVE_ENTRY_ACL_OTHER) return (1); if (qual != acl->qual) return (0); if (name == NULL) { if (acl->name == NULL || acl->name[0] == '\0') return (1); return (0); } if (acl->name == NULL) { if (name[0] == '\0') return (1); return (0); } return (0 == strcmp(name, acl->name)); } /* Compare ACLs */ int assertion_entry_compare_acls(const char *file, int line, struct archive_entry *ae, struct archive_test_acl_t *acls, int cnt, int want_type, int mode) { int *marker; int i, r, n, ret; int type, permset, tag, qual; int matched; const char *name; assertion_count(file, line); ret = 0; n = 0; marker = malloc(sizeof(marker[0]) * cnt); for (i = 0; i < cnt; i++) { if ((acls[i].type & want_type) != 0) { marker[n] = i; n++; } } if (n == 0) { failure_start(file, line, "No ACL's to compare, type mask: %d", want_type); return (1); } while (0 == (r = archive_entry_acl_next(ae, want_type, &type, &permset, &tag, &qual, &name))) { for (i = 0, matched = 0; i < n && !matched; i++) { if (archive_test_acl_match(&acls[marker[i]], type, permset, tag, qual, name)) { /* We found a match; remove it. */ marker[i] = marker[n - 1]; n--; matched = 1; } } if (type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS && tag == ARCHIVE_ENTRY_ACL_USER_OBJ) { if (!matched) { failure_start(file, line, "No match for " "user_obj perm"); failure_finish(NULL); ret = 1; } if ((permset << 6) != (mode & 0700)) { failure_start(file, line, "USER_OBJ permset " "(%02o) != user mode (%02o)", (unsigned int)permset, (unsigned int)(07 & (mode >> 6))); failure_finish(NULL); ret = 1; } } else if (type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS && tag == ARCHIVE_ENTRY_ACL_GROUP_OBJ) { if (!matched) { failure_start(file, line, "No match for " "group_obj perm"); failure_finish(NULL); ret = 1; } if ((permset << 3) != (mode & 0070)) { failure_start(file, line, "GROUP_OBJ permset " "(%02o) != group mode (%02o)", (unsigned int)permset, (unsigned int)(07 & (mode >> 3))); failure_finish(NULL); ret = 1; } } else if (type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS && tag == ARCHIVE_ENTRY_ACL_OTHER) { if (!matched) { failure_start(file, line, "No match for " "other perm"); failure_finish(NULL); ret = 1; } if ((permset << 0) != (mode & 0007)) { failure_start(file, line, "OTHER permset " "(%02o) != other mode (%02o)", (unsigned int)permset, (unsigned int)mode & 07); failure_finish(NULL); ret = 1; } } else if (matched != 1) { failure_start(file, line, "Could not find match for " "ACL (type=%#010x,permset=%#010x,tag=%d,qual=%d," "name=``%s'')", (unsigned int)type, (unsigned int)permset, tag, qual, name); failure_finish(NULL); ret = 1; } } if (r != ARCHIVE_EOF) { failure_start(file, line, "Should not exit before EOF"); failure_finish(NULL); ret = 1; } if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0 && (mode_t)(mode & 0777) != (archive_entry_mode(ae) & 0777)) { failure_start(file, line, "Mode (%02o) and entry mode (%02o) " "mismatch", (unsigned int)mode, (unsigned int)archive_entry_mode(ae)); failure_finish(NULL); ret = 1; } if (n != 0) { failure_start(file, line, "Could not find match for ACL " "(type=%#010x,permset=%#010x,tag=%d,qual=%d,name=``%s'')", (unsigned int)acls[marker[0]].type, (unsigned int)acls[marker[0]].permset, acls[marker[0]].tag, acls[marker[0]].qual, acls[marker[0]].name); failure_finish(NULL); ret = 1; /* Number of ACLs not matched should == 0 */ } free(marker); return (ret); } #endif /* !defined(PROGRAM) */ /* * * TEST management * */ /* * "list.h" is simply created by "grep DEFINE_TEST test_*.c"; it has * a line like * DEFINE_TEST(test_function) * for each test. */ struct test_list_t { void (*func)(void); const char *name; int failures; }; /* Use "list.h" to declare all of the test functions. */ #undef DEFINE_TEST #define DEFINE_TEST(name) void name(void); #include "list.h" /* Use "list.h" to create a list of all tests (functions and names). */ #undef DEFINE_TEST #define DEFINE_TEST(n) { n, #n, 0 }, static struct test_list_t tests[] = { #include "list.h" }; /* * Summarize repeated failures in the just-completed test. */ static void test_summarize(int failed, int skips_num) { unsigned int i; switch (verbosity) { case VERBOSITY_SUMMARY_ONLY: printf(failed ? "E" : "."); fflush(stdout); break; case VERBOSITY_PASSFAIL: printf(failed ? "FAIL\n" : skips_num ? "skipped\n" : "ok\n"); break; } log_console = (verbosity == VERBOSITY_LIGHT_REPORT); for (i = 0; i < sizeof(failed_lines)/sizeof(failed_lines[0]); i++) { if (failed_lines[i].count > 1 && !failed_lines[i].skip) logprintf("%s:%u: Summary: Failed %d times\n", failed_filename, i, failed_lines[i].count); } /* Clear the failure history for the next file. */ failed_filename = NULL; memset(failed_lines, 0, sizeof(failed_lines)); } /* * Set or unset environment variable. */ static void set_environment(const char *key, const char *value) { #if defined(_WIN32) && !defined(__CYGWIN__) if (!SetEnvironmentVariable(key, value)) { fprintf(stderr, "SetEnvironmentVariable failed with %d\n", (int)GetLastError()); } #else if (value == NULL) { if (unsetenv(key) == -1) fprintf(stderr, "unsetenv: %s\n", strerror(errno)); } else { if (setenv(key, value, 1) == -1) fprintf(stderr, "setenv: %s\n", strerror(errno)); } #endif } /* * Enforce C locale for (sub)processes. */ static void set_c_locale(void) { static const char *lcs[] = { "LC_ADDRESS", "LC_ALL", "LC_COLLATE", "LC_CTYPE", "LC_IDENTIFICATION", "LC_MEASUREMENT", "LC_MESSAGES", "LC_MONETARY", "LC_NAME", "LC_NUMERIC", "LC_PAPER", "LC_TELEPHONE", "LC_TIME", NULL }; size_t i; setlocale(LC_ALL, "C"); set_environment("LANG", "C"); for (i = 0; lcs[i] != NULL; i++) set_environment(lcs[i], NULL); } /* * Actually run a single test, with appropriate setup and cleanup. */ static int test_run(int i, const char *tmpdir) { #ifdef PATH_MAX char workdir[PATH_MAX * 2]; #else char workdir[1024 * 2]; #endif char logfilename[256]; int failures_before = failures; int skips_before = skips; int tmp; mode_t oldumask; switch (verbosity) { case VERBOSITY_SUMMARY_ONLY: /* No per-test reports at all */ break; case VERBOSITY_PASSFAIL: /* rest of line will include ok/FAIL marker */ printf("%3d: %-64s", i, tests[i].name); fflush(stdout); break; default: /* Title of test, details will follow */ printf("%3d: %s\n", i, tests[i].name); } /* Chdir to the top-level work directory. */ if (!assertChdir(tmpdir)) { fprintf(stderr, "ERROR: Can't chdir to top work dir %s\n", tmpdir); exit(1); } /* Create a log file for this test. */ tmp = snprintf(logfilename, sizeof(logfilename), "%s.log", tests[i].name); if (tmp < 0) { fprintf(stderr, "ERROR can't create %s.log: %s\n", tests[i].name, strerror(errno)); exit(1); } if ((size_t)tmp >= sizeof(logfilename)) { fprintf(stderr, "ERROR can't create %s.log: Name too long. " "Length %d; Max allowed length %zu\n", tests[i].name, tmp, sizeof(logfilename) - 1); exit(1); } logfile = fopen(logfilename, "w"); fprintf(logfile, "%s\n\n", tests[i].name); /* Chdir() to a work dir for this specific test. */ tmp = snprintf(workdir, sizeof(workdir), "%s/%s", tmpdir, tests[i].name); if (tmp < 0) { fprintf(stderr, "ERROR can't create %s/%s: %s\n", tmpdir, tests[i].name, strerror(errno)); exit(1); } if ((size_t)tmp >= sizeof(workdir)) { fprintf(stderr, "ERROR can't create %s/%s: Path too long. " "Length %d; Max allowed length %zu\n", tmpdir, tests[i].name, tmp, sizeof(workdir) - 1); exit(1); } testworkdir = workdir; if (!assertMakeDir(testworkdir, 0755) || !assertChdir(testworkdir)) { fprintf(stderr, "ERROR: Can't chdir to work dir %s\n", testworkdir); exit(1); } /* Explicitly reset the locale before each test. */ set_c_locale(); /* Record the umask before we run the test. */ umask(oldumask = umask(0)); /* * Run the actual test. */ (*tests[i].func)(); /* * Clean up and report afterwards. */ testworkdir = NULL; /* Restore umask */ umask(oldumask); /* Reset locale. */ set_c_locale(); /* Reset directory. */ if (!assertChdir(tmpdir)) { fprintf(stderr, "ERROR: Couldn't chdir to temp dir %s\n", tmpdir); exit(1); } /* Report per-test summaries. */ tests[i].failures = failures - failures_before; test_summarize(tests[i].failures, skips - skips_before); /* Close the per-test log file. */ fclose(logfile); logfile = NULL; /* If there were no failures, we can remove the work dir and logfile. */ if (tests[i].failures == 0) { if (!keep_temp_files && assertChdir(tmpdir)) { #if defined(_WIN32) && !defined(__CYGWIN__) /* Make sure not to leave empty directories. * Sometimes a processing of closing files used by tests * is not done, then rmdir will be failed and it will * leave a empty test directory. So we should wait a few * seconds and retry rmdir. */ int r, t; for (t = 0; t < 10; t++) { if (t > 0) Sleep(1000); r = systemf("rmdir /S /Q %s", tests[i].name); if (r == 0) break; } systemf("del %s", logfilename); #else systemf("rm -rf %s", tests[i].name); systemf("rm %s", logfilename); #endif } } /* Return appropriate status. */ return (tests[i].failures); } /* * * * MAIN and support routines. * * */ static void usage(const char *program) { static const int limit = nitems(tests); int i; printf("Usage: %s [options] ...\n", program); printf("Default is to run all tests.\n"); printf("Otherwise, specify the numbers of the tests you wish to run.\n"); printf("Options:\n"); printf(" -d Dump core after any failure, for debugging.\n"); printf(" -k Keep all temp files.\n"); printf(" Default: temp files for successful tests deleted.\n"); #ifdef PROGRAM printf(" -p Path to executable to be tested.\n"); printf(" Default: path taken from " ENVBASE " environment variable.\n"); #endif printf(" -q Quiet.\n"); printf(" -r Path to dir containing reference files.\n"); printf(" Default: Current directory.\n"); printf(" -s Exit with code 2 if any tests were skipped.\n"); printf(" -u Keep running specified tests until one fails.\n"); printf(" -v Verbose.\n"); printf("Available tests:\n"); for (i = 0; i < limit; i++) printf(" %d: %s\n", i, tests[i].name); exit(1); } static char * get_refdir(const char *d) { size_t tried_size, buff_size; char *buff, *tried, *pwd = NULL, *p = NULL; #ifdef PATH_MAX buff_size = PATH_MAX; #else buff_size = 8192; #endif buff = calloc(buff_size, 1); if (buff == NULL) { fprintf(stderr, "Unable to allocate memory\n"); exit(1); } /* Allocate a buffer to hold the various directories we checked. */ tried_size = buff_size * 2; tried = calloc(tried_size, 1); if (tried == NULL) { fprintf(stderr, "Unable to allocate memory\n"); exit(1); } /* If a dir was specified, try that */ if (d != NULL) { pwd = NULL; snprintf(buff, buff_size, "%s", d); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); goto failure; } /* Get the current dir. */ #if defined(PATH_MAX) && !defined(__GLIBC__) pwd = getcwd(NULL, PATH_MAX);/* Solaris getcwd needs the size. */ #else pwd = getcwd(NULL, 0); #endif while (pwd[strlen(pwd) - 1] == '\n') pwd[strlen(pwd) - 1] = '\0'; /* Look for a known file. */ snprintf(buff, buff_size, "%s", pwd); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); snprintf(buff, buff_size, "%s/test", pwd); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); #if defined(LIBRARY) snprintf(buff, buff_size, "%s/%s/test", pwd, LIBRARY); #else snprintf(buff, buff_size, "%s/%s/test", pwd, PROGRAM); #endif p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); #if defined(PROGRAM_ALIAS) snprintf(buff, buff_size, "%s/%s/test", pwd, PROGRAM_ALIAS); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); #endif if (memcmp(pwd, "/usr/obj", 8) == 0) { snprintf(buff, buff_size, "%s", pwd + 8); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); snprintf(buff, buff_size, "%s/test", pwd + 8); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); } failure: printf("Unable to locate known reference file %s\n", KNOWNREF); printf(" Checked following directories:\n%s\n", tried); printf("Use -r option to specify full path to reference directory\n"); #if defined(_WIN32) && !defined(__CYGWIN__) && defined(_DEBUG) DebugBreak(); #endif exit(1); success: free(p); free(pwd); free(tried); /* Copy result into a fresh buffer to reduce memory usage. */ p = strdup(buff); free(buff); return p; } /* Filter tests against a glob pattern. Returns non-zero if test matches * pattern, zero otherwise. A '^' at the beginning of the pattern negates * the return values (i.e. returns zero for a match, non-zero otherwise. */ static int test_filter(const char *pattern, const char *test) { int retval = 0; int negate = 0; const char *p = pattern; const char *t = test; if (p[0] == '^') { negate = 1; p++; } while (1) { if (p[0] == '\\') p++; else if (p[0] == '*') { while (p[0] == '*') p++; if (p[0] == '\\') p++; if ((t = strchr(t, p[0])) == 0) break; } if (p[0] != t[0]) break; if (p[0] == '\0') { retval = 1; break; } p++; t++; } return (negate) ? !retval : retval; } static int get_test_set(int *test_set, int limit, const char *test) { int start, end; int idx = 0; if (test == NULL) { /* Default: Run all tests. */ for (;idx < limit; idx++) test_set[idx] = idx; return (limit); } if (*test >= '0' && *test <= '9') { const char *vp = test; start = 0; while (*vp >= '0' && *vp <= '9') { start *= 10; start += *vp - '0'; ++vp; } if (*vp == '\0') { end = start; } else if (*vp == '-') { ++vp; if (*vp == '\0') { end = limit - 1; } else { end = 0; while (*vp >= '0' && *vp <= '9') { end *= 10; end += *vp - '0'; ++vp; } } } else return (-1); if (start < 0 || end >= limit || start > end) return (-1); while (start <= end) test_set[idx++] = start++; } else { for (start = 0; start < limit; ++start) { const char *name = tests[start].name; if (test_filter(test, name)) test_set[idx++] = start; } } return ((idx == 0)?-1:idx); } int main(int argc, char **argv) { static const int limit = nitems(tests); int test_set[nitems(tests)]; int i = 0, j = 0, tests_run = 0, tests_failed = 0, option; size_t testprogdir_len; size_t tmplen; #ifdef PROGRAM size_t tmp2_len; #endif time_t now; struct tm *tmptr; #if defined(HAVE_LOCALTIME_R) || defined(HAVE_LOCALTIME_S) struct tm tmbuf; #endif char *refdir_alloc = NULL; const char *progname; char **saved_argv; const char *tmp, *option_arg, *p; #ifdef PATH_MAX char tmpdir[PATH_MAX]; #else char tmpdir[256]; #endif char *pwd, *testprogdir, *tmp2 = NULL, *vlevel = NULL; char tmpdir_timestamp[32]; (void)argc; /* UNUSED */ /* Get the current dir. */ #if defined(PATH_MAX) && !defined(__GLIBC__) pwd = getcwd(NULL, PATH_MAX);/* Solaris getcwd needs the size. */ #else pwd = getcwd(NULL, 0); #endif while (pwd[strlen(pwd) - 1] == '\n') pwd[strlen(pwd) - 1] = '\0'; #if defined(HAVE__CrtSetReportMode) && !defined(__WATCOMC__) /* To stop to run the default invalid parameter handler. */ _set_invalid_parameter_handler(invalid_parameter_handler); /* Disable annoying assertion message box. */ _CrtSetReportMode(_CRT_ASSERT, 0); #endif /* * Name of this program, used to build root of our temp directory * tree. */ progname = p = argv[0]; testprogdir_len = strlen(progname) + 1; if ((testprogdir = malloc(testprogdir_len)) == NULL) { fprintf(stderr, "ERROR: Out of memory."); exit(1); } strncpy(testprogdir, progname, testprogdir_len); while (*p != '\0') { /* Support \ or / dir separators for Windows compat. */ if (*p == '/' || *p == '\\') { progname = p + 1; i = j; } ++p; j++; } testprogdir[i] = '\0'; #if defined(_WIN32) && !defined(__CYGWIN__) if (testprogdir[0] != '/' && testprogdir[0] != '\\' && !(((testprogdir[0] >= 'a' && testprogdir[0] <= 'z') || (testprogdir[0] >= 'A' && testprogdir[0] <= 'Z')) && testprogdir[1] == ':' && (testprogdir[2] == '/' || testprogdir[2] == '\\'))) #else if (testprogdir[0] != '/') #endif { /* Fixup path for relative directories. */ if ((testprogdir = realloc(testprogdir, strlen(pwd) + 1 + strlen(testprogdir) + 1)) == NULL) { fprintf(stderr, "ERROR: Out of memory."); exit(1); } memmove(testprogdir + strlen(pwd) + 1, testprogdir, strlen(testprogdir) + 1); memcpy(testprogdir, pwd, strlen(pwd)); testprogdir[strlen(pwd)] = '/'; } #ifdef PROGRAM /* Get the target program from environment, if available. */ testprogfile = getenv(ENVBASE); #endif if (getenv("TMPDIR") != NULL) tmp = getenv("TMPDIR"); else if (getenv("TMP") != NULL) tmp = getenv("TMP"); else if (getenv("TEMP") != NULL) tmp = getenv("TEMP"); else if (getenv("TEMPDIR") != NULL) tmp = getenv("TEMPDIR"); else tmp = "/tmp"; tmplen = strlen(tmp); while (tmplen > 0 && tmp[tmplen - 1] == '/') tmplen--; /* Allow -d to be controlled through the environment. */ if (getenv(ENVBASE "_DEBUG") != NULL) dump_on_failure = 1; /* Allow -v to be controlled through the environment. */ if (getenv("_VERBOSITY_LEVEL") != NULL) { vlevel = getenv("_VERBOSITY_LEVEL"); verbosity = atoi(vlevel); if (verbosity < VERBOSITY_SUMMARY_ONLY || verbosity > VERBOSITY_FULL) { /* Unsupported verbosity levels are silently ignored */ vlevel = NULL; verbosity = VERBOSITY_PASSFAIL; } } /* Get the directory holding test files from environment. */ refdir = getenv(ENVBASE "_TEST_FILES"); /* * Parse options, without using getopt(), which isn't available * on all platforms. */ ++argv; /* Skip program name */ while (*argv != NULL) { if (**argv != '-') break; p = *argv++; ++p; /* Skip '-' */ while (*p != '\0') { option = *p++; option_arg = NULL; /* If 'opt' takes an argument, parse that. */ if (option == 'p' || option == 'r') { if (*p != '\0') option_arg = p; else if (*argv == NULL) { fprintf(stderr, "Option -%c requires argument.\n", option); usage(progname); } else option_arg = *argv++; p = ""; /* End of this option word. */ } /* Now, handle the option. */ switch (option) { case 'd': dump_on_failure = 1; break; case 'k': keep_temp_files = 1; break; case 'p': #ifdef PROGRAM testprogfile = option_arg; #else fprintf(stderr, "-p option not permitted\n"); usage(progname); #endif break; case 'q': if (!vlevel) verbosity--; break; case 'r': refdir = option_arg; break; case 's': fail_if_tests_skipped = 1; break; case 'u': until_failure++; break; case 'v': if (!vlevel) verbosity++; break; default: fprintf(stderr, "Unrecognized option '%c'\n", option); usage(progname); } } } /* * Sanity-check that our options make sense. */ #ifdef PROGRAM if (testprogfile == NULL) { tmp2_len = strlen(testprogdir) + 1 + strlen(PROGRAM) + 1; +#if defined(_WIN32) && !defined(__CYGWIN__) + tmp2_len += 4; +#endif if ((tmp2 = malloc(tmp2_len)) == NULL) { fprintf(stderr, "ERROR: Out of memory."); exit(1); } strncpy(tmp2, testprogdir, tmp2_len); strncat(tmp2, "/", tmp2_len); strncat(tmp2, PROGRAM, tmp2_len); +#if defined(_WIN32) && !defined(__CYGWIN__) + strncat(tmp2, ".exe", tmp2_len); +#endif testprogfile = tmp2; } { char *testprg; size_t testprg_len; #if defined(_WIN32) && !defined(__CYGWIN__) /* Command.com sometimes rejects '/' separators. */ testprg = strdup(testprogfile); for (i = 0; testprg[i] != '\0'; i++) { if (testprg[i] == '/') testprg[i] = '\\'; } testprogfile = testprg; #endif /* Quote the name that gets put into shell command lines. */ testprg_len = strlen(testprogfile) + 3; testprg = malloc(testprg_len); strncpy(testprg, "\"", testprg_len); strncat(testprg, testprogfile, testprg_len); strncat(testprg, "\"", testprg_len); testprog = testprg; } /* Sanity check: reject a relative path for refdir. */ if (refdir != NULL) { #if defined(_WIN32) && !defined(__CYGWIN__) /* TODO: probably use PathIsRelative() from . */ #else if (refdir[0] != '/') { fprintf(stderr, "ERROR: Cannot use relative path for refdir\n"); exit(1); } #endif } #endif #if !defined(_WIN32) && defined(SIGPIPE) { /* Ignore SIGPIPE signals */ struct sigaction sa; sa.sa_handler = SIG_IGN; sigemptyset(&sa.sa_mask); sa.sa_flags = 0; sigaction(SIGPIPE, &sa, NULL); } #endif /* * Create a temp directory for the following tests. * Include the time the tests started as part of the name, * to make it easier to track the results of multiple tests. */ now = time(NULL); for (i = 0; ; i++) { #if defined(HAVE_LOCALTIME_S) tmptr = localtime_s(&tmbuf, &now) ? NULL : &tmbuf; #elif defined(HAVE_LOCALTIME_R) tmptr = localtime_r(&now, &tmbuf); #else tmptr = localtime(&now); #endif strftime(tmpdir_timestamp, sizeof(tmpdir_timestamp), "%Y-%m-%dT%H.%M.%S", tmptr); if (tmplen + 1 + strlen(progname) + 1 + strlen(tmpdir_timestamp) + 1 + 3 >= nitems(tmpdir)) { fprintf(stderr, "ERROR: Temp directory pathname too long\n"); exit(1); } snprintf(tmpdir, sizeof(tmpdir), "%.*s/%s.%s-%03d", (int)tmplen, tmp, progname, tmpdir_timestamp, i); if (assertMakeDir(tmpdir, 0755)) break; if (i >= 999) { fprintf(stderr, "ERROR: Unable to create temp directory %s\n", tmpdir); exit(1); } } /* * If the user didn't specify a directory for locating * reference files, try to find the reference files in * the "usual places." */ refdir = refdir_alloc = get_refdir(refdir); /* * Banner with basic information. */ printf("\n"); printf("If tests fail or crash, details will be in:\n"); printf(" %s\n", tmpdir); printf("\n"); if (verbosity > VERBOSITY_SUMMARY_ONLY) { printf("Reference files will be read from: %s\n", refdir); #ifdef PROGRAM printf("Running tests on: %s\n", testprog); #endif printf("Exercising: "); fflush(stdout); printf("%s\n", EXTRA_VERSION); } else { printf("Running "); fflush(stdout); } /* * Run some or all of the individual tests. */ saved_argv = argv; do { argv = saved_argv; do { int test_num; test_num = get_test_set(test_set, limit, *argv); if (test_num < 0) { printf("*** INVALID Test %s\n", *argv); free(refdir_alloc); free(testprogdir); usage(progname); } for (i = 0; i < test_num; i++) { tests_run++; if (test_run(test_set[i], tmpdir)) { tests_failed++; if (until_failure) goto finish; } } if (*argv != NULL) argv++; } while (*argv != NULL); } while (until_failure); finish: /* Must be freed after all tests run */ free(tmp2); free(testprogdir); free(pwd); /* * Report summary statistics. */ if (verbosity > VERBOSITY_SUMMARY_ONLY) { printf("\n"); printf("Totals:\n"); printf(" Tests run: %8d\n", tests_run); printf(" Tests failed: %8d\n", tests_failed); printf(" Assertions checked:%8d\n", assertions); printf(" Assertions failed: %8d\n", failures); printf(" Skips reported: %8d\n", skips); } if (failures) { printf("\n"); printf("Failing tests:\n"); for (i = 0; i < limit; ++i) { if (tests[i].failures) printf(" %d: %s (%d failures)\n", i, tests[i].name, tests[i].failures); } printf("\n"); printf("Details for failing tests: %s\n", tmpdir); printf("\n"); } else { if (verbosity == VERBOSITY_SUMMARY_ONLY) printf("\n"); printf("%d tests passed, no failures\n", tests_run); } free(refdir_alloc); /* If the final tmpdir is empty, we can remove it. */ /* This should be the usual case when all tests succeed. */ assertChdir(".."); rmdir(tmpdir); if (tests_failed) return 1; if (fail_if_tests_skipped == 1 && skips > 0) return 2; return 0; } diff --git a/contrib/libarchive/unzip/bsdunzip.c b/contrib/libarchive/unzip/bsdunzip.c index 621afbeb9d6a..1b520e841690 100644 --- a/contrib/libarchive/unzip/bsdunzip.c +++ b/contrib/libarchive/unzip/bsdunzip.c @@ -1,1266 +1,1279 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2009, 2010 Joerg Sonnenberger * Copyright (c) 2007-2008 Dag-Erling Smørgrav * All rights reserved. */ #include "bsdunzip_platform.h" #include "la_queue.h" #include "la_getline.h" #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_CTYPE_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_FNMATCH_H #include #endif #ifdef HAVE_LOCALE_H #include #endif +#ifdef HAVE_SIGNAL_H +#include +#endif #ifdef HAVE_STDARG_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #if ((!defined(HAVE_UTIMENSAT) && defined(HAVE_LUTIMES)) || \ (!defined(HAVE_FUTIMENS) && defined(HAVE_FUTIMES))) #ifdef HAVE_SYS_TIME_H #include #endif #endif #ifdef HAVE_GETOPT_OPTRESET #include #endif #include "bsdunzip.h" #include "passphrase.h" -#include "err.h" +#include "lafe_err.h" /* command-line options */ static int a_opt; /* convert EOL */ static int C_opt; /* match case-insensitively */ static int c_opt; /* extract to stdout */ static const char *d_arg; /* directory */ static int f_opt; /* update existing files only */ static const char *O_arg; /* encoding */ static int j_opt; /* junk directories */ static int L_opt; /* lowercase names */ static int n_opt; /* never overwrite */ static int o_opt; /* always overwrite */ static int p_opt; /* extract to stdout, quiet */ static const char *P_arg; /* passphrase */ static int q_opt; /* quiet */ static int t_opt; /* test */ static int u_opt; /* update */ static int v_opt; /* verbose/list */ static const char *y_str = ""; /* 4 digit year */ static int Z1_opt; /* zipinfo mode list files only */ static int version_opt; /* version string */ /* debug flag */ static int unzip_debug; /* zipinfo mode */ static int zipinfo_mode; /* running on tty? */ static int tty; /* processing exclude list */ static int unzip_exclude_mode = 0; int bsdunzip_optind; /* convenience macro */ /* XXX should differentiate between ARCHIVE_{WARN,FAIL,RETRY} */ #define ac(call) \ do { \ int acret = (call); \ if (acret != ARCHIVE_OK) \ errorx("%s", archive_error_string(a)); \ } while (0) /* * Indicates that last info() did not end with EOL. This helps error() et * al. avoid printing an error message on the same line as an incomplete * informational message. */ static int noeol; /* for an interactive passphrase input */ static char *passphrase_buf; /* fatal error message + errno */ static void __LA_NORETURN error(const char *fmt, ...) { va_list ap; if (noeol) fprintf(stdout, "\n"); fflush(stdout); fprintf(stderr, "unzip: "); va_start(ap, fmt); vfprintf(stderr, fmt, ap); va_end(ap); fprintf(stderr, ": %s\n", strerror(errno)); exit(EXIT_FAILURE); } /* fatal error message, no errno */ static void __LA_NORETURN errorx(const char *fmt, ...) { va_list ap; if (noeol) fprintf(stdout, "\n"); fflush(stdout); fprintf(stderr, "unzip: "); va_start(ap, fmt); vfprintf(stderr, fmt, ap); va_end(ap); fprintf(stderr, "\n"); exit(EXIT_FAILURE); } /* non-fatal error message + errno */ static void warning(const char *fmt, ...) { va_list ap; if (noeol) fprintf(stdout, "\n"); fflush(stdout); fprintf(stderr, "unzip: "); va_start(ap, fmt); vfprintf(stderr, fmt, ap); va_end(ap); fprintf(stderr, ": %s\n", strerror(errno)); } /* non-fatal error message, no errno */ static void warningx(const char *fmt, ...) { va_list ap; if (noeol) fprintf(stdout, "\n"); fflush(stdout); fprintf(stderr, "unzip: "); va_start(ap, fmt); vfprintf(stderr, fmt, ap); va_end(ap); fprintf(stderr, "\n"); } /* informational message (if not -q) */ static void info(const char *fmt, ...) { va_list ap; if (q_opt && !unzip_debug) return; va_start(ap, fmt); vfprintf(stdout, fmt, ap); va_end(ap); fflush(stdout); if (*fmt == '\0') noeol = 1; else noeol = fmt[strlen(fmt) - 1] != '\n'; } /* debug message (if unzip_debug) */ static void debug(const char *fmt, ...) { va_list ap; if (!unzip_debug) return; va_start(ap, fmt); vfprintf(stderr, fmt, ap); va_end(ap); fflush(stderr); if (*fmt == '\0') noeol = 1; else noeol = fmt[strlen(fmt) - 1] != '\n'; } /* duplicate a path name, possibly converting to lower case */ static char * pathdup(const char *path) { char *str; size_t i, len; if (path == NULL || path[0] == '\0') return (NULL); len = strlen(path); while (len && path[len - 1] == '/') len--; if ((str = malloc(len + 1)) == NULL) { errno = ENOMEM; error("malloc()"); } if (L_opt) { for (i = 0; i < len; ++i) str[i] = (char)tolower((unsigned char)path[i]); } else { memcpy(str, path, len); } str[len] = '\0'; return (str); } /* concatenate two path names */ static char * pathcat(const char *prefix, const char *path) { char *str; size_t prelen, len; prelen = prefix ? strlen(prefix) + 1 : 0; len = strlen(path) + 1; if ((str = malloc(prelen + len)) == NULL) { errno = ENOMEM; error("malloc()"); } if (prefix) { memcpy(str, prefix, prelen); /* includes zero */ str[prelen - 1] = '/'; /* splat zero */ } memcpy(str + prelen, path, len); /* includes zero */ return (str); } /* * Pattern lists for include / exclude processing */ struct pattern { STAILQ_ENTRY(pattern) link; char pattern[]; }; STAILQ_HEAD(pattern_list, pattern); static struct pattern_list include = STAILQ_HEAD_INITIALIZER(include); static struct pattern_list exclude = STAILQ_HEAD_INITIALIZER(exclude); /* * Add an entry to a pattern list */ static void add_pattern(struct pattern_list *list, const char *pattern) { struct pattern *entry; size_t len; debug("adding pattern '%s'\n", pattern); len = strlen(pattern); if ((entry = malloc(sizeof *entry + len + 1)) == NULL) { errno = ENOMEM; error("malloc()"); } memcpy(entry->pattern, pattern, len + 1); STAILQ_INSERT_TAIL(list, entry, link); } /* * Match a string against a list of patterns */ static int match_pattern(struct pattern_list *list, const char *str) { struct pattern *entry; STAILQ_FOREACH(entry, list, link) { #ifdef HAVE_FNMATCH if (fnmatch(entry->pattern, str, C_opt ? FNM_CASEFOLD : 0) == 0) return (1); #else #error "Unsupported platform: fnmatch() is required" #endif } return (0); } /* * Verify that a given pathname is in the include list and not in the * exclude list. */ static int accept_pathname(const char *pathname) { if (!STAILQ_EMPTY(&include) && !match_pattern(&include, pathname)) return (0); if (!STAILQ_EMPTY(&exclude) && match_pattern(&exclude, pathname)) return (0); return (1); } /* * Create the specified directory with the specified mode, taking certain * precautions on they way. */ static void make_dir(const char *path, int mode) { struct stat sb; if (lstat(path, &sb) == 0) { if (S_ISDIR(sb.st_mode)) return; /* * Normally, we should either ask the user about removing * the non-directory of the same name as a directory we * wish to create, or respect the -n or -o command-line * options. However, this may lead to a later failure or * even compromise (if this non-directory happens to be a * symlink to somewhere unsafe), so we don't. */ /* * Don't check unlink() result; failure will cause mkdir() * to fail later, which we will catch. */ (void)unlink(path); } if (mkdir(path, (mode_t)mode) != 0 && errno != EEXIST) error("mkdir('%s')", path); } /* * Ensure that all directories leading up to (but not including) the * specified path exist. * * XXX inefficient + modifies the file in-place */ static void make_parent(char *path) { struct stat sb; char *sep; sep = strrchr(path, '/'); if (sep == NULL || sep == path) return; *sep = '\0'; if (lstat(path, &sb) == 0) { if (S_ISDIR(sb.st_mode)) { *sep = '/'; return; } unlink(path); } make_parent(path); mkdir(path, 0755); *sep = '/'; #if 0 for (sep = path; (sep = strchr(sep, '/')) != NULL; sep++) { /* root in case of absolute d_arg */ if (sep == path) continue; *sep = '\0'; make_dir(path, 0755); *sep = '/'; } #endif } /* * Extract a directory. */ static void extract_dir(struct archive *a, struct archive_entry *e, const char *path) { int mode; /* * Dropbox likes to create '/' directory entries, just ignore * such junk. */ if (*path == '\0') return; mode = archive_entry_mode(e) & 0777; if (mode == 0) mode = 0755; /* * Some zipfiles contain directories with weird permissions such * as 0644 or 0444. This can cause strange issues such as being * unable to extract files into the directory we just created, or * the user being unable to remove the directory later without * first manually changing its permissions. Therefore, we whack * the permissions into shape, assuming that the user wants full * access and that anyone who gets read access also gets execute * access. */ mode |= 0700; if (mode & 0040) mode |= 0010; if (mode & 0004) mode |= 0001; info(" creating: %s/\n", path); make_dir(path, mode); ac(archive_read_data_skip(a)); } static unsigned char buffer[8192]; static char spinner[] = { '|', '/', '-', '\\' }; static int handle_existing_file(char **path) { size_t alen; ssize_t len; char buf[4]; for (;;) { fprintf(stderr, "replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", *path); if (fgets(buf, sizeof(buf), stdin) == NULL) goto stdin_err; switch (*buf) { case 'A': o_opt = 1; /* FALLTHROUGH */ case 'y': case 'Y': (void)unlink(*path); return 1; case 'N': n_opt = 1; /* FALLTHROUGH */ case 'n': return -1; case 'r': case 'R': printf("New name: "); fflush(stdout); free(*path); *path = NULL; alen = 0; len = getline(path, &alen, stdin); if (len < 1) goto stdin_err; if ((*path)[len - 1] == '\n') (*path)[len - 1] = '\0'; return 0; default: break; } } stdin_err: clearerr(stdin); printf("NULL\n(EOF or read error, " "treating as \"[N]one\"...)\n"); n_opt = 1; return -1; } /* * Detect binary files by a combination of character white list and * black list. NUL bytes and other control codes without use in text files * result directly in switching the file to binary mode. Otherwise, at least * one white-listed byte has to be found. * * Black-listed: 0..6, 14..25, 28..31 * 0xf3ffc07f = 11110011111111111100000001111111b * White-listed: 9..10, 13, >= 32 * 0x00002600 = 00000000000000000010011000000000b * * See the proginfo/txtvsbin.txt in the zip sources for a detailed discussion. */ #define BYTE_IS_BINARY(x) ((x) < 32 && (0xf3ffc07fU & (1U << (x)))) #define BYTE_IS_TEXT(x) ((x) >= 32 || (0x00002600U & (1U << (x)))) static int check_binary(const unsigned char *buf, size_t len) { int rv; for (rv = 1; len--; ++buf) { if (BYTE_IS_BINARY(*buf)) return 1; if (BYTE_IS_TEXT(*buf)) rv = 0; } return rv; } /* * Extract to a file descriptor */ static int extract2fd(struct archive *a, char *pathname, int fd) { int cr, text, warn; ssize_t len; unsigned char *p, *q, *end; text = a_opt; warn = 0; cr = 0; /* loop over file contents and write to fd */ for (int n = 0; ; n++) { if (fd != STDOUT_FILENO) if (tty && (n % 4) == 0) info(" %c\b\b", spinner[(n / 4) % sizeof spinner]); len = archive_read_data(a, buffer, sizeof buffer); if (len < 0) ac(len); /* left over CR from previous buffer */ if (a_opt && cr) { if (len == 0 || buffer[0] != '\n') if (write(fd, "\r", 1) != 1) error("write('%s')", pathname); cr = 0; } /* EOF */ if (len == 0) break; end = buffer + len; /* * Detect whether this is a text file. The correct way to * do this is to check the least significant bit of the * "internal file attributes" field of the corresponding * file header in the central directory, but libarchive * does not provide access to this field, so we have to * guess by looking for non-ASCII characters in the * buffer. Hopefully we won't guess wrong. If we do * guess wrong, we print a warning message later. */ if (a_opt && n == 0) { if (check_binary(buffer, len)) text = 0; } /* simple case */ if (!a_opt || !text) { if (write(fd, buffer, len) != len) error("write('%s')", pathname); continue; } /* hard case: convert \r\n to \n (sigh...) */ for (p = buffer; p < end; p = q + 1) { for (q = p; q < end; q++) { if (!warn && BYTE_IS_BINARY(*q)) { warningx("%s may be corrupted due" " to weak text file detection" " heuristic", pathname); warn = 1; } if (q[0] != '\r') continue; if (&q[1] == end) { cr = 1; break; } if (q[1] == '\n') break; } if (write(fd, p, q - p) != q - p) error("write('%s')", pathname); } } return text; } /* * Extract a regular file. */ static void extract_file(struct archive *a, struct archive_entry *e, char **path) { int mode; struct timespec mtime; struct stat sb; int fd, check, text; const char *linkname; #if defined(HAVE_UTIMENSAT) || defined(HAVE_FUTIMENS) struct timespec ts[2]; #endif #if ((!defined(HAVE_UTIMENSAT) && defined(HAVE_LUTIMES)) || \ (!defined(HAVE_FUTIMENS) && defined(HAVE_FUTIMES))) struct timeval times[2]; #endif mode = archive_entry_mode(e) & 0777; if (mode == 0) mode = 0644; mtime.tv_sec = archive_entry_mtime(e); mtime.tv_nsec = archive_entry_mtime_nsec(e); /* look for existing file of same name */ recheck: if (lstat(*path, &sb) == 0) { if (u_opt || f_opt) { /* check if up-to-date */ if (S_ISREG(sb.st_mode) && ( #if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC sb.st_mtimespec.tv_sec > mtime.tv_sec || (sb.st_mtimespec.tv_sec == mtime.tv_sec && sb.st_mtimespec.tv_nsec >= mtime.tv_nsec) #elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC sb.st_mtim.tv_sec > mtime.tv_sec || (sb.st_mtim.tv_sec == mtime.tv_sec && sb.st_mtim.tv_nsec >= mtime.tv_nsec) #elif HAVE_STRUCT_STAT_ST_MTIME_N sb.st_mtime > mtime.tv_sec || (sb.st_mtime == mtime.tv_sec && sb.st_mtime_n => mtime.tv_nsec) #elif HAVE_STRUCT_STAT_ST_MTIME_USEC sb.st_mtime > mtime.tv_sec || (sb.st_mtime == mtime.tv_sec && sb.st_mtime_usec => mtime.tv_nsec / 1000) #else sb.st_mtime > mtime.tv_sec #endif )) return; (void)unlink(*path); } else if (o_opt) { /* overwrite */ (void)unlink(*path); } else if (n_opt) { /* do not overwrite */ return; } else { check = handle_existing_file(path); if (check == 0) goto recheck; if (check == -1) return; /* do not overwrite */ } } else { if (f_opt) return; } #if defined(HAVE_UTIMENSAT) || defined(HAVE_FUTIMENS) ts[0].tv_sec = 0; ts[0].tv_nsec = UTIME_NOW; ts[1] = mtime; #endif #if ((!defined(HAVE_UTIMENSAT) && defined(HAVE_LUTIMES)) || \ (!defined(HAVE_FUTIMENS) && defined(HAVE_FUTIMES))) times[0].tv_sec = 0; times[0].tv_usec = -1; times[1].tv_sec = mtime.tv_sec; times[1].tv_usec = mtime.tv_nsec / 1000; #endif /* process symlinks */ linkname = archive_entry_symlink(e); if (linkname != NULL) { if (symlink(linkname, *path) != 0) error("symlink('%s')", *path); info(" extracting: %s -> %s\n", *path, linkname); #ifdef HAVE_LCHMOD if (lchmod(*path, (mode_t)mode) != 0) warning("Cannot set mode for '%s'", *path); #endif /* set access and modification time */ #if defined(HAVE_UTIMENSAT) if (utimensat(AT_FDCWD, *path, ts, AT_SYMLINK_NOFOLLOW) != 0) warning("utimensat('%s')", *path); #elif defined(HAVE_LUTIMES) gettimeofday(×[0], NULL); if (lutimes(*path, times) != 0) warning("lutimes('%s')", *path); #endif return; } if ((fd = open(*path, O_RDWR|O_CREAT|O_TRUNC, mode)) < 0) error("open('%s')", *path); info(" extracting: %s", *path); text = extract2fd(a, *path, fd); if (tty) info(" \b\b"); if (text) info(" (text)"); info("\n"); /* set access and modification time */ #if defined(HAVE_FUTIMENS) if (futimens(fd, ts) != 0) error("futimens('%s')", *path); #elif defined(HAVE_FUTIMES) gettimeofday(×[0], NULL); if (futimes(fd, times) != 0) error("futimes('%s')", *path); #endif if (close(fd) != 0) error("close('%s')", *path); } /* * Extract a zipfile entry: first perform some sanity checks to ensure * that it is either a directory or a regular file and that the path is * not absolute and does not try to break out of the current directory; * then call either extract_dir() or extract_file() as appropriate. * * This is complicated a bit by the various ways in which we need to * manipulate the path name. Case conversion (if requested by the -L * option) happens first, but the include / exclude patterns are applied * to the full converted path name, before the directory part of the path * is removed in accordance with the -j option. Sanity checks are * intentionally done earlier than they need to be, so the user will get a * warning about insecure paths even for files or directories which * wouldn't be extracted anyway. */ static void extract(struct archive *a, struct archive_entry *e) { char *pathname, *realpathname; mode_t filetype; char *p, *q; if ((pathname = pathdup(archive_entry_pathname(e))) == NULL) { warningx("skipping empty or unreadable filename entry"); ac(archive_read_data_skip(a)); return; } filetype = archive_entry_filetype(e); /* sanity checks */ if (pathname[0] == '/' || strncmp(pathname, "../", 3) == 0 || strstr(pathname, "/../") != NULL) { warningx("skipping insecure entry '%s'", pathname); ac(archive_read_data_skip(a)); free(pathname); return; } /* I don't think this can happen in a zipfile.. */ if (!S_ISDIR(filetype) && !S_ISREG(filetype) && !S_ISLNK(filetype)) { warningx("skipping non-regular entry '%s'", pathname); ac(archive_read_data_skip(a)); free(pathname); return; } /* skip directories in -j case */ if (S_ISDIR(filetype) && j_opt) { ac(archive_read_data_skip(a)); free(pathname); return; } /* apply include / exclude patterns */ if (!accept_pathname(pathname)) { ac(archive_read_data_skip(a)); free(pathname); return; } /* apply -j and -d */ if (j_opt) { for (p = q = pathname; *p; ++p) if (*p == '/') q = p + 1; realpathname = pathcat(d_arg, q); } else { realpathname = pathcat(d_arg, pathname); } /* ensure that parent directory exists */ make_parent(realpathname); if (S_ISDIR(filetype)) extract_dir(a, e, realpathname); else extract_file(a, e, &realpathname); free(realpathname); free(pathname); } static void extract_stdout(struct archive *a, struct archive_entry *e) { char *pathname; mode_t filetype; if ((pathname = pathdup(archive_entry_pathname(e))) == NULL) { warningx("skipping empty or unreadable filename entry"); ac(archive_read_data_skip(a)); return; } filetype = archive_entry_filetype(e); /* I don't think this can happen in a zipfile.. */ if (!S_ISDIR(filetype) && !S_ISREG(filetype) && !S_ISLNK(filetype)) { warningx("skipping non-regular entry '%s'", pathname); ac(archive_read_data_skip(a)); free(pathname); return; } /* skip directories in -j case */ if (S_ISDIR(filetype)) { ac(archive_read_data_skip(a)); free(pathname); return; } /* apply include / exclude patterns */ if (!accept_pathname(pathname)) { ac(archive_read_data_skip(a)); free(pathname); return; } if (c_opt) info("x %s\n", pathname); (void)extract2fd(a, pathname, STDOUT_FILENO); free(pathname); } /* * Print the name of an entry to stdout. */ static void list(struct archive *a, struct archive_entry *e) { char buf[20]; time_t mtime; struct tm *tm; const char *pathname; mtime = archive_entry_mtime(e); tm = localtime(&mtime); if (*y_str) strftime(buf, sizeof(buf), "%m-%d-%G %R", tm); else strftime(buf, sizeof(buf), "%m-%d-%g %R", tm); pathname = archive_entry_pathname(e); if (!pathname) pathname = ""; if (!zipinfo_mode) { if (v_opt == 1) { printf(" %8ju %s %s\n", (uintmax_t)archive_entry_size(e), buf, pathname); } else if (v_opt == 2) { printf("%8ju Stored %7ju 0%% %s %08x %s\n", (uintmax_t)archive_entry_size(e), (uintmax_t)archive_entry_size(e), buf, 0U, pathname); } } else { if (Z1_opt) printf("%s\n", pathname); } ac(archive_read_data_skip(a)); } /* * Extract to memory to check CRC */ static int test(struct archive *a, struct archive_entry *e) { ssize_t len; int error_count; error_count = 0; if (S_ISDIR(archive_entry_filetype(e))) return 0; info(" testing: %s\t", archive_entry_pathname(e)); while ((len = archive_read_data(a, buffer, sizeof buffer)) > 0) /* nothing */; if (len < 0) { info(" %s\n", archive_error_string(a)); ++error_count; } else { info(" OK\n"); } /* shouldn't be necessary, but it doesn't hurt */ ac(archive_read_data_skip(a)); return error_count; } /* * Callback function for reading passphrase. * Originally from cpio.c and passphrase.c, libarchive. */ #define PPBUFF_SIZE 1024 static const char * passphrase_callback(struct archive *a, void *_client_data) { char *p; (void)a; /* UNUSED */ (void)_client_data; /* UNUSED */ if (passphrase_buf == NULL) { passphrase_buf = malloc(PPBUFF_SIZE); if (passphrase_buf == NULL) { errno = ENOMEM; error("malloc()"); } } p = lafe_readpassphrase("\nEnter password: ", passphrase_buf, PPBUFF_SIZE); if (p == NULL && errno != EINTR) error("Error reading password"); return p; } /* * Main loop: open the zipfile, iterate over its contents and decide what * to do with each entry. */ static void unzip(const char *fn) { struct archive *a; struct archive_entry *e; int ret; uintmax_t total_size, file_count, error_count; if ((a = archive_read_new()) == NULL) error("archive_read_new failed"); ac(archive_read_support_format_zip(a)); if (O_arg) ac(archive_read_set_format_option(a, "zip", "hdrcharset", O_arg)); if (P_arg) archive_read_add_passphrase(a, P_arg); else archive_read_set_passphrase_callback(a, NULL, &passphrase_callback); ac(archive_read_open_filename(a, fn, 8192)); if (!zipinfo_mode) { if (!p_opt && !q_opt) printf("Archive: %s\n", fn); if (v_opt == 1) { printf(" Length %sDate Time Name\n", y_str); printf(" -------- %s---- ---- ----\n", y_str); } else if (v_opt == 2) { printf(" Length Method Size Ratio %sDate Time CRC-32 Name\n", y_str); printf("-------- ------ ------- ----- %s---- ---- ------ ----\n", y_str); } } total_size = 0; file_count = 0; error_count = 0; for (;;) { ret = archive_read_next_header(a, &e); if (ret == ARCHIVE_EOF) break; ac(ret); if (!zipinfo_mode) { if (t_opt) error_count += test(a, e); else if (v_opt) list(a, e); else if (p_opt || c_opt) extract_stdout(a, e); else extract(a, e); } else { if (Z1_opt) list(a, e); } total_size += archive_entry_size(e); ++file_count; } if (zipinfo_mode) { if (v_opt == 1) { printf(" -------- %s-------\n", y_str); printf(" %8ju %s%ju file%s\n", total_size, y_str, file_count, file_count != 1 ? "s" : ""); } else if (v_opt == 2) { printf("-------- ------- --- %s-------\n", y_str); printf("%8ju %7ju 0%% %s%ju file%s\n", total_size, total_size, y_str, file_count, file_count != 1 ? "s" : ""); } } ac(archive_read_free(a)); if (passphrase_buf != NULL) { memset(passphrase_buf, 0, PPBUFF_SIZE); free(passphrase_buf); } if (t_opt) { if (error_count > 0) { errorx("%ju checksum error(s) found.", error_count); } else { printf("No errors detected in compressed data of %s.\n", fn); } } } static void usage(void) { fprintf(stderr, "Usage: unzip [-aCcfjLlnopqtuvyZ1] [{-O|-I} encoding] [-d dir] [-x pattern] [-P password] zipfile\n" " [member ...]\n"); exit(EXIT_FAILURE); } static void version(void) { printf("bsdunzip %s - %s \n", BSDUNZIP_VERSION_STRING, archive_version_details()); exit(0); } static int getopts(int argc, char *argv[]) { struct bsdunzip *bsdunzip, bsdunzip_storage; int opt; bsdunzip_optind = 1; bsdunzip = &bsdunzip_storage; memset(bsdunzip, 0, sizeof(*bsdunzip)); bsdunzip->argv = argv; bsdunzip->argc = argc; while ((opt = bsdunzip_getopt(bsdunzip)) != -1) { unzip_exclude_mode = 0; switch (opt) { case 'a': a_opt = 1; break; case 'C': C_opt = 1; break; case 'c': c_opt = 1; break; case 'd': d_arg = bsdunzip->argument; break; case 'f': f_opt = 1; break; case 'I': case 'O': O_arg = bsdunzip->argument; break; case 'j': j_opt = 1; break; case 'L': L_opt = 1; break; case 'l': if (v_opt == 0) v_opt = 1; break; case 'n': n_opt = 1; break; case 'o': o_opt = 1; q_opt = 1; break; case 'p': p_opt = 1; break; case 'P': P_arg = bsdunzip->argument; break; case 'q': q_opt = 1; break; case 't': t_opt = 1; break; case 'u': u_opt = 1; break; case 'v': v_opt = 2; break; case 'x': add_pattern(&exclude, bsdunzip->argument); unzip_exclude_mode = 1; break; case 'y': y_str = " "; break; case 'Z': zipinfo_mode = 1; if (bsdunzip->argument != NULL && strcmp(bsdunzip->argument, "1") == 0) { Z1_opt = 1; } break; case OPTION_VERSION: version_opt = 1; break; case OPTION_NONE: break; default: usage(); } if (opt == OPTION_NONE) break; } return (bsdunzip_optind); } int main(int argc, char *argv[]) { const char *zipfile; int nopts; +#if defined(HAVE_SIGACTION) && defined(SIGCHLD) + { /* Do not ignore SIGCHLD. */ + struct sigaction sa; + sa.sa_handler = SIG_DFL; + sigemptyset(&sa.sa_mask); + sa.sa_flags = 0; + sigaction(SIGCHLD, &sa, NULL); + } +#endif + lafe_setprogname(*argv, "bsdunzip"); #if HAVE_SETLOCALE if (setlocale(LC_ALL, "") == NULL) lafe_warnc(0, "Failed to set default locale"); #endif if (isatty(STDOUT_FILENO)) tty = 1; if (getenv("UNZIP_DEBUG") != NULL) unzip_debug = 1; for (int i = 0; i < argc; ++i) debug("%s%c", argv[i], (i < argc - 1) ? ' ' : '\n'); #ifdef __GLIBC__ /* Prevent GNU getopt(3) from rearranging options. */ setenv("POSIXLY_CORRECT", "", 1); #endif /* * Info-ZIP's unzip(1) expects certain options to come before the * zipfile name, and others to come after - though it does not * enforce this. For simplicity, we accept *all* options both * before and after the zipfile name. */ nopts = getopts(argc, argv); if (version_opt == 1) version(); /* * When more of the zipinfo mode options are implemented, this * will need to change. */ if (zipinfo_mode && !Z1_opt) { printf("Zipinfo mode needs additional options\n"); exit(EXIT_FAILURE); } if (argc <= nopts) usage(); zipfile = argv[nopts++]; if (strcmp(zipfile, "-") == 0) zipfile = NULL; /* STDIN */ unzip_exclude_mode = 0; while (nopts < argc && *argv[nopts] != '-') add_pattern(&include, argv[nopts++]); nopts--; /* fake argv[0] */ nopts += getopts(argc - nopts, argv + nopts); /* * For compatibility with Info-ZIP's unzip(1) we need to treat * non-option arguments following an -x after the zipfile as * exclude list members. */ if (unzip_exclude_mode) { while (nopts < argc && *argv[nopts] != '-') add_pattern(&exclude, argv[nopts++]); nopts--; /* fake argv[0] */ nopts += getopts(argc - nopts, argv + nopts); } /* There may be residual arguments if we encountered -- */ while (nopts < argc) add_pattern(&include, argv[nopts++]); if (n_opt + o_opt + u_opt > 1) errorx("-n, -o and -u are contradictory"); unzip(zipfile); exit(EXIT_SUCCESS); } diff --git a/contrib/libarchive/unzip/cmdline.c b/contrib/libarchive/unzip/cmdline.c index 8afddedd5ead..8e6eedd2321c 100644 --- a/contrib/libarchive/unzip/cmdline.c +++ b/contrib/libarchive/unzip/cmdline.c @@ -1,235 +1,235 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2003-2008 Tim Kientzle * All rights reserved. */ /* * Command line parser for bsdunzip. */ #include "bsdunzip_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "bsdunzip.h" -#include "err.h" +#include "lafe_err.h" /* * Short options for bsdunzip. Please keep this sorted. */ static const char *short_options = "aCcd:fI:jLlnO:opP:qtuvx:yZ:"; /* * Long options for bsdunzip. Please keep this list sorted. * * The symbolic names for options that lack a short equivalent are * defined in bsdunzip.h. Also note that so far I've found no need * to support optional arguments to long options. That would be * a small change to the code below. */ static const struct bsdunzip_option { const char *name; int required; /* 1 if this option requires an argument. */ int equivalent; /* Equivalent short option. */ } bsdunzip_longopts[] = { { "version", 0, OPTION_VERSION }, { NULL, 0, 0 } }; /* * This getopt implementation has two key features that common * getopt_long() implementations lack. Apart from those, it's a * straightforward option parser, considerably simplified by not * needing to support the wealth of exotic getopt_long() features. It * has, of course, been shamelessly tailored for bsdunzip. (If you're * looking for a generic getopt_long() implementation for your * project, I recommend Gregory Pietsch's public domain getopt_long() * implementation.) The two additional features are: */ int bsdunzip_getopt(struct bsdunzip *bsdunzip) { enum { state_start = 0, state_next_word, state_short, state_long }; const struct bsdunzip_option *popt, *match, *match2; const char *p, *long_prefix; size_t optlength; int opt; int required; again: match = NULL; match2 = NULL; long_prefix = "--"; opt = OPTION_NONE; required = 0; bsdunzip->argument = NULL; /* First time through, initialize everything. */ if (bsdunzip->getopt_state == state_start) { /* Skip program name. */ ++bsdunzip->argv; --bsdunzip->argc; if (*bsdunzip->argv == NULL) return (-1); bsdunzip->getopt_state = state_next_word; } /* * We're ready to look at the next word in argv. */ if (bsdunzip->getopt_state == state_next_word) { /* No more arguments, so no more options. */ if (bsdunzip->argv[0] == NULL) return (-1); /* Doesn't start with '-', so no more options. */ if (bsdunzip->argv[0][0] != '-') return (-1); /* "--" marks end of options; consume it and return. */ if (strcmp(bsdunzip->argv[0], "--") == 0) { ++bsdunzip->argv; --bsdunzip->argc; bsdunzip_optind++; return (-1); } /* Get next word for parsing. */ bsdunzip->getopt_word = *bsdunzip->argv++; --bsdunzip->argc; bsdunzip_optind++; if (bsdunzip->getopt_word[1] == '-') { /* Set up long option parser. */ bsdunzip->getopt_state = state_long; bsdunzip->getopt_word += 2; /* Skip leading '--' */ } else { /* Set up short option parser. */ bsdunzip->getopt_state = state_short; ++bsdunzip->getopt_word; /* Skip leading '-' */ } } /* * We're parsing a group of POSIX-style single-character options. */ if (bsdunzip->getopt_state == state_short) { /* Peel next option off of a group of short options. */ opt = *bsdunzip->getopt_word++; if (opt == '\0') { /* End of this group; recurse to get next option. */ bsdunzip->getopt_state = state_next_word; goto again; } /* Does this option take an argument? */ p = strchr(short_options, opt); if (p == NULL) return ('?'); if (p[1] == ':') required = 1; /* If it takes an argument, parse that. */ if (required) { /* If arg is run-in, bsdunzip->getopt_word already points to it. */ if (bsdunzip->getopt_word[0] == '\0') { /* Otherwise, pick up the next word. */ bsdunzip->getopt_word = *bsdunzip->argv; if (bsdunzip->getopt_word == NULL) { lafe_warnc(0, "Option -%c requires an argument", opt); return ('?'); } ++bsdunzip->argv; --bsdunzip->argc; bsdunzip_optind++; } bsdunzip->getopt_state = state_next_word; bsdunzip->argument = bsdunzip->getopt_word; } } /* We're reading a long option */ if (bsdunzip->getopt_state == state_long) { /* After this long option, we'll be starting a new word. */ bsdunzip->getopt_state = state_next_word; /* Option name ends at '=' if there is one. */ p = strchr(bsdunzip->getopt_word, '='); if (p != NULL) { optlength = (size_t)(p - bsdunzip->getopt_word); bsdunzip->argument = (char *)(uintptr_t)(p + 1); } else { optlength = strlen(bsdunzip->getopt_word); } /* Search the table for an unambiguous match. */ for (popt = bsdunzip_longopts; popt->name != NULL; popt++) { /* Short-circuit if first chars don't match. */ if (popt->name[0] != bsdunzip->getopt_word[0]) continue; /* If option is a prefix of name in table, record it.*/ if (strncmp(bsdunzip->getopt_word, popt->name, optlength) == 0) { match2 = match; /* Record up to two matches. */ match = popt; /* If it's an exact match, we're done. */ if (strlen(popt->name) == optlength) { match2 = NULL; /* Forget the others. */ break; } } } /* Fail if there wasn't a unique match. */ if (match == NULL) { lafe_warnc(0, "Option %s%s is not supported", long_prefix, bsdunzip->getopt_word); return ('?'); } if (match2 != NULL) { lafe_warnc(0, "Ambiguous option %s%s (matches --%s and --%s)", long_prefix, bsdunzip->getopt_word, match->name, match2->name); return ('?'); } /* We've found a unique match; does it need an argument? */ if (match->required) { /* Argument required: get next word if necessary. */ if (bsdunzip->argument == NULL) { bsdunzip->argument = *bsdunzip->argv; if (bsdunzip->argument == NULL) { lafe_warnc(0, "Option %s%s requires an argument", long_prefix, match->name); return ('?'); } ++bsdunzip->argv; --bsdunzip->argc; bsdunzip_optind++; } } else { /* Argument forbidden: fail if there is one. */ if (bsdunzip->argument != NULL) { lafe_warnc(0, "Option %s%s does not allow an argument", long_prefix, match->name); return ('?'); } } return (match->equivalent); } return (opt); } diff --git a/contrib/libarchive/unzip/test/test_C.c b/contrib/libarchive/unzip/test/test_C.c index 66835c8406f3..d386bd61e85d 100644 --- a/contrib/libarchive/unzip/test/test_C.c +++ b/contrib/libarchive/unzip/test/test_C.c @@ -1,22 +1,26 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test C arg - match case-insensitive */ DEFINE_TEST(test_C) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s -C %s test_basic/caps >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("contents CAPS\n", "test_basic/CAPS"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_L.c b/contrib/libarchive/unzip/test/test_L.c index 5b004d5d5707..85b43f592542 100644 --- a/contrib/libarchive/unzip/test/test_L.c +++ b/contrib/libarchive/unzip/test/test_L.c @@ -1,25 +1,29 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test L arg - make names lowercase */ DEFINE_TEST(test_L) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s -L %s >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("contents a\n", "test_basic/a"); assertTextFileContents("contents b\n", "test_basic/b"); assertTextFileContents("contents c\n", "test_basic/c"); assertTextFileContents("contents CAPS\n", "test_basic/caps"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_P_encryption.c b/contrib/libarchive/unzip/test/test_P_encryption.c index e2cf3cda0ebd..ed08a068811b 100644 --- a/contrib/libarchive/unzip/test/test_P_encryption.c +++ b/contrib/libarchive/unzip/test/test_P_encryption.c @@ -1,22 +1,26 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test P arg - password protected */ DEFINE_TEST(test_P_encryption) { const char *reffile = "test_encrypted.zip"; int r; extract_reference_file(reffile); r = systemf("%s -P password %s >test.out 2>test.err", testprog, reffile); - assertEqualInt(0, r); - assertNonEmptyFile("test.out"); - assertEmptyFile("test.err"); + if (r == 256) { + assertTextFileContents("unzip: Decryption is unsupported due to lack of crypto library\n", "test.err"); + } else { + assertEqualInt(0, r); + assertNonEmptyFile("test.out"); + assertEmptyFile("test.err"); - assertTextFileContents("plaintext\n", "encrypted/file.txt"); + assertTextFileContents("plaintext\n", "encrypted/file.txt"); + } } diff --git a/contrib/libarchive/unzip/test/test_basic.c b/contrib/libarchive/unzip/test/test_basic.c index 1f37dcd416a3..3a884aa30e5d 100644 --- a/contrib/libarchive/unzip/test/test_basic.c +++ b/contrib/libarchive/unzip/test/test_basic.c @@ -1,25 +1,29 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* This test just does a basic zip decompression */ DEFINE_TEST(test_basic) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s %s >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("contents a\n", "test_basic/a"); assertTextFileContents("contents b\n", "test_basic/b"); assertTextFileContents("contents c\n", "test_basic/c"); assertTextFileContents("contents CAPS\n", "test_basic/CAPS"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_d.c b/contrib/libarchive/unzip/test/test_d.c index ea67246207f2..cd7c3dfd97ad 100644 --- a/contrib/libarchive/unzip/test/test_d.c +++ b/contrib/libarchive/unzip/test/test_d.c @@ -1,43 +1,51 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test d arg - extract to target dir - before zipfile argument */ DEFINE_TEST(test_d_before_zipfile) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s -d foobar %s >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("contents a\n", "foobar/test_basic/a"); assertTextFileContents("contents b\n", "foobar/test_basic/b"); assertTextFileContents("contents c\n", "foobar/test_basic/c"); assertTextFileContents("contents CAPS\n", "foobar/test_basic/CAPS"); +#else + skipping("zlib not available"); +#endif } /* Test d arg - extract to target dir - after zipfile argument */ DEFINE_TEST(test_d_after_zipfile) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s %s -d foobar >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("contents a\n", "foobar/test_basic/a"); assertTextFileContents("contents b\n", "foobar/test_basic/b"); assertTextFileContents("contents c\n", "foobar/test_basic/c"); assertTextFileContents("contents CAPS\n", "foobar/test_basic/CAPS"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_doubledash.c b/contrib/libarchive/unzip/test/test_doubledash.c index 4467213dbb89..db0445ec3c24 100644 --- a/contrib/libarchive/unzip/test/test_doubledash.c +++ b/contrib/libarchive/unzip/test/test_doubledash.c @@ -1,25 +1,29 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test double dash arg - swallow "--" and use next argument as file name */ DEFINE_TEST(test_doubledash) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s -- %s >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("contents a\n", "test_basic/a"); assertTextFileContents("contents b\n", "test_basic/b"); assertTextFileContents("contents c\n", "test_basic/c"); assertTextFileContents("contents CAPS\n", "test_basic/CAPS"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_glob.c b/contrib/libarchive/unzip/test/test_glob.c index b53aa16fd85c..589ff1c55ef3 100644 --- a/contrib/libarchive/unzip/test/test_glob.c +++ b/contrib/libarchive/unzip/test/test_glob.c @@ -1,25 +1,29 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test that the glob works */ DEFINE_TEST(test_glob) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s %s test_*/[ab] >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("contents a\n", "test_basic/a"); assertTextFileContents("contents b\n", "test_basic/b"); assertFileNotExists("test_basic/c"); assertFileNotExists("test_basic/CAPS"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_j.c b/contrib/libarchive/unzip/test/test_j.c index b87229f42e25..1fba8ca207ec 100644 --- a/contrib/libarchive/unzip/test/test_j.c +++ b/contrib/libarchive/unzip/test/test_j.c @@ -1,25 +1,29 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test j arg - don't make directories */ DEFINE_TEST(test_j) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s -j %s >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("contents a\n", "a"); assertTextFileContents("contents b\n", "b"); assertTextFileContents("contents c\n", "c"); assertTextFileContents("contents CAPS\n", "CAPS"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_n.c b/contrib/libarchive/unzip/test/test_n.c index bb75c5d7696d..a13623ce23df 100644 --- a/contrib/libarchive/unzip/test/test_n.c +++ b/contrib/libarchive/unzip/test/test_n.c @@ -1,29 +1,33 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test n arg - don't overwrite existing files */ DEFINE_TEST(test_n) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; assertMakeDir("test_basic", 0755); assertMakeFile("test_basic/a", 0644, "orig a\n"); assertMakeFile("test_basic/b", 0644, "orig b\n"); extract_reference_file(reffile); r = systemf("%s -n %s >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("orig a\n", "test_basic/a"); assertTextFileContents("orig b\n", "test_basic/b"); assertTextFileContents("contents c\n", "test_basic/c"); assertTextFileContents("contents CAPS\n", "test_basic/CAPS"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_o.c b/contrib/libarchive/unzip/test/test_o.c index 64f946774440..8c48348c41a4 100644 --- a/contrib/libarchive/unzip/test/test_o.c +++ b/contrib/libarchive/unzip/test/test_o.c @@ -1,28 +1,32 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test o arg - overwrite existing files */ DEFINE_TEST(test_o) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; assertMakeDir("test_basic", 0755); assertMakeFile("test_basic/a", 0644, "orig a\n"); assertMakeFile("test_basic/b", 0644, "orig b\n"); extract_reference_file(reffile); r = systemf("%s -o %s >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertEmptyFile("test.err"); assertTextFileContents("contents a\n", "test_basic/a"); assertTextFileContents("contents b\n", "test_basic/b"); assertTextFileContents("contents c\n", "test_basic/c"); assertTextFileContents("contents CAPS\n", "test_basic/CAPS"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_p.c b/contrib/libarchive/unzip/test/test_p.c index 8bfffbe5dc39..13a7765463ec 100644 --- a/contrib/libarchive/unzip/test/test_p.c +++ b/contrib/libarchive/unzip/test/test_p.c @@ -1,20 +1,24 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test p arg - Print to stdout */ DEFINE_TEST(test_p) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s -p %s >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertTextFileContents("contents a\ncontents b\ncontents c\ncontents CAPS\n", "test.out"); assertEmptyFile("test.err"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_q.c b/contrib/libarchive/unzip/test/test_q.c index 13222a483992..0579e8028d76 100644 --- a/contrib/libarchive/unzip/test/test_q.c +++ b/contrib/libarchive/unzip/test/test_q.c @@ -1,25 +1,29 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test q arg - Quiet */ DEFINE_TEST(test_q) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s -q %s >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("contents a\n", "test_basic/a"); assertTextFileContents("contents b\n", "test_basic/b"); assertTextFileContents("contents c\n", "test_basic/c"); assertTextFileContents("contents CAPS\n", "test_basic/CAPS"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_singlefile.c b/contrib/libarchive/unzip/test/test_singlefile.c index a72811f046d9..a5a35ecacc4d 100644 --- a/contrib/libarchive/unzip/test/test_singlefile.c +++ b/contrib/libarchive/unzip/test/test_singlefile.c @@ -1,22 +1,26 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Ensure single-file zips work */ DEFINE_TEST(test_singlefile) { +#ifdef HAVE_LIBZ const char *reffile = "test_singlefile.zip"; int r; extract_reference_file(reffile); r = systemf("%s %s >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("hello\n", "file.txt"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_t.c b/contrib/libarchive/unzip/test/test_t.c index 55a516fc636f..7565830915c3 100644 --- a/contrib/libarchive/unzip/test/test_t.c +++ b/contrib/libarchive/unzip/test/test_t.c @@ -1,20 +1,24 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test t arg - Test zip contents */ DEFINE_TEST(test_t) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s -t %s >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); +#else + skipping("zlib not available"); +#endif } diff --git a/contrib/libarchive/unzip/test/test_x.c b/contrib/libarchive/unzip/test/test_x.c index 959beb1950df..43a2085dc5b2 100644 --- a/contrib/libarchive/unzip/test/test_x.c +++ b/contrib/libarchive/unzip/test/test_x.c @@ -1,61 +1,73 @@ /* * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2023 Adrian Vovk * All rights reserved. */ #include "test.h" /* Test x arg with single exclude path */ DEFINE_TEST(test_x_single) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s %s -x test_basic/c >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("contents a\n", "test_basic/a"); assertTextFileContents("contents b\n", "test_basic/b"); assertFileNotExists("test_basic/c"); assertTextFileContents("contents CAPS\n", "test_basic/CAPS"); +#else + skipping("zlib not available"); +#endif } /* Test x arg with multiple exclude paths */ DEFINE_TEST(test_x_multiple) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s %s -x test_basic/c test_basic/b >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("contents a\n", "test_basic/a"); assertFileNotExists("test_basic/b"); assertFileNotExists("test_basic/c"); assertTextFileContents("contents CAPS\n", "test_basic/CAPS"); +#else + skipping("zlib not available"); +#endif } /* Test x arg with multiple exclude paths and a d arg afterwards */ DEFINE_TEST(test_x_multiple_with_d) { +#ifdef HAVE_LIBZ const char *reffile = "test_basic.zip"; int r; extract_reference_file(reffile); r = systemf("%s %s -x test_basic/c test_basic/b -d foobar >test.out 2>test.err", testprog, reffile); assertEqualInt(0, r); assertNonEmptyFile("test.out"); assertEmptyFile("test.err"); assertTextFileContents("contents a\n", "foobar/test_basic/a"); assertFileNotExists("foobar/test_basic/b"); assertFileNotExists("foobar/test_basic/c"); assertTextFileContents("contents CAPS\n", "foobar/test_basic/CAPS"); +#else + skipping("zlib not available"); +#endif } diff --git a/lib/libarchive/tests/Makefile b/lib/libarchive/tests/Makefile index 07c5fe24dd30..930250d974c5 100644 --- a/lib/libarchive/tests/Makefile +++ b/lib/libarchive/tests/Makefile @@ -1,742 +1,748 @@ .include PACKAGE= tests WARNS?= 3 _LIBARCHIVEDIR= ${SRCTOP}/contrib/libarchive ATF_TESTS_SH+= functional_test TEST_METADATA.functional_test+= timeout="600" BINDIR= ${TESTSDIR} PROGS+= libarchive_test CFLAGS+= -I${.CURDIR} -I${.CURDIR:H} -I${.OBJDIR} CFLAGS+= -I${_LIBARCHIVEDIR}/libarchive -I${_LIBARCHIVEDIR}/libarchive/test CFLAGS+= -I${_LIBARCHIVEDIR}/test_utils CFLAGS+= -I${SRCTOP}/sys/contrib/zstd/lib CFLAGS.test_utils.c+= -Wno-cast-align # Uncomment to link against dmalloc #LDADD+= -L/usr/local/lib -ldmalloc #CFLAGS+= -I/usr/local/include -DUSE_DMALLOC .PATH: ${_LIBARCHIVEDIR}/libarchive/test TESTS_SRCS= \ test_7zip_filename_encoding.c \ test_acl_nfs4.c \ test_acl_pax.c \ test_acl_platform_nfs4.c \ test_acl_platform_posix1e.c \ test_acl_posix1e.c \ test_acl_text.c \ test_ar_mode.c \ test_archive_api_feature.c \ test_archive_clear_error.c \ test_archive_cmdline.c \ test_archive_digest.c \ test_archive_match_time.c \ test_archive_match_owner.c \ test_archive_match_path.c \ test_archive_pathmatch.c \ test_archive_parse_date.c \ test_archive_read.c \ test_archive_read_add_passphrase.c \ test_archive_read_close_twice.c \ test_archive_read_close_twice_open_fd.c \ test_archive_read_close_twice_open_filename.c \ test_archive_read_multiple_data_objects.c \ test_archive_read_next_header_empty.c \ test_archive_read_next_header_raw.c \ test_archive_read_open2.c \ test_archive_read_set_filter_option.c \ test_archive_read_set_format_option.c \ test_archive_read_set_option.c \ test_archive_read_set_options.c \ test_archive_read_support.c \ test_archive_set_error.c \ test_archive_string.c \ test_archive_string_conversion.c \ test_archive_write_add_filter_by_name.c \ test_archive_write_set_filter_option.c \ test_archive_write_set_format_by_name.c \ test_archive_write_set_format_filter_by_ext.c \ test_archive_write_set_format_option.c \ test_archive_write_set_option.c \ test_archive_write_set_options.c \ test_archive_write_set_passphrase.c \ test_bad_fd.c \ test_compat_bzip2.c \ test_compat_cpio.c \ test_compat_gtar.c \ test_compat_gtar_large.c \ test_compat_gzip.c \ test_compat_lz4.c \ test_compat_lzip.c \ test_compat_lzma.c \ test_compat_lzop.c \ test_compat_mac.c \ test_compat_perl_archive_tar.c \ test_compat_plexus_archiver_tar.c \ test_compat_solaris_tar_acl.c \ test_compat_solaris_pax_sparse.c \ test_compat_star_acl.c \ test_compat_tar_directory.c \ test_compat_tar_hardlink.c \ test_compat_uudecode.c \ test_compat_uudecode_large.c \ test_compat_xz.c \ test_compat_zip.c \ test_compat_zstd.c \ test_empty_write.c \ test_entry.c \ test_entry_strmode.c \ test_extattr_freebsd.c \ test_filter_count.c \ test_gnutar_filename_encoding.c \ test_link_resolver.c \ test_open_fd.c \ test_open_failure.c \ test_open_file.c \ test_open_filename.c \ test_pax_filename_encoding.c \ test_pax_xattr_header.c \ test_read_data_large.c \ test_read_disk.c \ test_read_disk_directory_traversals.c \ test_read_disk_entry_from_file.c \ test_read_extract.c \ test_read_file_nonexistent.c \ test_read_filter_compress.c \ test_read_filter_grzip.c \ test_read_filter_gzip_recursive.c \ test_read_filter_lrzip.c \ test_read_filter_lzop.c \ test_read_filter_lzop_multiple_parts.c \ test_read_filter_program.c \ test_read_filter_program_signature.c \ test_read_filter_uudecode.c \ test_read_filter_uudecode_raw.c \ test_read_format_7zip.c \ test_read_format_7zip_encryption_data.c \ test_read_format_7zip_encryption_header.c \ test_read_format_7zip_encryption_partially.c \ test_read_format_7zip_malformed.c \ test_read_format_7zip_packinfo_digests.c \ test_read_format_ar.c \ test_read_format_cab.c \ test_read_format_cab_filename.c \ test_read_format_cpio_afio.c \ test_read_format_cpio_bin.c \ test_read_format_cpio_bin_Z.c \ test_read_format_cpio_bin_be.c \ test_read_format_cpio_bin_bz2.c \ test_read_format_cpio_bin_gz.c \ test_read_format_cpio_bin_le.c \ test_read_format_cpio_bin_lzip.c \ test_read_format_cpio_bin_lzma.c \ test_read_format_cpio_bin_xz.c \ test_read_format_cpio_filename.c \ test_read_format_cpio_odc.c \ test_read_format_cpio_svr4_gzip.c \ test_read_format_cpio_svr4c_Z.c \ test_read_format_cpio_svr4_bzip2_rpm.c \ test_read_format_cpio_svr4_gzip_rpm.c \ test_read_format_empty.c \ test_read_format_gtar_filename.c \ test_read_format_gtar_gz.c \ test_read_format_gtar_lzma.c \ test_read_format_gtar_redundant_L.c \ test_read_format_gtar_sparse.c \ test_read_format_gtar_sparse_length.c \ test_read_format_gtar_sparse_skip_entry.c \ test_read_format_huge_rpm.c \ test_read_format_iso_Z.c \ test_read_format_iso_multi_extent.c \ test_read_format_iso_xorriso.c \ test_read_format_isorr_rr_moved.c \ test_read_format_isojoliet_bz2.c \ test_read_format_isojoliet_long.c \ test_read_format_isojoliet_rr.c \ test_read_format_isojoliet_versioned.c \ test_read_format_isorr_bz2.c \ test_read_format_isorr_ce.c \ test_read_format_isorr_new_bz2.c \ test_read_format_isozisofs_bz2.c \ test_read_format_lha.c \ test_read_format_lha_bugfix_0.c \ test_read_format_lha_filename.c \ test_read_format_lha_filename_utf16.c \ test_read_format_mtree.c \ test_read_format_mtree_crash747.c \ test_read_format_pax_bz2.c \ test_read_format_rar.c \ test_read_format_rar5.c \ test_read_format_rar_encryption.c \ test_read_format_rar_encryption_data.c \ test_read_format_rar_encryption_header.c \ test_read_format_rar_encryption_partially.c \ test_read_format_rar_filter.c \ test_read_format_rar_invalid1.c \ test_read_format_rar_overflow.c \ test_read_format_raw.c \ test_read_format_tar.c \ + test_read_format_tar_V_negative_size.c \ test_read_format_tar_concatenated.c \ test_read_format_tar_empty_filename.c \ test_read_format_tar_empty_pax.c \ test_read_format_tar_empty_with_gnulabel.c \ test_read_format_tar_filename.c \ test_read_format_tar_invalid_pax_size.c \ test_read_format_tar_mac_metadata.c \ test_read_format_tar_pax_g_large.c \ test_read_format_tar_pax_large_attr.c \ test_read_format_tar_pax_negative_time.c \ test_read_format_tbz.c \ test_read_format_tgz.c \ test_read_format_tlz.c \ test_read_format_txz.c \ test_read_format_tz.c \ test_read_format_ustar_filename.c \ test_read_format_warc.c \ test_read_format_xar.c \ test_read_format_xar_doublelink.c \ test_read_format_zip.c \ test_read_format_zip_7075_utf8_paths.c \ test_read_format_zip_comment_stored.c \ test_read_format_zip_encryption_data.c \ test_read_format_zip_encryption_header.c \ test_read_format_zip_encryption_partially.c \ test_read_format_zip_extra_padding.c \ test_read_format_zip_filename.c \ test_read_format_zip_high_compression.c \ test_read_format_zip_jar.c \ test_read_format_zip_mac_metadata.c \ test_read_format_zip_malformed.c \ test_read_format_zip_msdos.c \ test_read_format_zip_nested.c \ test_read_format_zip_nofiletype.c \ test_read_format_zip_padded.c \ test_read_format_zip_sfx.c \ test_read_format_zip_traditional_encryption_data.c \ test_read_format_zip_winzip_aes.c \ test_read_format_zip_winzip_aes_large.c \ test_read_format_zip_with_invalid_traditional_eocd.c \ test_read_format_zip_zip64.c \ test_read_large.c \ test_read_pax_empty_val_no_nl.c \ test_read_pax_xattr_rht_security_selinux.c \ test_read_pax_xattr_schily.c \ test_read_pax_truncated.c \ test_read_position.c \ test_read_set_format.c \ test_read_too_many_filters.c \ test_read_truncated.c \ test_read_truncated_filter.c \ test_short_writes.c \ test_sparse_basic.c \ test_tar_filenames.c \ test_tar_large.c \ test_warn_missing_hardlink_target.c \ test_ustar_filenames.c \ test_ustar_filename_encoding.c \ test_write_disk.c \ test_write_disk_appledouble.c \ test_write_disk_failures.c \ test_write_disk_fixup.c \ test_write_disk_hardlink.c \ test_write_disk_hfs_compression.c \ test_write_disk_lookup.c \ test_write_disk_mac_metadata.c \ test_write_disk_no_hfs_compression.c \ test_write_disk_perms.c \ test_write_disk_secure.c \ test_write_disk_secure_noabsolutepaths.c \ test_write_disk_secure744.c \ test_write_disk_secure745.c \ test_write_disk_secure746.c \ test_write_disk_sparse.c \ test_write_disk_symlink.c \ test_write_disk_times.c \ test_write_filter_b64encode.c \ test_write_filter_bzip2.c \ test_write_filter_compress.c \ test_write_filter_gzip.c \ test_write_filter_gzip_timestamp.c \ test_write_filter_lrzip.c \ test_write_filter_lz4.c \ test_write_filter_lzip.c \ test_write_filter_lzma.c \ test_write_filter_lzop.c \ test_write_filter_program.c \ test_write_filter_uuencode.c \ test_write_filter_xz.c \ test_write_filter_zstd.c \ test_write_format_7zip.c \ test_write_format_7zip_empty.c \ test_write_format_7zip_large.c \ test_write_format_ar.c \ test_write_format_cpio.c \ test_write_format_cpio_empty.c \ test_write_format_cpio_newc.c \ test_write_format_cpio_odc.c \ test_write_format_gnutar.c \ test_write_format_gnutar_filenames.c \ test_write_format_iso9660.c \ test_write_format_iso9660_boot.c \ test_write_format_iso9660_empty.c \ test_write_format_iso9660_filename.c \ test_write_format_iso9660_zisofs.c \ test_write_format_mtree.c \ test_write_format_mtree_absolute_path.c \ test_write_format_mtree_classic.c \ test_write_format_mtree_classic_indent.c \ test_write_format_mtree_fflags.c \ test_write_format_mtree_no_separator.c \ test_write_format_mtree_preset_digests.c \ test_write_format_mtree_quoted_filename.c \ test_write_format_pax.c \ test_write_format_raw.c \ test_write_format_raw_b64.c \ test_write_format_shar_empty.c \ test_write_format_tar.c \ test_write_format_tar_empty.c \ test_write_format_tar_sparse.c \ test_write_format_tar_ustar.c \ test_write_format_tar_v7tar.c \ test_write_format_warc.c \ test_write_format_warc_empty.c \ test_write_format_xar.c \ test_write_format_xar_empty.c \ test_write_format_zip.c \ test_write_format_zip_compression_store.c \ test_write_format_zip_compression_zstd.c \ test_write_format_zip_compression_bzip2.c \ test_write_format_zip_compression_lzmaxz.c \ test_write_format_zip_empty.c \ test_write_format_zip_empty_zip64.c \ test_write_format_zip_entry_size_unset.c \ test_write_format_zip_file.c \ test_write_format_zip_file_zip64.c \ test_write_format_zip_large.c \ test_write_format_zip_zip64.c \ test_write_format_zip64_stream.c \ test_write_format_zip_stream.c \ test_write_format_zip_windows_path.c \ test_write_open_memory.c \ test_write_read_format_zip.c \ test_xattr_platform.c \ test_zip_filename_encoding.c # Deterministic failures: # Fails with `test_read_disk_directory_traversals.c:1094: File at has atime 886622, 1443306049 seconds ago` BROKEN_TESTS+= test_read_disk_directory_traversals .if 0 # test_fuzz.c is not a real test, but rather a simple fuzz-test using random(). # Since this is not a regression/unit test, we don't include it by default. # If you would still like to include it, comment out the `.if 0`. TEST_SCRCS+= test_fuzz.c ${PACKAGE}FILES+= test_fuzz.cab.uu ${PACKAGE}FILES+= test_fuzz.lzh.uu ${PACKAGE}FILES+= test_fuzz_1.iso.Z.uu # Non-deterministic failures: # (Times out?) [and] crashes BROKEN_TESTS+= test_fuzz_rar .endif # Build the test program. SRCS.libarchive_test= \ ${TESTS_SRCS} \ read_open_memory.c \ list.h LIBADD.libarchive_test= archive .PATH: ${_LIBARCHIVEDIR}/test_utils SRCS.libarchive_test+= test_main.c \ test_utils.c # list.h is just a list of all tests, as indicated by DEFINE_TEST macro lines list.h: ${TESTS_SRCS} Makefile @(cd ${_LIBARCHIVEDIR}/libarchive/test && \ grep -E -h ^DEFINE_TEST ${.ALLSRC:N*Makefile} | \ egrep -v '${BROKEN_TESTS:tW:C/ /|/g}') > ${.TARGET}.tmp @mv ${.TARGET}.tmp ${.TARGET} CLEANTESTS+= list.h list.h.tmp ${PACKAGE}FILES+= README ${PACKAGE}FILES+= test_acl_pax_posix1e.tar.uu ${PACKAGE}FILES+= test_acl_pax_nfs4.tar.uu ${PACKAGE}FILES+= test_archive_string_conversion.txt.Z.uu ${PACKAGE}FILES+= test_compat_bzip2_1.tbz.uu ${PACKAGE}FILES+= test_compat_bzip2_2.tbz.uu ${PACKAGE}FILES+= test_compat_cpio_1.cpio.uu ${PACKAGE}FILES+= test_compat_gtar_1.tar.uu ${PACKAGE}FILES+= test_compat_gtar_2.tar.uu ${PACKAGE}FILES+= test_compat_gzip_1.tgz.uu ${PACKAGE}FILES+= test_compat_gzip_2.tgz.uu ${PACKAGE}FILES+= test_compat_lz4_1.tar.lz4.uu ${PACKAGE}FILES+= test_compat_lz4_2.tar.lz4.uu ${PACKAGE}FILES+= test_compat_lz4_3.tar.lz4.uu ${PACKAGE}FILES+= test_compat_lz4_B4.tar.lz4.uu ${PACKAGE}FILES+= test_compat_lz4_B4BD.tar.lz4.uu ${PACKAGE}FILES+= test_compat_lz4_B4BDBX.tar.lz4.uu ${PACKAGE}FILES+= test_compat_lz4_B5.tar.lz4.uu ${PACKAGE}FILES+= test_compat_lz4_B5BD.tar.lz4.uu ${PACKAGE}FILES+= test_compat_lz4_B6.tar.lz4.uu ${PACKAGE}FILES+= test_compat_lz4_B6BD.tar.lz4.uu ${PACKAGE}FILES+= test_compat_lz4_B7.tar.lz4.uu ${PACKAGE}FILES+= test_compat_lz4_B7BD.tar.lz4.uu ${PACKAGE}FILES+= test_compat_lzip_1.tlz.uu ${PACKAGE}FILES+= test_compat_lzip_2.tlz.uu ${PACKAGE}FILES+= test_compat_lzip_3.lz.uu ${PACKAGE}FILES+= test_compat_lzip_4.tlz.uu ${PACKAGE}FILES+= test_compat_lzma_1.tlz.uu ${PACKAGE}FILES+= test_compat_lzma_2.tlz.uu ${PACKAGE}FILES+= test_compat_lzma_3.tlz.uu ${PACKAGE}FILES+= test_compat_lzop_1.tar.lzo.uu ${PACKAGE}FILES+= test_compat_lzop_2.tar.lzo.uu ${PACKAGE}FILES+= test_compat_lzop_3.tar.lzo.uu ${PACKAGE}FILES+= test_compat_mac-1.tar.Z.uu ${PACKAGE}FILES+= test_compat_mac-2.tar.Z.uu ${PACKAGE}FILES+= test_compat_perl_archive_tar.tar.uu ${PACKAGE}FILES+= test_compat_plexus_archiver_tar.tar.uu ${PACKAGE}FILES+= test_compat_solaris_pax_sparse_1.pax.Z.uu ${PACKAGE}FILES+= test_compat_solaris_pax_sparse_2.pax.Z.uu ${PACKAGE}FILES+= test_compat_solaris_tar_acl.tar.uu ${PACKAGE}FILES+= test_compat_star_acl_nfs4.tar.uu ${PACKAGE}FILES+= test_compat_star_acl_posix1e.tar.uu ${PACKAGE}FILES+= test_compat_tar_directory_1.tar.uu ${PACKAGE}FILES+= test_compat_tar_hardlink_1.tar.uu ${PACKAGE}FILES+= test_compat_uudecode_large.tar.Z.uu ${PACKAGE}FILES+= test_compat_xz_1.txz.uu ${PACKAGE}FILES+= test_compat_zip_1.zip.uu ${PACKAGE}FILES+= test_compat_zip_2.zip.uu ${PACKAGE}FILES+= test_compat_zip_3.zip.uu ${PACKAGE}FILES+= test_compat_zip_4.zip.uu ${PACKAGE}FILES+= test_compat_zip_5.zip.uu ${PACKAGE}FILES+= test_compat_zip_6.zip.uu ${PACKAGE}FILES+= test_compat_zip_7.xps.uu ${PACKAGE}FILES+= test_compat_zip_8.zip.uu ${PACKAGE}FILES+= test_compat_zstd_1.tar.zst.uu ${PACKAGE}FILES+= test_compat_zstd_2.tar.zst.uu ${PACKAGE}FILES+= test_pax_filename_encoding.tar.uu ${PACKAGE}FILES+= test_pax_xattr_header_all.tar.uu ${PACKAGE}FILES+= test_pax_xattr_header_libarchive.tar.uu ${PACKAGE}FILES+= test_pax_xattr_header_schily.tar.uu ${PACKAGE}FILES+= test_rar_multivolume_multiple_files.part1.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_multiple_files.part2.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_multiple_files.part3.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_multiple_files.part4.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_multiple_files.part5.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_multiple_files.part6.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_single_file.part1.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_single_file.part2.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_single_file.part3.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_uncompressed_files.part01.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_uncompressed_files.part02.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_uncompressed_files.part03.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_uncompressed_files.part04.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_uncompressed_files.part05.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_uncompressed_files.part06.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_uncompressed_files.part07.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_uncompressed_files.part08.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_uncompressed_files.part09.rar.uu ${PACKAGE}FILES+= test_rar_multivolume_uncompressed_files.part10.rar.uu ${PACKAGE}FILES+= test_read_filter_grzip.tar.grz.uu ${PACKAGE}FILES+= test_read_filter_gzip_recursive.gz.uu ${PACKAGE}FILES+= test_read_filter_lrzip.tar.lrz.uu ${PACKAGE}FILES+= test_read_filter_lzop.tar.lzo.uu ${PACKAGE}FILES+= test_read_filter_lzop_multiple_parts.tar.lzo.uu ${PACKAGE}FILES+= test_read_filter_uudecode_base64_raw.uu ${PACKAGE}FILES+= test_read_filter_uudecode_raw.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj2_bzip2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj2_copy_1.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj2_copy_2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj2_copy_lzma.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj2_deflate.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_deflate_arm64.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_deflate_powerpc.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj2_lzma1_1.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj2_lzma1_2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj2_lzma2_1.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj2_lzma2_2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj_bzip2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj_copy.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj_deflate.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj_lzma1.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bcj_lzma2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_bzip2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_copy.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_copy_2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_deflate.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_delta_lzma1.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_delta_lzma2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_delta4_lzma1.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_delta4_lzma2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_empty_archive.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_empty_file.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_encryption.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_encryption_header.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_encryption_partially.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_extract_second.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_lzma1.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_lzma1_2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_lzma1_lzma2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_lzma2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_lzma2_arm.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_lzma2_arm64.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_lzma2_powerpc.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_lzma2_riscv.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_lzma2_sparc.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_malformed.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_malformed2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_packinfo_digests.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_ppmd.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_sfx_elf.elf.uu ${PACKAGE}FILES+= test_read_format_7zip_sfx_modified_pe.exe.uu ${PACKAGE}FILES+= test_read_format_7zip_sfx_pe.exe.uu ${PACKAGE}FILES+= test_read_format_7zip_solid_zstd.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_symbolic_name.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_zstd.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_zstd_arm.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_zstd_bcj.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_zstd_nobcj.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_zstd_sparc.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_win_attrib.7z.uu ${PACKAGE}FILES+= test_read_format_ar.ar.uu ${PACKAGE}FILES+= test_read_format_cab_1.cab.uu ${PACKAGE}FILES+= test_read_format_cab_2.cab.uu ${PACKAGE}FILES+= test_read_format_cab_3.cab.uu ${PACKAGE}FILES+= test_read_format_cab_filename_cp932.cab.uu ${PACKAGE}FILES+= test_read_format_cpio_bin_be.cpio.uu ${PACKAGE}FILES+= test_read_format_cpio_bin_le.cpio.uu ${PACKAGE}FILES+= test_read_format_cpio_filename_cp866.cpio.uu ${PACKAGE}FILES+= test_read_format_cpio_filename_eucjp.cpio.uu ${PACKAGE}FILES+= test_read_format_cpio_filename_koi8r.cpio.uu ${PACKAGE}FILES+= test_read_format_cpio_filename_utf8_jp.cpio.uu ${PACKAGE}FILES+= test_read_format_cpio_filename_utf8_ru.cpio.uu ${PACKAGE}FILES+= test_read_format_cpio_svr4_bzip2_rpm.rpm.uu ${PACKAGE}FILES+= test_read_format_cpio_svr4_gzip_rpm.rpm.uu ${PACKAGE}FILES+= test_read_format_gtar_filename_cp866.tar.Z.uu ${PACKAGE}FILES+= test_read_format_gtar_filename_eucjp.tar.Z.uu ${PACKAGE}FILES+= test_read_format_gtar_filename_koi8r.tar.Z.uu ${PACKAGE}FILES+= test_read_format_gtar_redundant_L.tar.Z.uu ${PACKAGE}FILES+= test_read_format_gtar_sparse_1_13.tar.uu ${PACKAGE}FILES+= test_read_format_gtar_sparse_1_17.tar.uu ${PACKAGE}FILES+= test_read_format_gtar_sparse_1_17_posix00.tar.uu ${PACKAGE}FILES+= test_read_format_gtar_sparse_1_17_posix01.tar.uu ${PACKAGE}FILES+= test_read_format_gtar_sparse_1_17_posix10.tar.uu ${PACKAGE}FILES+= test_read_format_gtar_sparse_1_17_posix10_modified.tar.uu ${PACKAGE}FILES+= test_read_format_gtar_sparse_length.tar.Z.uu ${PACKAGE}FILES+= test_read_format_gtar_sparse_skip_entry.tar.Z.uu ${PACKAGE}FILES+= test_read_format_huge_rpm.rpm.uu ${PACKAGE}FILES+= test_read_format_iso.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_2.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_3.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_joliet.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_joliet_by_nero.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_joliet_long.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_joliet_rockridge.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_multi_extent.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_rockridge.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_rockridge_ce.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_rockridge_new.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_rockridge_rr_moved.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_xorriso.iso.Z.uu ${PACKAGE}FILES+= test_read_format_iso_zisofs.iso.Z.uu ${PACKAGE}FILES+= test_read_format_lha_bugfix_0.lzh.uu ${PACKAGE}FILES+= test_read_format_lha_filename_cp932.lzh.uu ${PACKAGE}FILES+= test_read_format_lha_filename_utf16.lzh.uu ${PACKAGE}FILES+= test_read_format_lha_header0.lzh.uu ${PACKAGE}FILES+= test_read_format_lha_header1.lzh.uu ${PACKAGE}FILES+= test_read_format_lha_header2.lzh.uu ${PACKAGE}FILES+= test_read_format_lha_header3.lzh.uu ${PACKAGE}FILES+= test_read_format_lha_lh0.lzh.uu ${PACKAGE}FILES+= test_read_format_lha_lh6.lzh.uu ${PACKAGE}FILES+= test_read_format_lha_lh7.lzh.uu ${PACKAGE}FILES+= test_read_format_lha_withjunk.lzh.uu ${PACKAGE}FILES+= test_read_format_mtree.mtree.uu ${PACKAGE}FILES+= test_read_format_mtree_crash747.mtree.bz2.uu ${PACKAGE}FILES+= test_read_format_mtree_nomagic.mtree.uu ${PACKAGE}FILES+= test_read_format_mtree_nomagic2.mtree.uu ${PACKAGE}FILES+= test_read_format_mtree_nomagic3.mtree.uu ${PACKAGE}FILES+= test_read_format_mtree_noprint.mtree.uu ${PACKAGE}FILES+= test_read_format_rar.rar.uu ${PACKAGE}FILES+= test_read_format_rar_binary_data.rar.uu ${PACKAGE}FILES+= test_read_format_rar_compress_best.rar.uu ${PACKAGE}FILES+= test_read_format_rar_compress_normal.rar.uu ${PACKAGE}FILES+= test_read_format_rar_encryption_data.rar.uu ${PACKAGE}FILES+= test_read_format_rar_encryption_header.rar.uu ${PACKAGE}FILES+= test_read_format_rar_encryption_partially.rar.uu ${PACKAGE}FILES+= test_read_format_rar_endarc_huge.rar.uu ${PACKAGE}FILES+= test_read_format_rar_filter.rar.uu ${PACKAGE}FILES+= test_read_format_rar_invalid1.rar.uu ${PACKAGE}FILES+= test_read_format_rar_multi_lzss_blocks.rar.uu ${PACKAGE}FILES+= test_read_format_rar_multivolume.part0001.rar.uu ${PACKAGE}FILES+= test_read_format_rar_multivolume.part0002.rar.uu ${PACKAGE}FILES+= test_read_format_rar_multivolume.part0003.rar.uu ${PACKAGE}FILES+= test_read_format_rar_multivolume.part0004.rar.uu ${PACKAGE}FILES+= test_read_format_rar_newsub_huge.rar.uu ${PACKAGE}FILES+= test_read_format_rar_noeof.rar.uu ${PACKAGE}FILES+= test_read_format_rar_overflow.rar.uu ${PACKAGE}FILES+= test_read_format_rar_ppmd_lzss_conversion.rar.uu ${PACKAGE}FILES+= test_read_format_rar_ppmd_use_after_free.rar.uu ${PACKAGE}FILES+= test_read_format_rar_ppmd_use_after_free2.rar.uu ${PACKAGE}FILES+= test_read_format_rar_sfx.exe.uu ${PACKAGE}FILES+= test_read_format_rar_subblock.rar.uu ${PACKAGE}FILES+= test_read_format_rar_symlink_huge.rar.uu ${PACKAGE}FILES+= test_read_format_rar_unicode.rar.uu ${PACKAGE}FILES+= test_read_format_rar_windows.rar.uu ${PACKAGE}FILES+= test_read_format_rar4_encrypted.rar.uu ${PACKAGE}FILES+= test_read_format_rar4_encrypted_filenames.rar.uu ${PACKAGE}FILES+= test_read_format_rar4_solid_encrypted.rar.uu ${PACKAGE}FILES+= test_read_format_rar4_solid_encrypted_filenames.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_arm.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_arm_filter_on_window_boundary.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_bad_window_sz_in_mltarc_file.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_data_ready_pointer_leak.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_blake2.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_block_size_is_too_small.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_compressed.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_decode_number_out_of_bounds_read.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_different_solid_window_size.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_different_window_size.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_different_winsize_on_merge.rar.uu +${PACKAGE}FILES+= test_read_format_rar5_dirdata.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_distance_overflow.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_encrypted.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_encrypted_filenames.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_solid_encrypted.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_solid_encrypted_filenames.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_extra_field_version.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_fileattr.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_hardlink.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_invalid_dict_reference.rar.uu +${PACKAGE}FILES+= test_read_format_rar5_invalid_hash_valid_htime_exfld.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_leftshift1.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_leftshift2.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiarchive.part01.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiarchive.part02.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiarchive.part03.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiarchive.part04.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiarchive.part05.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiarchive.part06.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiarchive.part07.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiarchive.part08.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiarchive_solid.part01.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiarchive_solid.part02.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiarchive_solid.part03.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiarchive_solid.part04.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiple_files.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_multiple_files_solid.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_nonempty_dir_stream.rar.uu +${PACKAGE}FILES+= test_read_format_rar5_only_crypt_exfld.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_owner.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_readtables_overflow.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_sfx.exe.uu ${PACKAGE}FILES+= test_read_format_rar5_solid.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_stored.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_stored_manyfiles.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_symlink.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_truncated_huff.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_unicode.rar.uu +${PACKAGE}FILES+= test_read_format_rar5_unsupported_exfld.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_win32.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_window_buf_and_size_desync.rar.uu ${PACKAGE}FILES+= test_read_format_raw.bufr.uu ${PACKAGE}FILES+= test_read_format_raw.data.Z.uu ${PACKAGE}FILES+= test_read_format_raw.data.gz.uu ${PACKAGE}FILES+= test_read_format_raw.data.uu +${PACKAGE}FILES+= test_read_format_tar_V_negative_size.tar.uu ${PACKAGE}FILES+= test_read_format_tar_concatenated.tar.uu ${PACKAGE}FILES+= test_read_format_tar_empty_filename.tar.uu ${PACKAGE}FILES+= test_read_format_tar_empty_with_gnulabel.tar.uu ${PACKAGE}FILES+= test_read_format_tar_empty_pax.tar.Z.uu ${PACKAGE}FILES+= test_read_format_tar_filename_koi8r.tar.Z.uu ${PACKAGE}FILES+= test_read_format_tar_invalid_pax_size.tar.uu ${PACKAGE}FILES+= test_read_format_tar_mac_metadata_1.tar.uu ${PACKAGE}FILES+= test_read_format_tar_pax_g_large.tar.uu ${PACKAGE}FILES+= test_read_format_tar_pax_large_attr.tar.Z.uu ${PACKAGE}FILES+= test_read_format_tar_pax_negative_time.tar.uu ${PACKAGE}FILES+= test_read_format_ustar_filename_cp866.tar.Z.uu ${PACKAGE}FILES+= test_read_format_ustar_filename_eucjp.tar.Z.uu ${PACKAGE}FILES+= test_read_format_ustar_filename_koi8r.tar.Z.uu ${PACKAGE}FILES+= test_read_format_warc.warc.uu ${PACKAGE}FILES+= test_read_format_warc_incomplete.warc.uu ${PACKAGE}FILES+= test_read_format_xar_doublelink.xar.uu ${PACKAGE}FILES+= test_read_format_xar_duplicate_filename_node.xar.uu ${PACKAGE}FILES+= test_read_format_zip.zip.uu ${PACKAGE}FILES+= test_read_format_zip_7075_utf8_paths.zip.uu ${PACKAGE}FILES+= test_read_format_zip_7z_deflate.zip.uu ${PACKAGE}FILES+= test_read_format_zip_7z_lzma.zip.uu ${PACKAGE}FILES+= test_read_format_zip_bz2_hang.zip.uu ${PACKAGE}FILES+= test_read_format_zip_bzip2.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_bzip2_multi.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_comment_stored_1.zip.uu ${PACKAGE}FILES+= test_read_format_zip_comment_stored_2.zip.uu ${PACKAGE}FILES+= test_read_format_zip_encryption_data.zip.uu ${PACKAGE}FILES+= test_read_format_zip_encryption_header.zip.uu ${PACKAGE}FILES+= test_read_format_zip_encryption_partially.zip.uu ${PACKAGE}FILES+= test_read_format_zip_extra_padding.zip.uu ${PACKAGE}FILES+= test_read_format_zip_filename_cp866.zip.uu ${PACKAGE}FILES+= test_read_format_zip_filename_cp932.zip.uu ${PACKAGE}FILES+= test_read_format_zip_filename_koi8r.zip.uu ${PACKAGE}FILES+= test_read_format_zip_filename_utf8_jp.zip.uu ${PACKAGE}FILES+= test_read_format_zip_filename_utf8_ru.zip.uu ${PACKAGE}FILES+= test_read_format_zip_filename_utf8_ru2.zip.uu ${PACKAGE}FILES+= test_read_format_zip_high_compression.zip.uu ${PACKAGE}FILES+= test_read_format_zip_jar.jar.uu ${PACKAGE}FILES+= test_read_format_zip_length_at_end.zip.uu ${PACKAGE}FILES+= test_read_format_zip_lzma_alone_leak.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_lzma.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_lzma_multi.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_lzma_stream_end.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_mac_metadata.zip.uu ${PACKAGE}FILES+= test_read_format_zip_malformed1.zip.uu ${PACKAGE}FILES+= test_read_format_zip_msdos.zip.uu ${PACKAGE}FILES+= test_read_format_zip_nested.zip.uu ${PACKAGE}FILES+= test_read_format_zip_nofiletype.zip.uu ${PACKAGE}FILES+= test_read_format_zip_padded1.zip.uu ${PACKAGE}FILES+= test_read_format_zip_padded2.zip.uu ${PACKAGE}FILES+= test_read_format_zip_padded3.zip.uu ${PACKAGE}FILES+= test_read_format_zip_ppmd8.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_ppmd8_crash_1.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_ppmd8_crash_2.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_ppmd8_multi.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_sfx.uu ${PACKAGE}FILES+= test_read_format_zip_symlink.zip.uu ${PACKAGE}FILES+= test_read_format_zip_traditional_encryption_data.zip.uu ${PACKAGE}FILES+= test_read_format_zip_ux.zip.uu ${PACKAGE}FILES+= test_read_format_zip_with_invalid_traditional_eocd.zip.uu ${PACKAGE}FILES+= test_read_format_zip_winzip_aes128.zip.uu ${PACKAGE}FILES+= test_read_format_zip_winzip_aes256.zip.uu ${PACKAGE}FILES+= test_read_format_zip_winzip_aes256_large.zip.uu ${PACKAGE}FILES+= test_read_format_zip_winzip_aes256_stored.zip.uu ${PACKAGE}FILES+= test_read_format_zip_xz_multi.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_zip64a.zip.uu ${PACKAGE}FILES+= test_read_format_zip_zip64b.zip.uu ${PACKAGE}FILES+= test_read_format_zip_zstd.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_zstd_multi.zipx.uu ${PACKAGE}FILES+= test_read_large_splitted_rar_aa.uu ${PACKAGE}FILES+= test_read_large_splitted_rar_ab.uu ${PACKAGE}FILES+= test_read_large_splitted_rar_ac.uu ${PACKAGE}FILES+= test_read_large_splitted_rar_ad.uu ${PACKAGE}FILES+= test_read_large_splitted_rar_ae.uu ${PACKAGE}FILES+= test_read_pax_empty_val_no_nl.tar.uu ${PACKAGE}FILES+= test_read_pax_xattr_rht_security_selinux.tar.uu ${PACKAGE}FILES+= test_read_pax_xattr_schily.tar.uu ${PACKAGE}FILES+= test_read_splitted_rar_aa.uu ${PACKAGE}FILES+= test_read_splitted_rar_ab.uu ${PACKAGE}FILES+= test_read_splitted_rar_ac.uu ${PACKAGE}FILES+= test_read_splitted_rar_ad.uu ${PACKAGE}FILES+= test_read_too_many_filters.gz.uu ${PACKAGE}FILES+= test_splitted_rar_seek_support_aa.uu ${PACKAGE}FILES+= test_splitted_rar_seek_support_ab.uu ${PACKAGE}FILES+= test_splitted_rar_seek_support_ac.uu ${PACKAGE}FILES+= test_write_disk_appledouble.cpio.gz.uu ${PACKAGE}FILES+= test_write_disk_appledouble_zip.zip.uu ${PACKAGE}FILES+= test_write_disk_hfs_compression.tgz.uu ${PACKAGE}FILES+= test_write_disk_mac_metadata.tar.gz.uu ${PACKAGE}FILES+= test_write_disk_no_hfs_compression.tgz.uu .include diff --git a/usr.bin/bsdcat/Makefile b/usr.bin/bsdcat/Makefile index 032207217be6..06081fc2b2f8 100644 --- a/usr.bin/bsdcat/Makefile +++ b/usr.bin/bsdcat/Makefile @@ -1,29 +1,29 @@ .include _LIBARCHIVEDIR= ${SRCTOP}/contrib/libarchive _LIBARCHIVECONFDIR= ${SRCTOP}/lib/libarchive PROG= bsdcat BSDCAT_VERSION_STRING!= sed -n '/define.*ARCHIVE_VERSION_ONLY_STRING/{s,[^0-9.],,gp;q;}' \ ${_LIBARCHIVEDIR}/libarchive/archive.h .PATH: ${_LIBARCHIVEDIR}/cat SRCS= bsdcat.c cmdline.c .PATH: ${_LIBARCHIVEDIR}/libarchive_fe -SRCS+= err.c +SRCS+= lafe_err.c CFLAGS+= -DBSDCAT_VERSION_STRING=\"${BSDCAT_VERSION_STRING}\" CFLAGS+= -DPLATFORM_CONFIG_H=\"${_LIBARCHIVECONFDIR}/config_freebsd.h\" CFLAGS+= -I${_LIBARCHIVEDIR}/cat -I${_LIBARCHIVEDIR}/libarchive_fe LIBADD= archive .if ${MK_ICONV} != "no" CFLAGS+= -DHAVE_ICONV=1 -DHAVE_ICONV_H=1 -DICONV_CONST=const .endif HAS_TESTS= SUBDIR.${MK_TESTS}+= tests .include diff --git a/usr.bin/cpio/Makefile b/usr.bin/cpio/Makefile index 46fe36d8c18e..31b25e4199da 100644 --- a/usr.bin/cpio/Makefile +++ b/usr.bin/cpio/Makefile @@ -1,32 +1,32 @@ .include _LIBARCHIVEDIR= ${SRCTOP}/contrib/libarchive _LIBARCHIVECONFDIR= ${SRCTOP}/lib/libarchive PROG= bsdcpio BSDCPIO_VERSION_STRING!= sed -n '/define.*ARCHIVE_VERSION_ONLY_STRING/{s,[^0-9.],,gp;q;}' \ ${_LIBARCHIVEDIR}/libarchive/archive.h .PATH: ${_LIBARCHIVEDIR}/cpio SRCS= cpio.c cmdline.c .PATH: ${_LIBARCHIVEDIR}/libarchive_fe -SRCS+= err.c line_reader.c passphrase.c +SRCS+= lafe_err.c line_reader.c passphrase.c CFLAGS+= -DBSDCPIO_VERSION_STRING=\"${BSDCPIO_VERSION_STRING}\" CFLAGS+= -DPLATFORM_CONFIG_H=\"${_LIBARCHIVECONFDIR}/config_freebsd.h\" CFLAGS+= -I${_LIBARCHIVEDIR}/cpio -I${_LIBARCHIVEDIR}/libarchive_fe LIBADD= archive .if ${MK_ICONV} != "no" CFLAGS+= -DHAVE_ICONV=1 -DHAVE_ICONV_H=1 -DICONV_CONST=const .endif SYMLINKS=bsdcpio ${BINDIR}/cpio MLINKS= bsdcpio.1 cpio.1 HAS_TESTS= SUBDIR.${MK_TESTS}+= tests .include diff --git a/usr.bin/cpio/tests/Makefile b/usr.bin/cpio/tests/Makefile index 9e1028c7eb58..ee4da15bc7e4 100644 --- a/usr.bin/cpio/tests/Makefile +++ b/usr.bin/cpio/tests/Makefile @@ -1,126 +1,126 @@ PACKAGE= tests _LIBARCHIVEDIR= ${SRCTOP}/contrib/libarchive ATF_TESTS_SH+= functional_test BINDIR= ${TESTSDIR} PROGS+= bsdcpio_test CFLAGS+= -DPLATFORM_CONFIG_H=\"${SRCTOP}/lib/libarchive/config_freebsd.h\" CFLAGS+= -I${SRCTOP}/lib/libarchive -I${.OBJDIR} CFLAGS+= -I${.OBJDIR} CFLAGS+= -I${_LIBARCHIVEDIR}/cpio -I${_LIBARCHIVEDIR}/cpio/test CFLAGS+= -I${_LIBARCHIVEDIR}/libarchive CFLAGS+= -I${_LIBARCHIVEDIR}/libarchive_fe -I${_LIBARCHIVEDIR}/test_utils # Uncomment to link against dmalloc #LDADD+= -L/usr/local/lib -ldmalloc #CFLAGS+= -I/usr/local/include -DUSE_DMALLOC CFLAGS.test_utils.c+= -Wno-cast-align .PATH: ${_LIBARCHIVEDIR}/cpio CPIO_SRCS= cmdline.c .PATH: ${_LIBARCHIVEDIR}/libarchive_fe -CPIO_SRCS+= err.c +CPIO_SRCS+= lafe_err.c .PATH: ${_LIBARCHIVEDIR}/cpio/test TESTS_SRCS= \ test_0.c \ test_basic.c \ test_cmdline.c \ test_extract_cpio_absolute_paths.c \ test_extract_cpio_Z.c \ test_extract_cpio_bz2.c \ test_extract_cpio_grz.c \ test_extract_cpio_gz.c \ test_extract_cpio_lrz.c \ test_extract_cpio_lz.c \ test_extract_cpio_lz4.c \ test_extract_cpio_lzma.c \ test_extract_cpio_lzo.c \ test_extract_cpio_xz.c \ test_extract_cpio_zstd.c \ test_format_newc.c \ test_gcpio_compat.c \ test_missing_file.c \ test_option_0.c \ test_option_B_upper.c \ test_option_C_upper.c \ test_option_J_upper.c \ test_option_L_upper.c \ test_option_Z_upper.c \ test_option_a.c \ test_option_b64encode.c \ test_option_c.c \ test_option_d.c \ test_option_f.c \ test_option_grzip.c \ test_option_help.c \ test_option_l.c \ test_option_lrzip.c \ test_option_lz4.c \ test_option_lzma.c \ test_option_lzop.c \ test_option_m.c \ test_option_passphrase.c \ test_option_t.c \ test_option_u.c \ test_option_uuencode.c \ test_option_version.c \ test_option_xz.c \ test_option_y.c \ test_option_z.c \ test_option_zstd.c \ test_owner_parse.c \ test_passthrough_dotdot.c \ test_passthrough_reverse.c SRCS.bsdcpio_test= list.h \ ${CPIO_SRCS} \ ${TESTS_SRCS} .PATH: ${_LIBARCHIVEDIR}/test_utils SRCS.bsdcpio_test+= test_main.c \ test_utils.c LIBADD.bsdcpio_test= archive list.h: ${TESTS_SRCS} Makefile @(cd ${_LIBARCHIVEDIR}/tar/test && \ grep -h DEFINE_TEST ${.ALLSRC:N*Makefile}) > ${.TARGET}.tmp @mv ${.TARGET}.tmp ${.TARGET} CLEANFILES+= list.h list.h.tmp ${PACKAGE}FILES+= test_extract.cpio.Z.uu ${PACKAGE}FILES+= test_extract.cpio.bz2.uu ${PACKAGE}FILES+= test_extract.cpio.grz.uu ${PACKAGE}FILES+= test_extract.cpio.gz.uu ${PACKAGE}FILES+= test_extract.cpio.lrz.uu ${PACKAGE}FILES+= test_extract.cpio.lz.uu ${PACKAGE}FILES+= test_extract.cpio.lz4.uu ${PACKAGE}FILES+= test_extract.cpio.lzma.uu ${PACKAGE}FILES+= test_extract.cpio.lzo.uu ${PACKAGE}FILES+= test_extract.cpio.xz.uu ${PACKAGE}FILES+= test_extract.cpio.zst.uu ${PACKAGE}FILES+= test_gcpio_compat_ref.bin.uu ${PACKAGE}FILES+= test_gcpio_compat_ref.crc.uu ${PACKAGE}FILES+= test_gcpio_compat_ref.newc.uu ${PACKAGE}FILES+= test_gcpio_compat_ref.ustar.uu ${PACKAGE}FILES+= test_gcpio_compat_ref_nosym.bin.uu ${PACKAGE}FILES+= test_gcpio_compat_ref_nosym.crc.uu ${PACKAGE}FILES+= test_gcpio_compat_ref_nosym.newc.uu ${PACKAGE}FILES+= test_gcpio_compat_ref_nosym.ustar.uu ${PACKAGE}FILES+= test_option_f.cpio.uu ${PACKAGE}FILES+= test_option_m.cpio.uu ${PACKAGE}FILES+= test_option_passphrase.zip.uu ${PACKAGE}FILES+= test_option_t.cpio.uu ${PACKAGE}FILES+= test_option_t.stdout.uu ${PACKAGE}FILES+= test_option_tv.stdout.uu .include diff --git a/usr.bin/tar/Makefile b/usr.bin/tar/Makefile index d3d29e03fd35..8b0d3e4a6cf0 100644 --- a/usr.bin/tar/Makefile +++ b/usr.bin/tar/Makefile @@ -1,39 +1,40 @@ .include _LIBARCHIVEDIR= ${SRCTOP}/contrib/libarchive +PACKAGE= runtime PROG= bsdtar BSDTAR_VERSION_STRING!= sed -n '/define.*ARCHIVE_VERSION_ONLY_STRING/{s,[^0-9.],,gp;q;}' \ ${_LIBARCHIVEDIR}/libarchive/archive.h .PATH: ${_LIBARCHIVEDIR}/tar SRCS= bsdtar.c \ cmdline.c \ creation_set.c \ read.c \ subst.c \ util.c \ write.c .PATH: ${_LIBARCHIVEDIR}/libarchive_fe -SRCS+= err.c \ +SRCS+= lafe_err.c \ line_reader.c \ passphrase.c LIBADD= archive .if ${MK_ICONV} != "no" CFLAGS+= -DHAVE_ICONV=1 -DHAVE_ICONV_H=1 -DICONV_CONST=const .endif CFLAGS+= -DBSDTAR_VERSION_STRING=\"${BSDTAR_VERSION_STRING}\" CFLAGS+= -DPLATFORM_CONFIG_H=\"${SRCTOP}/lib/libarchive/config_freebsd.h\" CFLAGS+= -I${_LIBARCHIVEDIR}/tar -I${_LIBARCHIVEDIR}/libarchive CFLAGS+= -I${_LIBARCHIVEDIR}/libarchive_fe SYMLINKS= bsdtar ${BINDIR}/tar MLINKS= bsdtar.1 tar.1 HAS_TESTS= SUBDIR.${MK_TESTS}+= tests .include diff --git a/usr.bin/tar/tests/Makefile b/usr.bin/tar/tests/Makefile index 929f8127f9b3..116425b0621f 100644 --- a/usr.bin/tar/tests/Makefile +++ b/usr.bin/tar/tests/Makefile @@ -1,133 +1,134 @@ PACKAGE= tests WARNS?= 3 _LIBARCHIVEDIR= ${SRCTOP}/contrib/libarchive ATF_TESTS_SH+= functional_test BINDIR= ${TESTSDIR} CFLAGS+= -DPLATFORM_CONFIG_H=\"${SRCTOP}/lib/libarchive/config_freebsd.h\" CFLAGS+= -I${SRCTOP}/lib/libarchive -I${.OBJDIR} CFLAGS+= -I${_LIBARCHIVEDIR}/libarchive CFLAGS+= -I${_LIBARCHIVEDIR}/tar -I${_LIBARCHIVEDIR}/tar/test CFLAGS+= -I${_LIBARCHIVEDIR}/test_utils CFLAGS.test_utils.c+= -Wno-cast-align # Uncomment to link against dmalloc #LDADD+= -L/usr/local/lib -ldmalloc #CFLAGS+= -I/usr/local/include -DUSE_DMALLOC PROGS+= bsdtar_test .PATH: ${_LIBARCHIVEDIR}/tar/test TESTS_SRCS= \ test_0.c \ test_basic.c \ test_copy.c \ + test_crlf_mtree.c \ test_empty_mtree.c \ test_extract_tar_bz2.c \ test_extract_tar_grz.c \ test_extract_tar_gz.c \ test_extract_tar_lrz.c \ test_extract_tar_lz.c \ test_extract_tar_lz4.c \ test_extract_tar_lzma.c \ test_extract_tar_lzo.c \ test_extract_tar_xz.c \ test_extract_tar_zstd.c \ test_format_newc.c \ test_help.c \ test_leading_slash.c \ test_list_item.c \ test_missing_file.c \ test_option_C_mtree.c \ test_option_C_upper.c \ test_option_H_upper.c \ test_option_L_upper.c \ test_option_O_upper.c \ test_option_P_upper.c \ test_option_T_upper.c \ test_option_U_upper.c \ test_option_X_upper.c \ test_option_a.c \ test_option_acls.c \ test_option_b.c \ test_option_b64encode.c \ test_option_exclude.c \ test_option_exclude_vcs.c \ test_option_fflags.c \ test_option_gid_gname.c \ test_option_grzip.c \ test_option_ignore_zeros.c \ test_option_j.c \ test_option_k.c \ test_option_keep_newer_files.c \ test_option_lrzip.c \ test_option_lz4.c \ test_option_lzma.c \ test_option_lzop.c \ test_option_mtime.c \ test_option_n.c \ test_option_newer_than.c \ test_option_nodump.c \ test_option_older_than.c \ test_option_passphrase.c \ test_option_q.c \ test_option_r.c \ test_option_s.c \ test_option_safe_writes.c \ test_option_uid_uname.c \ test_option_uuencode.c \ test_option_xattrs.c \ test_option_xz.c \ test_option_z.c \ test_option_zstd.c \ test_patterns.c \ test_print_longpath.c \ test_stdio.c \ test_strip_components.c \ test_symlink_dir.c \ test_version.c SRCS.bsdtar_test= \ ${TESTS_SRCS} \ list.h .PATH: ${_LIBARCHIVEDIR}/test_utils SRCS.bsdtar_test+= test_main.c \ test_utils.c LIBADD.bsdtar_test= archive list.h: ${TESTS_SRCS} Makefile @(cd ${_LIBARCHIVEDIR}/tar/test && \ grep -h DEFINE_TEST ${.ALLSRC:N*Makefile}) > ${.TARGET}.tmp @mv ${.TARGET}.tmp ${.TARGET} CLEANFILES+= list.h list.h.tmp ${PACKAGE}FILES+= test_extract.tar.Z.uu ${PACKAGE}FILES+= test_extract.tar.bz2.uu ${PACKAGE}FILES+= test_extract.tar.grz.uu ${PACKAGE}FILES+= test_extract.tar.gz.uu ${PACKAGE}FILES+= test_extract.tar.lrz.uu ${PACKAGE}FILES+= test_extract.tar.lz.uu ${PACKAGE}FILES+= test_extract.tar.lz4.uu ${PACKAGE}FILES+= test_extract.tar.lzma.uu ${PACKAGE}FILES+= test_extract.tar.lzo.uu ${PACKAGE}FILES+= test_extract.tar.xz.uu ${PACKAGE}FILES+= test_extract.tar.zst.uu ${PACKAGE}FILES+= test_leading_slash.tar.uu ${PACKAGE}FILES+= test_list_item.tar.uu ${PACKAGE}FILES+= test_option_keep_newer_files.tar.Z.uu ${PACKAGE}FILES+= test_option_passphrase.zip.uu ${PACKAGE}FILES+= test_option_s.tar.Z.uu ${PACKAGE}FILES+= test_patterns_2.tar.uu ${PACKAGE}FILES+= test_patterns_3.tar.uu ${PACKAGE}FILES+= test_patterns_4.tar.uu ${PACKAGE}FILES+= test_print_longpath.tar.Z.uu .include diff --git a/usr.bin/unzip/Makefile b/usr.bin/unzip/Makefile index 2db5e9ac4c99..63f49a203685 100644 --- a/usr.bin/unzip/Makefile +++ b/usr.bin/unzip/Makefile @@ -1,30 +1,30 @@ .include _LIBARCHIVEDIR= ${SRCTOP}/contrib/libarchive _LIBARCHIVECONFDIR= ${SRCTOP}/lib/libarchive PROG= bsdunzip BSDUNZIP_VERSION_STRING!= sed -n '/define.*ARCHIVE_VERSION_ONLY_STRING/{s,[^0-9.],,gp;q;}' \ ${_LIBARCHIVEDIR}/libarchive/archive.h .PATH: ${_LIBARCHIVEDIR}/unzip SRCS= bsdunzip.c .PATH: ${_LIBARCHIVEDIR}/libarchive_fe -SRCS+= cmdline.c err.c passphrase.c +SRCS+= cmdline.c lafe_err.c passphrase.c CFLAGS+= -DBSDUNZIP_VERSION_STRING=\"${BSDUNZIP_VERSION_STRING}\" CFLAGS+= -DPLATFORM_CONFIG_H=\"${_LIBARCHIVECONFDIR}/config_freebsd.h\" CFLAGS+= -I${_LIBARCHIVEDIR}/unzip -I${_LIBARCHIVEDIR}/libarchive_fe LIBADD= archive SYMLINKS=bsdunzip ${BINDIR}/unzip MLINKS= bsdunzip.1 unzip.1 HAS_TESTS= SUBDIR.${MK_TESTS}+= tests .include # DO NOT DELETE diff --git a/usr.bin/unzip/tests/Makefile b/usr.bin/unzip/tests/Makefile index 404a546410e4..fada172b1bd7 100644 --- a/usr.bin/unzip/tests/Makefile +++ b/usr.bin/unzip/tests/Makefile @@ -1,75 +1,75 @@ PACKAGE= tests _LIBARCHIVEDIR= ${SRCTOP}/contrib/libarchive ATF_TESTS_SH+= functional_test BINDIR= ${TESTSDIR} PROGS+= bsdunzip_test CFLAGS+= -DPLATFORM_CONFIG_H=\"${SRCTOP}/lib/libarchive/config_freebsd.h\" CFLAGS+= -I${SRCTOP}/lib/libarchive -I${.OBJDIR} CFLAGS+= -I${.OBJDIR} CFLAGS+= -I${_LIBARCHIVEDIR}/unzip -I${_LIBARCHIVEDIR}/unzip/test CFLAGS+= -I${_LIBARCHIVEDIR}/libarchive CFLAGS+= -I${_LIBARCHIVEDIR}/libarchive_fe -I${_LIBARCHIVEDIR}/test_utils # Uncomment to link against dmalloc #LDADD+= -L/usr/local/lib -ldmalloc #CFLAGS+= -I/usr/local/include -DUSE_DMALLOC CFLAGS.test_utils.c+= -Wno-cast-align .PATH: ${_LIBARCHIVEDIR}/libarchive_fe -UNZIP_SRCS+= err.c +UNZIP_SRCS+= lafe_err.c .PATH: ${_LIBARCHIVEDIR}/unzip/test TESTS_SRCS= \ test_0.c \ test_C.c \ test_I.c \ test_L.c \ test_P_encryption.c \ test_Z1.c \ test_basic.c \ test_d.c \ test_doubledash.c \ test_glob.c \ test_j.c \ test_n.c \ test_not_exist.c \ test_o.c \ test_p.c \ test_q.c \ test_singlefile.c \ test_t.c \ test_t_bad.c \ test_version.c \ test_x.c SRCS.bsdunzip_test= list.h \ ${UNZIP_SRCS} \ ${TESTS_SRCS} .PATH: ${_LIBARCHIVEDIR}/test_utils SRCS.bsdunzip_test+= test_main.c \ test_utils.c LIBADD.bsdunzip_test= archive list.h: ${TESTS_SRCS} Makefile @(cd ${_LIBARCHIVEDIR}/unzip/test && \ grep -h DEFINE_TEST ${.ALLSRC:N*Makefile}) > ${.TARGET}.tmp @mv ${.TARGET}.tmp ${.TARGET} CLEANFILES+= list.h list.h.tmp ${PACKAGE}FILES+= test_I.zip.uu ${PACKAGE}FILES+= test_basic.zip.uu ${PACKAGE}FILES+= test_encrypted.zip.uu ${PACKAGE}FILES+= test_singlefile.zip.uu ${PACKAGE}FILES+= test_t_bad.zip.uu .include