Index: head/contrib/libarchive/libarchive/archive_blake2sp_ref.c =================================================================== --- head/contrib/libarchive/libarchive/archive_blake2sp_ref.c (revision 345496) +++ head/contrib/libarchive/libarchive/archive_blake2sp_ref.c (revision 345497) @@ -1,359 +1,359 @@ /* BLAKE2 reference source code package - reference C implementations Copyright 2012, Samuel Neves . You may use this under the terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at your option. The terms of these licenses can be found at: - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0 - OpenSSL license : https://www.openssl.org/source/license.html - Apache 2.0 : http://www.apache.org/licenses/LICENSE-2.0 More information about the BLAKE2 hash function can be found at https://blake2.net. */ #include #include #include #if defined(_OPENMP) #include #endif #include "archive_blake2.h" #include "archive_blake2_impl.h" #define PARALLELISM_DEGREE 8 /* blake2sp_init_param defaults to setting the expecting output length from the digest_length parameter block field. In some cases, however, we do not want this, as the output length of these instances is given by inner_length instead. */ static int blake2sp_init_leaf_param( blake2s_state *S, const blake2s_param *P ) { int err = blake2s_init_param(S, P); S->outlen = P->inner_length; return err; } static int blake2sp_init_leaf( blake2s_state *S, size_t outlen, size_t keylen, uint32_t offset ) { blake2s_param P[1]; P->digest_length = (uint8_t)outlen; P->key_length = (uint8_t)keylen; P->fanout = PARALLELISM_DEGREE; P->depth = 2; store32( &P->leaf_length, 0 ); store32( &P->node_offset, offset ); store16( &P->xof_length, 0 ); P->node_depth = 0; P->inner_length = BLAKE2S_OUTBYTES; memset( P->salt, 0, sizeof( P->salt ) ); memset( P->personal, 0, sizeof( P->personal ) ); return blake2sp_init_leaf_param( S, P ); } static int blake2sp_init_root( blake2s_state *S, size_t outlen, size_t keylen ) { blake2s_param P[1]; P->digest_length = (uint8_t)outlen; P->key_length = (uint8_t)keylen; P->fanout = PARALLELISM_DEGREE; P->depth = 2; store32( &P->leaf_length, 0 ); store32( &P->node_offset, 0 ); store16( &P->xof_length, 0 ); P->node_depth = 1; P->inner_length = BLAKE2S_OUTBYTES; memset( P->salt, 0, sizeof( P->salt ) ); memset( P->personal, 0, sizeof( P->personal ) ); return blake2s_init_param( S, P ); } int blake2sp_init( blake2sp_state *S, size_t outlen ) { size_t i; if( !outlen || outlen > BLAKE2S_OUTBYTES ) return -1; memset( S->buf, 0, sizeof( S->buf ) ); S->buflen = 0; S->outlen = outlen; if( blake2sp_init_root( S->R, outlen, 0 ) < 0 ) return -1; for( i = 0; i < PARALLELISM_DEGREE; ++i ) - if( blake2sp_init_leaf( S->S[i], outlen, 0, i ) < 0 ) return -1; + if( blake2sp_init_leaf( S->S[i], outlen, 0, (uint32_t)i ) < 0 ) return -1; S->R->last_node = 1; S->S[PARALLELISM_DEGREE - 1]->last_node = 1; return 0; } int blake2sp_init_key( blake2sp_state *S, size_t outlen, const void *key, size_t keylen ) { size_t i; if( !outlen || outlen > BLAKE2S_OUTBYTES ) return -1; if( !key || !keylen || keylen > BLAKE2S_KEYBYTES ) return -1; memset( S->buf, 0, sizeof( S->buf ) ); S->buflen = 0; S->outlen = outlen; if( blake2sp_init_root( S->R, outlen, keylen ) < 0 ) return -1; for( i = 0; i < PARALLELISM_DEGREE; ++i ) - if( blake2sp_init_leaf( S->S[i], outlen, keylen, i ) < 0 ) return -1; + if( blake2sp_init_leaf( S->S[i], outlen, keylen, (uint32_t)i ) < 0 ) return -1; S->R->last_node = 1; S->S[PARALLELISM_DEGREE - 1]->last_node = 1; { uint8_t block[BLAKE2S_BLOCKBYTES]; memset( block, 0, BLAKE2S_BLOCKBYTES ); memcpy( block, key, keylen ); for( i = 0; i < PARALLELISM_DEGREE; ++i ) blake2s_update( S->S[i], block, BLAKE2S_BLOCKBYTES ); secure_zero_memory( block, BLAKE2S_BLOCKBYTES ); /* Burn the key from stack */ } return 0; } int blake2sp_update( blake2sp_state *S, const void *pin, size_t inlen ) { const unsigned char * in = (const unsigned char *)pin; size_t left = S->buflen; size_t fill = sizeof( S->buf ) - left; size_t i; if( left && inlen >= fill ) { memcpy( S->buf + left, in, fill ); for( i = 0; i < PARALLELISM_DEGREE; ++i ) blake2s_update( S->S[i], S->buf + i * BLAKE2S_BLOCKBYTES, BLAKE2S_BLOCKBYTES ); in += fill; inlen -= fill; left = 0; } #if defined(_OPENMP) #pragma omp parallel shared(S), num_threads(PARALLELISM_DEGREE) #else for( i = 0; i < PARALLELISM_DEGREE; ++i ) #endif { #if defined(_OPENMP) size_t i = omp_get_thread_num(); #endif size_t inlen__ = inlen; const unsigned char *in__ = ( const unsigned char * )in; in__ += i * BLAKE2S_BLOCKBYTES; while( inlen__ >= PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES ) { blake2s_update( S->S[i], in__, BLAKE2S_BLOCKBYTES ); in__ += PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES; inlen__ -= PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES; } } in += inlen - inlen % ( PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES ); inlen %= PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES; if( inlen > 0 ) memcpy( S->buf + left, in, inlen ); S->buflen = left + inlen; return 0; } int blake2sp_final( blake2sp_state *S, void *out, size_t outlen ) { uint8_t hash[PARALLELISM_DEGREE][BLAKE2S_OUTBYTES]; size_t i; if(out == NULL || outlen < S->outlen) { return -1; } for( i = 0; i < PARALLELISM_DEGREE; ++i ) { if( S->buflen > i * BLAKE2S_BLOCKBYTES ) { size_t left = S->buflen - i * BLAKE2S_BLOCKBYTES; if( left > BLAKE2S_BLOCKBYTES ) left = BLAKE2S_BLOCKBYTES; blake2s_update( S->S[i], S->buf + i * BLAKE2S_BLOCKBYTES, left ); } blake2s_final( S->S[i], hash[i], BLAKE2S_OUTBYTES ); } for( i = 0; i < PARALLELISM_DEGREE; ++i ) blake2s_update( S->R, hash[i], BLAKE2S_OUTBYTES ); return blake2s_final( S->R, out, S->outlen ); } int blake2sp( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen ) { uint8_t hash[PARALLELISM_DEGREE][BLAKE2S_OUTBYTES]; blake2s_state S[PARALLELISM_DEGREE][1]; blake2s_state FS[1]; size_t i; /* Verify parameters */ if ( NULL == in && inlen > 0 ) return -1; if ( NULL == out ) return -1; if ( NULL == key && keylen > 0) return -1; if( !outlen || outlen > BLAKE2S_OUTBYTES ) return -1; if( keylen > BLAKE2S_KEYBYTES ) return -1; for( i = 0; i < PARALLELISM_DEGREE; ++i ) - if( blake2sp_init_leaf( S[i], outlen, keylen, i ) < 0 ) return -1; + if( blake2sp_init_leaf( S[i], outlen, keylen, (uint32_t)i ) < 0 ) return -1; S[PARALLELISM_DEGREE - 1]->last_node = 1; /* mark last node */ if( keylen > 0 ) { uint8_t block[BLAKE2S_BLOCKBYTES]; memset( block, 0, BLAKE2S_BLOCKBYTES ); memcpy( block, key, keylen ); for( i = 0; i < PARALLELISM_DEGREE; ++i ) blake2s_update( S[i], block, BLAKE2S_BLOCKBYTES ); secure_zero_memory( block, BLAKE2S_BLOCKBYTES ); /* Burn the key from stack */ } #if defined(_OPENMP) #pragma omp parallel shared(S,hash), num_threads(PARALLELISM_DEGREE) #else for( i = 0; i < PARALLELISM_DEGREE; ++i ) #endif { #if defined(_OPENMP) size_t i = omp_get_thread_num(); #endif size_t inlen__ = inlen; const unsigned char *in__ = ( const unsigned char * )in; in__ += i * BLAKE2S_BLOCKBYTES; while( inlen__ >= PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES ) { blake2s_update( S[i], in__, BLAKE2S_BLOCKBYTES ); in__ += PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES; inlen__ -= PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES; } if( inlen__ > i * BLAKE2S_BLOCKBYTES ) { const size_t left = inlen__ - i * BLAKE2S_BLOCKBYTES; const size_t len = left <= BLAKE2S_BLOCKBYTES ? left : BLAKE2S_BLOCKBYTES; blake2s_update( S[i], in__, len ); } blake2s_final( S[i], hash[i], BLAKE2S_OUTBYTES ); } if( blake2sp_init_root( FS, outlen, keylen ) < 0 ) return -1; FS->last_node = 1; for( i = 0; i < PARALLELISM_DEGREE; ++i ) blake2s_update( FS, hash[i], BLAKE2S_OUTBYTES ); return blake2s_final( FS, out, outlen ); } #if defined(BLAKE2SP_SELFTEST) #include #include "blake2-kat.h" int main( void ) { uint8_t key[BLAKE2S_KEYBYTES]; uint8_t buf[BLAKE2_KAT_LENGTH]; size_t i, step; for( i = 0; i < BLAKE2S_KEYBYTES; ++i ) key[i] = ( uint8_t )i; for( i = 0; i < BLAKE2_KAT_LENGTH; ++i ) buf[i] = ( uint8_t )i; /* Test simple API */ for( i = 0; i < BLAKE2_KAT_LENGTH; ++i ) { uint8_t hash[BLAKE2S_OUTBYTES]; blake2sp( hash, BLAKE2S_OUTBYTES, buf, i, key, BLAKE2S_KEYBYTES ); if( 0 != memcmp( hash, blake2sp_keyed_kat[i], BLAKE2S_OUTBYTES ) ) { goto fail; } } /* Test streaming API */ for(step = 1; step < BLAKE2S_BLOCKBYTES; ++step) { for (i = 0; i < BLAKE2_KAT_LENGTH; ++i) { uint8_t hash[BLAKE2S_OUTBYTES]; blake2sp_state S; uint8_t * p = buf; size_t mlen = i; int err = 0; if( (err = blake2sp_init_key(&S, BLAKE2S_OUTBYTES, key, BLAKE2S_KEYBYTES)) < 0 ) { goto fail; } while (mlen >= step) { if ( (err = blake2sp_update(&S, p, step)) < 0 ) { goto fail; } mlen -= step; p += step; } if ( (err = blake2sp_update(&S, p, mlen)) < 0) { goto fail; } if ( (err = blake2sp_final(&S, hash, BLAKE2S_OUTBYTES)) < 0) { goto fail; } if (0 != memcmp(hash, blake2sp_keyed_kat[i], BLAKE2S_OUTBYTES)) { goto fail; } } } puts( "ok" ); return 0; fail: puts("error"); return -1; } #endif Index: head/contrib/libarchive/libarchive/archive_entry.c =================================================================== --- head/contrib/libarchive/libarchive/archive_entry.c (revision 345496) +++ head/contrib/libarchive/libarchive/archive_entry.c (revision 345497) @@ -1,2032 +1,2036 @@ /*- * Copyright (c) 2003-2007 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" __FBSDID("$FreeBSD$"); #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_SYS_TYPES_H #include #endif #if MAJOR_IN_MKDEV #include #define HAVE_MAJOR #elif MAJOR_IN_SYSMACROS #include #define HAVE_MAJOR #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LIMITS_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 /* for Linux file flags */ #endif #include #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_WCHAR_H #include #endif #include "archive.h" #include "archive_acl_private.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_entry_private.h" #if !defined(HAVE_MAJOR) && !defined(major) /* Replacement for major/minor/makedev. */ #define major(x) ((int)(0x00ff & ((x) >> 8))) #define minor(x) ((int)(0xffff00ff & (x))) #define makedev(maj,min) ((0xff00 & ((maj)<<8)) | (0xffff00ff & (min))) #endif /* Play games to come up with a suitable makedev() definition. */ #ifdef __QNXNTO__ /* QNX. */ #include #define ae_makedev(maj, min) makedev(ND_LOCAL_NODE, (maj), (min)) #elif defined makedev /* There's a "makedev" macro. */ #define ae_makedev(maj, min) makedev((maj), (min)) #elif defined mkdev || ((defined _WIN32 || defined __WIN32__) && !defined(__CYGWIN__)) /* Windows. */ #define ae_makedev(maj, min) mkdev((maj), (min)) #else /* There's a "makedev" function. */ #define ae_makedev(maj, min) makedev((maj), (min)) #endif /* * This adjustment is needed to support the following idiom for adding * 1000ns to the stored time: * archive_entry_set_atime(archive_entry_atime(), * archive_entry_atime_nsec() + 1000) * The additional if() here compensates for ambiguity in the C standard, * which permits two possible interpretations of a % b when a is negative. */ #define FIX_NS(t,ns) \ do { \ t += ns / 1000000000; \ ns %= 1000000000; \ if (ns < 0) { --t; ns += 1000000000; } \ } while (0) static char * ae_fflagstostr(unsigned long bitset, unsigned long bitclear); static const wchar_t *ae_wcstofflags(const wchar_t *stringp, unsigned long *setp, unsigned long *clrp); static const char *ae_strtofflags(const char *stringp, unsigned long *setp, unsigned long *clrp); #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 #ifndef HAVE_WCSLEN static size_t wcslen(const wchar_t *s) { const wchar_t *p = s; while (*p != L'\0') ++p; return p - s; } #endif #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 /**************************************************************************** * * Public Interface * ****************************************************************************/ struct archive_entry * archive_entry_clear(struct archive_entry *entry) { if (entry == NULL) return (NULL); archive_mstring_clean(&entry->ae_fflags_text); archive_mstring_clean(&entry->ae_gname); archive_mstring_clean(&entry->ae_hardlink); archive_mstring_clean(&entry->ae_pathname); archive_mstring_clean(&entry->ae_sourcepath); archive_mstring_clean(&entry->ae_symlink); archive_mstring_clean(&entry->ae_uname); archive_entry_copy_mac_metadata(entry, NULL, 0); archive_acl_clear(&entry->acl); archive_entry_xattr_clear(entry); archive_entry_sparse_clear(entry); free(entry->stat); memset(entry, 0, sizeof(*entry)); return entry; } struct archive_entry * archive_entry_clone(struct archive_entry *entry) { struct archive_entry *entry2; struct ae_xattr *xp; struct ae_sparse *sp; size_t s; const void *p; /* Allocate new structure and copy over all of the fields. */ /* TODO: Should we copy the archive over? Or require a new archive * as an argument? */ entry2 = archive_entry_new2(entry->archive); if (entry2 == NULL) return (NULL); entry2->ae_stat = entry->ae_stat; entry2->ae_fflags_set = entry->ae_fflags_set; entry2->ae_fflags_clear = entry->ae_fflags_clear; /* TODO: XXX If clone can have a different archive, what do we do here if * character sets are different? XXX */ archive_mstring_copy(&entry2->ae_fflags_text, &entry->ae_fflags_text); archive_mstring_copy(&entry2->ae_gname, &entry->ae_gname); archive_mstring_copy(&entry2->ae_hardlink, &entry->ae_hardlink); archive_mstring_copy(&entry2->ae_pathname, &entry->ae_pathname); archive_mstring_copy(&entry2->ae_sourcepath, &entry->ae_sourcepath); archive_mstring_copy(&entry2->ae_symlink, &entry->ae_symlink); entry2->ae_set = entry->ae_set; archive_mstring_copy(&entry2->ae_uname, &entry->ae_uname); /* Copy encryption status */ entry2->encryption = entry->encryption; /* Copy ACL data over. */ archive_acl_copy(&entry2->acl, &entry->acl); /* Copy Mac OS metadata. */ p = archive_entry_mac_metadata(entry, &s); archive_entry_copy_mac_metadata(entry2, p, s); /* Copy xattr data over. */ xp = entry->xattr_head; while (xp != NULL) { archive_entry_xattr_add_entry(entry2, xp->name, xp->value, xp->size); xp = xp->next; } /* Copy sparse data over. */ sp = entry->sparse_head; while (sp != NULL) { archive_entry_sparse_add_entry(entry2, sp->offset, sp->length); sp = sp->next; } return (entry2); } void archive_entry_free(struct archive_entry *entry) { archive_entry_clear(entry); free(entry); } struct archive_entry * archive_entry_new(void) { return archive_entry_new2(NULL); } struct archive_entry * archive_entry_new2(struct archive *a) { struct archive_entry *entry; entry = (struct archive_entry *)calloc(1, sizeof(*entry)); if (entry == NULL) return (NULL); entry->archive = a; return (entry); } /* * Functions for reading fields from an archive_entry. */ time_t archive_entry_atime(struct archive_entry *entry) { return (entry->ae_stat.aest_atime); } long archive_entry_atime_nsec(struct archive_entry *entry) { return (entry->ae_stat.aest_atime_nsec); } int archive_entry_atime_is_set(struct archive_entry *entry) { return (entry->ae_set & AE_SET_ATIME); } time_t archive_entry_birthtime(struct archive_entry *entry) { return (entry->ae_stat.aest_birthtime); } long archive_entry_birthtime_nsec(struct archive_entry *entry) { return (entry->ae_stat.aest_birthtime_nsec); } int archive_entry_birthtime_is_set(struct archive_entry *entry) { return (entry->ae_set & AE_SET_BIRTHTIME); } time_t archive_entry_ctime(struct archive_entry *entry) { return (entry->ae_stat.aest_ctime); } int archive_entry_ctime_is_set(struct archive_entry *entry) { return (entry->ae_set & AE_SET_CTIME); } long archive_entry_ctime_nsec(struct archive_entry *entry) { return (entry->ae_stat.aest_ctime_nsec); } dev_t archive_entry_dev(struct archive_entry *entry) { if (entry->ae_stat.aest_dev_is_broken_down) return ae_makedev(entry->ae_stat.aest_devmajor, entry->ae_stat.aest_devminor); else return (entry->ae_stat.aest_dev); } int archive_entry_dev_is_set(struct archive_entry *entry) { return (entry->ae_set & AE_SET_DEV); } dev_t archive_entry_devmajor(struct archive_entry *entry) { if (entry->ae_stat.aest_dev_is_broken_down) return (entry->ae_stat.aest_devmajor); else return major(entry->ae_stat.aest_dev); } dev_t archive_entry_devminor(struct archive_entry *entry) { if (entry->ae_stat.aest_dev_is_broken_down) return (entry->ae_stat.aest_devminor); else return minor(entry->ae_stat.aest_dev); } mode_t archive_entry_filetype(struct archive_entry *entry) { return (AE_IFMT & entry->acl.mode); } void archive_entry_fflags(struct archive_entry *entry, unsigned long *set, unsigned long *clear) { *set = entry->ae_fflags_set; *clear = entry->ae_fflags_clear; } /* * Note: if text was provided, this just returns that text. If you * really need the text to be rebuilt in a canonical form, set the * text, ask for the bitmaps, then set the bitmaps. (Setting the * bitmaps clears any stored text.) This design is deliberate: if * we're editing archives, we don't want to discard flags just because * they aren't supported on the current system. The bitmap<->text * conversions are platform-specific (see below). */ const char * archive_entry_fflags_text(struct archive_entry *entry) { const char *f; char *p; if (archive_mstring_get_mbs(entry->archive, &entry->ae_fflags_text, &f) == 0) { if (f != NULL) return (f); } else if (errno == ENOMEM) __archive_errx(1, "No memory"); if (entry->ae_fflags_set == 0 && entry->ae_fflags_clear == 0) return (NULL); p = ae_fflagstostr(entry->ae_fflags_set, entry->ae_fflags_clear); if (p == NULL) return (NULL); archive_mstring_copy_mbs(&entry->ae_fflags_text, p); free(p); if (archive_mstring_get_mbs(entry->archive, &entry->ae_fflags_text, &f) == 0) return (f); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } la_int64_t archive_entry_gid(struct archive_entry *entry) { return (entry->ae_stat.aest_gid); } const char * archive_entry_gname(struct archive_entry *entry) { const char *p; if (archive_mstring_get_mbs(entry->archive, &entry->ae_gname, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } const char * archive_entry_gname_utf8(struct archive_entry *entry) { const char *p; if (archive_mstring_get_utf8(entry->archive, &entry->ae_gname, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } const wchar_t * archive_entry_gname_w(struct archive_entry *entry) { const wchar_t *p; if (archive_mstring_get_wcs(entry->archive, &entry->ae_gname, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } int _archive_entry_gname_l(struct archive_entry *entry, const char **p, size_t *len, struct archive_string_conv *sc) { return (archive_mstring_get_mbs_l(&entry->ae_gname, p, len, sc)); } const char * archive_entry_hardlink(struct archive_entry *entry) { const char *p; if ((entry->ae_set & AE_SET_HARDLINK) == 0) return (NULL); if (archive_mstring_get_mbs( entry->archive, &entry->ae_hardlink, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } const char * archive_entry_hardlink_utf8(struct archive_entry *entry) { const char *p; if ((entry->ae_set & AE_SET_HARDLINK) == 0) return (NULL); if (archive_mstring_get_utf8( entry->archive, &entry->ae_hardlink, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } const wchar_t * archive_entry_hardlink_w(struct archive_entry *entry) { const wchar_t *p; if ((entry->ae_set & AE_SET_HARDLINK) == 0) return (NULL); if (archive_mstring_get_wcs( entry->archive, &entry->ae_hardlink, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } int _archive_entry_hardlink_l(struct archive_entry *entry, const char **p, size_t *len, struct archive_string_conv *sc) { if ((entry->ae_set & AE_SET_HARDLINK) == 0) { *p = NULL; *len = 0; return (0); } return (archive_mstring_get_mbs_l(&entry->ae_hardlink, p, len, sc)); } la_int64_t archive_entry_ino(struct archive_entry *entry) { return (entry->ae_stat.aest_ino); } int archive_entry_ino_is_set(struct archive_entry *entry) { return (entry->ae_set & AE_SET_INO); } la_int64_t archive_entry_ino64(struct archive_entry *entry) { return (entry->ae_stat.aest_ino); } mode_t archive_entry_mode(struct archive_entry *entry) { return (entry->acl.mode); } time_t archive_entry_mtime(struct archive_entry *entry) { return (entry->ae_stat.aest_mtime); } long archive_entry_mtime_nsec(struct archive_entry *entry) { return (entry->ae_stat.aest_mtime_nsec); } int archive_entry_mtime_is_set(struct archive_entry *entry) { return (entry->ae_set & AE_SET_MTIME); } unsigned int archive_entry_nlink(struct archive_entry *entry) { return (entry->ae_stat.aest_nlink); } const char * archive_entry_pathname(struct archive_entry *entry) { const char *p; if (archive_mstring_get_mbs( entry->archive, &entry->ae_pathname, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } const char * archive_entry_pathname_utf8(struct archive_entry *entry) { const char *p; if (archive_mstring_get_utf8( entry->archive, &entry->ae_pathname, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } const wchar_t * archive_entry_pathname_w(struct archive_entry *entry) { const wchar_t *p; if (archive_mstring_get_wcs( entry->archive, &entry->ae_pathname, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } int _archive_entry_pathname_l(struct archive_entry *entry, const char **p, size_t *len, struct archive_string_conv *sc) { return (archive_mstring_get_mbs_l(&entry->ae_pathname, p, len, sc)); } mode_t archive_entry_perm(struct archive_entry *entry) { return (~AE_IFMT & entry->acl.mode); } dev_t archive_entry_rdev(struct archive_entry *entry) { if (entry->ae_stat.aest_rdev_is_broken_down) return ae_makedev(entry->ae_stat.aest_rdevmajor, entry->ae_stat.aest_rdevminor); else return (entry->ae_stat.aest_rdev); } dev_t archive_entry_rdevmajor(struct archive_entry *entry) { if (entry->ae_stat.aest_rdev_is_broken_down) return (entry->ae_stat.aest_rdevmajor); else return major(entry->ae_stat.aest_rdev); } dev_t archive_entry_rdevminor(struct archive_entry *entry) { if (entry->ae_stat.aest_rdev_is_broken_down) return (entry->ae_stat.aest_rdevminor); else return minor(entry->ae_stat.aest_rdev); } la_int64_t archive_entry_size(struct archive_entry *entry) { return (entry->ae_stat.aest_size); } int archive_entry_size_is_set(struct archive_entry *entry) { return (entry->ae_set & AE_SET_SIZE); } const char * archive_entry_sourcepath(struct archive_entry *entry) { const char *p; if (archive_mstring_get_mbs( entry->archive, &entry->ae_sourcepath, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } const wchar_t * archive_entry_sourcepath_w(struct archive_entry *entry) { const wchar_t *p; if (archive_mstring_get_wcs( entry->archive, &entry->ae_sourcepath, &p) == 0) return (p); return (NULL); } const char * archive_entry_symlink(struct archive_entry *entry) { const char *p; if ((entry->ae_set & AE_SET_SYMLINK) == 0) return (NULL); if (archive_mstring_get_mbs( entry->archive, &entry->ae_symlink, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } const char * archive_entry_symlink_utf8(struct archive_entry *entry) { const char *p; if ((entry->ae_set & AE_SET_SYMLINK) == 0) return (NULL); if (archive_mstring_get_utf8( entry->archive, &entry->ae_symlink, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } const wchar_t * archive_entry_symlink_w(struct archive_entry *entry) { const wchar_t *p; if ((entry->ae_set & AE_SET_SYMLINK) == 0) return (NULL); if (archive_mstring_get_wcs( entry->archive, &entry->ae_symlink, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } int _archive_entry_symlink_l(struct archive_entry *entry, const char **p, size_t *len, struct archive_string_conv *sc) { if ((entry->ae_set & AE_SET_SYMLINK) == 0) { *p = NULL; *len = 0; return (0); } return (archive_mstring_get_mbs_l( &entry->ae_symlink, p, len, sc)); } la_int64_t archive_entry_uid(struct archive_entry *entry) { return (entry->ae_stat.aest_uid); } const char * archive_entry_uname(struct archive_entry *entry) { const char *p; if (archive_mstring_get_mbs(entry->archive, &entry->ae_uname, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } const char * archive_entry_uname_utf8(struct archive_entry *entry) { const char *p; if (archive_mstring_get_utf8(entry->archive, &entry->ae_uname, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } const wchar_t * archive_entry_uname_w(struct archive_entry *entry) { const wchar_t *p; if (archive_mstring_get_wcs(entry->archive, &entry->ae_uname, &p) == 0) return (p); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } int _archive_entry_uname_l(struct archive_entry *entry, const char **p, size_t *len, struct archive_string_conv *sc) { return (archive_mstring_get_mbs_l(&entry->ae_uname, p, len, sc)); } int archive_entry_is_data_encrypted(struct archive_entry *entry) { return ((entry->encryption & AE_ENCRYPTION_DATA) == AE_ENCRYPTION_DATA); } int archive_entry_is_metadata_encrypted(struct archive_entry *entry) { return ((entry->encryption & AE_ENCRYPTION_METADATA) == AE_ENCRYPTION_METADATA); } int archive_entry_is_encrypted(struct archive_entry *entry) { return (entry->encryption & (AE_ENCRYPTION_DATA|AE_ENCRYPTION_METADATA)); } /* * Functions to set archive_entry properties. */ void archive_entry_set_filetype(struct archive_entry *entry, unsigned int type) { entry->stat_valid = 0; entry->acl.mode &= ~AE_IFMT; entry->acl.mode |= AE_IFMT & type; } void archive_entry_set_fflags(struct archive_entry *entry, unsigned long set, unsigned long clear) { archive_mstring_clean(&entry->ae_fflags_text); entry->ae_fflags_set = set; entry->ae_fflags_clear = clear; } const char * archive_entry_copy_fflags_text(struct archive_entry *entry, const char *flags) { archive_mstring_copy_mbs(&entry->ae_fflags_text, flags); return (ae_strtofflags(flags, &entry->ae_fflags_set, &entry->ae_fflags_clear)); } const wchar_t * archive_entry_copy_fflags_text_w(struct archive_entry *entry, const wchar_t *flags) { archive_mstring_copy_wcs(&entry->ae_fflags_text, flags); return (ae_wcstofflags(flags, &entry->ae_fflags_set, &entry->ae_fflags_clear)); } void archive_entry_set_gid(struct archive_entry *entry, la_int64_t g) { entry->stat_valid = 0; entry->ae_stat.aest_gid = g; } void archive_entry_set_gname(struct archive_entry *entry, const char *name) { archive_mstring_copy_mbs(&entry->ae_gname, name); } void archive_entry_set_gname_utf8(struct archive_entry *entry, const char *name) { archive_mstring_copy_utf8(&entry->ae_gname, name); } void archive_entry_copy_gname(struct archive_entry *entry, const char *name) { archive_mstring_copy_mbs(&entry->ae_gname, name); } void archive_entry_copy_gname_w(struct archive_entry *entry, const wchar_t *name) { archive_mstring_copy_wcs(&entry->ae_gname, name); } int archive_entry_update_gname_utf8(struct archive_entry *entry, const char *name) { if (archive_mstring_update_utf8(entry->archive, &entry->ae_gname, name) == 0) return (1); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (0); } int _archive_entry_copy_gname_l(struct archive_entry *entry, const char *name, size_t len, struct archive_string_conv *sc) { return (archive_mstring_copy_mbs_len_l(&entry->ae_gname, name, len, sc)); } void archive_entry_set_ino(struct archive_entry *entry, la_int64_t ino) { entry->stat_valid = 0; entry->ae_set |= AE_SET_INO; entry->ae_stat.aest_ino = ino; } void archive_entry_set_ino64(struct archive_entry *entry, la_int64_t ino) { entry->stat_valid = 0; entry->ae_set |= AE_SET_INO; entry->ae_stat.aest_ino = ino; } void archive_entry_set_hardlink(struct archive_entry *entry, const char *target) { archive_mstring_copy_mbs(&entry->ae_hardlink, target); if (target != NULL) entry->ae_set |= AE_SET_HARDLINK; else entry->ae_set &= ~AE_SET_HARDLINK; } void archive_entry_set_hardlink_utf8(struct archive_entry *entry, const char *target) { archive_mstring_copy_utf8(&entry->ae_hardlink, target); if (target != NULL) entry->ae_set |= AE_SET_HARDLINK; else entry->ae_set &= ~AE_SET_HARDLINK; } void archive_entry_copy_hardlink(struct archive_entry *entry, const char *target) { archive_mstring_copy_mbs(&entry->ae_hardlink, target); if (target != NULL) entry->ae_set |= AE_SET_HARDLINK; else entry->ae_set &= ~AE_SET_HARDLINK; } void archive_entry_copy_hardlink_w(struct archive_entry *entry, const wchar_t *target) { archive_mstring_copy_wcs(&entry->ae_hardlink, target); if (target != NULL) entry->ae_set |= AE_SET_HARDLINK; else entry->ae_set &= ~AE_SET_HARDLINK; } int archive_entry_update_hardlink_utf8(struct archive_entry *entry, const char *target) { if (target != NULL) entry->ae_set |= AE_SET_HARDLINK; else entry->ae_set &= ~AE_SET_HARDLINK; if (archive_mstring_update_utf8(entry->archive, &entry->ae_hardlink, target) == 0) return (1); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (0); } int _archive_entry_copy_hardlink_l(struct archive_entry *entry, const char *target, size_t len, struct archive_string_conv *sc) { int r; r = archive_mstring_copy_mbs_len_l(&entry->ae_hardlink, target, len, sc); if (target != NULL && r == 0) entry->ae_set |= AE_SET_HARDLINK; else entry->ae_set &= ~AE_SET_HARDLINK; return (r); } void archive_entry_set_atime(struct archive_entry *entry, time_t t, long ns) { FIX_NS(t, ns); entry->stat_valid = 0; entry->ae_set |= AE_SET_ATIME; entry->ae_stat.aest_atime = t; entry->ae_stat.aest_atime_nsec = ns; } void archive_entry_unset_atime(struct archive_entry *entry) { archive_entry_set_atime(entry, 0, 0); entry->ae_set &= ~AE_SET_ATIME; } void archive_entry_set_birthtime(struct archive_entry *entry, time_t t, long ns) { FIX_NS(t, ns); entry->stat_valid = 0; entry->ae_set |= AE_SET_BIRTHTIME; entry->ae_stat.aest_birthtime = t; entry->ae_stat.aest_birthtime_nsec = ns; } void archive_entry_unset_birthtime(struct archive_entry *entry) { archive_entry_set_birthtime(entry, 0, 0); entry->ae_set &= ~AE_SET_BIRTHTIME; } void archive_entry_set_ctime(struct archive_entry *entry, time_t t, long ns) { FIX_NS(t, ns); entry->stat_valid = 0; entry->ae_set |= AE_SET_CTIME; entry->ae_stat.aest_ctime = t; entry->ae_stat.aest_ctime_nsec = ns; } void archive_entry_unset_ctime(struct archive_entry *entry) { archive_entry_set_ctime(entry, 0, 0); entry->ae_set &= ~AE_SET_CTIME; } void archive_entry_set_dev(struct archive_entry *entry, dev_t d) { entry->stat_valid = 0; entry->ae_set |= AE_SET_DEV; entry->ae_stat.aest_dev_is_broken_down = 0; entry->ae_stat.aest_dev = d; } void archive_entry_set_devmajor(struct archive_entry *entry, dev_t m) { entry->stat_valid = 0; entry->ae_set |= AE_SET_DEV; entry->ae_stat.aest_dev_is_broken_down = 1; entry->ae_stat.aest_devmajor = m; } void archive_entry_set_devminor(struct archive_entry *entry, dev_t m) { entry->stat_valid = 0; entry->ae_set |= AE_SET_DEV; entry->ae_stat.aest_dev_is_broken_down = 1; entry->ae_stat.aest_devminor = m; } /* Set symlink if symlink is already set, else set hardlink. */ void archive_entry_set_link(struct archive_entry *entry, const char *target) { if (entry->ae_set & AE_SET_SYMLINK) archive_mstring_copy_mbs(&entry->ae_symlink, target); else archive_mstring_copy_mbs(&entry->ae_hardlink, target); } void archive_entry_set_link_utf8(struct archive_entry *entry, const char *target) { if (entry->ae_set & AE_SET_SYMLINK) archive_mstring_copy_utf8(&entry->ae_symlink, target); else archive_mstring_copy_utf8(&entry->ae_hardlink, target); } /* Set symlink if symlink is already set, else set hardlink. */ void archive_entry_copy_link(struct archive_entry *entry, const char *target) { if (entry->ae_set & AE_SET_SYMLINK) archive_mstring_copy_mbs(&entry->ae_symlink, target); else archive_mstring_copy_mbs(&entry->ae_hardlink, target); } /* Set symlink if symlink is already set, else set hardlink. */ void archive_entry_copy_link_w(struct archive_entry *entry, const wchar_t *target) { if (entry->ae_set & AE_SET_SYMLINK) archive_mstring_copy_wcs(&entry->ae_symlink, target); else archive_mstring_copy_wcs(&entry->ae_hardlink, target); } int archive_entry_update_link_utf8(struct archive_entry *entry, const char *target) { int r; if (entry->ae_set & AE_SET_SYMLINK) r = archive_mstring_update_utf8(entry->archive, &entry->ae_symlink, target); else r = archive_mstring_update_utf8(entry->archive, &entry->ae_hardlink, target); if (r == 0) return (1); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (0); } int _archive_entry_copy_link_l(struct archive_entry *entry, const char *target, size_t len, struct archive_string_conv *sc) { int r; if (entry->ae_set & AE_SET_SYMLINK) r = archive_mstring_copy_mbs_len_l(&entry->ae_symlink, target, len, sc); else r = archive_mstring_copy_mbs_len_l(&entry->ae_hardlink, target, len, sc); return (r); } void archive_entry_set_mode(struct archive_entry *entry, mode_t m) { entry->stat_valid = 0; entry->acl.mode = m; } void archive_entry_set_mtime(struct archive_entry *entry, time_t t, long ns) { FIX_NS(t, ns); entry->stat_valid = 0; entry->ae_set |= AE_SET_MTIME; entry->ae_stat.aest_mtime = t; entry->ae_stat.aest_mtime_nsec = ns; } void archive_entry_unset_mtime(struct archive_entry *entry) { archive_entry_set_mtime(entry, 0, 0); entry->ae_set &= ~AE_SET_MTIME; } void archive_entry_set_nlink(struct archive_entry *entry, unsigned int nlink) { entry->stat_valid = 0; entry->ae_stat.aest_nlink = nlink; } void archive_entry_set_pathname(struct archive_entry *entry, const char *name) { archive_mstring_copy_mbs(&entry->ae_pathname, name); } void archive_entry_set_pathname_utf8(struct archive_entry *entry, const char *name) { archive_mstring_copy_utf8(&entry->ae_pathname, name); } void archive_entry_copy_pathname(struct archive_entry *entry, const char *name) { archive_mstring_copy_mbs(&entry->ae_pathname, name); } void archive_entry_copy_pathname_w(struct archive_entry *entry, const wchar_t *name) { archive_mstring_copy_wcs(&entry->ae_pathname, name); } int archive_entry_update_pathname_utf8(struct archive_entry *entry, const char *name) { if (archive_mstring_update_utf8(entry->archive, &entry->ae_pathname, name) == 0) return (1); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (0); } int _archive_entry_copy_pathname_l(struct archive_entry *entry, const char *name, size_t len, struct archive_string_conv *sc) { return (archive_mstring_copy_mbs_len_l(&entry->ae_pathname, name, len, sc)); } void archive_entry_set_perm(struct archive_entry *entry, mode_t p) { entry->stat_valid = 0; entry->acl.mode &= AE_IFMT; entry->acl.mode |= ~AE_IFMT & p; } void archive_entry_set_rdev(struct archive_entry *entry, dev_t m) { entry->stat_valid = 0; entry->ae_stat.aest_rdev = m; entry->ae_stat.aest_rdev_is_broken_down = 0; } void archive_entry_set_rdevmajor(struct archive_entry *entry, dev_t m) { entry->stat_valid = 0; entry->ae_stat.aest_rdev_is_broken_down = 1; entry->ae_stat.aest_rdevmajor = m; } void archive_entry_set_rdevminor(struct archive_entry *entry, dev_t m) { entry->stat_valid = 0; entry->ae_stat.aest_rdev_is_broken_down = 1; entry->ae_stat.aest_rdevminor = m; } void archive_entry_set_size(struct archive_entry *entry, la_int64_t s) { entry->stat_valid = 0; entry->ae_stat.aest_size = s; entry->ae_set |= AE_SET_SIZE; } void archive_entry_unset_size(struct archive_entry *entry) { archive_entry_set_size(entry, 0); entry->ae_set &= ~AE_SET_SIZE; } void archive_entry_copy_sourcepath(struct archive_entry *entry, const char *path) { archive_mstring_copy_mbs(&entry->ae_sourcepath, path); } void archive_entry_copy_sourcepath_w(struct archive_entry *entry, const wchar_t *path) { archive_mstring_copy_wcs(&entry->ae_sourcepath, path); } void archive_entry_set_symlink(struct archive_entry *entry, const char *linkname) { archive_mstring_copy_mbs(&entry->ae_symlink, linkname); if (linkname != NULL) entry->ae_set |= AE_SET_SYMLINK; else entry->ae_set &= ~AE_SET_SYMLINK; } void archive_entry_set_symlink_utf8(struct archive_entry *entry, const char *linkname) { archive_mstring_copy_utf8(&entry->ae_symlink, linkname); if (linkname != NULL) entry->ae_set |= AE_SET_SYMLINK; else entry->ae_set &= ~AE_SET_SYMLINK; } void archive_entry_copy_symlink(struct archive_entry *entry, const char *linkname) { archive_mstring_copy_mbs(&entry->ae_symlink, linkname); if (linkname != NULL) entry->ae_set |= AE_SET_SYMLINK; else entry->ae_set &= ~AE_SET_SYMLINK; } void archive_entry_copy_symlink_w(struct archive_entry *entry, const wchar_t *linkname) { archive_mstring_copy_wcs(&entry->ae_symlink, linkname); if (linkname != NULL) entry->ae_set |= AE_SET_SYMLINK; else entry->ae_set &= ~AE_SET_SYMLINK; } int archive_entry_update_symlink_utf8(struct archive_entry *entry, const char *linkname) { if (linkname != NULL) entry->ae_set |= AE_SET_SYMLINK; else entry->ae_set &= ~AE_SET_SYMLINK; if (archive_mstring_update_utf8(entry->archive, &entry->ae_symlink, linkname) == 0) return (1); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (0); } int _archive_entry_copy_symlink_l(struct archive_entry *entry, const char *linkname, size_t len, struct archive_string_conv *sc) { int r; r = archive_mstring_copy_mbs_len_l(&entry->ae_symlink, linkname, len, sc); if (linkname != NULL && r == 0) entry->ae_set |= AE_SET_SYMLINK; else entry->ae_set &= ~AE_SET_SYMLINK; return (r); } void archive_entry_set_uid(struct archive_entry *entry, la_int64_t u) { entry->stat_valid = 0; entry->ae_stat.aest_uid = u; } void archive_entry_set_uname(struct archive_entry *entry, const char *name) { archive_mstring_copy_mbs(&entry->ae_uname, name); } void archive_entry_set_uname_utf8(struct archive_entry *entry, const char *name) { archive_mstring_copy_utf8(&entry->ae_uname, name); } void archive_entry_copy_uname(struct archive_entry *entry, const char *name) { archive_mstring_copy_mbs(&entry->ae_uname, name); } void archive_entry_copy_uname_w(struct archive_entry *entry, const wchar_t *name) { archive_mstring_copy_wcs(&entry->ae_uname, name); } int archive_entry_update_uname_utf8(struct archive_entry *entry, const char *name) { if (archive_mstring_update_utf8(entry->archive, &entry->ae_uname, name) == 0) return (1); if (errno == ENOMEM) __archive_errx(1, "No memory"); return (0); } void archive_entry_set_is_data_encrypted(struct archive_entry *entry, char is_encrypted) { if (is_encrypted) { entry->encryption |= AE_ENCRYPTION_DATA; } else { entry->encryption &= ~AE_ENCRYPTION_DATA; } } void archive_entry_set_is_metadata_encrypted(struct archive_entry *entry, char is_encrypted) { if (is_encrypted) { entry->encryption |= AE_ENCRYPTION_METADATA; } else { entry->encryption &= ~AE_ENCRYPTION_METADATA; } } int _archive_entry_copy_uname_l(struct archive_entry *entry, const char *name, size_t len, struct archive_string_conv *sc) { return (archive_mstring_copy_mbs_len_l(&entry->ae_uname, name, len, sc)); } const void * archive_entry_mac_metadata(struct archive_entry *entry, size_t *s) { *s = entry->mac_metadata_size; return entry->mac_metadata; } void archive_entry_copy_mac_metadata(struct archive_entry *entry, const void *p, size_t s) { free(entry->mac_metadata); if (p == NULL || s == 0) { entry->mac_metadata = NULL; entry->mac_metadata_size = 0; } else { entry->mac_metadata_size = s; entry->mac_metadata = malloc(s); if (entry->mac_metadata == NULL) abort(); memcpy(entry->mac_metadata, p, s); } } /* * ACL management. The following would, of course, be a lot simpler * if: 1) the last draft of POSIX.1e were a really thorough and * complete standard that addressed the needs of ACL archiving and 2) * everyone followed it faithfully. Alas, neither is true, so the * following is a lot more complex than might seem necessary to the * uninitiated. */ struct archive_acl * archive_entry_acl(struct archive_entry *entry) { return &entry->acl; } void archive_entry_acl_clear(struct archive_entry *entry) { archive_acl_clear(&entry->acl); } /* * Add a single ACL entry to the internal list of ACL data. */ int archive_entry_acl_add_entry(struct archive_entry *entry, int type, int permset, int tag, int id, const char *name) { return archive_acl_add_entry(&entry->acl, type, permset, tag, id, name); } /* * As above, but with a wide-character name. */ int archive_entry_acl_add_entry_w(struct archive_entry *entry, int type, int permset, int tag, int id, const wchar_t *name) { return archive_acl_add_entry_w_len(&entry->acl, type, permset, tag, id, name, wcslen(name)); } /* * Return a bitmask of ACL types in an archive entry ACL list */ int archive_entry_acl_types(struct archive_entry *entry) { return (archive_acl_types(&entry->acl)); } /* * Return a count of entries matching "want_type". */ int archive_entry_acl_count(struct archive_entry *entry, int want_type) { return archive_acl_count(&entry->acl, want_type); } /* * 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_entry_acl_reset(struct archive_entry *entry, int want_type) { return archive_acl_reset(&entry->acl, want_type); } /* * Return the next ACL entry in the list. Fake entries for the * standard permissions and include them in the returned list. */ int archive_entry_acl_next(struct archive_entry *entry, int want_type, int *type, int *permset, int *tag, int *id, const char **name) { int r; r = archive_acl_next(entry->archive, &entry->acl, want_type, type, permset, tag, id, name); if (r == ARCHIVE_FATAL && errno == ENOMEM) __archive_errx(1, "No memory"); return (r); } /* * Generate a text version of the ACL. The flags parameter controls * the style of the generated ACL. */ wchar_t * archive_entry_acl_to_text_w(struct archive_entry *entry, la_ssize_t *len, int flags) { return (archive_acl_to_text_w(&entry->acl, len, flags, entry->archive)); } char * archive_entry_acl_to_text(struct archive_entry *entry, la_ssize_t *len, int flags) { return (archive_acl_to_text_l(&entry->acl, len, flags, NULL)); } char * _archive_entry_acl_to_text_l(struct archive_entry *entry, ssize_t *len, int flags, struct archive_string_conv *sc) { return (archive_acl_to_text_l(&entry->acl, len, flags, sc)); } /* * ACL text parser. */ int archive_entry_acl_from_text_w(struct archive_entry *entry, const wchar_t *wtext, int type) { return (archive_acl_from_text_w(&entry->acl, wtext, type)); } int archive_entry_acl_from_text(struct archive_entry *entry, const char *text, int type) { return (archive_acl_from_text_l(&entry->acl, text, type, NULL)); } int _archive_entry_acl_from_text_l(struct archive_entry *entry, const char *text, int type, struct archive_string_conv *sc) { return (archive_acl_from_text_l(&entry->acl, text, type, sc)); } /* Deprecated */ static int archive_entry_acl_text_compat(int *flags) { if ((*flags & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) == 0) return (1); /* ABI compat with old ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID */ if ((*flags & OLD_ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID) != 0) *flags |= ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID; /* ABI compat with old ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT */ if ((*flags & OLD_ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT) != 0) *flags |= ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT; *flags |= ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA; return (0); } /* Deprecated */ const wchar_t * archive_entry_acl_text_w(struct archive_entry *entry, int flags) { free(entry->acl.acl_text_w); entry->acl.acl_text_w = NULL; if (archive_entry_acl_text_compat(&flags) == 0) entry->acl.acl_text_w = archive_acl_to_text_w(&entry->acl, NULL, flags, entry->archive); return (entry->acl.acl_text_w); } /* Deprecated */ const char * archive_entry_acl_text(struct archive_entry *entry, int flags) { free(entry->acl.acl_text); entry->acl.acl_text = NULL; if (archive_entry_acl_text_compat(&flags) == 0) entry->acl.acl_text = archive_acl_to_text_l(&entry->acl, NULL, flags, NULL); return (entry->acl.acl_text); } /* Deprecated */ int _archive_entry_acl_text_l(struct archive_entry *entry, int flags, const char **acl_text, size_t *len, struct archive_string_conv *sc) { free(entry->acl.acl_text); entry->acl.acl_text = NULL; if (archive_entry_acl_text_compat(&flags) == 0) entry->acl.acl_text = archive_acl_to_text_l(&entry->acl, (ssize_t *)len, flags, sc); *acl_text = entry->acl.acl_text; return (0); } /* * Following code is modified from UC Berkeley sources, and * is subject to the following copyright notice. */ /*- * Copyright (c) 1993 * The Regents of the University of California. 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. * 3. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. */ +/* + * Supported file flags on FreeBSD and Mac OS: + * sappnd,sappend SF_APPEND + * arch,archived SF_ARCHIVED + * schg,schange,simmutable SF_IMMUTABLE + * sunlnk,sunlink SF_NOUNLINK (FreeBSD only) + * uappnd,uappend UF_APPEND + * compressed UF_COMPRESSED (Mac OS only) + * hidden,uhidden UF_HIDDEN + * uchg,uchange,uimmutable UF_IMMUTABLE + * nodump UF_NODUMP + * uunlnk,uunlink UF_NOUNLINK (FreeBSD only) + * offline,uoffline UF_OFFLINE (FreeBSD only) + * opaque UF_OPAQUE + * rdonly,urdonly,readonly UF_READONLY (FreeBSD only) + * reparse,ureparse UF_REPARSE (FreeBSD only) + * sparse,usparse UF_SPARSE (FreeBSD only) + * system,usystem UF_SYSTEM (FreeBSD only) + * + * See chflags(2) for more information + * + * Supported file attributes on Linux: + * a append only FS_APPEND_FL sappnd + * A no atime updates FS_NOATIME_FL atime + * c compress FS_COMPR_FL compress + * C no copy on write FS_NOCOW_FL cow + * d no dump FS_NODUMP_FL dump + * D synchronous directory updates FS_DIRSYNC_FL dirsync + * i immutable FS_IMMUTABLE_FL schg + * j data journalling FS_JOURNAL_DATA_FL journal + * P project hierarchy FS_PROJINHERIT_FL projinherit + * s secure deletion FS_SECRM_FL securedeletion + * S synchronous updates FS_SYNC_FL sync + * t no tail-merging FS_NOTAIL_FL tail + * T top of directory hierarchy FS_TOPDIR_FL topdir + * u undeletable FS_UNRM_FL undel + * + * See ioctl_iflags(2) for more information + * + * Equivalent file flags supported on FreeBSD / Mac OS and Linux: + * SF_APPEND FS_APPEND_FL sappnd + * SF_IMMUTABLE FS_IMMUTABLE_FL schg + * UF_NODUMP FS_NODUMP_FL nodump + */ + static const struct flag { const char *name; const wchar_t *wname; unsigned long set; unsigned long clear; } flags[] = { /* Preferred (shorter) names per flag first, all prefixed by "no" */ #ifdef SF_APPEND - { "nosappnd", L"nosappnd", SF_APPEND, 0 }, - { "nosappend", L"nosappend", SF_APPEND, 0 }, + { "nosappnd", L"nosappnd", SF_APPEND, 0}, + { "nosappend", L"nosappend", SF_APPEND, 0}, #endif #if defined(FS_APPEND_FL) /* 'a' */ - { "nosappnd", L"nosappnd", FS_APPEND_FL, 0 }, - { "nosappend", L"nosappend", FS_APPEND_FL, 0 }, + { "nosappnd", L"nosappnd", FS_APPEND_FL, 0}, + { "nosappend", L"nosappend", FS_APPEND_FL, 0}, #elif defined(EXT2_APPEND_FL) /* 'a' */ - { "nosappnd", L"nosappnd", EXT2_APPEND_FL, 0 }, - { "nosappend", L"nosappend", EXT2_APPEND_FL, 0 }, + { "nosappnd", L"nosappnd", EXT2_APPEND_FL, 0}, + { "nosappend", L"nosappend", EXT2_APPEND_FL, 0}, #endif #ifdef SF_ARCHIVED - { "noarch", L"noarch", SF_ARCHIVED, 0 }, - { "noarchived", L"noarchived", SF_ARCHIVED, 0 }, + { "noarch", L"noarch", SF_ARCHIVED, 0}, + { "noarchived", L"noarchived", SF_ARCHIVED, 0}, #endif #ifdef SF_IMMUTABLE - { "noschg", L"noschg", SF_IMMUTABLE, 0 }, - { "noschange", L"noschange", SF_IMMUTABLE, 0 }, - { "nosimmutable", L"nosimmutable", SF_IMMUTABLE, 0 }, + { "noschg", L"noschg", SF_IMMUTABLE, 0}, + { "noschange", L"noschange", SF_IMMUTABLE, 0}, + { "nosimmutable", L"nosimmutable", SF_IMMUTABLE, 0}, #endif #if defined(FS_IMMUTABLE_FL) /* 'i' */ - { "noschg", L"noschg", FS_IMMUTABLE_FL, 0 }, - { "noschange", L"noschange", FS_IMMUTABLE_FL, 0 }, - { "nosimmutable", L"nosimmutable", FS_IMMUTABLE_FL, 0 }, + { "noschg", L"noschg", FS_IMMUTABLE_FL, 0}, + { "noschange", L"noschange", FS_IMMUTABLE_FL, 0}, + { "nosimmutable", L"nosimmutable", FS_IMMUTABLE_FL, 0}, #elif defined(EXT2_IMMUTABLE_FL) /* 'i' */ - { "noschg", L"noschg", EXT2_IMMUTABLE_FL, 0 }, - { "noschange", L"noschange", EXT2_IMMUTABLE_FL, 0 }, - { "nosimmutable", L"nosimmutable", EXT2_IMMUTABLE_FL, 0 }, + { "noschg", L"noschg", EXT2_IMMUTABLE_FL, 0}, + { "noschange", L"noschange", EXT2_IMMUTABLE_FL, 0}, + { "nosimmutable", L"nosimmutable", EXT2_IMMUTABLE_FL, 0}, #endif #ifdef SF_NOUNLINK - { "nosunlnk", L"nosunlnk", SF_NOUNLINK, 0 }, - { "nosunlink", L"nosunlink", SF_NOUNLINK, 0 }, + { "nosunlnk", L"nosunlnk", SF_NOUNLINK, 0}, + { "nosunlink", L"nosunlink", SF_NOUNLINK, 0}, #endif -#ifdef SF_SNAPSHOT - { "nosnapshot", L"nosnapshot", SF_SNAPSHOT, 0 }, -#endif #ifdef UF_APPEND - { "nouappnd", L"nouappnd", UF_APPEND, 0 }, - { "nouappend", L"nouappend", UF_APPEND, 0 }, + { "nouappnd", L"nouappnd", UF_APPEND, 0}, + { "nouappend", L"nouappend", UF_APPEND, 0}, #endif #ifdef UF_IMMUTABLE - { "nouchg", L"nouchg", UF_IMMUTABLE, 0 }, - { "nouchange", L"nouchange", UF_IMMUTABLE, 0 }, - { "nouimmutable", L"nouimmutable", UF_IMMUTABLE, 0 }, + { "nouchg", L"nouchg", UF_IMMUTABLE, 0}, + { "nouchange", L"nouchange", UF_IMMUTABLE, 0}, + { "nouimmutable", L"nouimmutable", UF_IMMUTABLE, 0}, #endif #ifdef UF_NODUMP { "nodump", L"nodump", 0, UF_NODUMP}, #endif #if defined(FS_NODUMP_FL) /* 'd' */ { "nodump", L"nodump", 0, FS_NODUMP_FL}, -#elif defined(EXT2_NODUMP_FL) /* 'd' */ +#elif defined(EXT2_NODUMP_FL) { "nodump", L"nodump", 0, EXT2_NODUMP_FL}, #endif #ifdef UF_OPAQUE - { "noopaque", L"noopaque", UF_OPAQUE, 0 }, + { "noopaque", L"noopaque", UF_OPAQUE, 0}, #endif #ifdef UF_NOUNLINK - { "nouunlnk", L"nouunlnk", UF_NOUNLINK, 0 }, - { "nouunlink", L"nouunlink", UF_NOUNLINK, 0 }, + { "nouunlnk", L"nouunlnk", UF_NOUNLINK, 0}, + { "nouunlink", L"nouunlink", UF_NOUNLINK, 0}, #endif #ifdef UF_COMPRESSED - { "nocompressed",L"nocompressed", UF_COMPRESSED, 0 }, + /* Mac OS */ + { "nocompressed", L"nocompressed", UF_COMPRESSED, 0}, #endif #ifdef UF_HIDDEN - { "nohidden", L"nohidden", UF_HIDDEN, 0 }, + { "nohidden", L"nohidden", UF_HIDDEN, 0}, + { "nouhidden", L"nouhidden", UF_HIDDEN, 0}, #endif -#if defined(FS_UNRM_FL) - { "nouunlink", L"nouunlink", FS_UNRM_FL, 0}, -#elif defined(EXT2_UNRM_FL) - { "nouunlink", L"nouunlink", EXT2_UNRM_FL, 0}, +#ifdef UF_OFFLINE + { "nooffline", L"nooffline", UF_OFFLINE, 0}, + { "nouoffline", L"nouoffline", UF_OFFLINE, 0}, #endif - -#if defined(FS_BTREE_FL) - { "nobtree", L"nobtree", FS_BTREE_FL, 0 }, -#elif defined(EXT2_BTREE_FL) - { "nobtree", L"nobtree", EXT2_BTREE_FL, 0 }, +#ifdef UF_READONLY + { "nordonly", L"nordonly", UF_READONLY, 0}, + { "nourdonly", L"nourdonly", UF_READONLY, 0}, + { "noreadonly", L"noreadonly", UF_READONLY, 0}, #endif - -#if defined(FS_ECOMPR_FL) - { "nocomperr", L"nocomperr", FS_ECOMPR_FL, 0 }, -#elif defined(EXT2_ECOMPR_FL) - { "nocomperr", L"nocomperr", EXT2_ECOMPR_FL, 0 }, +#ifdef UF_SPARSE + { "nosparse", L"nosparse", UF_SPARSE, 0}, + { "nousparse", L"nousparse", UF_SPARSE, 0}, #endif - -#if defined(FS_COMPR_FL) /* 'c' */ - { "nocompress", L"nocompress", FS_COMPR_FL, 0 }, -#elif defined(EXT2_COMPR_FL) /* 'c' */ - { "nocompress", L"nocompress", EXT2_COMPR_FL, 0 }, +#ifdef UF_REPARSE + { "noreparse", L"noreparse", UF_REPARSE, 0}, + { "noureparse", L"noureparse", UF_REPARSE, 0}, #endif - -#if defined(FS_NOATIME_FL) /* 'A' */ - { "noatime", L"noatime", 0, FS_NOATIME_FL}, -#elif defined(EXT2_NOATIME_FL) /* 'A' */ - { "noatime", L"noatime", 0, EXT2_NOATIME_FL}, +#ifdef UF_SYSTEM + { "nosystem", L"nosystem", UF_SYSTEM, 0}, + { "nousystem", L"nousystem", UF_SYSTEM, 0}, #endif +#if defined(FS_UNRM_FL) /* 'u' */ + { "noundel", L"noundel", FS_UNRM_FL, 0}, +#elif defined(EXT2_UNRM_FL) + { "noundel", L"noundel", EXT2_UNRM_FL, 0}, +#endif -#if defined(FS_DIRTY_FL) - { "nocompdirty",L"nocompdirty", FS_DIRTY_FL, 0}, -#elif defined(EXT2_DIRTY_FL) - { "nocompdirty",L"nocompdirty", EXT2_DIRTY_FL, 0}, +#if defined(FS_COMPR_FL) /* 'c' */ + { "nocompress", L"nocompress", FS_COMPR_FL, 0}, +#elif defined(EXT2_COMPR_FL) + { "nocompress", L"nocompress", EXT2_COMPR_FL, 0}, #endif -#if defined(FS_COMPRBLK_FL) -#if defined(FS_NOCOMPR_FL) - { "nocomprblk", L"nocomprblk", FS_COMPRBLK_FL, FS_NOCOMPR_FL}, -#else - { "nocomprblk", L"nocomprblk", FS_COMPRBLK_FL, 0}, +#if defined(FS_NOATIME_FL) /* 'A' */ + { "noatime", L"noatime", 0, FS_NOATIME_FL}, +#elif defined(EXT2_NOATIME_FL) + { "noatime", L"noatime", 0, EXT2_NOATIME_FL}, #endif -#elif defined(EXT2_COMPRBLK_FL) -#if defined(EXT2_NOCOMPR_FL) - { "nocomprblk", L"nocomprblk", EXT2_COMPRBLK_FL, EXT2_NOCOMPR_FL}, -#else - { "nocomprblk", L"nocomprblk", EXT2_COMPRBLK_FL, 0}, -#endif -#endif -#if defined(FS_DIRSYNC_FL) - { "nodirsync", L"nodirsync", FS_DIRSYNC_FL, 0}, +#if defined(FS_DIRSYNC_FL) /* 'D' */ + { "nodirsync", L"nodirsync", FS_DIRSYNC_FL, 0}, #elif defined(EXT2_DIRSYNC_FL) - { "nodirsync", L"nodirsync", EXT2_DIRSYNC_FL, 0}, + { "nodirsync", L"nodirsync", EXT2_DIRSYNC_FL, 0}, #endif -#if defined(FS_INDEX_FL) - { "nohashidx", L"nohashidx", FS_INDEX_FL, 0}, -#elif defined(EXT2_INDEX_FL) - { "nohashidx", L"nohashidx", EXT2_INDEX_FL, 0}, -#endif -#if defined(FS_IMAGIC_FL) - { "noimagic", L"noimagic", FS_IMAGIC_FL, 0}, -#elif defined(EXT2_IMAGIC_FL) - { "noimagic", L"noimagic", EXT2_IMAGIC_FL, 0}, -#endif -#if defined(FS_JOURNAL_DATA_FL) - { "nojournal", L"nojournal", FS_JOURNAL_DATA_FL, 0}, +#if defined(FS_JOURNAL_DATA_FL) /* 'j' */ + { "nojournal-data",L"nojournal-data", FS_JOURNAL_DATA_FL, 0}, + { "nojournal", L"nojournal", FS_JOURNAL_DATA_FL, 0}, #elif defined(EXT3_JOURNAL_DATA_FL) - { "nojournal", L"nojournal", EXT3_JOURNAL_DATA_FL, 0}, + { "nojournal-data",L"nojournal-data", EXT3_JOURNAL_DATA_FL, 0}, + { "nojournal", L"nojournal", EXT3_JOURNAL_DATA_FL, 0}, #endif -#if defined(FS_SECRM_FL) - { "nosecuredeletion",L"nosecuredeletion",FS_SECRM_FL, 0}, +#if defined(FS_SECRM_FL) /* 's' */ + { "nosecdel", L"nosecdel", FS_SECRM_FL, 0}, + { "nosecuredeletion",L"nosecuredeletion",FS_SECRM_FL, 0}, #elif defined(EXT2_SECRM_FL) - { "nosecuredeletion",L"nosecuredeletion",EXT2_SECRM_FL, 0}, + { "nosecdel", L"nosecdel", EXT2_SECRM_FL, 0}, + { "nosecuredeletion",L"nosecuredeletion",EXT2_SECRM_FL, 0}, #endif -#if defined(FS_SYNC_FL) - { "nosync", L"nosync", FS_SYNC_FL, 0}, +#if defined(FS_SYNC_FL) /* 'S' */ + { "nosync", L"nosync", FS_SYNC_FL, 0}, #elif defined(EXT2_SYNC_FL) - { "nosync", L"nosync", EXT2_SYNC_FL, 0}, + { "nosync", L"nosync", EXT2_SYNC_FL, 0}, #endif -#if defined(FS_NOTAIL_FL) - { "notail", L"notail", 0, FS_NOTAIL_FL}, +#if defined(FS_NOTAIL_FL) /* 't' */ + { "notail", L"notail", 0, FS_NOTAIL_FL}, #elif defined(EXT2_NOTAIL_FL) - { "notail", L"notail", 0, EXT2_NOTAIL_FL}, + { "notail", L"notail", 0, EXT2_NOTAIL_FL}, #endif -#if defined(FS_TOPDIR_FL) - { "notopdir", L"notopdir", FS_TOPDIR_FL, 0}, +#if defined(FS_TOPDIR_FL) /* 'T' */ + { "notopdir", L"notopdir", FS_TOPDIR_FL, 0}, #elif defined(EXT2_TOPDIR_FL) - { "notopdir", L"notopdir", EXT2_TOPDIR_FL, 0}, + { "notopdir", L"notopdir", EXT2_TOPDIR_FL, 0}, #endif -#ifdef FS_ENCRYPT_FL - { "noencrypt", L"noencrypt", FS_ENCRYPT_FL, 0}, +#ifdef FS_NOCOW_FL /* 'C' */ + { "nocow", L"nocow", 0, FS_NOCOW_FL}, #endif -#ifdef FS_HUGE_FILE_FL - { "nohugefile", L"nohugefile", FS_HUGE_FILE_FL, 0}, +#ifdef FS_PROJINHERIT_FL /* 'P' */ + { "noprojinherit",L"noprojinherit", FS_PROJINHERIT_FL, 0}, #endif -#ifdef FS_EXTENT_FL - { "noextent", L"noextent", FS_EXTENT_FL, 0}, -#endif -#ifdef FS_EA_INODE_FL - { "noeainode", L"noeainode", FS_EA_INODE_FL, 0}, -#endif -#ifdef FS_EOFBLOCKS_FL - { "noeofblocks",L"noeofblocks", FS_EOFBLOCKS_FL, 0}, -#endif -#ifdef FS_NOCOW_FL - { "nocow", L"nocow", FS_NOCOW_FL, 0}, -#endif -#ifdef FS_INLINE_DATA_FL - { "noinlinedata",L"noinlinedata", FS_INLINE_DATA_FL, 0}, -#endif -#ifdef FS_PROJINHERIT_FL - { "noprojinherit",L"noprojinherit", FS_PROJINHERIT_FL, 0}, -#endif -#if defined(FS_RESERVED_FL) - { "noreserved", L"noreserved", FS_RESERVED_FL, 0}, -#elif defined(EXT2_RESERVED_FL) - { "noreserved", L"noreserved", EXT2_RESERVED_FL, 0}, -#endif - { NULL, NULL, 0, 0 } + { NULL, NULL, 0, 0} }; /* * fflagstostr -- * Convert file flags to a comma-separated string. If no flags * are set, return the empty string. */ static char * ae_fflagstostr(unsigned long bitset, unsigned long bitclear) { char *string, *dp; const char *sp; unsigned long bits; const struct flag *flag; size_t length; bits = bitset | bitclear; length = 0; for (flag = flags; flag->name != NULL; flag++) if (bits & (flag->set | flag->clear)) { length += strlen(flag->name) + 1; bits &= ~(flag->set | flag->clear); } if (length == 0) return (NULL); string = (char *)malloc(length); if (string == NULL) return (NULL); dp = string; for (flag = flags; flag->name != NULL; flag++) { if (bitset & flag->set || bitclear & flag->clear) { sp = flag->name + 2; } else if (bitset & flag->clear || bitclear & flag->set) { sp = flag->name; } else continue; bitset &= ~(flag->set | flag->clear); bitclear &= ~(flag->set | flag->clear); if (dp > string) *dp++ = ','; while ((*dp++ = *sp++) != '\0') ; dp--; } *dp = '\0'; return (string); } /* * strtofflags -- * Take string of arguments and return file flags. This * version works a little differently than strtofflags(3). * In particular, it always tests every token, skipping any * unrecognized tokens. It returns a pointer to the first * unrecognized token, or NULL if every token was recognized. * This version is also const-correct and does not modify the * provided string. */ static const char * ae_strtofflags(const char *s, unsigned long *setp, unsigned long *clrp) { const char *start, *end; const struct flag *flag; unsigned long set, clear; const char *failed; set = clear = 0; start = s; failed = NULL; /* Find start of first token. */ while (*start == '\t' || *start == ' ' || *start == ',') start++; while (*start != '\0') { size_t length; /* Locate end of token. */ end = start; while (*end != '\0' && *end != '\t' && *end != ' ' && *end != ',') end++; length = end - start; for (flag = flags; flag->name != NULL; flag++) { size_t flag_length = strlen(flag->name); if (length == flag_length && memcmp(start, flag->name, length) == 0) { /* Matched "noXXXX", so reverse the sense. */ clear |= flag->set; set |= flag->clear; break; } else if (length == flag_length - 2 && memcmp(start, flag->name + 2, length) == 0) { /* Matched "XXXX", so don't reverse. */ set |= flag->set; clear |= flag->clear; break; } } /* Ignore unknown flag names. */ if (flag->name == NULL && failed == NULL) failed = start; /* Find start of next token. */ start = end; while (*start == '\t' || *start == ' ' || *start == ',') start++; } if (setp) *setp = set; if (clrp) *clrp = clear; /* Return location of first failure. */ return (failed); } /* * wcstofflags -- * Take string of arguments and return file flags. This * version works a little differently than strtofflags(3). * In particular, it always tests every token, skipping any * unrecognized tokens. It returns a pointer to the first * unrecognized token, or NULL if every token was recognized. * This version is also const-correct and does not modify the * provided string. */ static const wchar_t * ae_wcstofflags(const wchar_t *s, unsigned long *setp, unsigned long *clrp) { const wchar_t *start, *end; const struct flag *flag; unsigned long set, clear; const wchar_t *failed; set = clear = 0; start = s; failed = NULL; /* Find start of first token. */ while (*start == L'\t' || *start == L' ' || *start == L',') start++; while (*start != L'\0') { size_t length; /* Locate end of token. */ end = start; while (*end != L'\0' && *end != L'\t' && *end != L' ' && *end != L',') end++; length = end - start; for (flag = flags; flag->wname != NULL; flag++) { size_t flag_length = wcslen(flag->wname); if (length == flag_length && wmemcmp(start, flag->wname, length) == 0) { /* Matched "noXXXX", so reverse the sense. */ clear |= flag->set; set |= flag->clear; break; } else if (length == flag_length - 2 && wmemcmp(start, flag->wname + 2, length) == 0) { /* Matched "XXXX", so don't reverse. */ set |= flag->set; clear |= flag->clear; break; } } /* Ignore unknown flag names. */ if (flag->wname == NULL && failed == NULL) failed = start; /* Find start of next token. */ start = end; while (*start == L'\t' || *start == L' ' || *start == L',') start++; } if (setp) *setp = set; if (clrp) *clrp = clear; /* Return location of first failure. */ return (failed); } #ifdef TEST #include int main(int argc, char **argv) { struct archive_entry *entry = archive_entry_new(); unsigned long set, clear; const wchar_t *remainder; remainder = archive_entry_copy_fflags_text_w(entry, L"nosappnd dump archive,,,,,,,"); archive_entry_fflags(entry, &set, &clear); wprintf(L"set=0x%lX clear=0x%lX remainder='%ls'\n", set, clear, remainder); wprintf(L"new flags='%s'\n", archive_entry_fflags_text(entry)); return (0); } #endif Index: head/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c (revision 345496) +++ head/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c (revision 345497) @@ -1,1041 +1,1044 @@ /*- * 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" __FBSDID("$FreeBSD"); /* 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); 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 (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 + 1; char *linkbuffer; int lnklen; linkbuffer = malloc(linkbuffer_len); 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; } 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; 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); 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"); 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(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { char buff[512]; char *list, *p; ssize_t list_size; const char *path; int namespace = EXTATTR_NAMESPACE_USER; 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) { 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; 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); } #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__) */ Index: head/contrib/libarchive/libarchive/archive_read_support_format_rar5.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_format_rar5.c (revision 345496) +++ head/contrib/libarchive/libarchive/archive_read_support_format_rar5.c (revision 345497) @@ -1,3496 +1,3497 @@ /*- * 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 #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" #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) #else #define DEBUG_CODE if(0) #endif /* Real RAR5 magic number is: * * 0x52, 0x61, 0x72, 0x21, 0x1a, 0x07, 0x01, 0x00 * "Rar!→•☺·\x00" * * It's stored in `rar5_signature` after 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[] = { 243, 192, 211, 128, 187, 166, 160, 161 }; static const ssize_t rar5_signature_size = sizeof(rar5_signature); /* static const size_t g_unpack_buf_chunk_size = 1024; */ static const size_t g_unpack_window_size = 0x20000; 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? */ /* Optional time fields. */ uint64_t e_mtime; uint64_t e_ctime; uint64_t e_atime; uint32_t e_unix_ns; /* Optional hash fields. */ uint32_t stored_crc32; uint32_t calculated_crc32; uint8_t blake2sp[32]; blake2sp_state b2state; char has_blake2; }; 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; int notused : 4; 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. */ size_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; 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 { 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; /* 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; }; /* Forward function declarations. */ 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); /* 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 struct rar5* get_context(struct archive_read* a) { return (struct rar5*) a->format->data; } /* Convenience functions used by filter implementations. */ static uint32_t read_filter_data(struct rar5* rar, uint32_t offset) { return archive_le32dec(&rar->cstate.window_buf[offset]); } static void write_filter_data(struct rar5* rar, uint32_t offset, uint32_t value) { archive_le32enc(&rar->cstate.filtered_buf[offset], value); } static void circular_memcpy(uint8_t* dst, uint8_t* window, const int 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)); } } /* 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 = (struct filter_info*) calloc(1, sizeof(struct filter_info)); 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; + const int mask = (int)rar->cstate.window_mask; circular_memcpy(rar->cstate.filtered_buf, rar->cstate.window_buf, - rar->cstate.window_mask, + 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]; + flt->block_start + i++) & 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, (rar->cstate.solid_offset + + 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, i, addr + file_size); + 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, i, naddr); + 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; - const int mask = rar->cstate.window_mask; + const int mask = (int)rar->cstate.window_mask; circular_memcpy(rar->cstate.filtered_buf, rar->cstate.window_buf, - rar->cstate.window_mask, + 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) & mask]; if(b[3] == 0xEB) { /* 0xEB = ARM's BL (branch + link) instruction. */ offset = read_filter_data(rar, (rar->cstate.solid_offset + flt->block_start + i) & mask) & 0x00ffffff; offset -= (uint32_t) ((i + flt->block_start) / 4); offset = (offset & 0x00ffffff) | 0xeb000000; - write_filter_data(rar, i, offset); + 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); 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%02x", 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 int wmask = rar->cstate.window_mask; + const int wmask = (int)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; 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) { if(!ptr) return 0; ssize_t avail = -1; *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 * the consuming process itself. */ 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 rar5* rar, const uint8_t* p, uint32_t* value) { uint32_t bits = 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 rar5* rar, const uint8_t* p, uint16_t* value) { int bits = (int) 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 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(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 int 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) ? 1 : 0; } static int 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) ? 1 : 0; } static int bid_standard(struct archive_read* a) { const uint8_t* p; if(!read_ahead(a, rar5_signature_size, &p)) return -1; if(!memcmp(rar5_signature, p, rar5_signature_size)) return 30; return -1; } 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; } 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"; } 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; if(!read_var(a, &locator_flags, NULL)) { return ARCHIVE_EOF; } enum LOCATOR_FLAGS { QLIST = 0x01, RECOVERY = 0x02, }; 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, ssize_t* extra_data_size) { size_t hash_type; size_t value_len; 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; } enum HASH_TYPE { BLAKE2sp = 0x00 }; /* 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%02x)", (int) hash_type); return ARCHIVE_FATAL; } return ARCHIVE_OK; } static uint64_t time_win_to_unix(uint64_t win_time) { const size_t ns_in_sec = 10000000; const uint64_t sec_to_unix = 11644473600LL; return win_time / ns_in_sec - sec_to_unix; } static int parse_htime_item(struct archive_read* a, char unix_time, uint64_t* where, ssize_t* extra_data_size) { if(unix_time) { uint32_t time_val; if(!read_u32(a, &time_val)) return ARCHIVE_EOF; *extra_data_size -= 4; *where = (uint64_t) time_val; } else { uint64_t windows_time; if(!read_u64(a, &windows_time)) return ARCHIVE_EOF; *where = time_win_to_unix(windows_time); *extra_data_size -= 8; } return ARCHIVE_OK; } static int parse_file_extra_htime(struct archive_read* a, struct archive_entry* e, struct rar5* rar, ssize_t* extra_data_size) { char unix_time = 0; size_t flags; 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; if(flags & HAS_MTIME) { parse_htime_item(a, unix_time, &rar->file.e_mtime, extra_data_size); archive_entry_set_mtime(e, rar->file.e_mtime, 0); } if(flags & HAS_CTIME) { parse_htime_item(a, unix_time, &rar->file.e_ctime, extra_data_size); archive_entry_set_ctime(e, rar->file.e_ctime, 0); } if(flags & HAS_ATIME) { parse_htime_item(a, unix_time, &rar->file.e_atime, extra_data_size); archive_entry_set_atime(e, rar->file.e_atime, 0); } if(flags & HAS_UNIX_NS) { if(!read_u32(a, &rar->file.e_unix_ns)) return ARCHIVE_EOF; *extra_data_size -= 4; } return ARCHIVE_OK; } static int process_head_file_extra(struct archive_read* a, struct archive_entry* e, struct rar5* rar, ssize_t extra_data_size) { size_t extra_field_size; size_t extra_field_id = 0; int ret = ARCHIVE_FATAL; size_t var_size; enum EXTRA { CRYPT = 0x01, HASH = 0x02, HTIME = 0x03, VERSION_ = 0x04, REDIR = 0x05, UOWNER = 0x06, SUBDATA = 0x07 }; while(extra_data_size > 0) { if(!read_var_sized(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_sized(a, &extra_field_id, &var_size)) return ARCHIVE_EOF; extra_data_size -= var_size; if(ARCHIVE_OK != consume(a, var_size)) { return ARCHIVE_EOF; } switch(extra_field_id) { case HASH: ret = parse_file_extra_hash(a, rar, &extra_data_size); break; case HTIME: ret = parse_file_extra_htime(a, e, rar, &extra_data_size); break; case CRYPT: /* fallthrough */ case VERSION_: /* fallthrough */ case REDIR: /* fallthrough */ case UOWNER: /* fallthrough */ case SUBDATA: /* fallthrough */ default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unknown extra field in file/service block: 0x%02x", (int) extra_field_id); return ARCHIVE_FATAL; } } if(ret != ARCHIVE_OK) { /* Attribute not implemented. */ return ret; } return ARCHIVE_OK; } static int process_head_file(struct archive_read* a, struct rar5* rar, struct archive_entry* entry, size_t block_flags) { ssize_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; uint32_t mtime = 0, crc = 0; int c_method = 0, c_version = 0, is_dir; char name_utf8_buf[2048 * 4]; const uint8_t* p; 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) { size_t edata_size = 0; if(!read_var_sized(a, &edata_size, NULL)) return ARCHIVE_EOF; /* Intentional type cast from unsigned to signed. */ extra_data_size = (ssize_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; } enum FILE_FLAGS { DIRECTORY = 0x0001, UTIME = 0x0002, CRC32 = 0x0004, UNKNOWN_UNPACKED_SIZE = 0x0008, }; enum COMP_INFO_FLAGS { SOLID = 0x0040, }; 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; } is_dir = (int) (file_flags & DIRECTORY); 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); rar->cstate.window_size = is_dir ? 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; rar->file.service = 0; if(!read_var_sized(a, &host_os, NULL)) return ARCHIVE_EOF; enum HOST_OS { HOST_WINDOWS = 0, HOST_UNIX = 1, }; if(host_os == HOST_WINDOWS) { /* Host OS is Windows */ unsigned short mode = 0660; if(is_dir) mode |= AE_IFDIR; else mode |= AE_IFREG; archive_entry_set_mode(entry, mode); } else if(host_os == HOST_UNIX) { /* Host OS is Unix */ archive_entry_set_mode(entry, (unsigned short) file_attr); } else { /* Unknown host OS */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported Host OS: 0x%02x", (int) host_os); return ARCHIVE_FATAL; } if(!read_var_sized(a, &name_size, NULL)) return ARCHIVE_EOF; if(!read_ahead(a, name_size, &p)) return ARCHIVE_EOF; if(name_size > 2047) { 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; } memcpy(name_utf8_buf, p, name_size); name_utf8_buf[name_size] = 0; if(ARCHIVE_OK != consume(a, name_size)) { return ARCHIVE_EOF; } if(extra_data_size > 0) { int ret = process_head_file_extra(a, entry, rar, extra_data_size); /* Sanity check. */ 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; 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; } archive_entry_update_pathname_utf8(entry, name_utf8_buf); 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) { (void) entry; int ret; size_t extra_data_size = 0; size_t extra_field_size = 0; size_t extra_field_id = 0; size_t archive_flags = 0; if(block_flags & HFL_EXTRA_DATA) { if(!read_var_sized(a, &extra_data_size, NULL)) return ARCHIVE_EOF; } else { extra_data_size = 0; } if(!read_var_sized(a, &archive_flags, NULL)) { return ARCHIVE_EOF; } 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 */ }; 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; } rar->main.vol_no = (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; } enum MAIN_EXTRA { // Just one attribute here. LOCATOR = 0x01, }; 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%02x)", (int) extra_field_id); return ARCHIVE_FATAL; } 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) { 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; /* Skip any unprocessed data for this file. */ if(rar->file.bytes_remaining) { ret = rar5_read_data_skip(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; } /* Sanity check, maximum header size for RAR5 is 2MB. */ if(raw_hdr_size > (2 * 1024 * 1024)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Base block header is too large"); return ARCHIVE_FATAL; } hdr_size = raw_hdr_size + hdr_size_len; /* 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) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Header CRC error"); return ARCHIVE_FATAL; } /* 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 = hdr_size; - rar->generic.last_header_id = header_id; + 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. */ enum HEADER_TYPE { HEAD_MARK = 0x00, HEAD_MAIN = 0x01, HEAD_FILE = 0x02, HEAD_SERVICE = 0x03, HEAD_CRYPT = 0x04, HEAD_ENDARC = 0x05, HEAD_UNKNOWN = 0xff, }; 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_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 { 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(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 rar5_read_header(struct archive_read *a, struct archive_entry *entry) { struct rar5* rar = get_context(a); int ret; if(rar->header_initialized == 0) { init_header(a); rar->header_initialized = 1; } if(rar->skipped_magic == 0) { if(ARCHIVE_OK != consume(a, rar5_signature_size)) { 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; if (rar->cstate.window_size) rar->cstate.window_mask = rar->cstate.window_size - 1; else rar->cstate.window_mask = 0; free(rar->cstate.window_buf); free(rar->cstate.filtered_buf); rar->cstate.window_buf = calloc(1, rar->cstate.window_size); rar->cstate.filtered_buf = calloc(1, rar->cstate.window_size); 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, 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 = 1 << table->quick_bits; + 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; uint16_t bitfield; uint32_t pos; struct rar5* rar = get_context(a); if(ARCHIVE_OK != read_bits_16(rar, p, &bitfield)) { return ARCHIVE_EOF; } 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;) { 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; 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++; continue; } if(num < 18) { /* 16..17: repeat previous code */ uint16_t n; if(ARCHIVE_OK != read_bits_16(rar, p, &n)) return ARCHIVE_EOF; 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; } continue; } /* other codes: fill with zeroes `n` times */ uint16_t n; if(ARCHIVE_OK != read_bits_16(rar, p, &n)) return ARCHIVE_EOF; 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) { 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. */ uint8_t 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) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Block checksum error: got 0x%02x, expected 0x%02x", hdr->block_cksum, calculated_cksum); return ARCHIVE_FATAL; } return ARCHIVE_OK; } /* Convenience function used during filter processing. */ static int parse_filter_data(struct rar5* rar, const uint8_t* p, uint32_t* filter_data) { int i, bytes; uint32_t data = 0; if(ARCHIVE_OK != read_consume_bits(rar, p, 2, &bytes)) return ARCHIVE_EOF; bytes++; for(i = 0; i < bytes; i++) { uint16_t byte; if(ARCHIVE_OK != read_bits_16(rar, p, &byte)) { return ARCHIVE_EOF; } data += (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 rar5* rar = get_context(ar); /* Read the parameters from the input stream. */ if(ARCHIVE_OK != parse_filter_data(rar, p, &block_start)) return ARCHIVE_EOF; if(ARCHIVE_OK != parse_filter_data(rar, p, &block_length)) return ARCHIVE_EOF; if(ARCHIVE_OK != read_bits_16(rar, p, &filter_type)) return ARCHIVE_EOF; 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. */ struct filter_info* 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 != read_consume_bits(rar, p, 5, &channels)) return ARCHIVE_EOF; filt->channels = channels + 1; } return ARCHIVE_OK; } static int decode_code_length(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(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 int cmask = rar->cstate.window_mask; + const int cmask = (int)rar->cstate.window_mask; const int64_t write_ptr = rar->cstate.write_ptr + rar->cstate.solid_offset; int i; /* 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 int cmask = rar->cstate.window_mask; + const int cmask = (int)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(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; dist += (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 != read_bits_32(rar, p, &add)) { /* Return EOF if we can't read more data. */ return ARCHIVE_EOF; } 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; } dist += low_dist; } else { /* dbits is one of [0,1,2,3] */ int add; if(ARCHIVE_OK != read_consume_bits(rar, p, dbits, &add)) { /* Return EOF if we can't read more data. */ return ARCHIVE_EOF; } 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 if(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(rar, p, len_slot); rar->cstate.last_len = len; if(ARCHIVE_OK != copy_string(a, len, dist)) return ARCHIVE_FATAL; continue; } /* The program counter shouldn't reach here. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported block code: 0x%02x", num); return ARCHIVE_FATAL; } 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; /* 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. */ while(1) { if(!read_ahead(a, chunk_size, &p)) return ARCHIVE_EOF; for(i = 0; i < chunk_size - rar5_signature_size; i++) { if(memcmp(&p[i], rar5_signature, rar5_signature_size) == 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 + rar5_signature_size); 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) { rar->main.endarc = 0; while(ARCHIVE_RETRY == skip_base_block(a)); 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); /* 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; /* 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) { ret = advance_multivolume(a); 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; rar->cstate.block_parsing_finished = 0; /* 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. */ ssize_t 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. */ ssize_t 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->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 { 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; } /* 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; } 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. */ 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) { 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; } } size_t 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: return uncompress_file(a); default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Compression method not supported: 0x%08x", 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: %08x, expected: %08x)\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 (%08x/%08x)\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); } 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(!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) { /* 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. */ 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) { /* 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); 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 0; } static int rar5_has_encrypted_entries(struct archive_read *_a) { (void) _a; /* Unsupported for now. */ return ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED; } static int rar5_init(struct rar5* rar) { ssize_t i; memset(rar, 0, sizeof(struct rar5)); /* Decrypt the magic signature pattern. Check the comment near the * `rar5_signature` symbol to read the rationale behind this. */ if(rar5_signature[0] == 243) { for(i = 0; i < rar5_signature_size; i++) { rar5_signature[i] ^= 0xA1; } } if(CDE_OK != cdeque_init(&rar->cstate.filters, 8192)) return ARCHIVE_FATAL; 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"); 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; } Index: head/contrib/libarchive/libarchive/archive_read_support_format_zip.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_format_zip.c (revision 345496) +++ head/contrib/libarchive/libarchive/archive_read_support_format_zip.c (revision 345497) @@ -1,3888 +1,3913 @@ /*- * 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" __FBSDID("$FreeBSD$"); /* * 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 #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_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) #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; 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 data */ unsigned long entry_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 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)) /* 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, NULL); + 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) */ {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 "??"; } /* Convert an MSDOS-style date/time into Unix-style time. */ static time_t zip_time(const char *p) { int msTime, msDate; struct tm ts; msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]); msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]); memset(&ts, 0, sizeof(ts)); ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ ts.tm_mday = msDate & 0x1f; /* Day of month. */ ts.tm_hour = (msTime >> 11) & 0x1f; ts.tm_min = (msTime >> 5) & 0x3f; ts.tm_sec = (msTime << 1) & 0x3e; ts.tm_isdst = -1; return mktime(&ts); } /* * 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, const char *p, size_t extra_length, struct zip_entry* zip_entry) { unsigned offset = 0; if (extra_length == 0) { return ARCHIVE_OK; } if (extra_length < 4) { 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; } 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 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; } if (offset != extra_length) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Malformed extra data: Consumed %d bytes of %d bytes", (int)offset, (int)extra_length); return ARCHIVE_FAILED; } 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->entry_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 = zip_time(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, 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 regularfile. */ 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, it's 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 = 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 = 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); 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_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_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) { /* 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); } /* * Read "uncompressed" data. There are three cases: * 1) We know the size of the data. This is always true for the * seeking reader (we've examined the Central Directory already). * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. * Info-ZIP seems to do this; we know the size but have to grab * the CRC from the data descriptor afterwards. * 3) We're streaming and ZIP_LENGTH_AT_END was specified and * we have no size information. In this case, we can do pretty * well by watching for the data descriptor record. The data * descriptor is 16 bytes and includes a computed CRC that should * provide a strong check. * * TODO: Technically, the PK\007\010 signature is optional. * In the original spec, the data descriptor contained CRC * and size fields but had no leading signature. In practice, * newer writers seem to provide the signature pretty consistently. * * For uncompressed data, the PK\007\010 marker seems essential * to be sure we've actually seen the end of the entry. * * 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; int r; (void)offset; /* UNUSED */ zip = (struct zip *)(a->format->data); if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { const char *p; ssize_t grabbing_bytes = 24; if (zip->hctx_valid) grabbing_bytes += AUTH_CODE_SIZE; /* 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; if (zip->hctx_valid) p += AUTH_CODE_SIZE; if (p[0] == 'P' && p[1] == 'K' && p[2] == '\007' && p[3] == '\010' && (archive_le32dec(p + 4) == zip->entry_crc32 || zip->ignore_crc32 || (zip->hctx_valid && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { if (zip->entry->flags & LA_USED_ZIP64) { uint64_t compressed, uncompressed; zip->entry->crc32 = archive_le32dec(p + 4); compressed = archive_le64dec(p + 8); uncompressed = archive_le64dec(p + 16); if (compressed > INT64_MAX || uncompressed > INT64_MAX) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Overflow of 64-bit file sizes"); return ARCHIVE_FAILED; } zip->entry->compressed_size = compressed; zip->entry->uncompressed_size = uncompressed; zip->unconsumed = 24; } else { zip->entry->crc32 = archive_le32dec(p + 4); zip->entry->compressed_size = archive_le32dec(p + 8); zip->entry->uncompressed_size = archive_le32dec(p + 12); zip->unconsumed = 16; } if (zip->hctx_valid) { r = check_authentication_code(a, buff); if (r != ARCHIVE_OK) return (r); } zip->end_of_entry = 1; 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') { if (zip->hctx_valid) p -= AUTH_CODE_SIZE; break; } else { p += 4; } } 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; } *size = bytes_avail; zip->entry_bytes_remaining -= bytes_avail; zip->entry_uncompressed_bytes_read += bytes_avail; zip->entry_compressed_bytes_read += bytes_avail; zip->unconsumed += bytes_avail; *_buff = buff; return (ARCHIVE_OK); } static int consume_optional_marker(struct archive_read *a, struct zip *zip) { if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { const char *p; if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP end-of-file record"); return (ARCHIVE_FATAL); } /* Consume the optional PK\007\010 marker. */ if (p[0] == 'P' && p[1] == 'K' && p[2] == '\007' && p[3] == '\010') { p += 4; zip->unconsumed = 4; } if (zip->entry->flags & LA_USED_ZIP64) { uint64_t compressed, uncompressed; zip->entry->crc32 = archive_le32dec(p); compressed = archive_le64dec(p + 4); uncompressed = archive_le64dec(p + 12); if (compressed > INT64_MAX || uncompressed > INT64_MAX) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Overflow of 64-bit file sizes"); return ARCHIVE_FAILED; } zip->entry->compressed_size = compressed; zip->entry->uncompressed_size = uncompressed; zip->unconsumed += 20; } else { zip->entry->crc32 = archive_le32dec(p); zip->entry->compressed_size = archive_le32dec(p + 4); zip->entry->uncompressed_size = archive_le32dec(p + 8); zip->unconsumed += 12; } } 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 = (uint8_t*) 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) /* 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((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 = (uint8_t*) 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 = 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 = 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 = zip->zipx_lzma_stream.total_out; *buff = zip->uncompressed_buffer; ret = consume_optional_marker(a, zip); if (ret != ARCHIVE_OK) return (ret); 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 = 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. */ 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); 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 = 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; } /* Return values. */ *size = zip->zipx_lzma_stream.total_out; *buff = zip->uncompressed_buffer; /* Behave the same way as during deflate decompression. */ ret = consume_optional_marker(a, zip); if (ret != ARCHIVE_OK) return (ret); /* Free lzma decoder handle because we'll no longer need it. */ if(zip->end_of_entry) { lzma_end(&zip->zipx_lzma_stream); zip->zipx_lzma_valid = 0; } /* 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=%d, " "restore=%d)", 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: %d 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 = (uint8_t*) 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 for libarchive. */ *buff = zip->uncompressed_buffer; *size = consumed_bytes; /* 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; } /* Seek for optional marker, same way as in each zip entry. */ ret = consume_optional_marker(a, zip); if (ret != ARCHIVE_OK) return ret; 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 + /* 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 = (uint8_t*) 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 = 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 = 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 = 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 = total_out; *buff = zip->uncompressed_buffer; /* Seek for optional marker, like in other entries. */ r = consume_optional_marker(a, zip); if(r != ARCHIVE_OK) return r; 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; 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 = (unsigned char *)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. */ bytes_avail = zip->stream.total_in; if (zip->tctx_valid || zip->cctx_valid) { zip->decrypted_bytes_remaining -= bytes_avail; if (zip->decrypted_bytes_remaining == 0) zip->decrypted_ptr = zip->decrypted_buffer; else zip->decrypted_ptr += bytes_avail; } /* Calculate compressed data as much as we used.*/ if (zip->hctx_valid) archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); __archive_read_consume(a, bytes_avail); zip->entry_bytes_remaining -= bytes_avail; zip->entry_compressed_bytes_read += bytes_avail; *size = zip->stream.total_out; zip->entry_uncompressed_bytes_read += zip->stream.total_out; *buff = zip->uncompressed_buffer; if (zip->end_of_entry && zip->hctx_valid) { r = check_authentication_code(a, NULL); if (r != ARCHIVE_OK) return (r); } r = consume_optional_marker(a, zip); if (r != ARCHIVE_OK) return (r); 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"); 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 /* 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); break; } if (r != ARCHIVE_OK) return (r); /* Update checksum */ if (*size) zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, (unsigned)*size); /* If we hit the end, swallow any end-of-data marker. */ if (zip->end_of_entry) { /* Check file size, CRC against these values. */ 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_WARN); } /* 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_WARN); } /* Check computed CRC against header */ if ((!zip->hctx_valid || zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && zip->entry->crc32 != zip->entry_crc32 && !zip->ignore_crc32) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP bad CRC: 0x%lx should be 0x%lx", (unsigned long)zip->entry_crc32, (unsigned long)zip->entry->crc32); return (ARCHIVE_WARN); } } 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 HAVA_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 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 = (struct 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) { /* Sanity-check the EOCD we've found. */ /* This must be the first volume. */ if (archive_le16dec(p + 4) != 0) return 0; /* Central directory must be on this volume. */ if (archive_le16dec(p + 4) != 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 (archive_le32dec(p + 16) + archive_le32dec(p + 12) > current_offset) return 0; /* Save the central directory location for later use. */ zip->central_directory_offset = archive_le32dec(p + 16); /* 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); 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 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, 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 = zip_time(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, 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. */ 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), filename_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, 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 = (struct 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:*/ Index: head/contrib/libarchive/libarchive/archive_string.c =================================================================== --- head/contrib/libarchive/libarchive/archive_string.c (revision 345496) +++ head/contrib/libarchive/libarchive/archive_string.c (revision 345497) @@ -1,4208 +1,4212 @@ /*- * 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" __FBSDID("$FreeBSD$"); /* * 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 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 is_big_endian(void); 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 *); 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 = (char *)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 count, ret = 0; 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 process. */ 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 < (int)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 = (int)utf16nbytes(s, length); } else { /* * UTF-8 NFD ===> UTF-16 NFC * UTF-8 NFC ===> UTF-16 NFD */ count = (int)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 = (int)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); int 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); int 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; buffsize = dest->length + length + 1; do { /* Allocate memory for WCS. */ if (NULL == archive_wstring_ensure(dest, buffsize)) return (-1); /* Convert MBS to WCS. */ count = MultiByteToWideChar(from_cp, mbflag, s, (int)length, dest->s + dest->length, (int)(dest->buffer_length >> 1) -1); if (count == 0 && GetLastError() == ERROR_INSUFFICIENT_BUFFER) { /* Expand the WCS buffer. */ buffsize = dest->buffer_length << 1; continue; } if (count == 0 && length != 0) ret = -1; 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 if (NULL == archive_wstring_ensure(dest, dest->length + wcs_length + 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) { 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, wcs_length, &shift_state); #else r = mbtowc(wcs, mbs, wcs_length); #endif if (r == (size_t)-1 || r == (size_t)-2) { ret_val = -1; if (errno == EILSEQ) { ++mbs; --mbs_length; continue; } else 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 count, ret = 0; UINT to_cp; int wslen = (int)len; if (sc != NULL) to_cp = sc->to_cp; else to_cp = get_current_codepage(); if (to_cp == CP_C_LOCALE) { /* * "C" locale special process. */ 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 < (int)len && *ws) { archive_be16enc(u16+count, *ws); ws++; count++; } } else { while (count < (int)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 { defchar_used = 0; if (to_cp == CP_UTF8 || sc == NULL) dp = NULL; else dp = &defchar_used; count = WideCharToMultiByte(to_cp, 0, ws, wslen, as->s + as->length, (int)as->buffer_length-1, NULL, dp); if (count == 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 (count == 0) ret = -1; 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 + len * 2 + 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, "Programing 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__) 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) + return CP_UTF8; cp = my_atoi(p+1); - if (cp <= 0) + 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; 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 */ int 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 >= (size_t)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 (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 ((int)n < 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 ((int)n < 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, 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, uc); return (2); } } /* * Copy UTF-8 string in checking surrogate pair. * If any surrogate pair are found, it would be canonicalized. */ static int strncat_from_utf8_to_utf8(struct archive_string *as, const void *_p, size_t len, struct archive_string_conv *sc) { const char *s; char *p, *endp; int n, ret = 0; (void)sc; /* UNUSED */ if (archive_string_ensure(as, as->length + len + 1) == NULL) return (-1); s = (const char *)_p; p = as->s + as->length; endp = as->s + as->buffer_length -1; do { uint32_t uc; const char *ss = s; size_t w; /* * Forward byte sequence until a conversion of that is needed. */ while ((n = utf8_to_unicode(&uc, s, len)) > 0) { s += n; len -= n; } if (ss < s) { if (p + (s - ss) > endp) { as->length = p - as->s; if (archive_string_ensure(as, as->buffer_length + len + 1) == NULL) return (-1); p = as->s + as->length; endp = as->s + as->buffer_length -1; } memcpy(p, ss, s - ss); p += s - ss; } /* * If n is negative, current byte sequence needs a replacement. */ if (n < 0) { if (n == -3 && IS_SURROGATE_PAIR_LA(uc)) { /* Current byte sequence may be CESU-8. */ n = cesu8_to_unicode(&uc, s, len); } if (n < 0) { ret = -1; n *= -1;/* Use a replaced unicode character. */ } /* Rebuild UTF-8 byte sequence. */ while ((w = unicode_to_utf8(p, endp - p, uc)) == 0) { as->length = p - as->s; if (archive_string_ensure(as, as->buffer_length + len + 1) == NULL) return (-1); p = as->s + as->length; endp = as->s + as->buffer_length -1; } p += w; s += n; len -= n; } } while (n > 0); as->length = p - as->s; as->s[as->length] = '\0'; return (ret); } 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; 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 + len * 2 + 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; int ll; BOOL defchar; char *mbs; size_t mbs_size, b; 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 { defchar = 0; ll = WideCharToMultiByte(sc->to_cp, 0, (LPCWSTR)u16, (int)bytes>>1, mbs, (int)mbs_size, NULL, &defchar); /* Exit loop if we succeeded */ if (ll != 0 || GetLastError() != ERROR_INSUFFICIENT_BUFFER) { break; } /* Else expand buffer and loop to try again. */ ll = WideCharToMultiByte(sc->to_cp, 0, (LPCWSTR)u16, (int)bytes, NULL, 0, NULL, NULL); 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)); } static int is_big_endian(void) { uint16_t d = 1; return (archive_be16dec(&d) == 1); } /* * 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 { count = MultiByteToWideChar(sc->from_cp, MB_PRECOMPOSED, s, (int)length, (LPWSTR)u16, (int)avail>>1); /* Exit loop if we succeeded */ if (count != 0 || GetLastError() != ERROR_INSUFFICIENT_BUFFER) { break; } /* Expand buffer and try again */ count = MultiByteToWideChar(sc->from_cp, MB_PRECOMPOSED, s, (int)length, NULL, 0); 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, c); else archive_le16enc(utf16, 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 (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) { int r, ret = 0; (void)a; /* UNUSED */ /* 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; } /* * Only a UTF-8 form cannot avail because its conversion already * failed at archive_mstring_update_utf8(). */ 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; /* 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_mstring *aes, const char **p, size_t *length, struct archive_string_conv *sc) { int r, ret = 0; #if defined(_WIN32) && !defined(__CYGWIN__) /* * 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 is a WCS 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 && (aes->aes_set & AES_SET_WCS) != 0) { archive_string_empty(&(aes->aes_mbs)); r = archive_string_append_from_wcs(&(aes->aes_mbs), aes->aes_wcs.s, aes->aes_wcs.length); if (r == 0) aes->aes_set |= AES_SET_MBS; else if (errno == ENOMEM) return (-1); else ret = -1; } /* 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. */ /* Try converting UTF-8 to MBS, return false on failure. */ sc = archive_string_conversion_from_charset(a, "UTF-8", 1); if (sc == NULL) return (-1);/* Couldn't allocate memory for sc. */ 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); aes->aes_set = AES_SET_UTF8 | AES_SET_WCS | AES_SET_MBS; /* All conversions succeeded. */ return (0); } Index: head/contrib/libarchive/libarchive/archive_write_disk_posix.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_disk_posix.c (revision 345496) +++ head/contrib/libarchive/libarchive/archive_write_disk_posix.c (revision 345497) @@ -1,4353 +1,4375 @@ /*- * 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" __FBSDID("$FreeBSD$"); #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 struct fixup_entry { struct fixup_entry *next; struct archive_acl acl; mode_t mode; 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' */ /* 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 void fsobj_error(int *, struct archive_string *, int, const char *, const char *); static int check_symlinks_fsobj(char *, int *, struct archive_string *, 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 struct archive_vtable *archive_write_disk_vtable(void); 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 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? */ if (lstat(a->name, &a->st) == 0) { a->pst = &a->st; return (ARCHIVE_OK); } archive_set_error(&a->archive, errno, "Couldn't stat file"); return (ARCHIVE_WARN); } static struct archive_vtable * archive_write_disk_vtable(void) { static struct archive_vtable av; static int inited = 0; if (!inited) { av.archive_close = _archive_write_disk_close; av.archive_filter_bytes = _archive_write_disk_filter_bytes; av.archive_free = _archive_write_disk_free; av.archive_write_header = _archive_write_disk_header; av.archive_write_finish_entry = _archive_write_disk_finish_entry; av.archive_write_data = _archive_write_disk_data; av.archive_write_data_block = _archive_write_disk_data_block; inited = 1; } return (&av); } 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; 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); /* * 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->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->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->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->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->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) { int64_t skip = a->offset - a->fd_offset; char nullblock[1024]; memset(nullblock, 0, sizeof(nullblock)); while (skip > 0) { if (skip > (int64_t)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) 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"); 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) 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"); return (ARCHIVE_FATAL); } if (write(a->fd, &nul, 1) < 0) { archive_set_error(&a->archive, errno, "Write to restore size failed"); 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 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 = (struct archive_write_disk *)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); } #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 = open(".", O_RDONLY | O_BINARY | O_CLOEXEC); __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 = 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)) r = lstat(a->name, &a->st); 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)) { /* A non-dir is in the way, unlink it. */ if (a->flags & ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS) (void)clear_nochange_fflags(a); 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); 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); r = link(linkname, a->name) ? errno : 0; /* * 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 = 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 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->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; int ret; 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) { a->pst = NULL; /* Mark stat cache as out-of-date. */ if (p->fixup & TODO_TIMES) { set_times(a, -1, 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) chmod(p->name, p->mode); if (p->fixup & TODO_ACLS) archive_write_disk_set_acls(&a->archive, -1, p->name, &p->acl, p->mode); if (p->fixup & TODO_FFLAGS) set_fflags_platform(a, -1, 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); next = p->next; archive_acl_clear(&p->acl); free(p->mac_metadata); free(p->name); 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->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 = (struct fixup_entry *)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->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. */ /* TODO: Extend this to support symlinks on Windows Vista and later. */ /* * 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) { #if !defined(HAVE_LSTAT) /* Platform doesn't have lstat, so we can't look for symlinks. */ (void)path; /* UNUSED */ (void)error_number; /* UNUSED */ (void)error_string; /* UNUSED */ (void)flags; /* UNUSED */ return (ARCHIVE_OK); #else int res = ARCHIVE_OK; char *tail; char *head; int last; char c; int r; struct stat st; int restore_pwd; /* 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 */ restore_pwd = open(".", O_RDONLY | O_BINARY | O_CLOEXEC); __archive_ensure_cloexec_flag(restore_pwd); - if (restore_pwd < 0) + if (restore_pwd < 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. */ r = lstat(head, &st); 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 (chdir(head) != 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) { /* * Last element is symlink; remove it * so we can overwrite it with the * item being extracted. */ if (unlink(head)) { 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 (unlink(head) != 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. */ r = stat(head, &st); 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 (chdir(head) != 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; #ifdef HAVE_FCHDIR /* If we changed directory above, restore it here. */ if (restore_pwd >= 0) { r = fchdir(restore_pwd); if (r != 0) { fsobj_error(a_eno, a_estr, errno, "chdir() failure", ""); } close(restore_pwd); restore_pwd = -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); 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 (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 (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) { -#ifndef __CYGWIN__ -/* unfortunately, on win32 there is no 'root' user with uid 0, - so we just have to try the chown and see if it works */ - - /* If we know we can't change it, don't bother trying. */ - if (a->user_uid != 0 && a->user_uid != a->uid) { +#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_nsec; /* 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) 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; 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) != 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", (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) { if (fchmod(a->fd, mode) != 0) { archive_set_error(&a->archive, errno, "Can't set permissions to 0%o", (int)mode); r = ARCHIVE_WARN; } } else #endif /* If this platform lacks fchmod(), then * we'll just use chmod(). */ if (chmod(a->name, mode) != 0) { archive_set_error(&a->archive, errno, "Can't set permissions to 0%o", (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; - int critical_flags; 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. */ - /* Hopefully, the compiler will optimize this mess into a constant. */ - critical_flags = 0; + const int critical_flags = 0 #ifdef SF_IMMUTABLE - critical_flags |= SF_IMMUTABLE; + | SF_IMMUTABLE #endif #ifdef UF_IMMUTABLE - critical_flags |= UF_IMMUTABLE; + | UF_IMMUTABLE #endif #ifdef SF_APPEND - critical_flags |= SF_APPEND; + | SF_APPEND #endif #ifdef UF_APPEND - critical_flags |= UF_APPEND; + | UF_APPEND #endif #if defined(FS_APPEND_FL) - critical_flags |= FS_APPEND_FL; + | FS_APPEND_FL #elif defined(EXT2_APPEND_FL) - critical_flags |= EXT2_APPEND_FL; + | EXT2_APPEND_FL #endif #if defined(FS_IMMUTABLE_FL) - critical_flags |= FS_IMMUTABLE_FL; + | FS_IMMUTABLE_FL #elif defined(EXT2_IMMUTABLE_FL) - critical_flags |= EXT2_IMMUTABLE_FL; + | EXT2_IMMUTABLE_FL #endif #ifdef FS_JOURNAL_DATA_FL - critical_flags |= 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->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) { - int nochange_flags; mode_t mode = archive_entry_mode(a->entry); - - /* Hopefully, the compiler will optimize this mess into a constant. */ - nochange_flags = 0; + const int nochange_flags = 0 #ifdef SF_IMMUTABLE - nochange_flags |= SF_IMMUTABLE; + | SF_IMMUTABLE #endif #ifdef UF_IMMUTABLE - nochange_flags |= UF_IMMUTABLE; + | UF_IMMUTABLE #endif #ifdef SF_APPEND - nochange_flags |= SF_APPEND; + | SF_APPEND #endif #ifdef UF_APPEND - nochange_flags |= UF_APPEND; + | UF_APPEND #endif #ifdef EXT2_APPEND_FL - nochange_flags |= EXT2_APPEND_FL; + | EXT2_APPEND_FL #endif #ifdef EXT2_IMMUTABLE_FL - nochange_flags |= 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; - int sf_mask = 0; - - 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); - __archive_ensure_cloexec_flag(myfd); - } - if (myfd < 0) - return (ARCHIVE_OK); - /* * 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) - sf_mask |= FS_IMMUTABLE_FL; + | FS_IMMUTABLE_FL #elif defined(EXT2_IMMUTABLE_FL) - sf_mask |= EXT2_IMMUTABLE_FL; + | EXT2_IMMUTABLE_FL #endif #if defined(FS_APPEND_FL) - sf_mask |= FS_APPEND_FL; + | FS_APPEND_FL #elif defined(EXT2_APPEND_FL) - sf_mask |= EXT2_APPEND_FL; + | EXT2_APPEND_FL #endif #if defined(FS_JOURNAL_DATA_FL) - sf_mask |= 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); + __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 *xattr_val_saved; 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; } xattr_val_saved = xattr_val; xattr_val = realloc(xattr_val, s); if (xattr_val == NULL) { archive_set_error(&a->archive, ENOMEM, "Failed to get metadata(xattr)"); ret = ARCHIVE_WARN; free(xattr_val_saved); goto exit_xattr; } 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, pathname); archive_strcat(&tmp, ".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 (lstat(datafork.s, &st) == -1 || (st.st_mode & AE_IFMT) != AE_IFREG) 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) { 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; if (strncmp(name, "user.", 5) == 0) { /* "user." attributes go to user namespace */ name += 5; namespace = EXTATTR_NAMESPACE_USER; } else { /* Other namespaces are unsupported */ archive_strcat(&errlist, name); archive_strappend_char(&errlist, ' '); fail = 1; ret = ARCHIVE_WARN; continue; } if (a->fd >= 0) { e = extattr_set_fd(a->fd, namespace, name, value, 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 #endif /* !_WIN32 || __CYGWIN__ */ Index: head/contrib/libarchive/libarchive/archive_write_set_format_7zip.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_7zip.c (revision 345496) +++ head/contrib/libarchive/libarchive/archive_write_set_format_7zip.c (revision 345497) @@ -1,2315 +1,2316 @@ /*- * 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" __FBSDID("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_BZLIB_H #include #endif #if HAVE_LZMA_H #include #endif #ifdef HAVE_ZLIB_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_write_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 /* * 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 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; 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; 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 *, unsigned, 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_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); 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; #else zip->opt_compression = _7Z_COPY; #endif zip->opt_compression_level = 6; 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, "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] >= '0' && value[0] <= '9') || value[1] != '\0') { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Illegal value `%s'", value); return (ARCHIVE_FAILED); } zip->opt_compression_level = value[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); } 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) { - file_free(file); + 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) { r = _7z_compression_init_encoder(a, zip->opt_compression, 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(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"); 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 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 r = _7z_compression_init_encoder(a, header_compression, 6); 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); } #define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) static uint64_t utcToFiletime(time_t t, long ns) { uint64_t fileTime; fileTime = t; fileTime *= 10000000; fileTime += ns / 100; fileTime += EPOC_TIME; return (fileTime); } 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, utcToFiletime(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; if (file->dir) attr = 0x8010; else attr = 0x8020; if ((file->mode & 0222) == 0) attr |= 1;/* Read Only. */ 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) file->size = strlen(archive_entry_symlink(entry)); 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) 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 = 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 = 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 = 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 = 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 > 6) level = 6; 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, unsigned 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); } /* * 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_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); } Index: head/contrib/libarchive/libarchive/test/test_entry.c =================================================================== --- head/contrib/libarchive/libarchive/test/test_entry.c (revision 345496) +++ head/contrib/libarchive/libarchive/test/test_entry.c (revision 345497) @@ -1,866 +1,866 @@ /*- * 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" __FBSDID("$FreeBSD$"); #include #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"); 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"); /* hardlink */ archive_entry_set_hardlink(e, "hardlinkname"); assertEqualString(archive_entry_hardlink(e), "hardlinkname"); strcpy(buff, "hardlinkname2"); archive_entry_copy_hardlink(e, buff); assertEqualString(archive_entry_hardlink(e), "hardlinkname2"); memset(buff, 0, sizeof(buff)); assertEqualString(archive_entry_hardlink(e), "hardlinkname2"); archive_entry_copy_hardlink(e, NULL); assertEqualString(archive_entry_hardlink(e), NULL); assertEqualWString(archive_entry_hardlink_w(e), NULL); wcscpy(wbuff, L"whardlink"); archive_entry_copy_hardlink_w(e, wbuff); assertEqualWString(archive_entry_hardlink_w(e), L"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); /* 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), "hardlink"); 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"); archive_entry_set_pathname(e, "path"); assertEqualString(archive_entry_pathname(e), "path"); strcpy(buff, "path2"); archive_entry_copy_pathname(e, buff); assertEqualString(archive_entry_pathname(e), "path2"); memset(buff, 0, sizeof(buff)); assertEqualString(archive_entry_pathname(e), "path2"); wcscpy(wbuff, L"wpath"); archive_entry_copy_pathname_w(e, wbuff); assertEqualWString(archive_entry_pathname_w(e), L"wpath"); memset(wbuff, 0, sizeof(wbuff)); assertEqualWString(archive_entry_pathname_w(e), L"wpath"); /* 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"); strcpy(buff, "symlinkname2"); archive_entry_copy_symlink(e, buff); assertEqualString(archive_entry_symlink(e), "symlinkname2"); memset(buff, 0, sizeof(buff)); assertEqualString(archive_entry_symlink(e), "symlinkname2"); archive_entry_copy_symlink_w(e, NULL); assertEqualWString(archive_entry_symlink_w(e), NULL); assertEqualString(archive_entry_symlink(e), NULL); archive_entry_copy_symlink_w(e, L"wsymlink"); assertEqualWString(archive_entry_symlink_w(e), L"wsymlink"); archive_entry_copy_symlink(e, NULL); assertEqualWString(archive_entry_symlink_w(e), NULL); assertEqualString(archive_entry_symlink(e), NULL); /* 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"); wcscpy(wbuff, L"wuser"); archive_entry_copy_gname_w(e, wbuff); assertEqualWString(archive_entry_gname_w(e), L"wuser"); memset(wbuff, 0, sizeof(wbuff)); assertEqualWString(archive_entry_gname_w(e), L"wuser"); /* 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"); + "uappnd,nouchg,nodump,noopaque,uunlnk,nosystem"); /* Test archive_entry_copy_fflags_text_w() */ archive_entry_copy_fflags_text_w(e, L" ,nouappnd, nouchg, dump,uunlnk"); archive_entry_fflags(e, &set, &clear); assertEqualInt(16, set); assertEqualInt(7, clear); /* Test archive_entry_copy_fflags_text() */ archive_entry_copy_fflags_text(e, " ,nouappnd, nouchg, dump,uunlnk"); archive_entry_fflags(e, &set, &clear); assertEqualInt(16, set); assertEqualInt(7, 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_symlink(e, "symlinkname"); 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), "symlinkname"); 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), "symlinkname"); 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); /* * 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); /* 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); } Index: head/contrib/libarchive/libarchive/test/test_read_format_rar5.c =================================================================== --- head/contrib/libarchive/libarchive/test/test_read_format_rar5.c (revision 345496) +++ head/contrib/libarchive/libarchive/test/test_read_format_rar5.c (revision 345497) @@ -1,770 +1,770 @@ /*- * 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 #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(archive_le32dec(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 = archive_entry_size(ae); + 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 = crc32(0, buf, fsize); assertEqualInt(computed_crc, crc); ret = 0; - + fn_exit: free(buf); return ret; } -DEFINE_TEST(test_read_format_rar5_stored) +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[DATA_SIZE]; + 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)); verify_data(buff, 0, DATA_SIZE); EPILOGUE(); } DEFINE_TEST(test_read_format_rar5_multiple_files) { const int DATA_SIZE = 4096; - uint8_t buff[DATA_SIZE]; + 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(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(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(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(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[DATA_SIZE]; + 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(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(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(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(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[proper_size]; + 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(crc32(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[proper_size]; + 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(crc32(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_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) +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) +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) +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("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)); EPILOGUE(); } 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]; int 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 = crc32(computed_crc, buf, bytes_read); } assertEqualInt(computed_crc, 0x7CCA70CD); EPILOGUE(); } Index: head/contrib/libarchive/libarchive/test/test_read_format_xar.c =================================================================== --- head/contrib/libarchive/libarchive/test/test_read_format_xar.c (revision 345496) +++ head/contrib/libarchive/libarchive/test/test_read_format_xar.c (revision 345497) @@ -1,863 +1,863 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * 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 "test.h" __FBSDID("$FreeBSD$"); #define UID 1001 #define UNAME "cue" #define GID 1001 #define GNAME "cue" /* Verify that a file records with hardlink. #How to make echo "hellohellohello" > f1 chown $UNAME:$GNAME f1 chmod 0644 f1 ln f1 hardlink chown $UNAME:$GNAME hardlink chmod 0644 hardlink env TZ=utc touch -afm -t 197001020000.01 f1 hardlink xar -cf archive1.xar f1 hardlink od -t x1 archive1.xar | sed -E -e 's/^0[0-9]+//;s/^ //;s/( )([0-9a-f]{2})/0x\2,/g;$ D' > archive1.xar.txt */ static unsigned char archive1[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x01,0xc6, 0x00,0x00,0x00,0x00,0x00,0x00,0x04,0x70,0x00,0x00,0x00,0x01,0x78,0xda,0xc4,0x54, 0xc9,0x6e,0xdb,0x30,0x14,0xbc,0xe7,0x2b,0x08,0xdd,0x55,0xae,0xb6,0x45,0x83,0x56, 0xd0,0x4b,0xd1,0x7b,0xd3,0x4b,0x6f,0x34,0x17,0x89,0x88,0x36,0x48,0x54,0xe0,0xe4, 0xeb,0x4b,0x52,0x52,0x0c,0xa7,0x71,0x6f,0x45,0x01,0x01,0x1a,0x0e,0x87,0xa3,0xa7, 0xf7,0x06,0x14,0x8f,0x97,0xb6,0x01,0x2f,0x66,0x9c,0x5c,0xdf,0x9d,0x32,0xfc,0x05, 0x65,0xc0,0x74,0xaa,0xd7,0xae,0xab,0x4e,0xd9,0xcf,0xa7,0x6f,0x79,0x91,0x3d,0x96, 0x0f,0xe2,0x22,0xc7,0xf2,0x01,0x08,0xdf,0xab,0xf0,0x02,0x42,0x8d,0x46,0xfa,0x70, 0x22,0xf7,0xae,0x35,0x25,0x41,0x88,0xe7,0x98,0xe4,0x88,0x3c,0x61,0x7a,0xa4,0xe8, 0x48,0xb9,0x80,0xb7,0x92,0x74,0xa8,0x36,0xea,0x79,0x9a,0x5b,0x30,0xf9,0xd7,0xc6, 0x9c,0xb2,0xa9,0x96,0x38,0x8b,0x3b,0x40,0xf4,0xd6,0x4e,0xc6,0x97,0x48,0xc0,0x15, 0x25,0x76,0x72,0x6f,0xd1,0x5c,0xc0,0x04,0xa2,0x05,0xdc,0x3c,0xd2,0xca,0xba,0xc6, 0x00,0xa7,0x4f,0x19,0x59,0x6d,0xd4,0x9d,0x72,0xd8,0xaf,0x70,0x72,0xab,0x03,0x88, 0x36,0x41,0xcc,0x0f,0x28,0x47,0x38,0xca,0x10,0x3a,0xc6,0x07,0x07,0x59,0x7b,0x95, 0xc9,0x7b,0x3f,0x17,0x64,0xf2,0x2a,0xab,0xc6,0x7e,0x1e,0x4a,0x35,0x1b,0x01,0x17, 0xb8,0xb0,0x4e,0x97,0x18,0x21,0x1c,0xc8,0x80,0x12,0x35,0x4f,0x66,0x5c,0x74,0x09, 0x2d,0xdc,0xbb,0x6c,0xde,0x64,0x6d,0xaf,0x4d,0x89,0xf6,0x8c,0x85,0x62,0x22,0x4c, 0xa4,0x7f,0x1d,0x0c,0x68,0x5c,0xf7,0x1c,0x66,0x94,0x95,0xb5,0x1c,0x75,0x5c,0x08, 0x18,0xf9,0x45,0xd1,0xc9,0x50,0xd0,0x75,0x23,0x2d,0x53,0xcb,0x62,0x97,0x6e,0xdb, 0xb5,0x75,0x5d,0x4b,0x2f,0x13,0x02,0xa2,0x31,0x5d,0xe5,0xeb,0x92,0x50,0x01,0x57, 0xb8,0xf0,0xeb,0x38,0xc8,0xed,0x64,0xd6,0xd1,0xe0,0xfd,0x75,0x34,0x81,0xdb,0x72, 0xb3,0xcd,0x57,0x0e,0x43,0xe3,0x54,0x0a,0x01,0xbc,0xe4,0xd5,0x9b,0x1b,0x32,0xb8, 0x4a,0xe5,0xa8,0x6a,0xf7,0x62,0x74,0xfe,0x31,0x13,0x3f,0xbe,0x7f,0x0d,0xd5,0xd9, 0x82,0x52,0x4d,0xac,0x56,0x98,0x53,0xc6,0xa9,0x3c,0xb3,0x82,0x4b,0x2d,0x09,0xb5, 0x85,0x3d,0x70,0x6c,0xf7,0xc4,0x2a,0xba,0xe7,0x45,0x98,0xc3,0x47,0xa3,0xad,0x96, 0x8b,0x1f,0xa5,0xf2,0x77,0xbf,0xb0,0xd3,0x07,0x76,0x56,0x67,0x75,0xe0,0x9a,0x5a, 0x7e,0xb6,0x4c,0xda,0xe0,0xcd,0x8a,0xa2,0x40,0x86,0xed,0xc8,0x7e,0xc7,0xac,0x41, 0x8a,0x87,0x1c,0xff,0xe9,0xb4,0x34,0x0f,0xbe,0x77,0xef,0x9f,0xc4,0xee,0x73,0xd9, 0x7f,0x8c,0x5d,0x3f,0xba,0xca,0x75,0xb2,0xf9,0x4b,0xfa,0x2c,0xfe,0x24,0x77,0x41, 0x15,0x2f,0x0d,0x01,0xd3,0x15,0xf2,0x1b,0x00,0x00,0xff,0xff,0x03,0x00,0x88,0x32, 0x49,0x7b,0x67,0xbf,0xc6,0x01,0x29,0xf2,0x1c,0x40,0x05,0x3c,0x49,0x25,0x9f,0xab, 0x7c,0x8e,0xc5,0xa5,0x79,0xe0,0x78,0xda,0xca,0x48,0xcd,0xc9,0xc9,0xcf,0x80,0x13, 0x5c,0x00,0x00,0x00,0x00,0xff,0xff,0x03,0x00,0x37,0xf7,0x06,0x47 }; static void verify0(struct archive *a, struct archive_entry *ae) { const void *p; size_t size; int64_t offset; assert(archive_entry_filetype(ae) == AE_IFREG); assertEqualInt(archive_entry_mode(ae) & 0777, 0644); assertEqualInt(archive_entry_uid(ae), UID); assertEqualInt(archive_entry_gid(ae), GID); assertEqualString(archive_entry_uname(ae), UNAME); assertEqualString(archive_entry_gname(ae), GNAME); assertEqualString(archive_entry_pathname(ae), "f1"); assert(archive_entry_hardlink(ae) == NULL); assert(archive_entry_symlink(ae) == NULL); assertEqualInt(archive_entry_mtime(ae), 86401); assertEqualInt(archive_entry_size(ae), 16); assertEqualInt(archive_read_data_block(a, &p, &size, &offset), 0); assertEqualInt((int)size, 16); assertEqualInt((int)offset, 0); assertEqualMem(p, "hellohellohello\n", 16); } static void verify1(struct archive *a, struct archive_entry *ae) { (void)a; /* UNUSED */ /* A hardlink is not a symlink. */ assert(archive_entry_filetype(ae) != AE_IFLNK); /* Nor is it a directory. */ assert(archive_entry_filetype(ae) != AE_IFDIR); assertEqualInt(archive_entry_mode(ae) & 0777, 0644); assertEqualInt(archive_entry_uid(ae), UID); assertEqualInt(archive_entry_gid(ae), GID); assertEqualString(archive_entry_uname(ae), UNAME); assertEqualString(archive_entry_gname(ae), GNAME); assertEqualString(archive_entry_pathname(ae), "hardlink"); assertEqualString(archive_entry_hardlink(ae), "f1"); assert(archive_entry_symlink(ae) == NULL); assertEqualInt(archive_entry_mtime(ae), 86401); assertEqualInt(archive_entry_nlink(ae), 2); } /* Verify that symlinks are read correctly. #How to make echo "hellohellohello" > f1 chown $UNAME:$GNAME f1 chmod 0644 f1 ln -s f1 symlink chown $UNAME:$GNAME symlink chmod 0644 symlink env TZ=utc touch -afm -t 197001020000.01 f1 symlink xar -cf archive2.xar f1 symlink od -t x1 archive2.xar | sed -E -e 's/^0[0-9]+//;s/^ //;s/( )([0-9a-f]{2})/0x\2,/g;$ D' > archive2.xar.txt */ static unsigned char archive2[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x01,0xe8, 0x00,0x00,0x00,0x00,0x00,0x00,0x04,0x84,0x00,0x00,0x00,0x01,0x78,0xda,0xb4,0x54, 0xcb,0x6e,0xa3,0x30,0x14,0xdd,0xf7,0x2b,0x90,0xf7,0x8c,0x1f,0x40,0x82,0x23,0xe3, 0xaa,0x9b,0x6a,0xf6,0xd3,0xd9,0xcc,0xce,0xf1,0x83,0x58,0xe5,0x25,0x70,0xaa,0xa4, 0x5f,0x3f,0xb6,0x09,0x4d,0xd3,0x30,0xdd,0x8d,0x84,0xc4,0xf5,0xf1,0xb9,0xc7,0x97, 0x7b,0x0f,0x66,0x8f,0xa7,0xb6,0x49,0xde,0xf4,0x38,0xd9,0xbe,0xab,0x00,0xfe,0x81, 0x40,0xa2,0x3b,0xd9,0x2b,0xdb,0xd5,0x15,0xf8,0xfd,0xf2,0x9c,0x96,0xe0,0x91,0x3f, 0xb0,0x93,0x18,0xf9,0x43,0xc2,0x5c,0x2f,0xfd,0x2b,0x61,0x72,0xd4,0xc2,0xf9,0x8c, 0xd4,0xd9,0x56,0x73,0x82,0x10,0x4d,0x31,0x49,0x11,0x79,0xc1,0xd9,0x2e,0x2b,0x76, 0xb8,0x60,0xf0,0x96,0x12,0x93,0x0e,0x5a,0xbe,0x4e,0xc7,0x36,0x99,0xdc,0xb9,0xd1, 0x15,0x98,0x0e,0x02,0x83,0xb0,0x93,0xb0,0xde,0x98,0x49,0x3b,0x8e,0x18,0xbc,0x44, 0x11,0x9d,0xec,0x7b,0x10,0x67,0x30,0x06,0x41,0x02,0x2e,0x1a,0x71,0x65,0x6c,0xa3, 0x13,0xab,0x2a,0x40,0x2e,0x32,0xf2,0xae,0x1c,0xb4,0xcb,0xd1,0x0e,0xd1,0x3f,0x3e, 0x73,0xa9,0x23,0x61,0xed,0x37,0xb4,0xf6,0x4a,0x13,0xdf,0xd0,0xc4,0x95,0x56,0x8f, 0xfd,0x71,0xe0,0xf2,0xa8,0x19,0x9c,0xc3,0x19,0xb5,0x8a,0x63,0x84,0xb0,0x07,0x7d, 0x14,0xa1,0xe3,0xa4,0xc7,0x99,0x17,0xa3,0x19,0xfb,0xa0,0x1d,0x17,0x5a,0xdb,0x2b, 0xcd,0xd1,0xb6,0xf0,0x3d,0x8c,0x61,0x04,0x1b,0xdb,0xbd,0x26,0xee,0x3c,0xf8,0xb6, 0x85,0xaf,0x06,0xdc,0xf8,0x94,0x00,0xce,0xdb,0x61,0x87,0x4f,0xe7,0x36,0x20,0x0c, 0xc6,0x55,0xc4,0x3b,0xd1,0x7e,0xc2,0xe3,0x2a,0xb6,0x31,0x68,0xdc,0xb6,0x70,0x99, 0x84,0x12,0x4e,0xc4,0xc8,0x9f,0xa9,0xbb,0xda,0x1d,0x38,0xc9,0xfc,0x49,0x73,0x38, 0xe3,0x97,0x11,0x91,0xdb,0x69,0x5d,0xc6,0x85,0x37,0xd7,0x71,0x79,0x6c,0xf1,0xd2, 0x32,0x73,0x31,0x0c,0x8d,0x95,0xd1,0x18,0xf0,0x94,0xd6,0xef,0x76,0x00,0xf0,0x42, 0x15,0xa3,0x3c,0xd8,0x37,0xad,0xd2,0xaf,0x3e,0xf9,0xf5,0xf3,0xc9,0x57,0x67,0xca, 0x2c,0x53,0xc4,0x28,0x89,0x69,0x96,0xd3,0x4c,0xec,0xf3,0x92,0x0a,0x25,0x48,0x66, 0x4a,0xb3,0xa5,0xd8,0x6c,0x88,0x91,0xd9,0x86,0x96,0x7e,0x36,0x5f,0x85,0x96,0x5a, 0x4e,0x6e,0x14,0xd2,0xfd,0xf3,0x84,0x42,0x6d,0xf3,0xbd,0xdc,0xcb,0x2d,0x55,0x99, 0xa1,0x7b,0x93,0x0b,0xe3,0xb5,0xf3,0xb2,0x2c,0x91,0xce,0x0b,0xb2,0x29,0x72,0xa3, 0x91,0xa4,0x94,0xc1,0x7b,0xa5,0xb9,0x79,0xf0,0xa3,0x7b,0x2b,0x56,0x9c,0xff,0x0c, 0xb2,0x66,0x45,0x4c,0xb7,0x28,0x45,0x38,0xd0,0x90,0x37,0x98,0x7f,0xf0,0x9a,0x15, 0xd7,0x69,0xff,0xdd,0x8a,0x9b,0x3c,0xff,0x6c,0xc5,0xe0,0xae,0x24,0x18,0xaa,0x02, 0xfd,0x68,0x6b,0xdb,0x89,0x06,0xf0,0x83,0x18,0xd5,0xaa,0xf9,0x82,0x4f,0xef,0x7c, 0xe7,0x59,0xe1,0x22,0x61,0x30,0x5e,0x2b,0x7f,0x01,0x00,0x00,0xff,0xff,0x03,0x00, 0x2b,0xab,0x4f,0xf9,0xbb,0xf7,0x90,0xb5,0x34,0x8f,0x7c,0xae,0x72,0xa0,0x80,0xd2, 0x69,0xc7,0xa2,0xe7,0x44,0x53,0xeb,0x75,0x78,0xda,0xca,0x48,0xcd,0xc9,0xc9,0xcf, 0x80,0x13,0x5c,0x00,0x00,0x00,0x00,0xff,0xff,0x03,0x00,0x37,0xf7,0x06,0x47 }; static void verify2(struct archive *a, struct archive_entry *ae) { (void)a; /* UNUSED */ assertEqualInt(archive_entry_filetype(ae), AE_IFLNK); assertEqualInt(archive_entry_mode(ae) & 0777, 0755); assertEqualInt(archive_entry_uid(ae), UID); assertEqualInt(archive_entry_gid(ae), GID); assertEqualString(archive_entry_uname(ae), UNAME); assertEqualString(archive_entry_gname(ae), GNAME); assertEqualString(archive_entry_pathname(ae), "symlink"); assertEqualString(archive_entry_symlink(ae), "f1"); assert(archive_entry_hardlink(ae) == NULL); } /* Character device node. #How to make mknod devchar c 0 30 chown $UNAME:$GNAME devchar chmod 0644 devchar env TZ=utc touch -afm -t 197001020000.01 devchar xar -cf archive3.xar devchar od -t x1 archive3.xar | sed -E -e 's/^0[0-9]+//;s/^ //;s/( )([0-9a-f]{2})/0x\2,/g;$ D' > archive3.xar.txt */ static unsigned char archive3[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x01,0x38, 0x00,0x00,0x00,0x00,0x00,0x00,0x02,0x3b,0x00,0x00,0x00,0x01,0x78,0xda,0x7c,0x92, 0x4d,0x6e,0xc3,0x20,0x10,0x85,0xf7,0x39,0x05,0xf2,0xde,0x05,0x9c,0x9f,0x36,0xd6, 0x84,0xec,0x7a,0x82,0x74,0xd3,0x1d,0xc2,0x93,0x98,0xd4,0x36,0x11,0xe0,0x28,0xe9, 0xe9,0x0b,0xe3,0xa4,0x69,0xa5,0xaa,0x92,0x25,0x1e,0x8f,0xef,0x8d,0x86,0xc1,0xb0, 0xbd,0xf4,0x1d,0x3b,0xa3,0x0f,0xd6,0x0d,0x9b,0x42,0x3e,0x89,0x82,0xe1,0x60,0x5c, 0x63,0x87,0xc3,0xa6,0x78,0xdb,0xbd,0x96,0x2f,0xc5,0x56,0xcd,0xe0,0xa2,0xbd,0x9a, 0x31,0x88,0xce,0xa4,0x85,0x81,0xf1,0xa8,0x63,0x4a,0x94,0xd1,0xf6,0xa8,0x2a,0x21, 0xd6,0xa5,0xac,0x4a,0x51,0xed,0xa4,0xa8,0xab,0x79,0x2d,0x57,0xc0,0x7f,0x23,0x14, 0x6a,0xd1,0x7c,0x84,0xb1,0x67,0x21,0x5e,0x3b,0xdc,0x14,0xa1,0xd5,0xb2,0xc8,0x27, 0x0c,0xdc,0x7e,0x1f,0x30,0x2a,0x01,0xfc,0xa6,0xc8,0x0d,0xf6,0x33,0x17,0x07,0x4e, 0x22,0x97,0xe0,0xf7,0x1a,0xb4,0xdb,0xdb,0x0e,0x99,0x6d,0x52,0xdb,0xb7,0x32,0xe6, 0xaf,0x76,0xaa,0x7a,0xb9,0x7c,0x4f,0xc9,0x7b,0x1f,0x0c,0x7a,0x92,0x72,0xfd,0x2c, 0x4a,0x21,0x33,0x26,0x44,0x9d,0x3f,0x99,0xb0,0xfe,0x81,0xe9,0x7f,0x30,0xfd,0xc0, 0x0e,0xde,0x8d,0x27,0x65,0x46,0x04,0x3e,0xc9,0xc9,0xb5,0x8d,0x92,0x42,0xc8,0x64, 0x26,0x45,0xd6,0x18,0xd0,0x4f,0x1c,0xa9,0xc9,0xfb,0xc6,0xc6,0x3b,0xd6,0xbb,0x06, 0x95,0x58,0x2d,0x16,0xa9,0x99,0x2c,0xc9,0x6c,0xf0,0x6c,0xcd,0xa4,0x13,0x61,0x07, 0xe7,0xd5,0x3c,0x0d,0x66,0x52,0x37,0x57,0x1f,0x93,0xce,0x26,0x09,0x8a,0xf1,0x1f, 0x39,0x88,0xd7,0x13,0x2a,0xd3,0x6a,0xaf,0x4d,0x44,0xcf,0xc2,0x09,0x8d,0xd5,0x1d, 0x70,0xf2,0x89,0x18,0x74,0xba,0x54,0x8a,0x64,0x08,0x38,0xed,0x68,0xea,0x79,0xd0, 0xf9,0xf9,0x39,0xbd,0x3f,0x70,0xfa,0x1b,0xbe,0x00,0x00,0x00,0xff,0xff,0x03,0x00, 0xab,0x43,0xa3,0xac,0x76,0x40,0x1e,0x8b,0x95,0x0d,0x28,0x79,0x79,0x43,0x49,0x4e, 0x16,0xa1,0x56,0x99,0x1f,0x83,0x77,0x41 }; static void verify3(struct archive *a, struct archive_entry *ae) { (void)a; /* UNUSED */ assertEqualInt(archive_entry_filetype(ae), AE_IFCHR); assertEqualInt(archive_entry_mode(ae) & 0777, 0644); assertEqualInt(archive_entry_uid(ae), UID); assertEqualInt(archive_entry_gid(ae), GID); assertEqualString(archive_entry_uname(ae), UNAME); assertEqualString(archive_entry_gname(ae), GNAME); assertEqualString(archive_entry_pathname(ae), "devchar"); assert(archive_entry_symlink(ae) == NULL); assert(archive_entry_hardlink(ae) == NULL); assertEqualInt(archive_entry_mtime(ae), 86401); } /* Block device node. #How to make mknod devblock b 0 30 chown $UNAME:$GNAME devblock chmod 0644 devblock env TZ=utc touch -afm -t 197001020000.01 devblock xar -cf archive4.xar devblock od -t x1 archive4.xar | sed -E -e 's/^0[0-9]+//;s/^ //;s/( )([0-9a-f]{2})/0x\2,/g;$ D' > archive4.xar.txt */ static unsigned char archive4[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x01,0x34, 0x00,0x00,0x00,0x00,0x00,0x00,0x02,0x38,0x00,0x00,0x00,0x01,0x78,0xda,0x7c,0x92, 0xc1,0x6e,0xc2,0x30,0x0c,0x86,0xef,0x3c,0x45,0xd4,0x7b,0x17,0x07,0xd0,0x0a,0x95, 0x09,0xb7,0x3d,0x01,0xbb,0xec,0x96,0xa5,0x06,0x32,0xda,0xa6,0x6a,0x5a,0x04,0x7b, 0xfa,0x25,0x2e,0x8c,0x4d,0x9a,0x26,0x55,0xea,0x97,0x3f,0x9f,0x2d,0x37,0x29,0x6e, 0x2f,0x4d,0x2d,0xce,0xd4,0x07,0xe7,0xdb,0x4d,0xa6,0x9e,0x20,0x13,0xd4,0x5a,0x5f, 0xb9,0xf6,0xb0,0xc9,0x5e,0x77,0x2f,0xf9,0x2a,0xdb,0xea,0x19,0x5e,0x4c,0xaf,0x67, 0x02,0x07,0x6f,0xe3,0x4b,0xa0,0xed,0xc9,0x0c,0xb1,0x22,0x1f,0x5c,0x43,0x7a,0x0e, 0xb0,0xce,0xd5,0x3c,0x87,0xf9,0x4e,0x41,0xb9,0x58,0x95,0xaa,0x40,0xf9,0x5b,0xe1, 0xa2,0x23,0xd9,0x53,0x18,0x1b,0x11,0x86,0x6b,0x4d,0x9b,0x2c,0x1c,0x8d,0xca,0xd2, 0x8e,0x40,0xbf,0xdf,0x07,0x1a,0x34,0xa0,0xbc,0x11,0xa7,0xc1,0x7d,0xa6,0xe6,0x28, 0x19,0x52,0x0b,0x79,0xef,0xc1,0xab,0xbd,0xab,0x49,0xb8,0x2a,0x8e,0x7d,0x6b,0x63, 0xff,0x1e,0x07,0x8a,0xb7,0x58,0x79,0x9f,0x43,0x60,0xc3,0xa8,0xd6,0x05,0xe4,0xa0, 0x92,0x06,0x50,0xa6,0x47,0x45,0xad,0x79,0x68,0xe6,0x1f,0xcd,0x3c,0xb4,0x43,0xef, 0xc7,0x4e,0xdb,0x91,0x50,0x4e,0x38,0xa5,0xae,0xd2,0x0a,0x40,0xc5,0x30,0x12,0x47, 0x63,0xa0,0x7e,0xf2,0x98,0xa6,0xec,0x5b,0x1b,0xef,0x5a,0xe3,0x2b,0xd2,0xf0,0xbc, 0x5c,0xc6,0x61,0x12,0x72,0x58,0xd1,0xd9,0xd9,0x89,0xa3,0xe1,0x5a,0xdf,0xeb,0x45, 0x3c,0x98,0x89,0x6e,0xa9,0xf9,0x88,0x9c,0x42,0x06,0x2e,0x93,0x3f,0xea,0x70,0xb8, 0x76,0xa4,0xdf,0x6b,0x6f,0x4f,0x22,0x74,0x64,0x9d,0xa9,0x51,0x72,0xc6,0xbb,0xad, 0x89,0x1f,0x14,0x75,0x16,0x50,0xf2,0x92,0x8f,0x3c,0x9d,0x72,0xba,0x7b,0xc9,0x97, 0x8f,0x92,0x7f,0x85,0x2f,0x00,0x00,0x00,0xff,0xff,0x03,0x00,0xbe,0x66,0xa2,0x82, 0x3a,0x54,0xd3,0x61,0xaa,0x8e,0x30,0x4c,0xc8,0x36,0x3b,0x7a,0xa4,0xb9,0xef,0xfc, 0x7a,0x5d,0x21,0xde }; static void verify4(struct archive *a, struct archive_entry *ae) { (void)a; /* UNUSED */ assertEqualInt(archive_entry_filetype(ae), AE_IFBLK); assertEqualInt(archive_entry_mode(ae) & 0777, 0644); assertEqualInt(archive_entry_uid(ae), UID); assertEqualInt(archive_entry_gid(ae), GID); assertEqualString(archive_entry_uname(ae), UNAME); assertEqualString(archive_entry_gname(ae), GNAME); assertEqualString(archive_entry_pathname(ae), "devblock"); assert(archive_entry_symlink(ae) == NULL); assert(archive_entry_hardlink(ae) == NULL); assertEqualInt(archive_entry_mtime(ae), 86401); } /* Directory. #How to make mkdir dir1 chown $UNAME:$GNAME dir1 chmod 0755 dir1 env TZ=utc touch -afm -t 197001020000.01 dir1 xar -cf archive5.xar dir1 od -t x1 archive5.xar | sed -E -e 's/^0[0-9]+//;s/^ //;s/( )([0-9a-f]{2})/0x\2,/g;$ D' > archive5.xar.txt */ static unsigned char archive5[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x01,0x16, 0x00,0x00,0x00,0x00,0x00,0x00,0x01,0xec,0x00,0x00,0x00,0x01,0x78,0xda,0x74,0x91, 0xc1,0x6e,0xc2,0x30,0x0c,0x86,0xef,0x3c,0x45,0xd4,0x7b,0x17,0xa7,0x83,0x31,0xaa, 0x34,0xdc,0xf6,0x04,0xec,0xb2,0x5b,0x95,0x1a,0x88,0x68,0x1a,0x94,0xa4,0x13,0xdd, 0xd3,0x2f,0x71,0xe9,0xd0,0xa4,0x4d,0xaa,0xd4,0x3f,0xbf,0x3f,0xff,0xb2,0x6c,0xb9, 0xbf,0xd9,0x9e,0x7d,0xa2,0x0f,0xc6,0x0d,0x4d,0x21,0x9e,0xa0,0x60,0x38,0x68,0xd7, 0x99,0xe1,0xd4,0x14,0xef,0x87,0xb7,0xf2,0xb5,0xd8,0xab,0x95,0xbc,0xb5,0x5e,0xad, 0x98,0x8c,0x4e,0xa7,0x1f,0x93,0xda,0x63,0x1b,0x53,0x47,0x19,0x8d,0x45,0x55,0x01, 0xec,0x4a,0x51,0x95,0x50,0x1d,0x04,0xd4,0x6b,0x51,0xaf,0x37,0x92,0xff,0x46,0xa8, 0xe9,0x8c,0xfa,0x12,0x46,0xcb,0x42,0x9c,0x7a,0x6c,0x8a,0x70,0x6e,0x45,0x91,0x2b, 0x4c,0xba,0xe3,0x31,0x60,0x54,0x20,0xf9,0x5d,0x91,0x1b,0xcc,0x57,0x0e,0x97,0x9c, 0x44,0x8e,0xe0,0x4b,0x06,0xbd,0x8e,0xa6,0x47,0x66,0xba,0x34,0xf6,0x3d,0x46,0xff, 0x3d,0xce,0x33,0x7c,0xa4,0xce,0x65,0x0e,0x26,0x2d,0x49,0xb1,0xdb,0x42,0x09,0x22, 0x63,0x00,0x75,0xfe,0x44,0xc2,0xec,0x03,0x6b,0xff,0x49,0x7b,0x49,0x58,0xfb,0xc0, 0x4e,0xde,0x8d,0x57,0xa5,0x47,0x94,0x7c,0x96,0xb3,0x6b,0x3a,0x25,0x00,0x44,0x32, 0x93,0x22,0x6b,0x0c,0xe8,0x67,0x8e,0xd4,0xec,0xfd,0x60,0xe3,0x82,0x59,0xd7,0xa1, 0x82,0xed,0x26,0xed,0x90,0x24,0x99,0x71,0xba,0xa2,0xea,0x8c,0x47,0x1d,0x9d,0x9f, 0x24,0xa7,0x37,0x55,0x86,0xd6,0x52,0x25,0x45,0x90,0xa4,0x35,0xe5,0xcd,0xe4,0x7b, 0x71,0x3a,0x98,0xe4,0x74,0xbe,0x6f,0x00,0x00,0x00,0xff,0xff,0x03,0x00,0x23,0x7a, 0x8c,0x2f,0x78,0xe9,0x69,0x28,0x93,0x14,0x72,0x68,0x8d,0xeb,0x42,0x7b,0xf6,0x0f, 0x70,0x64,0xa3,0xff,0xb9,0x35 }; static void verify5(struct archive *a, struct archive_entry *ae) { (void)a; /* UNUSED */ assertEqualInt(archive_entry_filetype(ae), AE_IFDIR); assertEqualInt(archive_entry_mtime(ae), 86401); assertEqualInt(archive_entry_mode(ae) & 0777, 0755); assertEqualInt(archive_entry_uid(ae), UID); assertEqualInt(archive_entry_gid(ae), GID); assertEqualString(archive_entry_uname(ae), UNAME); assertEqualString(archive_entry_gname(ae), GNAME); } /* fifo #How to make mkfifo -m 0755 fifo chown $UNAME:$GNAME fifo env TZ=utc touch -afm -t 197001020000.01 fifo xar -cf archive6.xar fifo od -t x1 archive6.xar | sed -E -e 's/^0[0-9]+//;s/^ //;s/( )([0-9a-f]{2})/0x\2,/g;$ D' > archive6.xar.txt */ static unsigned char archive6[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x01,0x0e, 0x00,0x00,0x00,0x00,0x00,0x00,0x01,0xe7,0x00,0x00,0x00,0x01,0x78,0xda,0x7c,0x91, 0xc1,0x6e,0xc3,0x20,0x0c,0x86,0xef,0x7d,0x0a,0xc4,0x3d,0xc3,0x64,0xab,0xda,0x46, 0x94,0xde,0xf6,0x04,0xdd,0x65,0x37,0x44,0x9c,0x16,0x2d,0x84,0x2a,0x90,0xa9,0xdd, 0xd3,0x0f,0x9c,0x66,0xd5,0xa4,0x69,0x12,0x52,0xbe,0xfc,0x7c,0xb6,0x2c,0xac,0x0e, 0x57,0xdf,0xb3,0x4f,0x1c,0xa3,0x0b,0xc3,0x9e,0xcb,0x27,0xe0,0x0c,0x07,0x1b,0x5a, 0x37,0x9c,0xf6,0xfc,0xed,0xf8,0x5a,0x6d,0xf9,0x41,0xaf,0xd4,0xd5,0x8c,0x7a,0xc5, 0x54,0x0a,0x36,0x7f,0x98,0xb2,0x23,0x9a,0x94,0x2b,0xaa,0xe4,0x3c,0xea,0x1a,0x60, 0x57,0xc9,0xba,0x82,0xfa,0x28,0x65,0xf3,0x02,0x4d,0xbd,0x55,0xe2,0xb7,0x42,0x45, 0x67,0xb4,0x1f,0x71,0xf2,0x2c,0xa6,0x5b,0x8f,0x7b,0x1e,0xcf,0x46,0xf2,0x72,0xc3, 0x54,0xe8,0xba,0x88,0x49,0x83,0x12,0x77,0xa2,0x34,0xba,0xaf,0xd2,0x5c,0x09,0x82, 0xd2,0x42,0x2c,0x3d,0xe8,0xaf,0x73,0x3d,0x32,0xd7,0xe6,0xb1,0xef,0x6d,0xec,0xdf, 0xe3,0xc8,0xe7,0xf7,0x5c,0xb9,0xcc,0xc1,0x94,0x27,0x94,0xbb,0x0d,0x54,0x20,0x8b, 0x06,0xd0,0x94,0x23,0xb3,0xe6,0x1f,0x9a,0xf9,0x47,0x33,0x0f,0xed,0x34,0x86,0xe9, 0xa2,0xed,0x84,0x4a,0xcc,0x38,0xa7,0xae,0xd5,0x12,0x40,0xe6,0x30,0x13,0x45,0x53, 0xc4,0x71,0xf6,0x88,0xe6,0xec,0x47,0x9b,0x16,0xcd,0x87,0x16,0x35,0x6c,0xd6,0xeb, 0x3c,0x4c,0x41,0x0a,0xd3,0xed,0x82,0xba,0x73,0x5d,0x50,0x82,0x90,0xc2,0xc1,0xf8, 0x25,0x24,0xa4,0x17,0x2a,0x8f,0x52,0x56,0x25,0x68,0x57,0x4a,0xd0,0xe6,0xbe,0x01, 0x00,0x00,0xff,0xff,0x03,0x00,0x44,0x19,0x8a,0x2a,0x82,0xbc,0x8c,0xae,0x97,0xa7, 0x7d,0x65,0xa5,0x82,0xdb,0xaa,0xc2,0xcb,0xbe,0xf0,0x1f,0xd1,0xf9,0x56 }; static void verify6(struct archive *a, struct archive_entry *ae) { (void)a; /* UNUSED */ assertEqualInt(archive_entry_filetype(ae), AE_IFIFO); assertEqualInt(archive_entry_mode(ae) & 0777, 0755); assertEqualInt(archive_entry_uid(ae), UID); assertEqualInt(archive_entry_gid(ae), GID); assertEqualString(archive_entry_uname(ae), UNAME); assertEqualString(archive_entry_gname(ae), GNAME); assertEqualString(archive_entry_pathname(ae), "fifo"); assert(archive_entry_symlink(ae) == NULL); assert(archive_entry_hardlink(ae) == NULL); assertEqualInt(archive_entry_mtime(ae), 86401); } /* Verify that a file records with directory name. #How to make mkdir dir1 echo "hellohellohello" > dir1/f1 chown $UNAME:$GNAME dir1/f1 chmod 0644 dir1/f1 env TZ=utc touch -afm -t 197001020000.01 dir1/f1 xar -cf archive7.xar dir1/f1 od -t x1 archive7.xar | sed -E -e 's/^0[0-9]+//;s/^ //;s/( )([0-9a-f]{2})/0x\2,/g;$ D' > archive7.xar.txt */ static unsigned char archive7[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x01,0xbb, 0x00,0x00,0x00,0x00,0x00,0x00,0x03,0x8a,0x00,0x00,0x00,0x01,0x78,0xda,0x7c,0x53, 0xc9,0x6e,0xdb,0x30,0x14,0xbc,0xe7,0x2b,0x04,0xdd,0x55,0x2e,0xa2,0x16,0x1a,0xb4, 0x82,0x5e,0x8a,0xdc,0x93,0x5e,0x7a,0xa3,0xb9,0xd8,0x44,0xb5,0x41,0xa2,0x02,0x3b, 0x5f,0x5f,0x92,0xa2,0x1c,0xbb,0x59,0x00,0x01,0x1a,0x3e,0xce,0x1b,0x0d,0x9f,0x86, 0xec,0xf1,0xdc,0xb5,0xc9,0xab,0x9a,0x66,0x33,0xf4,0xfb,0x14,0xfd,0x80,0x69,0xa2, 0x7a,0x31,0x48,0xd3,0x1f,0xf7,0xe9,0xef,0x97,0x5f,0x59,0x9d,0x3e,0x36,0x0f,0xec, 0xcc,0xa7,0xe6,0x21,0x61,0x76,0x10,0xee,0x95,0x30,0x31,0x29,0x6e,0x5d,0x47,0x66, 0x4d,0xa7,0x1a,0x0c,0x21,0xcd,0x10,0xce,0x20,0x7e,0x41,0x68,0x57,0xe0,0x5d,0x51, 0x31,0x70,0x4f,0x09,0x4d,0x27,0x25,0xfe,0xce,0x4b,0x97,0xcc,0xf6,0xd2,0xaa,0x7d, 0x3a,0x9f,0x38,0x4a,0xfd,0x4e,0xc2,0x06,0xad,0x67,0x65,0x1b,0xc8,0x40,0x44,0xa1, 0x3a,0x9b,0x37,0x2f,0xce,0x40,0x00,0x5e,0x02,0x6c,0x1a,0x61,0xa5,0x4d,0xab,0x12, 0x23,0x9d,0xed,0x28,0x63,0x2f,0xa3,0x6a,0xa4,0x99,0x94,0xb0,0xc3,0x74,0x61,0x20, 0xac,0xc3,0x4e,0xcf,0xbb,0xb0,0x83,0x18,0x08,0x30,0x14,0xaf,0xfd,0x78,0xed,0x4f, 0x98,0xe4,0x96,0xaf,0x30,0x61,0xad,0xea,0x8f,0xf6,0xd4,0xe0,0x9c,0x81,0x08,0xe3, 0x46,0xb4,0x88,0xef,0xdd,0x6e,0x7e,0x51,0xf9,0xee,0xd7,0x17,0xb7,0x69,0x6e,0xa7, 0xe6,0xe3,0xd8,0x1a,0x11,0x46,0x03,0xce,0xd9,0xf1,0xcd,0x8c,0x29,0xd8,0xb8,0x7c, 0x12,0x27,0xf3,0xaa,0x64,0xf6,0xff,0xa8,0x9e,0x9f,0x7e,0xba,0x33,0xea,0x3a,0xcf, 0x25,0xd6,0x52,0x20,0x9a,0x13,0x9a,0xf3,0x03,0xa9,0x29,0x97,0x1c,0xe7,0xba,0xd6, 0x15,0x45,0xba,0xc4,0x5a,0xe4,0x25,0xad,0x19,0xf8,0x20,0x74,0x75,0x73,0xb6,0x13, 0x17,0xf6,0xcb,0x4f,0x14,0xb2,0x22,0x07,0x71,0x10,0x15,0x95,0xb9,0xa6,0x07,0x4d, 0xb8,0x76,0xe2,0xa4,0xae,0x6b,0xa8,0x48,0x81,0xcb,0x82,0x68,0x05,0x05,0xa5,0x0c, 0x7c,0x54,0x8a,0x33,0x04,0xef,0x43,0x64,0xe2,0xf3,0x7c,0x90,0xfa,0x8f,0xfb,0x95, 0x5b,0x30,0x1c,0xaf,0x0b,0x18,0xd1,0x0a,0x66,0x10,0x79,0x1e,0x84,0x3b,0xff,0x20, 0xc7,0xeb,0x6e,0x78,0xfc,0x1b,0x1e,0xbf,0xe1,0x1d,0xa7,0x61,0x19,0x1b,0xb1,0x28, 0x06,0x56,0x18,0xcb,0x46,0x36,0x08,0x42,0x17,0x02,0x8f,0xd6,0xda,0x32,0xab,0x69, 0x65,0x06,0x14,0x8b,0x57,0xe2,0x72,0x25,0x76,0x83,0x54,0x0d,0x2c,0x09,0x71,0x96, 0x3c,0x5c,0xab,0x21,0x62,0x3e,0x48,0x37,0x69,0x8b,0x71,0xd3,0x77,0x61,0x03,0x9e, 0xb4,0x86,0x38,0x22,0xd7,0xe1,0xaf,0x13,0x03,0xe1,0x72,0xfd,0x03,0x00,0x00,0xff, 0xff,0x03,0x00,0x8d,0xb1,0x06,0x76,0xa6,0x7a,0xc3,0xbb,0x13,0x3d,0x45,0xe2,0x2b, 0x3b,0xd0,0x88,0xc7,0x58,0x7b,0xbd,0x30,0x9d,0x01,0x44,0x78,0xda,0xca,0x48,0xcd, 0xc9,0xc9,0xcf,0x80,0x13,0x5c,0x00,0x00,0x00,0x00,0xff,0xff,0x03,0x00,0x37,0xf7, 0x06,0x47 }; static void verify7(struct archive *a, struct archive_entry *ae) { (void)a; /* UNUSED */ assert(archive_entry_filetype(ae) == AE_IFREG); assertEqualInt(archive_entry_mode(ae) & 0777, 0644); assertEqualInt(archive_entry_uid(ae), UID); assertEqualInt(archive_entry_gid(ae), GID); assertEqualString(archive_entry_uname(ae), UNAME); assertEqualString(archive_entry_gname(ae), GNAME); assertEqualString(archive_entry_pathname(ae), "dir1/f1"); assert(archive_entry_hardlink(ae) == NULL); assert(archive_entry_symlink(ae) == NULL); assertEqualInt(archive_entry_mtime(ae), 86401); } /* Verify that a file records with bzip2 compression #How to make echo "hellohellohello" > f1 chown $UNAME:$GNAME f1 chmod 0644 f1 env TZ=utc touch -afm -t 197001020000.01 f1 xar --compression bzip2 -cf archive8.xar f1 od -t x1 archive8.xar | sed -E -e 's/^0[0-9]+//;s/^ //;s/( )([0-9a-f]{2})/0x\2,/g;$ D' > archive8.xar.txt */ static unsigned char archive8[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x01,0xb1, 0x00,0x00,0x00,0x00,0x00,0x00,0x03,0x42,0x00,0x00,0x00,0x01,0x78,0xda,0x7c,0x53, 0xcb,0x6e,0xdc,0x20,0x14,0xdd,0xe7,0x2b,0x90,0xf7,0x0e,0x60,0xe3,0x07,0x23,0x86, 0xa8,0x9b,0xa8,0xfb,0x4e,0x37,0xdd,0x61,0x1e,0x63,0x14,0xbf,0x64,0xe3,0x68,0x92, 0xaf,0x2f,0x60,0x3b,0xa3,0x34,0x6d,0x25,0x4b,0x3e,0x1c,0x0e,0xe7,0x5e,0xee,0xe5, 0xb2,0xa7,0x5b,0xdf,0x81,0x57,0x3d,0x2f,0x76,0x1c,0xce,0x09,0x7e,0x44,0x09,0xd0, 0x83,0x1c,0x95,0x1d,0xae,0xe7,0xe4,0xe7,0xe5,0x39,0xad,0x93,0x27,0xfe,0xc0,0x6e, 0x62,0xe6,0x0f,0x80,0xb9,0x51,0xfa,0x1f,0x60,0x72,0xd6,0xc2,0xf9,0x13,0xa9,0xb3, 0xbd,0xe6,0x19,0x42,0x34,0xc5,0x59,0x8a,0xc8,0x05,0xd1,0x13,0xc6,0x27,0x9c,0x33, 0xf8,0x59,0x12,0x0f,0xb5,0x5a,0xbe,0x2c,0x6b,0x0f,0x16,0xf7,0xd6,0xe9,0x73,0xb2, 0xb4,0x02,0x27,0x61,0x07,0xb0,0xd1,0x98,0x45,0x3b,0x8e,0x18,0xdc,0x51,0x64,0x17, 0xfb,0x1e,0xcc,0x19,0x8c,0x20,0x58,0xc0,0xc3,0x23,0xae,0x8c,0xed,0x34,0xb0,0xca, 0xa7,0xbd,0xdb,0x28,0xe1,0x44,0x44,0x80,0x75,0x7a,0xb8,0xba,0x96,0x13,0xc2,0xe0, 0x0e,0x37,0x7e,0xf7,0xcf,0x3e,0x87,0xda,0x63,0xe1,0xf2,0x1e,0xcb,0x73,0x47,0x21, 0x8e,0x84,0xc5,0x34,0x75,0x56,0xc6,0x5b,0xc1,0x5b,0xda,0xbc,0xdb,0x29,0x4b,0xe0, 0xae,0x15,0xb3,0x6c,0xed,0xab,0x56,0xe9,0x9f,0xb7,0xfc,0xf1,0xfd,0x9b,0x4f,0xcf, 0xe4,0xa4,0x28,0x4a,0x94,0xcb,0x3a,0xcf,0x9b,0x26,0x93,0xaa,0x92,0xba,0x29,0xa8, 0x2a,0x89,0x29,0xa8,0x50,0x22,0x97,0x45,0xa1,0x71,0xe5,0xeb,0xf6,0xc5,0xe8,0x48, 0xe6,0xe6,0x66,0x21,0xdd,0x3f,0x23,0x14,0xaa,0x22,0x8d,0x6c,0x64,0x45,0x55,0x6e, 0x68,0x63,0x88,0x30,0xa6,0x36,0xa4,0xae,0x6b,0xa4,0x49,0x91,0x95,0x05,0x31,0x1a, 0x49,0x4a,0x19,0xfc,0xea,0xb4,0x55,0x0f,0x7e,0x94,0x8f,0xc9,0xbf,0xf7,0x15,0xd5, 0xbf,0x7c,0x0b,0x8e,0x86,0x02,0xd6,0x47,0x88,0x69,0x85,0x52,0x84,0x53,0x94,0x5d, 0x10,0x3a,0x85,0x0f,0x7b,0x59,0x7f,0x97,0x89,0xff,0xc8,0xc4,0x5d,0x76,0x9d,0xc7, 0x75,0xe2,0x72,0xd5,0x0c,0x6e,0x70,0x63,0xad,0xe2,0x18,0x21,0xec,0x49,0x8f,0x22, 0xb5,0x2e,0x7a,0xde,0x74,0x11,0x6d,0xdc,0x87,0x6c,0x3d,0x64,0xfd,0xa8,0x34,0x47, 0x65,0x78,0x02,0x11,0x46,0xd2,0xbd,0x4d,0x1a,0x74,0x76,0x78,0x39,0x27,0xe3,0x6c, 0xaf,0x76,0x10,0x5d,0xc2,0x5b,0x31,0xab,0xc0,0x31,0x18,0xb6,0x37,0xe1,0x20,0x7c, 0x5e,0xc6,0xfb,0x45,0x10,0x1f,0x5f,0x78,0x6f,0x61,0x0a,0x60,0x1c,0x03,0x06,0xe3, 0x50,0xfc,0x06,0x00,0x00,0xff,0xff,0x03,0x00,0x19,0xcf,0xf5,0xc0,0xf9,0x65,0xe8, 0x78,0xc3,0xfa,0x5f,0x0a,0xf6,0x09,0x17,0xd8,0xb0,0x54,0xb9,0x02,0x8d,0x91,0x31, 0x9c,0x42,0x5a,0x68,0x39,0x31,0x41,0x59,0x26,0x53,0x59,0xc1,0x52,0x36,0xf7,0x00, 0x00,0x03,0x41,0x00,0x00,0x10,0x02,0x44,0xa0,0x00,0x21,0xb4,0x01,0x9a,0x0d,0x46, 0xa5,0x32,0x38,0xbb,0x92,0x29,0xc2,0x84,0x86,0x0a,0x91,0xb7,0xb8 }; /* Verify that a file records with no compression #How to make echo "hellohellohello" > f1 chown $UNAME:$GNAME f1 chmod 0644 f1 env TZ=utc touch -afm -t 197001020000.01 f1 xar --compression none -cf archive9.xar f1 od -t x1 archive9.xar | sed -E -e 's/^0[0-9]+//;s/^ //;s/( )([0-9a-f]{2})/0x\2,/g;$ D' > archive9.xar.txt */ static unsigned char archive9[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x01,0x98, 0x00,0x00,0x00,0x00,0x00,0x00,0x03,0x47,0x00,0x00,0x00,0x01,0x78,0xda,0xa4,0x53, 0x4d,0x6f,0xe3,0x20,0x14,0xbc,0xf7,0x57,0x20,0xee,0x5e,0xc0,0x25,0x89,0x1d,0x11, 0xaa,0x5e,0xaa,0xbd,0x6f,0xf6,0xb2,0x37,0x02,0x8f,0x18,0xc5,0x1f,0x11,0xc6,0x55, 0xba,0xbf,0x7e,0x01,0xdb,0xad,0xba,0x55,0x7b,0xa9,0x64,0xc9,0xe3,0x61,0xde,0x78, 0x78,0xf0,0xc4,0xc3,0xad,0x6b,0xd1,0x33,0xf8,0xd1,0x0d,0xfd,0x01,0xb3,0x1f,0x14, 0x23,0xe8,0xf5,0x60,0x5c,0x7f,0x3e,0xe0,0xdf,0xc7,0xa7,0xa2,0xc2,0x0f,0xf2,0x4e, 0xdc,0x94,0x97,0x77,0x48,0x84,0x41,0xc7,0x17,0x12,0xda,0x83,0x0a,0xb1,0xa2,0x08, 0xae,0x03,0x59,0x52,0x5a,0x17,0xac,0x2c,0x28,0x3f,0xd2,0x7a,0xcf,0xaa,0x3d,0xaf, 0x05,0x79,0x2f,0xc9,0x45,0x0d,0xe8,0xcb,0x38,0x75,0x68,0x0c,0x2f,0x2d,0x1c,0xf0, 0xd8,0x28,0x86,0xd3,0x0a,0x12,0x83,0xb5,0x23,0x04,0x49,0x05,0x59,0x50,0x66,0x47, 0xf7,0x37,0x99,0x0b,0x92,0x41,0xb2,0x20,0xab,0x47,0xfe,0xb2,0xae,0x05,0xe4,0x4c, 0x8c,0xbd,0xd8,0x18,0x15,0x54,0x46,0x48,0xb4,0xd0,0x9f,0x43,0x23,0xd9,0x56,0x90, 0x05,0xce,0xfc,0xba,0xb9,0x35,0x84,0xba,0x5e,0x5b,0xa7,0x73,0x52,0x32,0xe8,0x00, 0xa1,0x18,0x43,0x4c,0xde,0x61,0xb2,0x14,0x2c,0x81,0xca,0xf7,0xd9,0x96,0x70,0xc9, 0x7e,0x0d,0x17,0x39,0xe5,0x75,0xe3,0x9e,0xc1,0x14,0xff,0x6f,0xf5,0xd7,0xcf,0xc7, 0x98,0x71,0x63,0x76,0xfc,0xa4,0x4f,0x7a,0x57,0x9b,0x7b,0x5b,0x9f,0x2c,0x57,0xd6, 0x56,0x96,0x57,0x55,0x45,0x81,0x6f,0xca,0xed,0x86,0x5b,0xa0,0xba,0x8e,0xcd,0xfb, 0x60,0xb4,0xa6,0xbf,0x05,0xaf,0x62,0xca,0xef,0xff,0xe1,0xa3,0xd3,0xdc,0x42,0xf2, 0xda,0x43,0xa1,0x3f,0x39,0xdc,0xed,0x9f,0x78,0x0e,0xeb,0xa9,0x22,0xd1,0x65,0xc8, 0xea,0x1d,0x2d,0x28,0x2b,0x68,0x79,0xa4,0x74,0x9f,0x1e,0x16,0x65,0xdd,0x9b,0x4c, 0x7d,0x21,0x53,0x6f,0xb2,0xb3,0x1f,0xa6,0xab,0xd4,0x13,0x08,0x32,0xc3,0x99,0x75, 0x46,0x32,0x4a,0x59,0x24,0x23,0xca,0xd4,0x34,0x82,0x9f,0x75,0x19,0xcd,0xdc,0xab, 0x6c,0x5a,0x65,0xdd,0x60,0x40,0xd2,0x2d,0xe7,0x31,0x4c,0x82,0x99,0x0c,0x2f,0x57, 0x40,0xad,0xeb,0x2f,0x07,0x3c,0x78,0x77,0x76,0xbd,0x6a,0xb1,0x6c,0x94,0x37,0x89, 0x13,0x24,0x2d,0xcf,0xc2,0x5e,0xc5,0x5c,0x36,0xfa,0x65,0x90,0x6f,0x60,0xba,0x74, 0x69,0x14,0x48,0x9e,0x05,0x41,0xf2,0x64,0xfc,0x03,0x00,0x00,0xff,0xff,0x03,0x00, 0xee,0x8e,0xf8,0x75,0xa1,0xaf,0x74,0x71,0x3f,0x40,0x08,0xab,0x13,0x7d,0xc0,0x82, 0x3a,0x56,0xeb,0x4e,0x35,0xf1,0x35,0xb7,0x68,0x65,0x6c,0x6c,0x6f,0x68,0x65,0x6c, 0x6c,0x6f,0x68,0x65,0x6c,0x6c,0x6f,0x0a }; /* Verify that a file records with md5 hashing algorithm #How to make echo "hellohellohello" > f1 chown $UNAME:$GNAME f1 chmod 0644 f1 env TZ=utc touch -afm -t 197001020000.01 f1 xar --toc-cksum md5 -cf archive10.xar f1 od -t x1 archive10.xar | sed -E -e 's/^0[0-9]+//;s/^ //;s/( )([0-9a-f]{2})/0x\2,/g;$ D' > archive10.xar.txt */ static unsigned char archive10[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x01,0xaf, 0x00,0x00,0x00,0x00,0x00,0x00,0x03,0x40,0x00,0x00,0x00,0x02,0x78,0xda,0x7c,0x53, 0x4d,0x6f,0xdc,0x20,0x10,0xbd,0xe7,0x57,0x20,0xee,0x0e,0x60,0xb3,0xb6,0x59,0xb1, 0x44,0xbd,0x44,0xbd,0x77,0x7b,0xe9,0x8d,0xe5,0xc3,0x8b,0xe2,0x2f,0x61,0x1c,0x6d, 0xf2,0xeb,0x8b,0xb1,0x9d,0xb4,0x4d,0x52,0xc9,0x92,0x1f,0x8f,0xc7,0x9b,0x61,0x86, 0xe1,0x0f,0xb7,0xae,0x05,0xcf,0xc6,0x4f,0x6e,0xe8,0x4f,0x90,0xdc,0x63,0x08,0x4c, 0xaf,0x06,0xed,0xfa,0xe6,0x04,0x7f,0x9e,0x1f,0xb3,0x1a,0x3e,0x88,0x3b,0x7e,0x93, 0x5e,0xdc,0x01,0x1e,0x06,0x15,0x7f,0x80,0x2b,0x6f,0x64,0x88,0x27,0xb2,0xe0,0x3a, 0x23,0x72,0x8c,0x59,0x46,0xf2,0x0c,0xd3,0x33,0x66,0xc7,0x02,0x1f,0x69,0xcd,0xd1, 0xdf,0x92,0x74,0xe8,0x6a,0xd4,0xd3,0x34,0x77,0x60,0x0a,0x2f,0xad,0x39,0xc1,0x4e, 0x1f,0xe0,0xb2,0x01,0xf8,0x60,0xed,0x64,0x82,0xc0,0x1c,0x6d,0x28,0xb1,0x93,0x7b, 0x35,0x82,0x94,0x1c,0x25,0xb0,0x38,0xa0,0xdd,0x22,0xad,0xac,0x6b,0x0d,0x70,0x3a, 0x66,0xbd,0xd9,0x68,0x19,0x64,0x42,0x80,0xb7,0xa6,0x6f,0xc2,0x55,0xe4,0x05,0x47, 0x1b,0x5c,0xf9,0xcd,0x7f,0x71,0xfd,0x23,0xd4,0x27,0xb1,0x22,0xb7,0xd7,0x61,0xcf, 0x57,0x8e,0x63,0xeb,0x54,0xba,0x14,0xba,0x65,0xcd,0xab,0x1b,0x21,0xda,0xa4,0xd2, 0xab,0xab,0x7b,0x36,0x3a,0xfb,0xf7,0x8e,0x3f,0xbe,0x7f,0x8b,0xd9,0xd9,0xba,0x28, 0x74,0x6e,0xb5,0x22,0xac,0xa0,0xac,0x90,0x17,0x5a,0x33,0xa9,0x65,0x5e,0xd8,0xda, 0x56,0x8c,0xd8,0x32,0xb7,0xaa,0x28,0x59,0xac,0xda,0x07,0xa3,0x3d,0x97,0x5b,0xf0, 0x52,0x85,0x2f,0x23,0x1c,0x74,0x45,0x2f,0xea,0xa2,0x2a,0xa6,0x0b,0xcb,0x2e,0x96, 0x4a,0x1b,0xbd,0x69,0x5d,0xd7,0xd8,0xd0,0x43,0x5e,0x1e,0xa8,0x35,0x58,0x31,0xc6, 0xd1,0x47,0xa7,0xb5,0x78,0xe8,0xad,0x7a,0x5c,0x7d,0xd1,0xd5,0xea,0x57,0xec,0xc0, 0xde,0x4e,0xc0,0xbb,0x04,0x09,0xab,0x70,0x86,0x49,0x86,0xf3,0x33,0xc6,0xc7,0xe5, 0x23,0x51,0xd6,0xbd,0xcb,0xe4,0x7f,0x64,0xf2,0x5d,0xd6,0xf8,0x61,0x1e,0x85,0x9a, 0x0d,0x47,0x2b,0x5c,0x59,0xa7,0x05,0xc1,0x98,0x44,0x32,0xa2,0x44,0xcd,0x93,0xf1, 0xab,0x2e,0xa1,0x95,0x7b,0x93,0xcd,0xbb,0xac,0x1b,0xb4,0x11,0xb8,0xa4,0x34,0x26, 0xb3,0xc0,0x44,0x86,0x97,0xd1,0x80,0xd6,0xf5,0x4f,0x27,0x38,0x78,0xd7,0xb8,0x5e, 0xb6,0x50,0x5c,0xa5,0xd7,0x0b,0xc7,0xd1,0xb2,0xbd,0x0a,0x7b,0x19,0xf3,0xb2,0xd1, 0x2f,0x81,0xf4,0xf6,0x96,0xe7,0xb6,0xcc,0x00,0x4a,0x43,0xc0,0x51,0x1a,0x89,0xdf, 0x00,0x00,0x00,0xff,0xff,0x03,0x00,0x27,0xf8,0xf5,0x28,0x87,0x01,0xb1,0xb7,0x18, 0xe8,0x34,0x20,0x06,0x5c,0x66,0x9a,0x43,0x26,0xe7,0x94,0x78,0xda,0xca,0x48,0xcd, 0xc9,0xc9,0xcf,0x80,0x13,0x5c,0x00,0x00,0x00,0x00,0xff,0xff,0x03,0x00,0x37,0xf7, 0x06,0x47 }; /* Verify that a file records with no hashing algorithm #How to make echo "hellohellohello" > f1 chown $UNAME:$GNAME f1 chmod 0644 f1 env TZ=utc touch -afm -t 197001020000.01 f1 xar --toc-cksum none -cf archive11.xar f1 od -t x1 archive11.xar | sed -E -e 's/^0[0-9]+//;s/^ //;s/( )([0-9a-f]{2})/0x\2,/g;$ D' > archive11.xar.txt */ static unsigned char archive11[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x01,0x98, 0x00,0x00,0x00,0x00,0x00,0x00,0x02,0xef,0x00,0x00,0x00,0x00,0x78,0xda,0x7c,0x52, 0xcb,0x6e,0xeb,0x20,0x14,0xdc,0xf7,0x2b,0x10,0x7b,0x17,0xb0,0x89,0x63,0x22,0x42, 0x75,0x37,0x55,0xf7,0xcd,0xdd,0x74,0x47,0x78,0x38,0xa8,0x7e,0xc9,0xc6,0x55,0xda, 0xaf,0xbf,0x3c,0xe2,0x56,0x55,0xd5,0x2b,0x21,0x79,0x3c,0xcc,0x39,0x67,0x74,0x18, 0xfe,0x70,0xed,0x3b,0xf0,0x66,0xe6,0xc5,0x8d,0xc3,0x11,0x92,0x7b,0x0c,0x81,0x19, 0xd4,0xa8,0xdd,0xd0,0x1e,0xe1,0xdf,0xd3,0x63,0xd1,0xc0,0x07,0x71,0xc7,0xaf,0x72, 0x16,0x77,0x80,0xfb,0x51,0x85,0x0f,0xe0,0x6a,0x36,0xd2,0x87,0x8a,0xc2,0xbb,0xde, 0x88,0x12,0x63,0x56,0x90,0xb2,0xc0,0xf4,0x44,0xf0,0x81,0x54,0x07,0x5a,0x73,0xf4, 0x5d,0x12,0x8b,0xac,0xeb,0x0c,0x70,0x3a,0x4c,0x81,0xf1,0x1f,0x70,0x2d,0xbd,0x4c, 0x08,0xf0,0xce,0x0c,0xad,0xbf,0x88,0xb2,0xe2,0xe8,0x06,0x33,0x3f,0x5a,0xbb,0x18, 0x2f,0x30,0x47,0x37,0x94,0xe9,0xc5,0x7d,0x18,0x41,0xc2,0x94,0x04,0x32,0xb7,0xd9, 0x06,0x8b,0x7f,0xef,0xcc,0x11,0xca,0x69,0xea,0x9c,0x4a,0x1e,0xd0,0xb5,0x68,0x3f, 0xdc,0x04,0xd1,0x4d,0x2a,0x67,0x75,0x71,0x6f,0x46,0x17,0xea,0x62,0xd4,0xeb,0xb2, 0xf6,0x5b,0xcd,0xf3,0xd3,0x9f,0x60,0xce,0x36,0x55,0xa5,0x4b,0xab,0x15,0x61,0x15, 0x65,0x95,0x3c,0xd3,0x86,0x49,0x2d,0xcb,0xca,0x36,0x76,0xcf,0x88,0xad,0x4b,0xab, 0xaa,0x9a,0x35,0x1c,0xfd,0x68,0xb4,0x79,0xb9,0xfa,0x59,0x2a,0xff,0xeb,0x84,0x9d, 0xde,0xd3,0xb3,0x3a,0xab,0x3d,0xd3,0x95,0x65,0x67,0x4b,0xa5,0x0d,0xbd,0x69,0xd3, 0x34,0xd8,0xd0,0x5d,0x59,0xef,0xa8,0x35,0x58,0x31,0xc6,0xd1,0xcf,0x4e,0x79,0x77, 0xe8,0x73,0x79,0x5c,0xfd,0xf2,0x08,0xe4,0x25,0xbc,0xc2,0xb6,0x7d,0xc0,0xfb,0x04, 0x09,0xdb,0xe3,0x02,0x93,0x02,0x97,0x27,0x8c,0x0f,0xf1,0x44,0x59,0xff,0x25,0x93, 0xff,0x91,0xc9,0x2f,0x59,0x3b,0x8f,0xeb,0x24,0xd4,0x6a,0x38,0xca,0x30,0xb3,0x4e, 0x0b,0x82,0x31,0x09,0x64,0x40,0x89,0x5a,0x17,0x33,0x67,0x5d,0x42,0x99,0xfb,0x94, 0xad,0x9b,0xac,0x1f,0xb5,0x11,0xb8,0xa6,0x34,0x98,0x89,0x30,0x91,0xfe,0x7d,0x32, 0xa0,0x73,0xc3,0xeb,0x11,0x8e,0xb3,0x6b,0xdd,0x20,0x3b,0x28,0x2e,0x72,0xd6,0x91, 0xe3,0x28,0x5e,0x67,0xe1,0x20,0x83,0x2f,0x1b,0xfa,0x25,0x10,0xc3,0x86,0x62,0xda, 0x62,0x64,0x51,0xca,0x2c,0x47,0x29,0xc1,0xff,0x00,0x00,0x00,0xff,0xff,0x03,0x00, 0xf1,0x18,0xdc,0x71,0x78,0xda,0xca,0x48,0xcd,0xc9,0xc9,0xcf,0x80,0x13,0x5c,0x00, 0x00,0x00,0x00,0xff,0xff,0x03,0x00,0x37,0xf7,0x06,0x47 }; /* Verify that a file which is missing timestamp information has the corresponding timestamps unset. #How to make e.g. struct archive *a = archive_write_new(); archive_write_set_format_xar(a); archive_write_add_filter_none(a); size_t used, buffsize = 1500; char *buff = (char*) malloc(buffsize); archive_write_open_memory(a, buff, buffsize, &used); struct archive_entry *ae = archive_entry_new(); archive_entry_copy_pathname(ae, "file"); archive_entry_set_size(ae, 8); archive_write_header(a, ae); archive_entry_free(ae); archive_write_data(a, "12345678", 9); archive_write_free(a); std::cout << std::string(buff, used); free(buff); $./a.out > f12.xar Verify toc has no mtime/atime sections $ xar --dump-toc=- -f f12.xar Dump contents $ ./a.out | od -t x1 | sed -E -e 's/^0[0-9]+//;s/^ //;s/( )([0-9a-f]{2})/0x\2,/g;$ D' */ static unsigned char archive12[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0xd1, 0x00,0x00,0x00,0x00,0x00,0x00,0x01,0x56,0x00,0x00,0x00,0x01,0x78,0x9c,0x55,0x90, 0x41,0x8e,0xc2,0x30,0x0c,0x45,0xf7,0x9c,0x22,0xca,0xbe,0x38,0x2d,0x2c,0x10,0x72, 0xdb,0xdd,0x9c,0x00,0x0e,0x50,0xb5,0x6e,0x89,0x68,0x12,0xd4,0x84,0x11,0x70,0x7a, 0x62,0x17,0x46,0x9a,0x28,0x92,0xbf,0x5f,0xec,0xef,0xc8,0xd8,0x3e,0xdc,0xac,0x7e, 0x69,0x89,0x36,0xf8,0x5a,0x97,0x5b,0xa3,0x15,0xf9,0x3e,0x0c,0xd6,0x4f,0xb5,0x3e, 0x9f,0x7e,0x8a,0x83,0x6e,0x9b,0x0d,0x3e,0xba,0xa5,0xd9,0x28,0x4c,0xa1,0xcf,0x41, 0x61,0xbf,0x50,0x97,0x72,0x47,0x91,0xac,0xa3,0xa6,0x32,0xe5,0xa1,0x28,0x4d,0x61, 0xaa,0x53,0x65,0x8e,0xf9,0xee,0x77,0x08,0xff,0x4b,0xa4,0xe9,0x42,0xfd,0x35,0xde, 0x9d,0x8a,0xe9,0x39,0x53,0xad,0xe3,0xa5,0x2b,0x35,0xbf,0x28,0x0c,0xe3,0x18,0x29, 0x35,0x06,0xe1,0xa3,0x84,0x46,0xfb,0x62,0x73,0x04,0x11,0x6c,0x01,0x5f,0x0f,0xc9, 0x46,0x3b,0x93,0xb2,0x43,0xfe,0xf6,0xc7,0xc6,0x77,0x79,0x14,0x53,0x04,0x91,0x02, 0xd3,0xf3,0xf6,0x85,0x22,0x05,0x5a,0x1f,0x06,0xe2,0x79,0xab,0x10,0xe6,0x04,0xe5, 0x83,0xe0,0xfe,0xe0,0xdd,0x0e,0x5c,0xc6,0x41,0xf2,0x69,0xcd,0xa7,0x35,0x47,0x60, 0x63,0xde,0x0c,0xc8,0x6a,0x10,0x64,0x51,0x6f,0x2a,0x6b,0x63,0x9a,0x01,0x79,0x57, 0x93,0xd4,0x55,0xd4,0x06,0x1c,0x76,0x99,0x10,0x31,0x87,0x52,0x2b,0x16,0xff,0x5b, 0x36,0x78,0x9c,0x55,0x90,0x41,0x8e,0xc2,0x30,0x0c,0x45,0xf7,0x9c,0x22,0xca,0xbe, 0x38,0x2d,0x2c,0x10,0x72 }; static void verify12(struct archive *a, struct archive_entry *ae) { (void)a; /* UNUSED */ assertEqualInt(archive_entry_mtime_is_set(ae), 0); assertEqualInt(archive_entry_atime_is_set(ae), 0); assertEqualInt(archive_entry_mtime(ae), 0); assertEqualInt(archive_entry_atime(ae), 0); } /* #How to make echo "onetwothree" > f1 echo "fourfivesize" > f2 xar -cf archive13.xar f1 f2 od -t x1 archive13.xar | sed -E -e 's/^0[0-9]+//' | sed -E -e 's/^ //' | sed -E -e 's/( )([0-9a-f]{2})/0x\2,/g;$ D' */ static unsigned char archive13[] = { 0x78,0x61,0x72,0x21,0x00,0x1c,0x00,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x02,0x31, 0x00,0x00,0x00,0x00,0x00,0x00,0x06,0xaa,0x00,0x00,0x00,0x01,0x78,0xda,0xcc,0x95, 0xcb,0x8e,0x9b,0x30,0x14,0x86,0xf7,0x79,0x0a,0xc4,0x9e,0xe2,0x0b,0xbe,0x10,0x39, 0xcc,0xae,0x4f,0x30,0xdd,0x74,0x67,0xec,0x43,0x82,0x86,0x4b,0x04,0x24,0xcd,0xcc, 0xd3,0xd7,0x36,0x30,0x69,0x34,0xd3,0x51,0xda,0x4a,0xa3,0x8a,0x05,0xbf,0x8f,0x7f, 0x9f,0xe3,0x63,0x3e,0x64,0xf5,0x70,0x69,0x9b,0xe8,0x0c,0xc3,0x58,0xf7,0xdd,0x2e, 0xc6,0x5f,0x50,0x1c,0x41,0x67,0x7a,0x5b,0x77,0xfb,0x5d,0xfc,0xed,0xf1,0x6b,0x22, 0xe3,0x87,0x62,0xa3,0x2e,0x7a,0x28,0x36,0x91,0x9a,0x7a,0xe3,0x5e,0x91,0x32,0x03, 0xe8,0xc9,0xad,0x48,0xa6,0xba,0x85,0x82,0x20,0x2c,0x13,0x44,0x12,0xc2,0x1f,0x71, 0xbe,0xc5,0x68,0x4b,0xf3,0xef,0x2a,0xbd,0xf5,0x84,0x55,0x07,0x30,0x4f,0xe3,0xa9, 0x8d,0xc6,0xe9,0xb9,0x81,0x5d,0x3c,0x1e,0x34,0x8e,0xfd,0x4c,0xa4,0xfa,0xaa,0x1a, 0x61,0x2a,0x90,0x4a,0x17,0x15,0xa2,0x63,0xfd,0xe2,0xb3,0xab,0x34,0x08,0x9f,0x22, 0x5d,0x73,0x84,0x51,0x55,0x37,0x10,0xd5,0x76,0x17,0x93,0x25,0x8d,0xd5,0x93,0x0e, 0x2a,0x52,0x0d,0x74,0xfb,0xe9,0x10,0x56,0x2f,0x72,0x8e,0x2f,0xf9,0xb3,0xdb,0x52, 0x4b,0x2d,0x4c,0xae,0xb5,0x5c,0x6c,0x3d,0x89,0x75,0xc3,0xfa,0x78,0x6c,0x6a,0x13, 0xba,0x4a,0x2f,0xc9,0xfe,0xa5,0x3e,0xc6,0xe9,0x62,0xd5,0x83,0x39,0xd4,0x67,0xb0, 0xc9,0xfb,0x4d,0x66,0x92,0x96,0x58,0x70,0x4e,0x09,0xa3,0x02,0x73,0x28,0x05,0xa9, 0x20,0xe7,0xb9,0x34,0x98,0x94,0x0c,0x33,0x8c,0xa4,0xd5,0x22,0xd3,0x2a,0x7d,0x93, 0x68,0xdd,0xcb,0x65,0x1a,0xb4,0x99,0x7e,0x5b,0x81,0xcb,0x9c,0xba,0x22,0x3a,0x93, 0x40,0xb8,0x70,0x0f,0xd7,0x39,0x42,0x34,0x77,0xe9,0x89,0x28,0x2d,0xe7,0x6e,0x20, 0xb4,0xc1,0x2a,0x7d,0x9b,0x69,0x3e,0xbc,0xf4,0xf5,0xf4,0x7c,0x31,0xf2,0x0b,0x15, 0x8c,0x0b,0x21,0x08,0x47,0xd9,0xd2,0xae,0xd2,0xa6,0x59,0xfb,0x36,0x06,0xc6,0xb1, 0x38,0x8d,0x30,0x6c,0xb7,0xc3,0x8f,0x64,0xb3,0x1f,0xfa,0xd3,0xd1,0xc9,0x24,0xd9, 0xf4,0xd3,0x21,0x44,0x9d,0x74,0x7d,0xcd,0xc6,0xb9,0xd2,0xba,0x5e,0x99,0x77,0x00, 0x42,0x72,0x8b,0x99,0x07,0x68,0x05,0x27,0x52,0xed,0x07,0xb6,0xf6,0x6a,0xd3,0x1f, 0xd8,0xf4,0xd5,0xb6,0xaf,0x6d,0x81,0x11,0x72,0x04,0x78,0x15,0x42,0x7e,0xff,0x85, 0xee,0x26,0x78,0x52,0x69,0xd0,0x73,0xf4,0xd5,0x78,0x5a,0x8d,0x6d,0x6f,0xa1,0x40, 0x3c,0xcb,0x5c,0x61,0x2f,0x67,0xec,0xe0,0x5c,0x1b,0xe8,0x7a,0x57,0x9a,0x51,0x77, 0x8e,0xeb,0x30,0x4c,0xd6,0x9d,0xf7,0xe5,0x88,0xe5,0xc2,0xc3,0x38,0x0f,0xc3,0xcc, 0xf4,0x7c,0x84,0xc2,0x33,0xac,0xd2,0x20,0x43,0xb0,0xd3,0x6e,0x9b,0x95,0xe3,0x30, 0x88,0xc0,0xbc,0xb7,0xdc,0xf2,0x8e,0xff,0x98,0x77,0xf2,0xc9,0xbc,0x3b,0x98,0x31, 0x94,0x3c,0x67,0x8e,0xfb,0x8c,0x49,0xc0,0xd2,0x20,0xcc,0x40,0x0b,0x69,0x08,0x35, 0xbc,0x92,0xc2,0x7d,0x25,0x66,0xff,0x9e,0x77,0x60,0x19,0xa6,0xd6,0xd2,0x4c,0xd3, 0x8a,0x72,0x6b,0x44,0x25,0x01,0xa8,0x75,0xbf,0x81,0x81,0x32,0xd3,0xbc,0xd2,0xa2, 0xc4,0xc4,0xfe,0x13,0xef,0xf4,0xd3,0x78,0x47,0xf7,0xf1,0x8e,0xee,0xe3,0x1d,0xfd, 0x27,0xbc,0xe3,0xfc,0x5e,0xde,0xf1,0x3b,0xbc,0x3b,0x97,0xbf,0x6e,0x54,0x1a,0x2e, 0x9f,0x9f,0x00,0x00,0x00,0xff,0xff,0x03,0x00,0xe5,0xbe,0xed,0xcc,0x2a,0x10,0x73, 0x9e,0x84,0x13,0x2e,0x27,0xe1,0x8d,0x6f,0xf8,0x9e,0xae,0xd0,0x0a,0x8f,0x2a,0xca, 0xf7,0x78,0xda,0xcb,0xcf,0x4b,0x2d,0x29,0xcf,0x2f,0xc9,0x28,0x4a,0x4d,0xe5,0x02, 0x00,0x21,0x4c,0x04,0xbf,0x78,0xda,0x4b,0xcb,0x2f,0x2d,0x4a,0xcb,0x2c,0x4b,0x2d, 0xce,0xac,0xe0,0x02,0x00,0x20,0xfa,0x04,0xc5 }; enum enc { GZIP, BZIP2 }; static void verify(unsigned char *d, size_t s, void (*f1)(struct archive *, struct archive_entry *), void (*f2)(struct archive *, struct archive_entry *), enum enc etype) { struct archive_entry *ae; struct archive *a; unsigned char *buff; int r; assert((a = archive_read_new()) != NULL); switch (etype) { case BZIP2: /* This is only check whether bzip is supported or not. * This filter won't be used this test. */ if (ARCHIVE_OK != archive_read_support_filter_bzip2(a)) { skipping("Unsupported bzip2"); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); return; } break; case GZIP: /* This gzip must be needed. archive_read_support_format_xar() * will return a warning if gzip is unsupported. */ break; } assertA(0 == archive_read_support_filter_all(a)); r = archive_read_support_format_xar(a); if (r == ARCHIVE_WARN) { skipping("xar reading not fully supported on this platform"); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); return; } assert((buff = malloc(100000)) != NULL); if (buff == NULL) return; memcpy(buff, d, s); memset(buff + s, 0, 2048); assertA(0 == archive_read_support_format_all(a)); assertA(0 == archive_read_open_memory(a, buff, s + 1024)); assertA(0 == archive_read_next_header(a, &ae)); assertEqualInt(archive_filter_code(a, 0), ARCHIVE_FILTER_NONE); assertEqualInt(archive_format(a), ARCHIVE_FORMAT_XAR); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED); /* Verify the only entry. */ f1(a, ae); if (f2) { assertA(0 == archive_read_next_header(a, &ae)); assertEqualInt(archive_filter_code(a, 0), ARCHIVE_FILTER_NONE); assertEqualInt(archive_format(a), ARCHIVE_FORMAT_XAR); /* Verify the only entry. */ f2(a, ae); assertEqualInt(2, archive_file_count(a)); } else { assertEqualInt(1, archive_file_count(a)); } /* End of archive. */ assertEqualInt(ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); free(buff); } static void verifyB(unsigned char *d, size_t s) { struct archive* a; struct archive_entry *entry = NULL; - la_int64_t buf_size; + size_t buf_size; unsigned char *buf; assert((a = archive_read_new()) != NULL); if(ARCHIVE_OK != archive_read_support_filter_gzip(a)) { skipping("Unsupported gzip"); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); return; } if(ARCHIVE_OK != archive_read_support_filter_bzip2(a)) { skipping("Unsupported bzip2"); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); return; } if(ARCHIVE_OK != archive_read_support_format_xar(a)) { skipping("Unsupported xar"); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); return; } assertA(0 == archive_read_open_memory(a, d, s)); // f1, content "onetwothree\n", size 12 bytes assertA(0 == archive_read_next_header(a, &entry)); - buf_size = archive_entry_size(entry); + buf_size = (size_t) archive_entry_size(entry); assertA(buf_size == 12); buf = (unsigned char*) malloc(buf_size); assertA(NULL != buf); - assertA(buf_size == archive_read_data(a, buf, buf_size)); + assertA(buf_size == (size_t) archive_read_data(a, buf, buf_size)); free(buf); // f2, content "fourfivesix\n", size 12 bytes assertA(0 == archive_read_next_header(a, &entry)); - buf_size = archive_entry_size(entry); + buf_size = (size_t) archive_entry_size(entry); assertA(buf_size == 12); buf = (unsigned char*) malloc(buf_size); assertA(NULL != buf); - assertA(buf_size == archive_read_data(a, buf, buf_size)); + assertA(buf_size == (size_t) archive_read_data(a, buf, buf_size)); free(buf); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_xar) { verify(archive1, sizeof(archive1), verify0, verify1, GZIP); verify(archive2, sizeof(archive2), verify0, verify2, GZIP); verify(archive3, sizeof(archive3), verify3, NULL, GZIP); verify(archive4, sizeof(archive4), verify4, NULL, GZIP); verify(archive5, sizeof(archive5), verify5, NULL, GZIP); verify(archive6, sizeof(archive6), verify6, NULL, GZIP); verify(archive7, sizeof(archive7), verify7, NULL, GZIP); verify(archive8, sizeof(archive8), verify0, NULL, BZIP2); verify(archive9, sizeof(archive9), verify0, NULL, GZIP); verify(archive10, sizeof(archive10), verify0, NULL, GZIP); verify(archive11, sizeof(archive11), verify0, NULL, GZIP); verify(archive12, sizeof(archive12), verify12, NULL, GZIP); verifyB(archive13, sizeof(archive13)); } Index: head/contrib/libarchive/libarchive/test/test_read_format_zip.c =================================================================== --- head/contrib/libarchive/libarchive/test/test_read_format_zip.c (revision 345496) +++ head/contrib/libarchive/libarchive/test/test_read_format_zip.c (revision 345497) @@ -1,761 +1,825 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * 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" __FBSDID("$FreeBSD$"); #define __LIBARCHIVE_BUILD #include 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 = archive_entry_size(ae); + 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 = crc32(0, buf, fsize); assertEqualInt(computed_crc, crc); ret = 0; fn_exit: free(buf); return ret; } static int extract_one_using_blocks(struct archive* a, int block_size, uint32_t crc) { uint8_t* buf; int ret = 1; uint32_t computed_crc = 0; la_ssize_t bytes_read; buf = malloc(block_size); if(buf == NULL) return 1; while(1) { bytes_read = archive_read_data(a, buf, block_size); if(bytes_read == ARCHIVE_RETRY) continue; else if(bytes_read == 0) break; else if(bytes_read < 0) { /* If we're here, it means the decompressor has failed * to properly decode test file. */ assertA(0); ret = 1; goto fn_exit; } else { /* ok */ } computed_crc = crc32(computed_crc, buf, bytes_read); } assertEqualInt(computed_crc, crc); ret = 0; fn_exit: free(buf); return ret; } /* * The reference file for this has been manually tweaked so that: * * file2 has length-at-end but file1 does not * * file2 has an invalid CRC */ static void verify_basic(struct archive *a, int seek_checks) { struct archive_entry *ae; char *buff[128]; const void *pv; size_t s; int64_t o; assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 1.0 (uncompressed)", archive_format_name(a)); assertEqualString("dir/", archive_entry_pathname(ae)); assertEqualInt(1179604249, archive_entry_mtime(ae)); assertEqualInt(0, archive_entry_size(ae)); if (seek_checks) assertEqualInt(AE_IFDIR | 0755, archive_entry_mode(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualIntA(a, ARCHIVE_EOF, archive_read_data_block(a, &pv, &s, &o)); assertEqualInt((int)s, 0); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 2.0 (deflation)", archive_format_name(a)); assertEqualString("file1", archive_entry_pathname(ae)); assertEqualInt(1179604289, archive_entry_mtime(ae)); if (seek_checks) assertEqualInt(AE_IFREG | 0755, archive_entry_mode(ae)); assertEqualInt(18, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); failure("archive_read_data() returns number of bytes read"); if (archive_zlib_version() != NULL) { assertEqualInt(18, archive_read_data(a, buff, 19)); assertEqualMem(buff, "hello\nhello\nhello\n", 18); } else { assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, 19)); assertEqualString(archive_error_string(a), "Unsupported ZIP compression method (deflation)"); assert(archive_errno(a) != 0); } assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 2.0 (deflation)", archive_format_name(a)); assertEqualString("file2", archive_entry_pathname(ae)); assertEqualInt(1179605932, archive_entry_mtime(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); if (seek_checks) { assertEqualInt(AE_IFREG | 0755, archive_entry_mode(ae)); } assert(archive_entry_size_is_set(ae)); assertEqualInt(18, archive_entry_size(ae)); if (archive_zlib_version() != NULL) { failure("file2 has a bad CRC, so read should fail and not change buff"); memset(buff, 'a', 19); assertEqualInt(ARCHIVE_WARN, archive_read_data(a, buff, 19)); assertEqualMem(buff, "aaaaaaaaaaaaaaaaaaa", 19); } else { assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, 19)); assertEqualString(archive_error_string(a), "Unsupported ZIP compression method (deflation)"); assert(archive_errno(a) != 0); } assertEqualInt(ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualString("ZIP 2.0 (deflation)", archive_format_name(a)); /* Verify the number of files read. */ failure("the archive file has three files"); assertEqualInt(3, archive_file_count(a)); assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_ZIP, archive_format(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } static void test_basic(void) { const char *refname = "test_read_format_zip.zip"; struct archive *a; char *p; size_t s; extract_reference_file(refname); /* Verify with seeking reader. */ 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_basic(a, 1); /* Verify with streaming reader. */ p = slurpfile(&s, 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, read_open_memory(a, p, s, 31)); verify_basic(a, 0); free(p); } /* * Read Info-ZIP New Unix Extra Field 0x7875 "ux". * Currently stores Unix UID/GID up to 32 bits. */ static void verify_info_zip_ux(struct archive *a, int seek_checks) { struct archive_entry *ae; char *buff[128]; assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("file1", archive_entry_pathname(ae)); assertEqualInt(1300668680, archive_entry_mtime(ae)); assertEqualInt(18, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); if (seek_checks) assertEqualInt(AE_IFREG | 0644, archive_entry_mode(ae)); failure("zip reader should read Info-ZIP New Unix Extra Field"); assertEqualInt(1001, archive_entry_uid(ae)); assertEqualInt(1001, archive_entry_gid(ae)); if (archive_zlib_version() != NULL) { failure("archive_read_data() returns number of bytes read"); assertEqualInt(18, archive_read_data(a, buff, 19)); assertEqualMem(buff, "hello\nhello\nhello\n", 18); } else { assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, 19)); assertEqualString(archive_error_string(a), "Unsupported ZIP compression method (deflation)"); assert(archive_errno(a) != 0); } assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify the number of files read. */ failure("the archive file has just one file"); assertEqualInt(1, archive_file_count(a)); assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_ZIP, archive_format(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } static void test_info_zip_ux(void) { const char *refname = "test_read_format_zip_ux.zip"; struct archive *a; char *p; size_t s; extract_reference_file(refname); /* Verify with seeking reader. */ 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_info_zip_ux(a, 1); /* Verify with streaming reader. */ p = slurpfile(&s, 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, read_open_memory(a, p, s, 108)); verify_info_zip_ux(a, 0); free(p); } /* * Verify that test_read_extract correctly works with * Zip entries that use length-at-end. */ static void verify_extract_length_at_end(struct archive *a, int seek_checks) { struct archive_entry *ae; assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualString("hello.txt", archive_entry_pathname(ae)); if (seek_checks) { assertEqualInt(AE_IFREG | 0644, archive_entry_mode(ae)); assert(archive_entry_size_is_set(ae)); assertEqualInt(6, archive_entry_size(ae)); } else { assert(!archive_entry_size_is_set(ae)); assertEqualInt(0, archive_entry_size(ae)); } if (archive_zlib_version() != NULL) { assertEqualIntA(a, ARCHIVE_OK, archive_read_extract(a, ae, 0)); assertFileContents("hello\x0A", 6, "hello.txt"); } else { assertEqualIntA(a, ARCHIVE_FAILED, archive_read_extract(a, ae, 0)); assertEqualString(archive_error_string(a), "Unsupported ZIP compression method (deflation)"); assert(archive_errno(a) != 0); } assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } static void test_extract_length_at_end(void) { const char *refname = "test_read_format_zip_length_at_end.zip"; char *p; size_t s; struct archive *a; extract_reference_file(refname); /* Verify extraction with seeking reader. */ 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_extract_length_at_end(a, 1); /* Verify extraction with streaming reader. */ p = slurpfile(&s, 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, read_open_memory(a, p, s, 108)); verify_extract_length_at_end(a, 0); free(p); } static void test_symlink(void) { const char *refname = "test_read_format_zip_symlink.zip"; char *p; size_t s; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); p = slurpfile(&s, refname); /* Symlinks can only be extracted with the seeking reader. */ assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, read_open_memory_seek(a, p, s, 1)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("file", archive_entry_pathname(ae)); assertEqualInt(AE_IFREG, archive_entry_filetype(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("symlink", archive_entry_pathname(ae)); assertEqualInt(AE_IFLNK, archive_entry_filetype(ae)); assertEqualInt(0, archive_entry_size(ae)); assertEqualString("file", archive_entry_symlink(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), 0); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); free(p); } DEFINE_TEST(test_read_format_zip) { test_basic(); test_info_zip_ux(); test_extract_length_at_end(); test_symlink(); } DEFINE_TEST(test_read_format_zip_ppmd_one_file) { const char *refname = "test_read_format_zip_ppmd8.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (ppmd-1)", archive_format_name(a)); assertEqualString("vimrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0xBA8E3BAA)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_ppmd_one_file_blockread) { const char *refname = "test_read_format_zip_ppmd8.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (ppmd-1)", archive_format_name(a)); assertEqualString("vimrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 13, 0xBA8E3BAA)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_ppmd_multi) { const char *refname = "test_read_format_zip_ppmd8_multi.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (ppmd-1)", archive_format_name(a)); assertEqualString("smartd.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0x8DD7379E)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (ppmd-1)", archive_format_name(a)); assertEqualString("ts.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0x7AE59B31)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (ppmd-1)", archive_format_name(a)); assertEqualString("vimrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0xBA8E3BAA)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_ppmd_multi_blockread) { const char *refname = "test_read_format_zip_ppmd8_multi.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (ppmd-1)", archive_format_name(a)); assertEqualString("smartd.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 12, 0x8DD7379E)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (ppmd-1)", archive_format_name(a)); assertEqualString("ts.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 13, 0x7AE59B31)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (ppmd-1)", archive_format_name(a)); assertEqualString("vimrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 14, 0xBA8E3BAA)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_lzma_one_file) { const char *refname = "test_read_format_zip_lzma.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (lzma)", archive_format_name(a)); assertEqualString("vimrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0xBA8E3BAA)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_lzma_one_file_blockread) { const char *refname = "test_read_format_zip_lzma.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (lzma)", archive_format_name(a)); assertEqualString("vimrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 13, 0xBA8E3BAA)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_lzma_multi) { const char *refname = "test_read_format_zip_lzma_multi.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (lzma)", archive_format_name(a)); assertEqualString("smartd.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0x8DD7379E)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (lzma)", archive_format_name(a)); assertEqualString("ts.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0x7AE59B31)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (lzma)", archive_format_name(a)); assertEqualString("vimrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0xBA8E3BAA)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_lzma_multi_blockread) { const char *refname = "test_read_format_zip_lzma_multi.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (lzma)", archive_format_name(a)); assertEqualString("smartd.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 12, 0x8DD7379E)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (lzma)", archive_format_name(a)); assertEqualString("ts.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 13, 0x7AE59B31)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 6.3 (lzma)", archive_format_name(a)); assertEqualString("vimrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 14, 0xBA8E3BAA)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_bzip2_one_file) { const char *refname = "test_read_format_zip_bzip2.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 4.6 (bzip)", archive_format_name(a)); assertEqualString("vimrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0xBA8E3BAA)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_bzip2_one_file_blockread) { const char *refname = "test_read_format_zip_bzip2.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 4.6 (bzip)", archive_format_name(a)); assertEqualString("vimrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 13, 0xBA8E3BAA)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_bzip2_multi) { const char *refname = "test_read_format_zip_bzip2_multi.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 4.6 (bzip)", archive_format_name(a)); assertEqualString("smartd.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0x8DD7379E)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 4.6 (bzip)", archive_format_name(a)); assertEqualString("ts.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0x7AE59B31)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 4.6 (bzip)", archive_format_name(a)); assertEqualString("vimrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0xBA8E3BAA)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_bzip2_multi_blockread) { const char *refname = "test_read_format_zip_bzip2_multi.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 4.6 (bzip)", archive_format_name(a)); assertEqualString("smartd.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 12, 0x8DD7379E)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 4.6 (bzip)", archive_format_name(a)); assertEqualString("ts.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 13, 0x7AE59B31)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 4.6 (bzip)", archive_format_name(a)); assertEqualString("vimrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 14, 0xBA8E3BAA)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_xz_multi) { const char *refname = "test_read_format_zip_xz_multi.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 2.0 (xz)", archive_format_name(a)); assertEqualString("bash.bashrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0xF751B8C9)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 2.0 (xz)", archive_format_name(a)); assertEqualString("pacman.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0xB20B7F88)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 2.0 (xz)", archive_format_name(a)); assertEqualString("profile", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one(a, ae, 0x2329F054)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_xz_multi_blockread) { const char *refname = "test_read_format_zip_xz_multi.zipx"; struct archive *a; struct archive_entry *ae; extract_reference_file(refname); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 2.0 (xz)", archive_format_name(a)); assertEqualString("bash.bashrc", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 12, 0xF751B8C9)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 2.0 (xz)", archive_format_name(a)); assertEqualString("pacman.conf", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 13, 0xB20B7F88)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualString("ZIP 2.0 (xz)", archive_format_name(a)); assertEqualString("profile", archive_entry_pathname(ae)); assertEqualIntA(a, 0, extract_one_using_blocks(a, 14, 0x2329F054)); assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); +} + +DEFINE_TEST(test_read_format_zip_ppmd8_crash_1) +{ + const char *refname = "test_read_format_zip_ppmd8_crash_2.zipx"; + struct archive *a; + struct archive_entry *ae; + char buf[64]; + + extract_reference_file(refname); + + assert((a = archive_read_new()) != NULL); + assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 100)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); + + /* This file shouldn't be properly decompressed, because it's invalid. + * However, unpacker should return an error during unpacking. Without the + * proper fix, the unpacker was entering an unlimited loop. */ + assertEqualIntA(a, ARCHIVE_FATAL, archive_read_data(a, buf, 1)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); +} + +DEFINE_TEST(test_read_format_zip_bz2_hang_on_invalid) +{ + const char *refname = "test_read_format_zip_bz2_hang.zip"; + struct archive *a; + struct archive_entry *ae; + char buf[8]; + + extract_reference_file(refname); + + assert((a = archive_read_new()) != NULL); + assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); + + /* The file `refname` is invalid in this case, so this call should fail. + * But it shouldn't crash. */ + assertEqualIntA(a, ARCHIVE_FATAL, archive_read_data(a, buf, 64)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); +} + +DEFINE_TEST(test_read_format_zip_ppmd8_crash_2) +{ + const char *refname = "test_read_format_zip_ppmd8_crash_2.zipx"; + struct archive *a; + struct archive_entry *ae; + char buf[64]; + + extract_reference_file(refname); + + assert((a = archive_read_new()) != NULL); + assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_zip(a)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 37)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); + + /* The file `refname` is invalid in this case, so this call should fail. + * But it shouldn't crash. */ + assertEqualIntA(a, ARCHIVE_FATAL, archive_read_data(a, buf, 64)); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_free(a)); } Index: head/contrib/libarchive/libarchive/test/test_read_format_zip_bz2_hang.zip.uu =================================================================== --- head/contrib/libarchive/libarchive/test/test_read_format_zip_bz2_hang.zip.uu (nonexistent) +++ head/contrib/libarchive/libarchive/test/test_read_format_zip_bz2_hang.zip.uu (revision 345497) @@ -0,0 +1,5 @@ +begin 644 test_read_format_zip_bz2_hang.zip +M4$L#!)LP,#`,,#`P,#`P,#`P,#`P,#`P,#`$`!P`,#`P,#`P"0`P,#`P,#`P +1,#!U>`L`8(0P,#`P,#`P,#`` +` +end Index: head/contrib/libarchive/libarchive/test/test_read_format_zip_ppmd8_crash_1.zipx.uu =================================================================== --- head/contrib/libarchive/libarchive/test/test_read_format_zip_ppmd8_crash_1.zipx.uu (nonexistent) +++ head/contrib/libarchive/libarchive/test/test_read_format_zip_ppmd8_crash_1.zipx.uu (revision 345497) @@ -0,0 +1,4 @@ +begin 644 test_read_format_zip_ppmd8_crash_1.zipx +K4$L'"(=02P,$\+N.O&*A>*\+."U``$H`````````@``````#````6(0````` +` +end Index: head/contrib/libarchive/libarchive/test/test_read_format_zip_ppmd8_crash_2.zipx.uu =================================================================== --- head/contrib/libarchive/libarchive/test/test_read_format_zip_ppmd8_crash_2.zipx.uu (nonexistent) +++ head/contrib/libarchive/libarchive/test/test_read_format_zip_ppmd8_crash_2.zipx.uu (revision 345497) @@ -0,0 +1,4 @@ +begin 644 test_read_format_zip_ppmd8_crash_2.zipx +L4$L'"(=02P,$\+N.O&*A>*\+.2U`@$H`````````@``````#```````````` +` +end Index: head/contrib/libarchive =================================================================== --- head/contrib/libarchive (revision 345496) +++ head/contrib/libarchive (revision 345497) Property changes on: head/contrib/libarchive ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /vendor/libarchive/dist:r345495 Index: head/lib/libarchive/tests/Makefile =================================================================== --- head/lib/libarchive/tests/Makefile (revision 345496) +++ head/lib/libarchive/tests/Makefile (revision 345497) @@ -1,600 +1,603 @@ # $FreeBSD$ PACKAGE= tests _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+= -DHAVE_LIBLZMA=1 -DHAVE_LZMA_H=1 # 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_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_archive_api_feature.c \ test_archive_clear_error.c \ test_archive_cmdline.c \ test_archive_digest.c \ test_archive_getdate.c \ test_archive_match_time.c \ test_archive_match_owner.c \ test_archive_match_path.c \ test_archive_pathmatch.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_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_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_fuzz.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_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_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_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_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_sparse.c \ test_read_format_gtar_sparse_skip_entry.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_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_data.c \ test_read_format_rar_encryption_header.c \ test_read_format_rar_encryption_partially.c \ test_read_format_rar_invalid1.c \ test_read_format_raw.c \ test_read_format_tar.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_filename.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_zip.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_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_schily_xattr.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_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_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_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_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_empty.c \ test_write_format_zip_empty_zip64.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_open_memory.c \ test_write_read_format_zip.c \ test_xattr_platform.c \ test_zip_filename_encoding.c # Deterministic failures: # Crashes with SIGBUS BROKEN_TESTS+= test_archive_rmd160 # Fails with `libarchive/test/test_archive_crypto.c:121: md != actualmd` BROKEN_TESTS+= test_archive_sha384 # Fails with `test_read_disk_directory_traversals.c:1094: File at has atime 886622, 1443306049 seconds ago` BROKEN_TESTS+= test_read_disk_directory_traversals # Non-deterministic failures: # (Times out?) [and] crashes BROKEN_TESTS+= test_fuzz_rar # 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_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_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_fuzz.cab.uu ${PACKAGE}FILES+= test_fuzz.lzh.uu ${PACKAGE}FILES+= test_fuzz_1.iso.Z.uu ${PACKAGE}FILES+= test_pax_filename_encoding.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_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_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_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_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_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_malformed.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_malformed2.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_ppmd.7z.uu ${PACKAGE}FILES+= test_read_format_7zip_symbolic_name.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_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_skip_entry.tar.Z.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_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_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_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_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_noeof.rar.uu ${PACKAGE}FILES+= test_read_format_rar_ppmd_lzss_conversion.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_unicode.rar.uu ${PACKAGE}FILES+= test_read_format_rar_windows.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_arm.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_blake2.rar.uu ${PACKAGE}FILES+= test_read_format_rar5_compressed.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_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_win32.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.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_pax.tar.Z.uu ${PACKAGE}FILES+= test_read_format_tar_filename_koi8r.tar.Z.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 ${PACKAGE}FILES+= test_read_format_zip.zip.uu +${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_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.zipx.uu ${PACKAGE}FILES+= test_read_format_zip_lzma_multi.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_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_schily_xattr.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_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