Index: head/contrib/libarchive/cat/test/main.c =================================================================== --- head/contrib/libarchive/cat/test/main.c (revision 309589) +++ head/contrib/libarchive/cat/test/main.c (revision 309590) @@ -1,3072 +1,3072 @@ /* * Copyright (c) 2003-2009 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" #include "test_utils.h" #ifdef HAVE_SYS_IOCTL_H #include #endif #ifdef HAVE_SYS_TIME_H #include #endif #include #ifdef HAVE_ICONV_H #include #endif /* * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. * As the include guards don't agree, the order of include is important. */ #ifdef HAVE_LINUX_EXT2_FS_H #include /* for Linux file flags */ #endif #if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) #include /* Linux file flags, broken on Cygwin */ #endif #include #include #ifdef HAVE_SIGNAL_H #include #endif #include #include /* * This same file is used pretty much verbatim for all test harnesses. * * The next few lines are the only differences. * TODO: Move this into a separate configuration header, have all test * suites share one copy of this file. */ #define KNOWNREF "test_expand.Z.uu" #define ENVBASE "BSDCAT" /* Prefix for environment variables. */ #define PROGRAM "bsdcat" /* Name of program being tested. */ #define PROGRAM_ALIAS "cat" /* Generic alias for program */ #undef LIBRARY /* Not testing a library. */ #undef EXTRA_DUMP /* How to dump extra data */ #undef EXTRA_ERRNO /* How to dump errno */ /* How to generate extra version info. */ #define EXTRA_VERSION (systemf("%s --version", testprog) ? "" : "") /* * * Windows support routines * * Note: Configuration is a tricky issue. Using HAVE_* feature macros * in the test harness is dangerous because they cover up * configuration errors. The classic example of this is omitting a * configure check. If libarchive and libarchive_test both look for * the same feature macro, such errors are hard to detect. Platform * macros (e.g., _WIN32 or __GNUC__) are a little better, but can * easily lead to very messy code. It's best to limit yourself * to only the most generic programming techniques in the test harness * and thus avoid conditionals altogether. Where that's not possible, * try to minimize conditionals by grouping platform-specific tests in * one place (e.g., test_acl_freebsd) or by adding new assert() * functions (e.g., assertMakeHardlink()) to cover up platform * differences. Platform-specific coding in libarchive_test is often * a symptom that some capability is missing from libarchive itself. */ #if defined(_WIN32) && !defined(__CYGWIN__) #include #include #include #ifndef F_OK #define F_OK (0) #endif #ifndef S_ISDIR #define S_ISDIR(m) ((m) & _S_IFDIR) #endif #ifndef S_ISREG #define S_ISREG(m) ((m) & _S_IFREG) #endif #if !defined(__BORLANDC__) #define access _access #undef chdir #define chdir _chdir #endif #ifndef fileno #define fileno _fileno #endif /*#define fstat _fstat64*/ #if !defined(__BORLANDC__) #define getcwd _getcwd #endif #define lstat stat /*#define lstat _stat64*/ /*#define stat _stat64*/ #define rmdir _rmdir #if !defined(__BORLANDC__) #define strdup _strdup #define umask _umask #endif #define int64_t __int64 #endif #if defined(HAVE__CrtSetReportMode) # include #endif mode_t umasked(mode_t expected_mode) { mode_t mode = umask(0); umask(mode); return expected_mode & ~mode; } /* Path to working directory for current test */ const char *testworkdir; #ifdef PROGRAM /* Pathname of exe to be tested. */ const char *testprogfile; /* Name of exe to use in printf-formatted command strings. */ /* On Windows, this includes leading/trailing quotes. */ const char *testprog; #endif #if defined(_WIN32) && !defined(__CYGWIN__) static void *GetFunctionKernel32(const char *); static int my_CreateSymbolicLinkA(const char *, const char *, int); static int my_CreateHardLinkA(const char *, const char *); static int my_GetFileInformationByName(const char *, BY_HANDLE_FILE_INFORMATION *); static void * GetFunctionKernel32(const char *name) { static HINSTANCE lib; static int set; if (!set) { set = 1; lib = LoadLibrary("kernel32.dll"); } if (lib == NULL) { fprintf(stderr, "Can't load kernel32.dll?!\n"); exit(1); } return (void *)GetProcAddress(lib, name); } static int my_CreateSymbolicLinkA(const char *linkname, const char *target, int flags) { static BOOLEAN (WINAPI *f)(LPCSTR, LPCSTR, DWORD); static int set; if (!set) { set = 1; f = GetFunctionKernel32("CreateSymbolicLinkA"); } return f == NULL ? 0 : (*f)(linkname, target, flags); } static int my_CreateHardLinkA(const char *linkname, const char *target) { static BOOLEAN (WINAPI *f)(LPCSTR, LPCSTR, LPSECURITY_ATTRIBUTES); static int set; if (!set) { set = 1; f = GetFunctionKernel32("CreateHardLinkA"); } return f == NULL ? 0 : (*f)(linkname, target, NULL); } static int my_GetFileInformationByName(const char *path, BY_HANDLE_FILE_INFORMATION *bhfi) { HANDLE h; int r; memset(bhfi, 0, sizeof(*bhfi)); h = CreateFile(path, FILE_READ_ATTRIBUTES, 0, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); if (h == INVALID_HANDLE_VALUE) return (0); r = GetFileInformationByHandle(h, bhfi); CloseHandle(h); return (r); } #endif #if defined(HAVE__CrtSetReportMode) && !defined(__WATCOMC__) static void invalid_parameter_handler(const wchar_t * expression, const wchar_t * function, const wchar_t * file, unsigned int line, uintptr_t pReserved) { /* nop */ } #endif /* * * OPTIONS FLAGS * */ /* Enable core dump on failure. */ static int dump_on_failure = 0; /* Default is to remove temp dirs and log data for successful tests. */ static int keep_temp_files = 0; /* Default is to run the specified tests once and report errors. */ static int until_failure = 0; /* Default is to just report pass/fail for each test. */ static int verbosity = 0; #define VERBOSITY_SUMMARY_ONLY -1 /* -q */ #define VERBOSITY_PASSFAIL 0 /* Default */ #define VERBOSITY_LIGHT_REPORT 1 /* -v */ #define VERBOSITY_FULL 2 /* -vv */ /* A few places generate even more output for verbosity > VERBOSITY_FULL, * mostly for debugging the test harness itself. */ /* Cumulative count of assertion failures. */ static int failures = 0; /* Cumulative count of reported skips. */ static int skips = 0; /* Cumulative count of assertions checked. */ static int assertions = 0; /* Directory where uuencoded reference files can be found. */ static const char *refdir; /* * Report log information selectively to console and/or disk log. */ static int log_console = 0; static FILE *logfile; static void vlogprintf(const char *fmt, va_list ap) { #ifdef va_copy va_list lfap; va_copy(lfap, ap); #endif if (log_console) vfprintf(stdout, fmt, ap); if (logfile != NULL) #ifdef va_copy vfprintf(logfile, fmt, lfap); va_end(lfap); #else vfprintf(logfile, fmt, ap); #endif } static void logprintf(const char *fmt, ...) { va_list ap; va_start(ap, fmt); vlogprintf(fmt, ap); va_end(ap); } /* Set up a message to display only if next assertion fails. */ static char msgbuff[4096]; static const char *msg, *nextmsg; void failure(const char *fmt, ...) { va_list ap; if (fmt == NULL) { nextmsg = NULL; } else { va_start(ap, fmt); vsprintf(msgbuff, fmt, ap); va_end(ap); nextmsg = msgbuff; } } /* * Copy arguments into file-local variables. * This was added to permit vararg assert() functions without needing * variadic wrapper macros. Turns out that the vararg capability is almost * never used, so almost all of the vararg assertions can be simplified * by removing the vararg capability and reworking the wrapper macro to * pass __FILE__, __LINE__ directly into the function instead of using * this hook. I suspect this machinery is used so rarely that we * would be better off just removing it entirely. That would simplify * the code here noticeably. */ static const char *skipping_filename; static int skipping_line; void skipping_setup(const char *filename, int line) { skipping_filename = filename; skipping_line = line; } /* Called at the beginning of each assert() function. */ static void assertion_count(const char *file, int line) { (void)file; /* UNUSED */ (void)line; /* UNUSED */ ++assertions; /* Proper handling of "failure()" message. */ msg = nextmsg; nextmsg = NULL; /* Uncomment to print file:line after every assertion. * Verbose, but occasionally useful in tracking down crashes. */ /* printf("Checked %s:%d\n", file, line); */ } /* * For each test source file, we remember how many times each * assertion was reported. Cleared before each new test, * used by test_summarize(). */ static struct line { int count; int skip; } failed_lines[10000]; const char *failed_filename; /* Count this failure, setup up log destination and handle initial report. */ static void failure_start(const char *filename, int line, const char *fmt, ...) { va_list ap; /* Record another failure for this line. */ ++failures; failed_filename = filename; failed_lines[line].count++; /* Determine whether to log header to console. */ switch (verbosity) { case VERBOSITY_LIGHT_REPORT: log_console = (failed_lines[line].count < 2); break; default: log_console = (verbosity >= VERBOSITY_FULL); } /* Log file:line header for this failure */ va_start(ap, fmt); #if _MSC_VER logprintf("%s(%d): ", filename, line); #else logprintf("%s:%d: ", filename, line); #endif vlogprintf(fmt, ap); va_end(ap); logprintf("\n"); if (msg != NULL && msg[0] != '\0') { logprintf(" Description: %s\n", msg); msg = NULL; } /* Determine whether to log details to console. */ if (verbosity == VERBOSITY_LIGHT_REPORT) log_console = 0; } /* Complete reporting of failed tests. */ /* * The 'extra' hook here is used by libarchive to include libarchive * error messages with assertion failures. It could also be used * to add strerror() output, for example. Just define the EXTRA_DUMP() * macro appropriately. */ static void failure_finish(void *extra) { (void)extra; /* UNUSED (maybe) */ #ifdef EXTRA_DUMP if (extra != NULL) { logprintf(" errno: %d\n", EXTRA_ERRNO(extra)); logprintf(" detail: %s\n", EXTRA_DUMP(extra)); } #endif if (dump_on_failure) { fprintf(stderr, " *** forcing core dump so failure can be debugged ***\n"); abort(); } } /* Inform user that we're skipping some checks. */ void test_skipping(const char *fmt, ...) { char buff[1024]; va_list ap; va_start(ap, fmt); vsprintf(buff, fmt, ap); va_end(ap); /* Use failure() message if set. */ msg = nextmsg; nextmsg = NULL; /* failure_start() isn't quite right, but is awfully convenient. */ failure_start(skipping_filename, skipping_line, "SKIPPING: %s", buff); --failures; /* Undo failures++ in failure_start() */ /* Don't failure_finish() here. */ /* Mark as skip, so doesn't count as failed test. */ failed_lines[skipping_line].skip = 1; ++skips; } /* * * ASSERTIONS * */ /* Generic assert() just displays the failed condition. */ int assertion_assert(const char *file, int line, int value, const char *condition, void *extra) { assertion_count(file, line); if (!value) { failure_start(file, line, "Assertion failed: %s", condition); failure_finish(extra); } return (value); } /* chdir() and report any errors */ int assertion_chdir(const char *file, int line, const char *pathname) { assertion_count(file, line); if (chdir(pathname) == 0) return (1); failure_start(file, line, "chdir(\"%s\")", pathname); failure_finish(NULL); return (0); } /* Verify two integers are equal. */ int assertion_equal_int(const char *file, int line, long long v1, const char *e1, long long v2, const char *e2, void *extra) { assertion_count(file, line); if (v1 == v2) return (1); failure_start(file, line, "%s != %s", e1, e2); logprintf(" %s=%lld (0x%llx, 0%llo)\n", e1, v1, v1, v1); logprintf(" %s=%lld (0x%llx, 0%llo)\n", e2, v2, v2, v2); failure_finish(extra); return (0); } /* * Utility to convert a single UTF-8 sequence. */ static int _utf8_to_unicode(uint32_t *pwc, const char *s, size_t n) { static const char utf8_count[256] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 00 - 0F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 10 - 1F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 20 - 2F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 30 - 3F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 40 - 4F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 50 - 5F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 60 - 6F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 70 - 7F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 80 - 8F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 90 - 9F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* A0 - AF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* B0 - BF */ 0, 0, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* C0 - CF */ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* D0 - DF */ 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,/* E0 - EF */ 4, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 /* F0 - FF */ }; int ch; int cnt; uint32_t wc; *pwc = 0; /* Sanity check. */ if (n == 0) return (0); /* * Decode 1-4 bytes depending on the value of the first byte. */ ch = (unsigned char)*s; if (ch == 0) return (0); /* Standard: return 0 for end-of-string. */ cnt = utf8_count[ch]; - /* Invalide sequence or there are not plenty bytes. */ + /* Invalid sequence or there are not plenty bytes. */ if (n < (size_t)cnt) return (-1); /* Make a Unicode code point from a single UTF-8 sequence. */ switch (cnt) { case 1: /* 1 byte sequence. */ *pwc = ch & 0x7f; return (cnt); case 2: /* 2 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) return (-1); *pwc = ((ch & 0x1f) << 6) | (s[1] & 0x3f); return (cnt); case 3: /* 3 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) return (-1); if ((s[2] & 0xc0) != 0x80) return (-1); wc = ((ch & 0x0f) << 12) | ((s[1] & 0x3f) << 6) | (s[2] & 0x3f); if (wc < 0x800) return (-1);/* Overlong sequence. */ break; case 4: /* 4 bytes sequence. */ if (n < 4) return (-1); if ((s[1] & 0xc0) != 0x80) return (-1); if ((s[2] & 0xc0) != 0x80) return (-1); if ((s[3] & 0xc0) != 0x80) return (-1); wc = ((ch & 0x07) << 18) | ((s[1] & 0x3f) << 12) | ((s[2] & 0x3f) << 6) | (s[3] & 0x3f); if (wc < 0x10000) return (-1);/* Overlong sequence. */ break; default: return (-1); } - /* The code point larger than 0x10FFFF is not leagal + /* The code point larger than 0x10FFFF is not legal * Unicode values. */ if (wc > 0x10FFFF) return (-1); /* Correctly gets a Unicode, returns used bytes. */ *pwc = wc; return (cnt); } static void strdump(const char *e, const char *p, int ewidth, int utf8) { const char *q = p; logprintf(" %*s = ", ewidth, e); if (p == NULL) { logprintf("NULL\n"); return; } logprintf("\""); while (*p != '\0') { unsigned int c = 0xff & *p++; switch (c) { case '\a': logprintf("\\a"); break; case '\b': logprintf("\\b"); break; case '\n': logprintf("\\n"); break; case '\r': logprintf("\\r"); break; default: if (c >= 32 && c < 127) logprintf("%c", c); else logprintf("\\x%02X", c); } } logprintf("\""); logprintf(" (length %d)", q == NULL ? -1 : (int)strlen(q)); /* * If the current string is UTF-8, dump its code points. */ if (utf8) { size_t len; uint32_t uc; int n; int cnt = 0; p = q; len = strlen(p); logprintf(" ["); while ((n = _utf8_to_unicode(&uc, p, len)) > 0) { if (p != q) logprintf(" "); logprintf("%04X", uc); p += n; len -= n; cnt++; } logprintf("]"); logprintf(" (count %d", cnt); if (n < 0) { logprintf(",unknown %d bytes", len); } logprintf(")"); } logprintf("\n"); } /* Verify two strings are equal, dump them if not. */ int assertion_equal_string(const char *file, int line, const char *v1, const char *e1, const char *v2, const char *e2, void *extra, int utf8) { int l1, l2; assertion_count(file, line); if (v1 == v2 || (v1 != NULL && v2 != NULL && strcmp(v1, v2) == 0)) return (1); failure_start(file, line, "%s != %s", e1, e2); l1 = (int)strlen(e1); l2 = (int)strlen(e2); if (l1 < l2) l1 = l2; strdump(e1, v1, l1, utf8); strdump(e2, v2, l1, utf8); failure_finish(extra); return (0); } static void wcsdump(const char *e, const wchar_t *w) { logprintf(" %s = ", e); if (w == NULL) { logprintf("(null)"); return; } logprintf("\""); while (*w != L'\0') { unsigned int c = *w++; if (c >= 32 && c < 127) logprintf("%c", c); else if (c < 256) logprintf("\\x%02X", c); else if (c < 0x10000) logprintf("\\u%04X", c); else logprintf("\\U%08X", c); } logprintf("\"\n"); } #ifndef HAVE_WCSCMP static int wcscmp(const wchar_t *s1, const wchar_t *s2) { while (*s1 == *s2++) { if (*s1++ == L'\0') return 0; } if (*s1 > *--s2) return 1; else return -1; } #endif /* Verify that two wide strings are equal, dump them if not. */ int assertion_equal_wstring(const char *file, int line, const wchar_t *v1, const char *e1, const wchar_t *v2, const char *e2, void *extra) { assertion_count(file, line); if (v1 == v2) return (1); if (v1 != NULL && v2 != NULL && wcscmp(v1, v2) == 0) return (1); failure_start(file, line, "%s != %s", e1, e2); wcsdump(e1, v1); wcsdump(e2, v2); failure_finish(extra); return (0); } /* * Pretty standard hexdump routine. As a bonus, if ref != NULL, then * any bytes in p that differ from ref will be highlighted with '_' * before and after the hex value. */ static void hexdump(const char *p, const char *ref, size_t l, size_t offset) { size_t i, j; char sep; if (p == NULL) { logprintf("(null)\n"); return; } for(i=0; i < l; i+=16) { logprintf("%04x", (unsigned)(i + offset)); sep = ' '; for (j = 0; j < 16 && i + j < l; j++) { if (ref != NULL && p[i + j] != ref[i + j]) sep = '_'; logprintf("%c%02x", sep, 0xff & (int)p[i+j]); if (ref != NULL && p[i + j] == ref[i + j]) sep = ' '; } for (; j < 16; j++) { logprintf("%c ", sep); sep = ' '; } logprintf("%c", sep); for (j=0; j < 16 && i + j < l; j++) { int c = p[i + j]; if (c >= ' ' && c <= 126) logprintf("%c", c); else logprintf("."); } logprintf("\n"); } } /* Verify that two blocks of memory are the same, display the first * block of differences if they're not. */ int assertion_equal_mem(const char *file, int line, const void *_v1, const char *e1, const void *_v2, const char *e2, size_t l, const char *ld, void *extra) { const char *v1 = (const char *)_v1; const char *v2 = (const char *)_v2; size_t offset; assertion_count(file, line); if (v1 == v2 || (v1 != NULL && v2 != NULL && memcmp(v1, v2, l) == 0)) return (1); if (v1 == NULL || v2 == NULL) return (0); failure_start(file, line, "%s != %s", e1, e2); logprintf(" size %s = %d\n", ld, (int)l); /* Dump 48 bytes (3 lines) so that the first difference is * in the second line. */ offset = 0; while (l > 64 && memcmp(v1, v2, 32) == 0) { /* Two lines agree, so step forward one line. */ v1 += 16; v2 += 16; l -= 16; offset += 16; } logprintf(" Dump of %s\n", e1); hexdump(v1, v2, l < 128 ? l : 128, offset); logprintf(" Dump of %s\n", e2); hexdump(v2, v1, l < 128 ? l : 128, offset); logprintf("\n"); failure_finish(extra); return (0); } /* Verify that a block of memory is filled with the specified byte. */ int assertion_memory_filled_with(const char *file, int line, const void *_v1, const char *vd, size_t l, const char *ld, char b, const char *bd, void *extra) { const char *v1 = (const char *)_v1; size_t c = 0; size_t i; (void)ld; /* UNUSED */ assertion_count(file, line); for (i = 0; i < l; ++i) { if (v1[i] == b) { ++c; } } if (c == l) return (1); failure_start(file, line, "%s (size %d) not filled with %s", vd, (int)l, bd); logprintf(" Only %d bytes were correct\n", (int)c); failure_finish(extra); return (0); } /* Verify that the named file exists and is empty. */ int assertion_empty_file(const char *filename, int line, const char *f1) { char buff[1024]; struct stat st; ssize_t s; FILE *f; assertion_count(filename, line); if (stat(f1, &st) != 0) { failure_start(filename, line, "Stat failed: %s", f1); failure_finish(NULL); return (0); } if (st.st_size == 0) return (1); failure_start(filename, line, "File should be empty: %s", f1); logprintf(" File size: %d\n", (int)st.st_size); logprintf(" Contents:\n"); f = fopen(f1, "rb"); if (f == NULL) { logprintf(" Unable to open %s\n", f1); } else { s = ((off_t)sizeof(buff) < st.st_size) ? (ssize_t)sizeof(buff) : (ssize_t)st.st_size; s = fread(buff, 1, s, f); hexdump(buff, NULL, s, 0); fclose(f); } failure_finish(NULL); return (0); } /* Verify that the named file exists and is not empty. */ int assertion_non_empty_file(const char *filename, int line, const char *f1) { struct stat st; assertion_count(filename, line); if (stat(f1, &st) != 0) { failure_start(filename, line, "Stat failed: %s", f1); failure_finish(NULL); return (0); } if (st.st_size == 0) { failure_start(filename, line, "File empty: %s", f1); failure_finish(NULL); return (0); } return (1); } /* Verify that two files have the same contents. */ /* TODO: hexdump the first bytes that actually differ. */ int assertion_equal_file(const char *filename, int line, const char *fn1, const char *fn2) { char buff1[1024]; char buff2[1024]; FILE *f1, *f2; int n1, n2; assertion_count(filename, line); f1 = fopen(fn1, "rb"); f2 = fopen(fn2, "rb"); if (f1 == NULL || f2 == NULL) { if (f1) fclose(f1); if (f2) fclose(f2); return (0); } for (;;) { n1 = (int)fread(buff1, 1, sizeof(buff1), f1); n2 = (int)fread(buff2, 1, sizeof(buff2), f2); if (n1 != n2) break; if (n1 == 0 && n2 == 0) { fclose(f1); fclose(f2); return (1); } if (memcmp(buff1, buff2, n1) != 0) break; } fclose(f1); fclose(f2); failure_start(filename, line, "Files not identical"); logprintf(" file1=\"%s\"\n", fn1); logprintf(" file2=\"%s\"\n", fn2); failure_finish(NULL); return (0); } /* Verify that the named file does exist. */ int assertion_file_exists(const char *filename, int line, const char *f) { assertion_count(filename, line); #if defined(_WIN32) && !defined(__CYGWIN__) if (!_access(f, 0)) return (1); #else if (!access(f, F_OK)) return (1); #endif failure_start(filename, line, "File should exist: %s", f); failure_finish(NULL); return (0); } /* Verify that the named file doesn't exist. */ int assertion_file_not_exists(const char *filename, int line, const char *f) { assertion_count(filename, line); #if defined(_WIN32) && !defined(__CYGWIN__) if (_access(f, 0)) return (1); #else if (access(f, F_OK)) return (1); #endif failure_start(filename, line, "File should not exist: %s", f); failure_finish(NULL); return (0); } /* Compare the contents of a file to a block of memory. */ int assertion_file_contents(const char *filename, int line, const void *buff, int s, const char *fn) { char *contents; FILE *f; int n; assertion_count(filename, line); f = fopen(fn, "rb"); if (f == NULL) { failure_start(filename, line, "File should exist: %s", fn); failure_finish(NULL); return (0); } contents = malloc(s * 2); n = (int)fread(contents, 1, s * 2, f); fclose(f); if (n == s && memcmp(buff, contents, s) == 0) { free(contents); return (1); } failure_start(filename, line, "File contents don't match"); logprintf(" file=\"%s\"\n", fn); if (n > 0) hexdump(contents, buff, n > 512 ? 512 : n, 0); else { logprintf(" File empty, contents should be:\n"); hexdump(buff, NULL, s > 512 ? 512 : s, 0); } failure_finish(NULL); free(contents); return (0); } /* Check the contents of a text file, being tolerant of line endings. */ int assertion_text_file_contents(const char *filename, int line, const char *buff, const char *fn) { char *contents; const char *btxt, *ftxt; FILE *f; int n, s; assertion_count(filename, line); f = fopen(fn, "r"); if (f == NULL) { failure_start(filename, line, "File doesn't exist: %s", fn); failure_finish(NULL); return (0); } s = (int)strlen(buff); contents = malloc(s * 2 + 128); n = (int)fread(contents, 1, s * 2 + 128 - 1, f); if (n >= 0) contents[n] = '\0'; fclose(f); /* Compare texts. */ btxt = buff; ftxt = (const char *)contents; while (*btxt != '\0' && *ftxt != '\0') { if (*btxt == *ftxt) { ++btxt; ++ftxt; continue; } if (btxt[0] == '\n' && ftxt[0] == '\r' && ftxt[1] == '\n') { /* Pass over different new line characters. */ ++btxt; ftxt += 2; continue; } break; } if (*btxt == '\0' && *ftxt == '\0') { free(contents); return (1); } failure_start(filename, line, "Contents don't match"); logprintf(" file=\"%s\"\n", fn); if (n > 0) { hexdump(contents, buff, n, 0); logprintf(" expected\n", fn); hexdump(buff, contents, s, 0); } else { logprintf(" File empty, contents should be:\n"); hexdump(buff, NULL, s, 0); } failure_finish(NULL); free(contents); return (0); } /* Verify that a text file contains the specified lines, regardless of order */ /* This could be more efficient if we sorted both sets of lines, etc, but * since this is used only for testing and only ever deals with a dozen or so * lines at a time, this relatively crude approach is just fine. */ int assertion_file_contains_lines_any_order(const char *file, int line, const char *pathname, const char *lines[]) { char *buff; size_t buff_size; size_t expected_count, actual_count, i, j; char **expected = NULL; char *p, **actual = NULL; char c; int expected_failure = 0, actual_failure = 0; assertion_count(file, line); buff = slurpfile(&buff_size, "%s", pathname); if (buff == NULL) { failure_start(pathname, line, "Can't read file: %s", pathname); failure_finish(NULL); return (0); } /* Make a copy of the provided lines and count up the expected * file size. */ for (i = 0; lines[i] != NULL; ++i) { } expected_count = i; if (expected_count) { expected = malloc(sizeof(char *) * expected_count); if (expected == NULL) { failure_start(pathname, line, "Can't allocate memory"); failure_finish(NULL); free(expected); return (0); } for (i = 0; lines[i] != NULL; ++i) { expected[i] = strdup(lines[i]); } } /* Break the file into lines */ actual_count = 0; for (c = '\0', p = buff; p < buff + buff_size; ++p) { if (*p == '\x0d' || *p == '\x0a') *p = '\0'; if (c == '\0' && *p != '\0') ++actual_count; c = *p; } if (actual_count) { actual = calloc(sizeof(char *), actual_count); if (actual == NULL) { failure_start(pathname, line, "Can't allocate memory"); failure_finish(NULL); free(expected); return (0); } for (j = 0, p = buff; p < buff + buff_size; p += 1 + strlen(p)) { if (*p != '\0') { actual[j] = p; ++j; } } } /* Erase matching lines from both lists */ for (i = 0; i < expected_count; ++i) { if (expected[i] == NULL) continue; for (j = 0; j < actual_count; ++j) { if (actual[j] == NULL) continue; if (strcmp(expected[i], actual[j]) == 0) { free(expected[i]); expected[i] = NULL; actual[j] = NULL; break; } } } /* If there's anything left, it's a failure */ for (i = 0; i < expected_count; ++i) { if (expected[i] != NULL) ++expected_failure; } for (j = 0; j < actual_count; ++j) { if (actual[j] != NULL) ++actual_failure; } if (expected_failure == 0 && actual_failure == 0) { free(buff); free(expected); free(actual); return (1); } failure_start(file, line, "File doesn't match: %s", pathname); for (i = 0; i < expected_count; ++i) { if (expected[i] != NULL) { logprintf(" Expected but not present: %s\n", expected[i]); free(expected[i]); } } for (j = 0; j < actual_count; ++j) { if (actual[j] != NULL) logprintf(" Present but not expected: %s\n", actual[j]); } failure_finish(NULL); free(buff); free(expected); free(actual); return (0); } /* Verify that a text file does not contains the specified strings */ int assertion_file_contains_no_invalid_strings(const char *file, int line, const char *pathname, const char *strings[]) { char *buff; int i; buff = slurpfile(NULL, "%s", pathname); if (buff == NULL) { failure_start(file, line, "Can't read file: %s", pathname); failure_finish(NULL); return (0); } for (i = 0; strings[i] != NULL; ++i) { if (strstr(buff, strings[i]) != NULL) { failure_start(file, line, "Invalid string in %s: %s", pathname, strings[i]); failure_finish(NULL); free(buff); return(0); } } free(buff); return (0); } /* Test that two paths point to the same file. */ /* As a side-effect, asserts that both files exist. */ static int is_hardlink(const char *file, int line, const char *path1, const char *path2) { #if defined(_WIN32) && !defined(__CYGWIN__) BY_HANDLE_FILE_INFORMATION bhfi1, bhfi2; int r; assertion_count(file, line); r = my_GetFileInformationByName(path1, &bhfi1); if (r == 0) { failure_start(file, line, "File %s can't be inspected?", path1); failure_finish(NULL); return (0); } r = my_GetFileInformationByName(path2, &bhfi2); if (r == 0) { failure_start(file, line, "File %s can't be inspected?", path2); failure_finish(NULL); return (0); } return (bhfi1.dwVolumeSerialNumber == bhfi2.dwVolumeSerialNumber && bhfi1.nFileIndexHigh == bhfi2.nFileIndexHigh && bhfi1.nFileIndexLow == bhfi2.nFileIndexLow); #else struct stat st1, st2; int r; assertion_count(file, line); r = lstat(path1, &st1); if (r != 0) { failure_start(file, line, "File should exist: %s", path1); failure_finish(NULL); return (0); } r = lstat(path2, &st2); if (r != 0) { failure_start(file, line, "File should exist: %s", path2); failure_finish(NULL); return (0); } return (st1.st_ino == st2.st_ino && st1.st_dev == st2.st_dev); #endif } int assertion_is_hardlink(const char *file, int line, const char *path1, const char *path2) { if (is_hardlink(file, line, path1, path2)) return (1); failure_start(file, line, "Files %s and %s are not hardlinked", path1, path2); failure_finish(NULL); return (0); } int assertion_is_not_hardlink(const char *file, int line, const char *path1, const char *path2) { if (!is_hardlink(file, line, path1, path2)) return (1); failure_start(file, line, "Files %s and %s should not be hardlinked", path1, path2); failure_finish(NULL); return (0); } /* Verify a/b/mtime of 'pathname'. */ /* If 'recent', verify that it's within last 10 seconds. */ static int assertion_file_time(const char *file, int line, const char *pathname, long t, long nsec, char type, int recent) { long long filet, filet_nsec; int r; #if defined(_WIN32) && !defined(__CYGWIN__) #define EPOC_TIME (116444736000000000ULL) FILETIME fxtime, fbirthtime, fatime, fmtime; ULARGE_INTEGER wintm; HANDLE h; fxtime.dwLowDateTime = 0; fxtime.dwHighDateTime = 0; assertion_count(file, line); /* Note: FILE_FLAG_BACKUP_SEMANTICS applies to open * a directory file. If not, CreateFile() will fail when * the pathname is a directory. */ h = CreateFile(pathname, FILE_READ_ATTRIBUTES, 0, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); if (h == INVALID_HANDLE_VALUE) { failure_start(file, line, "Can't access %s\n", pathname); failure_finish(NULL); return (0); } r = GetFileTime(h, &fbirthtime, &fatime, &fmtime); switch (type) { case 'a': fxtime = fatime; break; case 'b': fxtime = fbirthtime; break; case 'm': fxtime = fmtime; break; } CloseHandle(h); if (r == 0) { failure_start(file, line, "Can't GetFileTime %s\n", pathname); failure_finish(NULL); return (0); } wintm.LowPart = fxtime.dwLowDateTime; wintm.HighPart = fxtime.dwHighDateTime; filet = (wintm.QuadPart - EPOC_TIME) / 10000000; filet_nsec = ((wintm.QuadPart - EPOC_TIME) % 10000000) * 100; nsec = (nsec / 100) * 100; /* Round the request */ #else struct stat st; assertion_count(file, line); r = lstat(pathname, &st); if (r != 0) { failure_start(file, line, "Can't stat %s\n", pathname); failure_finish(NULL); return (0); } switch (type) { case 'a': filet = st.st_atime; break; case 'm': filet = st.st_mtime; break; case 'b': filet = 0; break; default: fprintf(stderr, "INTERNAL: Bad type %c for file time", type); exit(1); } #if defined(__FreeBSD__) switch (type) { case 'a': filet_nsec = st.st_atimespec.tv_nsec; break; case 'b': filet = st.st_birthtime; /* FreeBSD filesystems that don't support birthtime * (e.g., UFS1) always return -1 here. */ if (filet == -1) { return (1); } filet_nsec = st.st_birthtimespec.tv_nsec; break; case 'm': filet_nsec = st.st_mtimespec.tv_nsec; break; default: fprintf(stderr, "INTERNAL: Bad type %c for file time", type); exit(1); } /* FreeBSD generally only stores to microsecond res, so round. */ filet_nsec = (filet_nsec / 1000) * 1000; nsec = (nsec / 1000) * 1000; #else filet_nsec = nsec = 0; /* Generic POSIX only has whole seconds. */ if (type == 'b') return (1); /* Generic POSIX doesn't have birthtime */ #if defined(__HAIKU__) if (type == 'a') return (1); /* Haiku doesn't have atime. */ #endif #endif #endif if (recent) { /* Check that requested time is up-to-date. */ time_t now = time(NULL); if (filet < now - 10 || filet > now + 1) { failure_start(file, line, "File %s has %ctime %lld, %lld seconds ago\n", pathname, type, filet, now - filet); failure_finish(NULL); return (0); } } else if (filet != t || filet_nsec != nsec) { failure_start(file, line, "File %s has %ctime %lld.%09lld, expected %lld.%09lld", pathname, type, filet, filet_nsec, t, nsec); failure_finish(NULL); return (0); } return (1); } /* Verify atime of 'pathname'. */ int assertion_file_atime(const char *file, int line, const char *pathname, long t, long nsec) { return assertion_file_time(file, line, pathname, t, nsec, 'a', 0); } /* Verify atime of 'pathname' is up-to-date. */ int assertion_file_atime_recent(const char *file, int line, const char *pathname) { return assertion_file_time(file, line, pathname, 0, 0, 'a', 1); } /* Verify birthtime of 'pathname'. */ int assertion_file_birthtime(const char *file, int line, const char *pathname, long t, long nsec) { return assertion_file_time(file, line, pathname, t, nsec, 'b', 0); } /* Verify birthtime of 'pathname' is up-to-date. */ int assertion_file_birthtime_recent(const char *file, int line, const char *pathname) { return assertion_file_time(file, line, pathname, 0, 0, 'b', 1); } /* Verify mode of 'pathname'. */ int assertion_file_mode(const char *file, int line, const char *pathname, int expected_mode) { int mode; int r; assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) failure_start(file, line, "assertFileMode not yet implemented for Windows"); (void)mode; /* UNUSED */ (void)r; /* UNUSED */ #else { struct stat st; r = lstat(pathname, &st); mode = (int)(st.st_mode & 0777); } if (r == 0 && mode == expected_mode) return (1); failure_start(file, line, "File %s has mode %o, expected %o", pathname, mode, expected_mode); #endif failure_finish(NULL); return (0); } /* Verify mtime of 'pathname'. */ int assertion_file_mtime(const char *file, int line, const char *pathname, long t, long nsec) { return assertion_file_time(file, line, pathname, t, nsec, 'm', 0); } /* Verify mtime of 'pathname' is up-to-date. */ int assertion_file_mtime_recent(const char *file, int line, const char *pathname) { return assertion_file_time(file, line, pathname, 0, 0, 'm', 1); } /* Verify number of links to 'pathname'. */ int assertion_file_nlinks(const char *file, int line, const char *pathname, int nlinks) { #if defined(_WIN32) && !defined(__CYGWIN__) BY_HANDLE_FILE_INFORMATION bhfi; int r; assertion_count(file, line); r = my_GetFileInformationByName(pathname, &bhfi); if (r != 0 && bhfi.nNumberOfLinks == (DWORD)nlinks) return (1); failure_start(file, line, "File %s has %d links, expected %d", pathname, bhfi.nNumberOfLinks, nlinks); failure_finish(NULL); return (0); #else struct stat st; int r; assertion_count(file, line); r = lstat(pathname, &st); if (r == 0 && (int)st.st_nlink == nlinks) return (1); failure_start(file, line, "File %s has %d links, expected %d", pathname, st.st_nlink, nlinks); failure_finish(NULL); return (0); #endif } /* Verify size of 'pathname'. */ int assertion_file_size(const char *file, int line, const char *pathname, long size) { int64_t filesize; int r; assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) { BY_HANDLE_FILE_INFORMATION bhfi; r = !my_GetFileInformationByName(pathname, &bhfi); filesize = ((int64_t)bhfi.nFileSizeHigh << 32) + bhfi.nFileSizeLow; } #else { struct stat st; r = lstat(pathname, &st); filesize = st.st_size; } #endif if (r == 0 && filesize == size) return (1); failure_start(file, line, "File %s has size %ld, expected %ld", pathname, (long)filesize, (long)size); failure_finish(NULL); return (0); } /* Assert that 'pathname' is a dir. If mode >= 0, verify that too. */ int assertion_is_dir(const char *file, int line, const char *pathname, int mode) { struct stat st; int r; #if defined(_WIN32) && !defined(__CYGWIN__) (void)mode; /* UNUSED */ #endif assertion_count(file, line); r = lstat(pathname, &st); if (r != 0) { failure_start(file, line, "Dir should exist: %s", pathname); failure_finish(NULL); return (0); } if (!S_ISDIR(st.st_mode)) { failure_start(file, line, "%s is not a dir", pathname); failure_finish(NULL); return (0); } #if !defined(_WIN32) || defined(__CYGWIN__) /* Windows doesn't handle permissions the same way as POSIX, * so just ignore the mode tests. */ /* TODO: Can we do better here? */ if (mode >= 0 && (mode_t)mode != (st.st_mode & 07777)) { failure_start(file, line, "Dir %s has wrong mode", pathname); logprintf(" Expected: 0%3o\n", mode); logprintf(" Found: 0%3o\n", st.st_mode & 07777); failure_finish(NULL); return (0); } #endif return (1); } /* Verify that 'pathname' is a regular file. If 'mode' is >= 0, * verify that too. */ int assertion_is_reg(const char *file, int line, const char *pathname, int mode) { struct stat st; int r; #if defined(_WIN32) && !defined(__CYGWIN__) (void)mode; /* UNUSED */ #endif assertion_count(file, line); r = lstat(pathname, &st); if (r != 0 || !S_ISREG(st.st_mode)) { failure_start(file, line, "File should exist: %s", pathname); failure_finish(NULL); return (0); } #if !defined(_WIN32) || defined(__CYGWIN__) /* Windows doesn't handle permissions the same way as POSIX, * so just ignore the mode tests. */ /* TODO: Can we do better here? */ if (mode >= 0 && (mode_t)mode != (st.st_mode & 07777)) { failure_start(file, line, "File %s has wrong mode", pathname); logprintf(" Expected: 0%3o\n", mode); logprintf(" Found: 0%3o\n", st.st_mode & 07777); failure_finish(NULL); return (0); } #endif return (1); } /* Check whether 'pathname' is a symbolic link. If 'contents' is * non-NULL, verify that the symlink has those contents. */ static int is_symlink(const char *file, int line, const char *pathname, const char *contents) { #if defined(_WIN32) && !defined(__CYGWIN__) (void)pathname; /* UNUSED */ (void)contents; /* UNUSED */ assertion_count(file, line); /* Windows sort-of has real symlinks, but they're only usable * by privileged users and are crippled even then, so there's * really not much point in bothering with this. */ return (0); #else char buff[300]; struct stat st; ssize_t linklen; int r; assertion_count(file, line); r = lstat(pathname, &st); if (r != 0) { failure_start(file, line, "Symlink should exist: %s", pathname); failure_finish(NULL); return (0); } if (!S_ISLNK(st.st_mode)) return (0); if (contents == NULL) return (1); linklen = readlink(pathname, buff, sizeof(buff)); if (linklen < 0) { failure_start(file, line, "Can't read symlink %s", pathname); failure_finish(NULL); return (0); } buff[linklen] = '\0'; if (strcmp(buff, contents) != 0) return (0); return (1); #endif } /* Assert that path is a symlink that (optionally) contains contents. */ int assertion_is_symlink(const char *file, int line, const char *path, const char *contents) { if (is_symlink(file, line, path, contents)) return (1); if (contents) failure_start(file, line, "File %s is not a symlink to %s", path, contents); else failure_start(file, line, "File %s is not a symlink", path); failure_finish(NULL); return (0); } /* Create a directory and report any errors. */ int assertion_make_dir(const char *file, int line, const char *dirname, int mode) { assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) (void)mode; /* UNUSED */ if (0 == _mkdir(dirname)) return (1); #else if (0 == mkdir(dirname, mode)) { if (0 == chmod(dirname, mode)) { assertion_file_mode(file, line, dirname, mode); return (1); } } #endif failure_start(file, line, "Could not create directory %s", dirname); failure_finish(NULL); return(0); } /* Create a file with the specified contents and report any failures. */ int assertion_make_file(const char *file, int line, const char *path, int mode, int csize, const void *contents) { #if defined(_WIN32) && !defined(__CYGWIN__) /* TODO: Rework this to set file mode as well. */ FILE *f; (void)mode; /* UNUSED */ assertion_count(file, line); f = fopen(path, "wb"); if (f == NULL) { failure_start(file, line, "Could not create file %s", path); failure_finish(NULL); return (0); } if (contents != NULL) { size_t wsize; if (csize < 0) wsize = strlen(contents); else wsize = (size_t)csize; if (wsize != fwrite(contents, 1, wsize, f)) { fclose(f); failure_start(file, line, "Could not write file %s", path); failure_finish(NULL); return (0); } } fclose(f); return (1); #else int fd; assertion_count(file, line); fd = open(path, O_CREAT | O_WRONLY, mode >= 0 ? mode : 0644); if (fd < 0) { failure_start(file, line, "Could not create %s", path); failure_finish(NULL); return (0); } if (0 != chmod(path, mode)) { failure_start(file, line, "Could not chmod %s", path); failure_finish(NULL); close(fd); return (0); } if (contents != NULL) { ssize_t wsize; if (csize < 0) wsize = (ssize_t)strlen(contents); else wsize = (ssize_t)csize; if (wsize != write(fd, contents, wsize)) { close(fd); failure_start(file, line, "Could not write to %s", path); failure_finish(NULL); close(fd); return (0); } } close(fd); assertion_file_mode(file, line, path, mode); return (1); #endif } /* Create a hardlink and report any failures. */ int assertion_make_hardlink(const char *file, int line, const char *newpath, const char *linkto) { int succeeded; assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) succeeded = my_CreateHardLinkA(newpath, linkto); #elif HAVE_LINK succeeded = !link(linkto, newpath); #else succeeded = 0; #endif if (succeeded) return (1); failure_start(file, line, "Could not create hardlink"); logprintf(" New link: %s\n", newpath); logprintf(" Old name: %s\n", linkto); failure_finish(NULL); return(0); } /* Create a symlink and report any failures. */ int assertion_make_symlink(const char *file, int line, const char *newpath, const char *linkto) { #if defined(_WIN32) && !defined(__CYGWIN__) int targetIsDir = 0; /* TODO: Fix this */ assertion_count(file, line); if (my_CreateSymbolicLinkA(newpath, linkto, targetIsDir)) return (1); #elif HAVE_SYMLINK assertion_count(file, line); if (0 == symlink(linkto, newpath)) return (1); #endif failure_start(file, line, "Could not create symlink"); logprintf(" New link: %s\n", newpath); logprintf(" Old name: %s\n", linkto); failure_finish(NULL); return(0); } /* Set umask, report failures. */ int assertion_umask(const char *file, int line, int mask) { assertion_count(file, line); (void)file; /* UNUSED */ (void)line; /* UNUSED */ umask(mask); return (1); } /* Set times, report failures. */ int assertion_utimes(const char *file, int line, const char *pathname, long at, long at_nsec, long mt, long mt_nsec) { int r; #if defined(_WIN32) && !defined(__CYGWIN__) #define WINTIME(sec, nsec) ((Int32x32To64(sec, 10000000) + EPOC_TIME)\ + (((nsec)/1000)*10)) HANDLE h; ULARGE_INTEGER wintm; FILETIME fatime, fmtime; FILETIME *pat, *pmt; assertion_count(file, line); h = CreateFileA(pathname,GENERIC_READ | GENERIC_WRITE, FILE_SHARE_READ | FILE_SHARE_WRITE, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); if (h == INVALID_HANDLE_VALUE) { failure_start(file, line, "Can't access %s\n", pathname); failure_finish(NULL); return (0); } if (at > 0 || at_nsec > 0) { wintm.QuadPart = WINTIME(at, at_nsec); fatime.dwLowDateTime = wintm.LowPart; fatime.dwHighDateTime = wintm.HighPart; pat = &fatime; } else pat = NULL; if (mt > 0 || mt_nsec > 0) { wintm.QuadPart = WINTIME(mt, mt_nsec); fmtime.dwLowDateTime = wintm.LowPart; fmtime.dwHighDateTime = wintm.HighPart; pmt = &fmtime; } else pmt = NULL; if (pat != NULL || pmt != NULL) r = SetFileTime(h, NULL, pat, pmt); else r = 1; CloseHandle(h); if (r == 0) { failure_start(file, line, "Can't SetFileTime %s\n", pathname); failure_finish(NULL); return (0); } return (1); #else /* defined(_WIN32) && !defined(__CYGWIN__) */ struct stat st; struct timeval times[2]; #if !defined(__FreeBSD__) mt_nsec = at_nsec = 0; /* Generic POSIX only has whole seconds. */ #endif if (mt == 0 && mt_nsec == 0 && at == 0 && at_nsec == 0) return (1); r = lstat(pathname, &st); if (r < 0) { failure_start(file, line, "Can't stat %s\n", pathname); failure_finish(NULL); return (0); } if (mt == 0 && mt_nsec == 0) { mt = st.st_mtime; #if defined(__FreeBSD__) mt_nsec = st.st_mtimespec.tv_nsec; /* FreeBSD generally only stores to microsecond res, so round. */ mt_nsec = (mt_nsec / 1000) * 1000; #endif } if (at == 0 && at_nsec == 0) { at = st.st_atime; #if defined(__FreeBSD__) at_nsec = st.st_atimespec.tv_nsec; /* FreeBSD generally only stores to microsecond res, so round. */ at_nsec = (at_nsec / 1000) * 1000; #endif } times[1].tv_sec = mt; times[1].tv_usec = mt_nsec / 1000; times[0].tv_sec = at; times[0].tv_usec = at_nsec / 1000; #ifdef HAVE_LUTIMES r = lutimes(pathname, times); #else r = utimes(pathname, times); #endif if (r < 0) { failure_start(file, line, "Can't utimes %s\n", pathname); failure_finish(NULL); return (0); } return (1); #endif /* defined(_WIN32) && !defined(__CYGWIN__) */ } /* Set nodump, report failures. */ int assertion_nodump(const char *file, int line, const char *pathname) { #if defined(HAVE_STRUCT_STAT_ST_FLAGS) && defined(UF_NODUMP) int r; assertion_count(file, line); r = chflags(pathname, UF_NODUMP); if (r < 0) { failure_start(file, line, "Can't set nodump %s\n", pathname); failure_finish(NULL); return (0); } #elif defined(EXT2_IOC_GETFLAGS) && defined(HAVE_WORKING_EXT2_IOC_GETFLAGS)\ && defined(EXT2_NODUMP_FL) int fd, r, flags; assertion_count(file, line); fd = open(pathname, O_RDONLY | O_NONBLOCK); if (fd < 0) { failure_start(file, line, "Can't open %s\n", pathname); failure_finish(NULL); return (0); } r = ioctl(fd, EXT2_IOC_GETFLAGS, &flags); if (r < 0) { failure_start(file, line, "Can't get flags %s\n", pathname); failure_finish(NULL); return (0); } flags |= EXT2_NODUMP_FL; r = ioctl(fd, EXT2_IOC_SETFLAGS, &flags); if (r < 0) { failure_start(file, line, "Can't set nodump %s\n", pathname); failure_finish(NULL); return (0); } close(fd); #else (void)pathname; /* UNUSED */ assertion_count(file, line); #endif return (1); } /* * * UTILITIES for use by tests. * */ /* * Check whether platform supports symlinks. This is intended * for tests to use in deciding whether to bother testing symlink * support; if the platform doesn't support symlinks, there's no point * in checking whether the program being tested can create them. * * Note that the first time this test is called, we actually go out to * disk to create and verify a symlink. This is necessary because * symlink support is actually a property of a particular filesystem * and can thus vary between directories on a single system. After * the first call, this returns the cached result from memory, so it's * safe to call it as often as you wish. */ int canSymlink(void) { /* Remember the test result */ static int value = 0, tested = 0; if (tested) return (value); ++tested; assertion_make_file(__FILE__, __LINE__, "canSymlink.0", 0644, 1, "a"); /* Note: Cygwin has its own symlink() emulation that does not * use the Win32 CreateSymbolicLink() function. */ #if defined(_WIN32) && !defined(__CYGWIN__) value = my_CreateSymbolicLinkA("canSymlink.1", "canSymlink.0", 0) && is_symlink(__FILE__, __LINE__, "canSymlink.1", "canSymlink.0"); #elif HAVE_SYMLINK value = (0 == symlink("canSymlink.0", "canSymlink.1")) && is_symlink(__FILE__, __LINE__, "canSymlink.1","canSymlink.0"); #endif return (value); } /* Platform-dependent options for hiding the output of a subcommand. */ #if defined(_WIN32) && !defined(__CYGWIN__) static const char *redirectArgs = ">NUL 2>NUL"; /* Win32 cmd.exe */ #else static const char *redirectArgs = ">/dev/null 2>/dev/null"; /* POSIX 'sh' */ #endif /* * Can this platform run the bzip2 program? */ int canBzip2(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("bzip2 -d -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the grzip program? */ int canGrzip(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("grzip -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the gzip program? */ int canGzip(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("gzip -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lrzip program? */ int canRunCommand(const char *cmd) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("%s %s", cmd, redirectArgs) == 0) value = 1; } return (value); } int canLrzip(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lrzip -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lz4 program? */ int canLz4(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lz4 -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lzip program? */ int canLzip(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lzip -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lzma program? */ int canLzma(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lzma -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lzop program? */ int canLzop(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lzop -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the xz program? */ int canXz(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("xz -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this filesystem handle nodump flags. */ #if defined(HAVE_STRUCT_STAT_ST_FLAGS) && defined(UF_NODUMP) int canNodump(void) { const char *path = "cannodumptest"; struct stat sb; assertion_make_file(__FILE__, __LINE__, path, 0644, 0, NULL); if (chflags(path, UF_NODUMP) < 0) return (0); if (stat(path, &sb) < 0) return (0); if (sb.st_flags & UF_NODUMP) return (1); return (0); } #elif defined(EXT2_IOC_GETFLAGS) && defined(HAVE_WORKING_EXT2_IOC_GETFLAGS)\ && defined(EXT2_NODUMP_FL) int canNodump(void) { const char *path = "cannodumptest"; int fd, r, flags; assertion_make_file(__FILE__, __LINE__, path, 0644, 0, NULL); fd = open(path, O_RDONLY | O_NONBLOCK); if (fd < 0) return (0); r = ioctl(fd, EXT2_IOC_GETFLAGS, &flags); if (r < 0) return (0); flags |= EXT2_NODUMP_FL; r = ioctl(fd, EXT2_IOC_SETFLAGS, &flags); if (r < 0) return (0); close(fd); fd = open(path, O_RDONLY | O_NONBLOCK); if (fd < 0) return (0); r = ioctl(fd, EXT2_IOC_GETFLAGS, &flags); if (r < 0) return (0); close(fd); if (flags & EXT2_NODUMP_FL) return (1); return (0); } #else int canNodump() { return (0); } #endif /* * Sleep as needed; useful for verifying disk timestamp changes by * ensuring that the wall-clock time has actually changed before we * go back to re-read something from disk. */ void sleepUntilAfter(time_t t) { while (t >= time(NULL)) #if defined(_WIN32) && !defined(__CYGWIN__) Sleep(500); #else sleep(1); #endif } /* * Call standard system() call, but build up the command line using * sprintf() conventions. */ int systemf(const char *fmt, ...) { char buff[8192]; va_list ap; int r; va_start(ap, fmt); vsprintf(buff, fmt, ap); if (verbosity > VERBOSITY_FULL) logprintf("Cmd: %s\n", buff); r = system(buff); va_end(ap); return (r); } /* * Slurp a file into memory for ease of comparison and testing. * Returns size of file in 'sizep' if non-NULL, null-terminates * data in memory for ease of use. */ char * slurpfile(size_t * sizep, const char *fmt, ...) { char filename[8192]; struct stat st; va_list ap; char *p; ssize_t bytes_read; FILE *f; int r; va_start(ap, fmt); vsprintf(filename, fmt, ap); va_end(ap); f = fopen(filename, "rb"); if (f == NULL) { /* Note: No error; non-existent file is okay here. */ return (NULL); } r = fstat(fileno(f), &st); if (r != 0) { logprintf("Can't stat file %s\n", filename); fclose(f); return (NULL); } p = malloc((size_t)st.st_size + 1); if (p == NULL) { logprintf("Can't allocate %ld bytes of memory to read file %s\n", (long int)st.st_size, filename); fclose(f); return (NULL); } bytes_read = fread(p, 1, (size_t)st.st_size, f); if (bytes_read < st.st_size) { logprintf("Can't read file %s\n", filename); fclose(f); free(p); return (NULL); } p[st.st_size] = '\0'; if (sizep != NULL) *sizep = (size_t)st.st_size; fclose(f); return (p); } /* * Slurp a file into memory for ease of comparison and testing. * Returns size of file in 'sizep' if non-NULL, null-terminates * data in memory for ease of use. */ void dumpfile(const char *filename, void *data, size_t len) { ssize_t bytes_written; FILE *f; f = fopen(filename, "wb"); if (f == NULL) { logprintf("Can't open file %s for writing\n", filename); return; } bytes_written = fwrite(data, 1, len, f); if (bytes_written < (ssize_t)len) logprintf("Can't write file %s\n", filename); fclose(f); } /* Read a uuencoded file from the reference directory, decode, and * write the result into the current directory. */ #define VALID_UUDECODE(c) (c >= 32 && c <= 96) #define UUDECODE(c) (((c) - 0x20) & 0x3f) void extract_reference_file(const char *name) { char buff[1024]; FILE *in, *out; sprintf(buff, "%s/%s.uu", refdir, name); in = fopen(buff, "r"); failure("Couldn't open reference file %s", buff); assert(in != NULL); if (in == NULL) return; /* Read up to and including the 'begin' line. */ for (;;) { if (fgets(buff, sizeof(buff), in) == NULL) { /* TODO: This is a failure. */ return; } if (memcmp(buff, "begin ", 6) == 0) break; } /* Now, decode the rest and write it. */ out = fopen(name, "wb"); while (fgets(buff, sizeof(buff), in) != NULL) { char *p = buff; int bytes; if (memcmp(buff, "end", 3) == 0) break; bytes = UUDECODE(*p++); while (bytes > 0) { int n = 0; /* Write out 1-3 bytes from that. */ if (bytes > 0) { assert(VALID_UUDECODE(p[0])); assert(VALID_UUDECODE(p[1])); n = UUDECODE(*p++) << 18; n |= UUDECODE(*p++) << 12; fputc(n >> 16, out); --bytes; } if (bytes > 0) { assert(VALID_UUDECODE(p[0])); n |= UUDECODE(*p++) << 6; fputc((n >> 8) & 0xFF, out); --bytes; } if (bytes > 0) { assert(VALID_UUDECODE(p[0])); n |= UUDECODE(*p++); fputc(n & 0xFF, out); --bytes; } } } fclose(out); fclose(in); } void copy_reference_file(const char *name) { char buff[1024]; FILE *in, *out; size_t rbytes; sprintf(buff, "%s/%s", refdir, name); in = fopen(buff, "rb"); failure("Couldn't open reference file %s", buff); assert(in != NULL); if (in == NULL) return; /* Now, decode the rest and write it. */ /* Not a lot of error checking here; the input better be right. */ out = fopen(name, "wb"); while ((rbytes = fread(buff, 1, sizeof(buff), in)) > 0) { if (fwrite(buff, 1, rbytes, out) != rbytes) { logprintf("Error: fwrite\n"); break; } } fclose(out); fclose(in); } int is_LargeInode(const char *file) { #if defined(_WIN32) && !defined(__CYGWIN__) BY_HANDLE_FILE_INFORMATION bhfi; int r; r = my_GetFileInformationByName(file, &bhfi); if (r != 0) return (0); return (bhfi.nFileIndexHigh & 0x0000FFFFUL); #else struct stat st; int64_t ino; if (stat(file, &st) < 0) return (0); ino = (int64_t)st.st_ino; return (ino > 0xffffffff); #endif } void extract_reference_files(const char **names) { while (names && *names) extract_reference_file(*names++); } /* * * TEST management * */ /* * "list.h" is simply created by "grep DEFINE_TEST test_*.c"; it has * a line like * DEFINE_TEST(test_function) * for each test. */ /* Use "list.h" to declare all of the test functions. */ #undef DEFINE_TEST #define DEFINE_TEST(name) void name(void); #include "list.h" /* Use "list.h" to create a list of all tests (functions and names). */ #undef DEFINE_TEST #define DEFINE_TEST(n) { n, #n, 0 }, struct test_list_t tests[] = { #include "list.h" }; /* * Summarize repeated failures in the just-completed test. */ static void test_summarize(int failed, int skips_num) { unsigned int i; switch (verbosity) { case VERBOSITY_SUMMARY_ONLY: printf(failed ? "E" : "."); fflush(stdout); break; case VERBOSITY_PASSFAIL: printf(failed ? "FAIL\n" : skips_num ? "ok (S)\n" : "ok\n"); break; } log_console = (verbosity == VERBOSITY_LIGHT_REPORT); for (i = 0; i < sizeof(failed_lines)/sizeof(failed_lines[0]); i++) { if (failed_lines[i].count > 1 && !failed_lines[i].skip) logprintf("%s:%d: Summary: Failed %d times\n", failed_filename, i, failed_lines[i].count); } /* Clear the failure history for the next file. */ failed_filename = NULL; memset(failed_lines, 0, sizeof(failed_lines)); } /* * Actually run a single test, with appropriate setup and cleanup. */ static int test_run(int i, const char *tmpdir) { char workdir[1024]; char logfilename[64]; int failures_before = failures; int skips_before = skips; int oldumask; switch (verbosity) { case VERBOSITY_SUMMARY_ONLY: /* No per-test reports at all */ break; case VERBOSITY_PASSFAIL: /* rest of line will include ok/FAIL marker */ printf("%3d: %-64s", i, tests[i].name); fflush(stdout); break; default: /* Title of test, details will follow */ printf("%3d: %s\n", i, tests[i].name); } /* Chdir to the top-level work directory. */ if (!assertChdir(tmpdir)) { fprintf(stderr, "ERROR: Can't chdir to top work dir %s\n", tmpdir); exit(1); } /* Create a log file for this test. */ sprintf(logfilename, "%s.log", tests[i].name); logfile = fopen(logfilename, "w"); fprintf(logfile, "%s\n\n", tests[i].name); /* Chdir() to a work dir for this specific test. */ snprintf(workdir, sizeof(workdir), "%s/%s", tmpdir, tests[i].name); testworkdir = workdir; if (!assertMakeDir(testworkdir, 0755) || !assertChdir(testworkdir)) { fprintf(stderr, "ERROR: Can't chdir to work dir %s\n", testworkdir); exit(1); } /* Explicitly reset the locale before each test. */ setlocale(LC_ALL, "C"); /* Record the umask before we run the test. */ umask(oldumask = umask(0)); /* * Run the actual test. */ (*tests[i].func)(); /* * Clean up and report afterwards. */ testworkdir = NULL; /* Restore umask */ umask(oldumask); /* Reset locale. */ setlocale(LC_ALL, "C"); /* Reset directory. */ if (!assertChdir(tmpdir)) { fprintf(stderr, "ERROR: Couldn't chdir to temp dir %s\n", tmpdir); exit(1); } /* Report per-test summaries. */ tests[i].failures = failures - failures_before; test_summarize(tests[i].failures, skips - skips_before); /* Close the per-test log file. */ fclose(logfile); logfile = NULL; /* If there were no failures, we can remove the work dir and logfile. */ if (tests[i].failures == 0) { if (!keep_temp_files && assertChdir(tmpdir)) { #if defined(_WIN32) && !defined(__CYGWIN__) /* Make sure not to leave empty directories. * Sometimes a processing of closing files used by tests * is not done, then rmdir will be failed and it will * leave a empty test directory. So we should wait a few * seconds and retry rmdir. */ int r, t; for (t = 0; t < 10; t++) { if (t > 0) Sleep(1000); r = systemf("rmdir /S /Q %s", tests[i].name); if (r == 0) break; } systemf("del %s", logfilename); #else systemf("rm -rf %s", tests[i].name); systemf("rm %s", logfilename); #endif } } /* Return appropriate status. */ return (tests[i].failures); } /* * * * MAIN and support routines. * * */ static void usage(const char *program) { static const int limit = sizeof(tests) / sizeof(tests[0]); int i; printf("Usage: %s [options] ...\n", program); printf("Default is to run all tests.\n"); printf("Otherwise, specify the numbers of the tests you wish to run.\n"); printf("Options:\n"); printf(" -d Dump core after any failure, for debugging.\n"); printf(" -k Keep all temp files.\n"); printf(" Default: temp files for successful tests deleted.\n"); #ifdef PROGRAM printf(" -p Path to executable to be tested.\n"); printf(" Default: path taken from " ENVBASE " environment variable.\n"); #endif printf(" -q Quiet.\n"); printf(" -r Path to dir containing reference files.\n"); printf(" Default: Current directory.\n"); printf(" -u Keep running specifies tests until one fails.\n"); printf(" -v Verbose.\n"); printf("Available tests:\n"); for (i = 0; i < limit; i++) printf(" %d: %s\n", i, tests[i].name); exit(1); } static char * get_refdir(const char *d) { size_t tried_size, buff_size; char *buff, *tried, *pwd = NULL, *p = NULL; #ifdef PATH_MAX buff_size = PATH_MAX; #else buff_size = 8192; #endif buff = calloc(buff_size, 1); if (buff == NULL) { fprintf(stderr, "Unable to allocate memory\n"); exit(1); } /* Allocate a buffer to hold the various directories we checked. */ tried_size = buff_size * 2; tried = calloc(tried_size, 1); if (tried == NULL) { fprintf(stderr, "Unable to allocate memory\n"); exit(1); } /* If a dir was specified, try that */ if (d != NULL) { pwd = NULL; snprintf(buff, buff_size, "%s", d); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); goto failure; } /* Get the current dir. */ #ifdef PATH_MAX pwd = getcwd(NULL, PATH_MAX);/* Solaris getcwd needs the size. */ #else pwd = getcwd(NULL, 0); #endif while (pwd[strlen(pwd) - 1] == '\n') pwd[strlen(pwd) - 1] = '\0'; /* Look for a known file. */ snprintf(buff, buff_size, "%s", pwd); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); snprintf(buff, buff_size, "%s/test", pwd); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); #if defined(LIBRARY) snprintf(buff, buff_size, "%s/%s/test", pwd, LIBRARY); #else snprintf(buff, buff_size, "%s/%s/test", pwd, PROGRAM); #endif p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); #if defined(PROGRAM_ALIAS) snprintf(buff, buff_size, "%s/%s/test", pwd, PROGRAM_ALIAS); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); #endif if (memcmp(pwd, "/usr/obj", 8) == 0) { snprintf(buff, buff_size, "%s", pwd + 8); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); snprintf(buff, buff_size, "%s/test", pwd + 8); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); } failure: printf("Unable to locate known reference file %s\n", KNOWNREF); printf(" Checked following directories:\n%s\n", tried); printf("Use -r option to specify full path to reference directory\n"); #if defined(_WIN32) && !defined(__CYGWIN__) && defined(_DEBUG) DebugBreak(); #endif exit(1); success: free(p); free(pwd); free(tried); /* Copy result into a fresh buffer to reduce memory usage. */ p = strdup(buff); free(buff); return p; } int main(int argc, char **argv) { static const int limit = sizeof(tests) / sizeof(tests[0]); int test_set[sizeof(tests) / sizeof(tests[0])]; int i = 0, j = 0, tests_run = 0, tests_failed = 0, option; time_t now; char *refdir_alloc = NULL; const char *progname; char **saved_argv; const char *tmp, *option_arg, *p; char tmpdir[256], *pwd, *testprogdir, *tmp2 = NULL, *vlevel = NULL; char tmpdir_timestamp[256]; (void)argc; /* UNUSED */ /* Get the current dir. */ #ifdef PATH_MAX pwd = getcwd(NULL, PATH_MAX);/* Solaris getcwd needs the size. */ #else pwd = getcwd(NULL, 0); #endif while (pwd[strlen(pwd) - 1] == '\n') pwd[strlen(pwd) - 1] = '\0'; #if defined(HAVE__CrtSetReportMode) && !defined(__WATCOMC__) /* To stop to run the default invalid parameter handler. */ _set_invalid_parameter_handler(invalid_parameter_handler); /* Disable annoying assertion message box. */ _CrtSetReportMode(_CRT_ASSERT, 0); #endif /* * Name of this program, used to build root of our temp directory * tree. */ progname = p = argv[0]; if ((testprogdir = (char *)malloc(strlen(progname) + 1)) == NULL) { fprintf(stderr, "ERROR: Out of memory."); exit(1); } strcpy(testprogdir, progname); while (*p != '\0') { /* Support \ or / dir separators for Windows compat. */ if (*p == '/' || *p == '\\') { progname = p + 1; i = j; } ++p; j++; } testprogdir[i] = '\0'; #if defined(_WIN32) && !defined(__CYGWIN__) if (testprogdir[0] != '/' && testprogdir[0] != '\\' && !(((testprogdir[0] >= 'a' && testprogdir[0] <= 'z') || (testprogdir[0] >= 'A' && testprogdir[0] <= 'Z')) && testprogdir[1] == ':' && (testprogdir[2] == '/' || testprogdir[2] == '\\'))) #else if (testprogdir[0] != '/') #endif { /* Fixup path for relative directories. */ if ((testprogdir = (char *)realloc(testprogdir, strlen(pwd) + 1 + strlen(testprogdir) + 1)) == NULL) { fprintf(stderr, "ERROR: Out of memory."); exit(1); } memmove(testprogdir + strlen(pwd) + 1, testprogdir, strlen(testprogdir) + 1); memcpy(testprogdir, pwd, strlen(pwd)); testprogdir[strlen(pwd)] = '/'; } #ifdef PROGRAM /* Get the target program from environment, if available. */ testprogfile = getenv(ENVBASE); #endif if (getenv("TMPDIR") != NULL) tmp = getenv("TMPDIR"); else if (getenv("TMP") != NULL) tmp = getenv("TMP"); else if (getenv("TEMP") != NULL) tmp = getenv("TEMP"); else if (getenv("TEMPDIR") != NULL) tmp = getenv("TEMPDIR"); else tmp = "/tmp"; /* Allow -d to be controlled through the environment. */ if (getenv(ENVBASE "_DEBUG") != NULL) dump_on_failure = 1; /* Allow -v to be controlled through the environment. */ if (getenv("_VERBOSITY_LEVEL") != NULL) { vlevel = getenv("_VERBOSITY_LEVEL"); verbosity = atoi(vlevel); if (verbosity < VERBOSITY_SUMMARY_ONLY || verbosity > VERBOSITY_FULL) { /* Unsupported verbosity levels are silently ignored */ vlevel = NULL; verbosity = VERBOSITY_PASSFAIL; } } /* Get the directory holding test files from environment. */ refdir = getenv(ENVBASE "_TEST_FILES"); /* * Parse options, without using getopt(), which isn't available * on all platforms. */ ++argv; /* Skip program name */ while (*argv != NULL) { if (**argv != '-') break; p = *argv++; ++p; /* Skip '-' */ while (*p != '\0') { option = *p++; option_arg = NULL; /* If 'opt' takes an argument, parse that. */ if (option == 'p' || option == 'r') { if (*p != '\0') option_arg = p; else if (*argv == NULL) { fprintf(stderr, "Option -%c requires argument.\n", option); usage(progname); } else option_arg = *argv++; p = ""; /* End of this option word. */ } /* Now, handle the option. */ switch (option) { case 'd': dump_on_failure = 1; break; case 'k': keep_temp_files = 1; break; case 'p': #ifdef PROGRAM testprogfile = option_arg; #else fprintf(stderr, "-p option not permitted\n"); usage(progname); #endif break; case 'q': if (!vlevel) verbosity--; break; case 'r': refdir = option_arg; break; case 'u': until_failure++; break; case 'v': if (!vlevel) verbosity++; break; default: fprintf(stderr, "Unrecognized option '%c'\n", option); usage(progname); } } } /* * Sanity-check that our options make sense. */ #ifdef PROGRAM if (testprogfile == NULL) { if ((tmp2 = (char *)malloc(strlen(testprogdir) + 1 + strlen(PROGRAM) + 1)) == NULL) { fprintf(stderr, "ERROR: Out of memory."); exit(1); } strcpy(tmp2, testprogdir); strcat(tmp2, "/"); strcat(tmp2, PROGRAM); testprogfile = tmp2; } { char *testprg; #if defined(_WIN32) && !defined(__CYGWIN__) /* Command.com sometimes rejects '/' separators. */ testprg = strdup(testprogfile); for (i = 0; testprg[i] != '\0'; i++) { if (testprg[i] == '/') testprg[i] = '\\'; } testprogfile = testprg; #endif /* Quote the name that gets put into shell command lines. */ testprg = malloc(strlen(testprogfile) + 3); strcpy(testprg, "\""); strcat(testprg, testprogfile); strcat(testprg, "\""); testprog = testprg; } #endif #if !defined(_WIN32) && defined(SIGPIPE) { /* Ignore SIGPIPE signals */ struct sigaction sa; sa.sa_handler = SIG_IGN; sigemptyset(&sa.sa_mask); sa.sa_flags = 0; sigaction(SIGPIPE, &sa, NULL); } #endif /* * Create a temp directory for the following tests. * Include the time the tests started as part of the name, * to make it easier to track the results of multiple tests. */ now = time(NULL); for (i = 0; ; i++) { strftime(tmpdir_timestamp, sizeof(tmpdir_timestamp), "%Y-%m-%dT%H.%M.%S", localtime(&now)); sprintf(tmpdir, "%s/%s.%s-%03d", tmp, progname, tmpdir_timestamp, i); if (assertMakeDir(tmpdir,0755)) break; if (i >= 999) { fprintf(stderr, "ERROR: Unable to create temp directory %s\n", tmpdir); exit(1); } } /* * If the user didn't specify a directory for locating * reference files, try to find the reference files in * the "usual places." */ refdir = refdir_alloc = get_refdir(refdir); /* * Banner with basic information. */ printf("\n"); printf("If tests fail or crash, details will be in:\n"); printf(" %s\n", tmpdir); printf("\n"); if (verbosity > VERBOSITY_SUMMARY_ONLY) { printf("Reference files will be read from: %s\n", refdir); #ifdef PROGRAM printf("Running tests on: %s\n", testprog); #endif printf("Exercising: "); fflush(stdout); printf("%s\n", EXTRA_VERSION); } else { printf("Running "); fflush(stdout); } /* * Run some or all of the individual tests. */ saved_argv = argv; do { argv = saved_argv; do { int test_num; test_num = get_test_set(test_set, limit, *argv, tests); if (test_num < 0) { printf("*** INVALID Test %s\n", *argv); free(refdir_alloc); free(testprogdir); usage(progname); return (1); } for (i = 0; i < test_num; i++) { tests_run++; if (test_run(test_set[i], tmpdir)) { tests_failed++; if (until_failure) goto finish; } } if (*argv != NULL) argv++; } while (*argv != NULL); } while (until_failure); finish: /* Must be freed after all tests run */ free(tmp2); free(testprogdir); free(pwd); /* * Report summary statistics. */ if (verbosity > VERBOSITY_SUMMARY_ONLY) { printf("\n"); printf("Totals:\n"); printf(" Tests run: %8d\n", tests_run); printf(" Tests failed: %8d\n", tests_failed); printf(" Assertions checked:%8d\n", assertions); printf(" Assertions failed: %8d\n", failures); printf(" Skips reported: %8d\n", skips); } if (failures) { printf("\n"); printf("Failing tests:\n"); for (i = 0; i < limit; ++i) { if (tests[i].failures) printf(" %d: %s (%d failures)\n", i, tests[i].name, tests[i].failures); } printf("\n"); printf("Details for failing tests: %s\n", tmpdir); printf("\n"); } else { if (verbosity == VERBOSITY_SUMMARY_ONLY) printf("\n"); printf("%d tests passed, no failures\n", tests_run); } free(refdir_alloc); /* If the final tmpdir is empty, we can remove it. */ /* This should be the usual case when all tests succeed. */ assertChdir(".."); rmdir(tmpdir); return (tests_failed ? 1 : 0); } Index: head/contrib/libarchive/cpio/test/main.c =================================================================== --- head/contrib/libarchive/cpio/test/main.c (revision 309589) +++ head/contrib/libarchive/cpio/test/main.c (revision 309590) @@ -1,3073 +1,3073 @@ /* * Copyright (c) 2003-2009 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" #include "test_utils.h" #ifdef HAVE_SYS_IOCTL_H #include #endif #ifdef HAVE_SYS_TIME_H #include #endif #include #ifdef HAVE_ICONV_H #include #endif /* * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. * As the include guards don't agree, the order of include is important. */ #ifdef HAVE_LINUX_EXT2_FS_H #include /* for Linux file flags */ #endif #if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) #include /* Linux file flags, broken on Cygwin */ #endif #include #include #ifdef HAVE_SIGNAL_H #include #endif #include #include /* * This same file is used pretty much verbatim for all test harnesses. * * The next few lines are the only differences. * TODO: Move this into a separate configuration header, have all test * suites share one copy of this file. */ __FBSDID("$FreeBSD$"); #define KNOWNREF "test_option_f.cpio.uu" #define ENVBASE "BSDCPIO" /* Prefix for environment variables. */ #define PROGRAM "bsdcpio" /* Name of program being tested. */ #define PROGRAM_ALIAS "cpio" /* Generic alias for program */ #undef LIBRARY /* Not testing a library. */ #undef EXTRA_DUMP /* How to dump extra data */ #undef EXTRA_ERRNO /* How to dump errno */ /* How to generate extra version info. */ #define EXTRA_VERSION (systemf("%s --version", testprog) ? "" : "") /* * * Windows support routines * * Note: Configuration is a tricky issue. Using HAVE_* feature macros * in the test harness is dangerous because they cover up * configuration errors. The classic example of this is omitting a * configure check. If libarchive and libarchive_test both look for * the same feature macro, such errors are hard to detect. Platform * macros (e.g., _WIN32 or __GNUC__) are a little better, but can * easily lead to very messy code. It's best to limit yourself * to only the most generic programming techniques in the test harness * and thus avoid conditionals altogether. Where that's not possible, * try to minimize conditionals by grouping platform-specific tests in * one place (e.g., test_acl_freebsd) or by adding new assert() * functions (e.g., assertMakeHardlink()) to cover up platform * differences. Platform-specific coding in libarchive_test is often * a symptom that some capability is missing from libarchive itself. */ #if defined(_WIN32) && !defined(__CYGWIN__) #include #include #include #ifndef F_OK #define F_OK (0) #endif #ifndef S_ISDIR #define S_ISDIR(m) ((m) & _S_IFDIR) #endif #ifndef S_ISREG #define S_ISREG(m) ((m) & _S_IFREG) #endif #if !defined(__BORLANDC__) #define access _access #undef chdir #define chdir _chdir #endif #ifndef fileno #define fileno _fileno #endif /*#define fstat _fstat64*/ #if !defined(__BORLANDC__) #define getcwd _getcwd #endif #define lstat stat /*#define lstat _stat64*/ /*#define stat _stat64*/ #define rmdir _rmdir #if !defined(__BORLANDC__) #define strdup _strdup #define umask _umask #endif #define int64_t __int64 #endif #if defined(HAVE__CrtSetReportMode) # include #endif mode_t umasked(mode_t expected_mode) { mode_t mode = umask(0); umask(mode); return expected_mode & ~mode; } /* Path to working directory for current test */ const char *testworkdir; #ifdef PROGRAM /* Pathname of exe to be tested. */ const char *testprogfile; /* Name of exe to use in printf-formatted command strings. */ /* On Windows, this includes leading/trailing quotes. */ const char *testprog; #endif #if defined(_WIN32) && !defined(__CYGWIN__) static void *GetFunctionKernel32(const char *); static int my_CreateSymbolicLinkA(const char *, const char *, int); static int my_CreateHardLinkA(const char *, const char *); static int my_GetFileInformationByName(const char *, BY_HANDLE_FILE_INFORMATION *); static void * GetFunctionKernel32(const char *name) { static HINSTANCE lib; static int set; if (!set) { set = 1; lib = LoadLibrary("kernel32.dll"); } if (lib == NULL) { fprintf(stderr, "Can't load kernel32.dll?!\n"); exit(1); } return (void *)GetProcAddress(lib, name); } static int my_CreateSymbolicLinkA(const char *linkname, const char *target, int flags) { static BOOLEAN (WINAPI *f)(LPCSTR, LPCSTR, DWORD); static int set; if (!set) { set = 1; f = GetFunctionKernel32("CreateSymbolicLinkA"); } return f == NULL ? 0 : (*f)(linkname, target, flags); } static int my_CreateHardLinkA(const char *linkname, const char *target) { static BOOLEAN (WINAPI *f)(LPCSTR, LPCSTR, LPSECURITY_ATTRIBUTES); static int set; if (!set) { set = 1; f = GetFunctionKernel32("CreateHardLinkA"); } return f == NULL ? 0 : (*f)(linkname, target, NULL); } static int my_GetFileInformationByName(const char *path, BY_HANDLE_FILE_INFORMATION *bhfi) { HANDLE h; int r; memset(bhfi, 0, sizeof(*bhfi)); h = CreateFile(path, FILE_READ_ATTRIBUTES, 0, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); if (h == INVALID_HANDLE_VALUE) return (0); r = GetFileInformationByHandle(h, bhfi); CloseHandle(h); return (r); } #endif #if defined(HAVE__CrtSetReportMode) && !defined(__WATCOMC__) static void invalid_parameter_handler(const wchar_t * expression, const wchar_t * function, const wchar_t * file, unsigned int line, uintptr_t pReserved) { /* nop */ } #endif /* * * OPTIONS FLAGS * */ /* Enable core dump on failure. */ static int dump_on_failure = 0; /* Default is to remove temp dirs and log data for successful tests. */ static int keep_temp_files = 0; /* Default is to run the specified tests once and report errors. */ static int until_failure = 0; /* Default is to just report pass/fail for each test. */ static int verbosity = 0; #define VERBOSITY_SUMMARY_ONLY -1 /* -q */ #define VERBOSITY_PASSFAIL 0 /* Default */ #define VERBOSITY_LIGHT_REPORT 1 /* -v */ #define VERBOSITY_FULL 2 /* -vv */ /* A few places generate even more output for verbosity > VERBOSITY_FULL, * mostly for debugging the test harness itself. */ /* Cumulative count of assertion failures. */ static int failures = 0; /* Cumulative count of reported skips. */ static int skips = 0; /* Cumulative count of assertions checked. */ static int assertions = 0; /* Directory where uuencoded reference files can be found. */ static const char *refdir; /* * Report log information selectively to console and/or disk log. */ static int log_console = 0; static FILE *logfile; static void vlogprintf(const char *fmt, va_list ap) { #ifdef va_copy va_list lfap; va_copy(lfap, ap); #endif if (log_console) vfprintf(stdout, fmt, ap); if (logfile != NULL) #ifdef va_copy vfprintf(logfile, fmt, lfap); va_end(lfap); #else vfprintf(logfile, fmt, ap); #endif } static void logprintf(const char *fmt, ...) { va_list ap; va_start(ap, fmt); vlogprintf(fmt, ap); va_end(ap); } /* Set up a message to display only if next assertion fails. */ static char msgbuff[4096]; static const char *msg, *nextmsg; void failure(const char *fmt, ...) { va_list ap; if (fmt == NULL) { nextmsg = NULL; } else { va_start(ap, fmt); vsprintf(msgbuff, fmt, ap); va_end(ap); nextmsg = msgbuff; } } /* * Copy arguments into file-local variables. * This was added to permit vararg assert() functions without needing * variadic wrapper macros. Turns out that the vararg capability is almost * never used, so almost all of the vararg assertions can be simplified * by removing the vararg capability and reworking the wrapper macro to * pass __FILE__, __LINE__ directly into the function instead of using * this hook. I suspect this machinery is used so rarely that we * would be better off just removing it entirely. That would simplify * the code here noticeably. */ static const char *skipping_filename; static int skipping_line; void skipping_setup(const char *filename, int line) { skipping_filename = filename; skipping_line = line; } /* Called at the beginning of each assert() function. */ static void assertion_count(const char *file, int line) { (void)file; /* UNUSED */ (void)line; /* UNUSED */ ++assertions; /* Proper handling of "failure()" message. */ msg = nextmsg; nextmsg = NULL; /* Uncomment to print file:line after every assertion. * Verbose, but occasionally useful in tracking down crashes. */ /* printf("Checked %s:%d\n", file, line); */ } /* * For each test source file, we remember how many times each * assertion was reported. Cleared before each new test, * used by test_summarize(). */ static struct line { int count; int skip; } failed_lines[10000]; const char *failed_filename; /* Count this failure, setup up log destination and handle initial report. */ static void failure_start(const char *filename, int line, const char *fmt, ...) { va_list ap; /* Record another failure for this line. */ ++failures; failed_filename = filename; failed_lines[line].count++; /* Determine whether to log header to console. */ switch (verbosity) { case VERBOSITY_LIGHT_REPORT: log_console = (failed_lines[line].count < 2); break; default: log_console = (verbosity >= VERBOSITY_FULL); } /* Log file:line header for this failure */ va_start(ap, fmt); #if _MSC_VER logprintf("%s(%d): ", filename, line); #else logprintf("%s:%d: ", filename, line); #endif vlogprintf(fmt, ap); va_end(ap); logprintf("\n"); if (msg != NULL && msg[0] != '\0') { logprintf(" Description: %s\n", msg); msg = NULL; } /* Determine whether to log details to console. */ if (verbosity == VERBOSITY_LIGHT_REPORT) log_console = 0; } /* Complete reporting of failed tests. */ /* * The 'extra' hook here is used by libarchive to include libarchive * error messages with assertion failures. It could also be used * to add strerror() output, for example. Just define the EXTRA_DUMP() * macro appropriately. */ static void failure_finish(void *extra) { (void)extra; /* UNUSED (maybe) */ #ifdef EXTRA_DUMP if (extra != NULL) { logprintf(" errno: %d\n", EXTRA_ERRNO(extra)); logprintf(" detail: %s\n", EXTRA_DUMP(extra)); } #endif if (dump_on_failure) { fprintf(stderr, " *** forcing core dump so failure can be debugged ***\n"); abort(); } } /* Inform user that we're skipping some checks. */ void test_skipping(const char *fmt, ...) { char buff[1024]; va_list ap; va_start(ap, fmt); vsprintf(buff, fmt, ap); va_end(ap); /* Use failure() message if set. */ msg = nextmsg; nextmsg = NULL; /* failure_start() isn't quite right, but is awfully convenient. */ failure_start(skipping_filename, skipping_line, "SKIPPING: %s", buff); --failures; /* Undo failures++ in failure_start() */ /* Don't failure_finish() here. */ /* Mark as skip, so doesn't count as failed test. */ failed_lines[skipping_line].skip = 1; ++skips; } /* * * ASSERTIONS * */ /* Generic assert() just displays the failed condition. */ int assertion_assert(const char *file, int line, int value, const char *condition, void *extra) { assertion_count(file, line); if (!value) { failure_start(file, line, "Assertion failed: %s", condition); failure_finish(extra); } return (value); } /* chdir() and report any errors */ int assertion_chdir(const char *file, int line, const char *pathname) { assertion_count(file, line); if (chdir(pathname) == 0) return (1); failure_start(file, line, "chdir(\"%s\")", pathname); failure_finish(NULL); return (0); } /* Verify two integers are equal. */ int assertion_equal_int(const char *file, int line, long long v1, const char *e1, long long v2, const char *e2, void *extra) { assertion_count(file, line); if (v1 == v2) return (1); failure_start(file, line, "%s != %s", e1, e2); logprintf(" %s=%lld (0x%llx, 0%llo)\n", e1, v1, v1, v1); logprintf(" %s=%lld (0x%llx, 0%llo)\n", e2, v2, v2, v2); failure_finish(extra); return (0); } /* * Utility to convert a single UTF-8 sequence. */ static int _utf8_to_unicode(uint32_t *pwc, const char *s, size_t n) { static const char utf8_count[256] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 00 - 0F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 10 - 1F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 20 - 2F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 30 - 3F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 40 - 4F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 50 - 5F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 60 - 6F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 70 - 7F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 80 - 8F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 90 - 9F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* A0 - AF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* B0 - BF */ 0, 0, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* C0 - CF */ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* D0 - DF */ 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,/* E0 - EF */ 4, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 /* F0 - FF */ }; int ch; int cnt; uint32_t wc; *pwc = 0; /* Sanity check. */ if (n == 0) return (0); /* * Decode 1-4 bytes depending on the value of the first byte. */ ch = (unsigned char)*s; if (ch == 0) return (0); /* Standard: return 0 for end-of-string. */ cnt = utf8_count[ch]; - /* Invalide sequence or there are not plenty bytes. */ + /* Invalid sequence or there are not plenty bytes. */ if (n < (size_t)cnt) return (-1); /* Make a Unicode code point from a single UTF-8 sequence. */ switch (cnt) { case 1: /* 1 byte sequence. */ *pwc = ch & 0x7f; return (cnt); case 2: /* 2 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) return (-1); *pwc = ((ch & 0x1f) << 6) | (s[1] & 0x3f); return (cnt); case 3: /* 3 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) return (-1); if ((s[2] & 0xc0) != 0x80) return (-1); wc = ((ch & 0x0f) << 12) | ((s[1] & 0x3f) << 6) | (s[2] & 0x3f); if (wc < 0x800) return (-1);/* Overlong sequence. */ break; case 4: /* 4 bytes sequence. */ if (n < 4) return (-1); if ((s[1] & 0xc0) != 0x80) return (-1); if ((s[2] & 0xc0) != 0x80) return (-1); if ((s[3] & 0xc0) != 0x80) return (-1); wc = ((ch & 0x07) << 18) | ((s[1] & 0x3f) << 12) | ((s[2] & 0x3f) << 6) | (s[3] & 0x3f); if (wc < 0x10000) return (-1);/* Overlong sequence. */ break; default: return (-1); } - /* The code point larger than 0x10FFFF is not leagal + /* The code point larger than 0x10FFFF is not legal * Unicode values. */ if (wc > 0x10FFFF) return (-1); /* Correctly gets a Unicode, returns used bytes. */ *pwc = wc; return (cnt); } static void strdump(const char *e, const char *p, int ewidth, int utf8) { const char *q = p; logprintf(" %*s = ", ewidth, e); if (p == NULL) { logprintf("NULL\n"); return; } logprintf("\""); while (*p != '\0') { unsigned int c = 0xff & *p++; switch (c) { case '\a': logprintf("\\a"); break; case '\b': logprintf("\\b"); break; case '\n': logprintf("\\n"); break; case '\r': logprintf("\\r"); break; default: if (c >= 32 && c < 127) logprintf("%c", c); else logprintf("\\x%02X", c); } } logprintf("\""); logprintf(" (length %d)", q == NULL ? -1 : (int)strlen(q)); /* * If the current string is UTF-8, dump its code points. */ if (utf8) { size_t len; uint32_t uc; int n; int cnt = 0; p = q; len = strlen(p); logprintf(" ["); while ((n = _utf8_to_unicode(&uc, p, len)) > 0) { if (p != q) logprintf(" "); logprintf("%04X", uc); p += n; len -= n; cnt++; } logprintf("]"); logprintf(" (count %d", cnt); if (n < 0) { logprintf(",unknown %d bytes", len); } logprintf(")"); } logprintf("\n"); } /* Verify two strings are equal, dump them if not. */ int assertion_equal_string(const char *file, int line, const char *v1, const char *e1, const char *v2, const char *e2, void *extra, int utf8) { int l1, l2; assertion_count(file, line); if (v1 == v2 || (v1 != NULL && v2 != NULL && strcmp(v1, v2) == 0)) return (1); failure_start(file, line, "%s != %s", e1, e2); l1 = (int)strlen(e1); l2 = (int)strlen(e2); if (l1 < l2) l1 = l2; strdump(e1, v1, l1, utf8); strdump(e2, v2, l1, utf8); failure_finish(extra); return (0); } static void wcsdump(const char *e, const wchar_t *w) { logprintf(" %s = ", e); if (w == NULL) { logprintf("(null)"); return; } logprintf("\""); while (*w != L'\0') { unsigned int c = *w++; if (c >= 32 && c < 127) logprintf("%c", c); else if (c < 256) logprintf("\\x%02X", c); else if (c < 0x10000) logprintf("\\u%04X", c); else logprintf("\\U%08X", c); } logprintf("\"\n"); } #ifndef HAVE_WCSCMP static int wcscmp(const wchar_t *s1, const wchar_t *s2) { while (*s1 == *s2++) { if (*s1++ == L'\0') return 0; } if (*s1 > *--s2) return 1; else return -1; } #endif /* Verify that two wide strings are equal, dump them if not. */ int assertion_equal_wstring(const char *file, int line, const wchar_t *v1, const char *e1, const wchar_t *v2, const char *e2, void *extra) { assertion_count(file, line); if (v1 == v2) return (1); if (v1 != NULL && v2 != NULL && wcscmp(v1, v2) == 0) return (1); failure_start(file, line, "%s != %s", e1, e2); wcsdump(e1, v1); wcsdump(e2, v2); failure_finish(extra); return (0); } /* * Pretty standard hexdump routine. As a bonus, if ref != NULL, then * any bytes in p that differ from ref will be highlighted with '_' * before and after the hex value. */ static void hexdump(const char *p, const char *ref, size_t l, size_t offset) { size_t i, j; char sep; if (p == NULL) { logprintf("(null)\n"); return; } for(i=0; i < l; i+=16) { logprintf("%04x", (unsigned)(i + offset)); sep = ' '; for (j = 0; j < 16 && i + j < l; j++) { if (ref != NULL && p[i + j] != ref[i + j]) sep = '_'; logprintf("%c%02x", sep, 0xff & (int)p[i+j]); if (ref != NULL && p[i + j] == ref[i + j]) sep = ' '; } for (; j < 16; j++) { logprintf("%c ", sep); sep = ' '; } logprintf("%c", sep); for (j=0; j < 16 && i + j < l; j++) { int c = p[i + j]; if (c >= ' ' && c <= 126) logprintf("%c", c); else logprintf("."); } logprintf("\n"); } } /* Verify that two blocks of memory are the same, display the first * block of differences if they're not. */ int assertion_equal_mem(const char *file, int line, const void *_v1, const char *e1, const void *_v2, const char *e2, size_t l, const char *ld, void *extra) { const char *v1 = (const char *)_v1; const char *v2 = (const char *)_v2; size_t offset; assertion_count(file, line); if (v1 == v2 || (v1 != NULL && v2 != NULL && memcmp(v1, v2, l) == 0)) return (1); if (v1 == NULL || v2 == NULL) return (0); failure_start(file, line, "%s != %s", e1, e2); logprintf(" size %s = %d\n", ld, (int)l); /* Dump 48 bytes (3 lines) so that the first difference is * in the second line. */ offset = 0; while (l > 64 && memcmp(v1, v2, 32) == 0) { /* Two lines agree, so step forward one line. */ v1 += 16; v2 += 16; l -= 16; offset += 16; } logprintf(" Dump of %s\n", e1); hexdump(v1, v2, l < 128 ? l : 128, offset); logprintf(" Dump of %s\n", e2); hexdump(v2, v1, l < 128 ? l : 128, offset); logprintf("\n"); failure_finish(extra); return (0); } /* Verify that a block of memory is filled with the specified byte. */ int assertion_memory_filled_with(const char *file, int line, const void *_v1, const char *vd, size_t l, const char *ld, char b, const char *bd, void *extra) { const char *v1 = (const char *)_v1; size_t c = 0; size_t i; (void)ld; /* UNUSED */ assertion_count(file, line); for (i = 0; i < l; ++i) { if (v1[i] == b) { ++c; } } if (c == l) return (1); failure_start(file, line, "%s (size %d) not filled with %s", vd, (int)l, bd); logprintf(" Only %d bytes were correct\n", (int)c); failure_finish(extra); return (0); } /* Verify that the named file exists and is empty. */ int assertion_empty_file(const char *filename, int line, const char *f1) { char buff[1024]; struct stat st; ssize_t s; FILE *f; assertion_count(filename, line); if (stat(f1, &st) != 0) { failure_start(filename, line, "Stat failed: %s", f1); failure_finish(NULL); return (0); } if (st.st_size == 0) return (1); failure_start(filename, line, "File should be empty: %s", f1); logprintf(" File size: %d\n", (int)st.st_size); logprintf(" Contents:\n"); f = fopen(f1, "rb"); if (f == NULL) { logprintf(" Unable to open %s\n", f1); } else { s = ((off_t)sizeof(buff) < st.st_size) ? (ssize_t)sizeof(buff) : (ssize_t)st.st_size; s = fread(buff, 1, s, f); hexdump(buff, NULL, s, 0); fclose(f); } failure_finish(NULL); return (0); } /* Verify that the named file exists and is not empty. */ int assertion_non_empty_file(const char *filename, int line, const char *f1) { struct stat st; assertion_count(filename, line); if (stat(f1, &st) != 0) { failure_start(filename, line, "Stat failed: %s", f1); failure_finish(NULL); return (0); } if (st.st_size == 0) { failure_start(filename, line, "File empty: %s", f1); failure_finish(NULL); return (0); } return (1); } /* Verify that two files have the same contents. */ /* TODO: hexdump the first bytes that actually differ. */ int assertion_equal_file(const char *filename, int line, const char *fn1, const char *fn2) { char buff1[1024]; char buff2[1024]; FILE *f1, *f2; int n1, n2; assertion_count(filename, line); f1 = fopen(fn1, "rb"); f2 = fopen(fn2, "rb"); if (f1 == NULL || f2 == NULL) { if (f1) fclose(f1); if (f2) fclose(f2); return (0); } for (;;) { n1 = (int)fread(buff1, 1, sizeof(buff1), f1); n2 = (int)fread(buff2, 1, sizeof(buff2), f2); if (n1 != n2) break; if (n1 == 0 && n2 == 0) { fclose(f1); fclose(f2); return (1); } if (memcmp(buff1, buff2, n1) != 0) break; } fclose(f1); fclose(f2); failure_start(filename, line, "Files not identical"); logprintf(" file1=\"%s\"\n", fn1); logprintf(" file2=\"%s\"\n", fn2); failure_finish(NULL); return (0); } /* Verify that the named file does exist. */ int assertion_file_exists(const char *filename, int line, const char *f) { assertion_count(filename, line); #if defined(_WIN32) && !defined(__CYGWIN__) if (!_access(f, 0)) return (1); #else if (!access(f, F_OK)) return (1); #endif failure_start(filename, line, "File should exist: %s", f); failure_finish(NULL); return (0); } /* Verify that the named file doesn't exist. */ int assertion_file_not_exists(const char *filename, int line, const char *f) { assertion_count(filename, line); #if defined(_WIN32) && !defined(__CYGWIN__) if (_access(f, 0)) return (1); #else if (access(f, F_OK)) return (1); #endif failure_start(filename, line, "File should not exist: %s", f); failure_finish(NULL); return (0); } /* Compare the contents of a file to a block of memory. */ int assertion_file_contents(const char *filename, int line, const void *buff, int s, const char *fn) { char *contents; FILE *f; int n; assertion_count(filename, line); f = fopen(fn, "rb"); if (f == NULL) { failure_start(filename, line, "File should exist: %s", fn); failure_finish(NULL); return (0); } contents = malloc(s * 2); n = (int)fread(contents, 1, s * 2, f); fclose(f); if (n == s && memcmp(buff, contents, s) == 0) { free(contents); return (1); } failure_start(filename, line, "File contents don't match"); logprintf(" file=\"%s\"\n", fn); if (n > 0) hexdump(contents, buff, n > 512 ? 512 : n, 0); else { logprintf(" File empty, contents should be:\n"); hexdump(buff, NULL, s > 512 ? 512 : s, 0); } failure_finish(NULL); free(contents); return (0); } /* Check the contents of a text file, being tolerant of line endings. */ int assertion_text_file_contents(const char *filename, int line, const char *buff, const char *fn) { char *contents; const char *btxt, *ftxt; FILE *f; int n, s; assertion_count(filename, line); f = fopen(fn, "r"); if (f == NULL) { failure_start(filename, line, "File doesn't exist: %s", fn); failure_finish(NULL); return (0); } s = (int)strlen(buff); contents = malloc(s * 2 + 128); n = (int)fread(contents, 1, s * 2 + 128 - 1, f); if (n >= 0) contents[n] = '\0'; fclose(f); /* Compare texts. */ btxt = buff; ftxt = (const char *)contents; while (*btxt != '\0' && *ftxt != '\0') { if (*btxt == *ftxt) { ++btxt; ++ftxt; continue; } if (btxt[0] == '\n' && ftxt[0] == '\r' && ftxt[1] == '\n') { /* Pass over different new line characters. */ ++btxt; ftxt += 2; continue; } break; } if (*btxt == '\0' && *ftxt == '\0') { free(contents); return (1); } failure_start(filename, line, "Contents don't match"); logprintf(" file=\"%s\"\n", fn); if (n > 0) { hexdump(contents, buff, n, 0); logprintf(" expected\n", fn); hexdump(buff, contents, s, 0); } else { logprintf(" File empty, contents should be:\n"); hexdump(buff, NULL, s, 0); } failure_finish(NULL); free(contents); return (0); } /* Verify that a text file contains the specified lines, regardless of order */ /* This could be more efficient if we sorted both sets of lines, etc, but * since this is used only for testing and only ever deals with a dozen or so * lines at a time, this relatively crude approach is just fine. */ int assertion_file_contains_lines_any_order(const char *file, int line, const char *pathname, const char *lines[]) { char *buff; size_t buff_size; size_t expected_count, actual_count, i, j; char **expected = NULL; char *p, **actual = NULL; char c; int expected_failure = 0, actual_failure = 0; assertion_count(file, line); buff = slurpfile(&buff_size, "%s", pathname); if (buff == NULL) { failure_start(pathname, line, "Can't read file: %s", pathname); failure_finish(NULL); return (0); } /* Make a copy of the provided lines and count up the expected * file size. */ for (i = 0; lines[i] != NULL; ++i) { } expected_count = i; if (expected_count) { expected = malloc(sizeof(char *) * expected_count); if (expected == NULL) { failure_start(pathname, line, "Can't allocate memory"); failure_finish(NULL); free(expected); return (0); } for (i = 0; lines[i] != NULL; ++i) { expected[i] = strdup(lines[i]); } } /* Break the file into lines */ actual_count = 0; for (c = '\0', p = buff; p < buff + buff_size; ++p) { if (*p == '\x0d' || *p == '\x0a') *p = '\0'; if (c == '\0' && *p != '\0') ++actual_count; c = *p; } if (actual_count) { actual = calloc(sizeof(char *), actual_count); if (actual == NULL) { failure_start(pathname, line, "Can't allocate memory"); failure_finish(NULL); free(expected); return (0); } for (j = 0, p = buff; p < buff + buff_size; p += 1 + strlen(p)) { if (*p != '\0') { actual[j] = p; ++j; } } } /* Erase matching lines from both lists */ for (i = 0; i < expected_count; ++i) { if (expected[i] == NULL) continue; for (j = 0; j < actual_count; ++j) { if (actual[j] == NULL) continue; if (strcmp(expected[i], actual[j]) == 0) { free(expected[i]); expected[i] = NULL; actual[j] = NULL; break; } } } /* If there's anything left, it's a failure */ for (i = 0; i < expected_count; ++i) { if (expected[i] != NULL) ++expected_failure; } for (j = 0; j < actual_count; ++j) { if (actual[j] != NULL) ++actual_failure; } if (expected_failure == 0 && actual_failure == 0) { free(buff); free(expected); free(actual); return (1); } failure_start(file, line, "File doesn't match: %s", pathname); for (i = 0; i < expected_count; ++i) { if (expected[i] != NULL) { logprintf(" Expected but not present: %s\n", expected[i]); free(expected[i]); } } for (j = 0; j < actual_count; ++j) { if (actual[j] != NULL) logprintf(" Present but not expected: %s\n", actual[j]); } failure_finish(NULL); free(buff); free(expected); free(actual); return (0); } /* Verify that a text file does not contains the specified strings */ int assertion_file_contains_no_invalid_strings(const char *file, int line, const char *pathname, const char *strings[]) { char *buff; int i; buff = slurpfile(NULL, "%s", pathname); if (buff == NULL) { failure_start(file, line, "Can't read file: %s", pathname); failure_finish(NULL); return (0); } for (i = 0; strings[i] != NULL; ++i) { if (strstr(buff, strings[i]) != NULL) { failure_start(file, line, "Invalid string in %s: %s", pathname, strings[i]); failure_finish(NULL); free(buff); return(0); } } free(buff); return (0); } /* Test that two paths point to the same file. */ /* As a side-effect, asserts that both files exist. */ static int is_hardlink(const char *file, int line, const char *path1, const char *path2) { #if defined(_WIN32) && !defined(__CYGWIN__) BY_HANDLE_FILE_INFORMATION bhfi1, bhfi2; int r; assertion_count(file, line); r = my_GetFileInformationByName(path1, &bhfi1); if (r == 0) { failure_start(file, line, "File %s can't be inspected?", path1); failure_finish(NULL); return (0); } r = my_GetFileInformationByName(path2, &bhfi2); if (r == 0) { failure_start(file, line, "File %s can't be inspected?", path2); failure_finish(NULL); return (0); } return (bhfi1.dwVolumeSerialNumber == bhfi2.dwVolumeSerialNumber && bhfi1.nFileIndexHigh == bhfi2.nFileIndexHigh && bhfi1.nFileIndexLow == bhfi2.nFileIndexLow); #else struct stat st1, st2; int r; assertion_count(file, line); r = lstat(path1, &st1); if (r != 0) { failure_start(file, line, "File should exist: %s", path1); failure_finish(NULL); return (0); } r = lstat(path2, &st2); if (r != 0) { failure_start(file, line, "File should exist: %s", path2); failure_finish(NULL); return (0); } return (st1.st_ino == st2.st_ino && st1.st_dev == st2.st_dev); #endif } int assertion_is_hardlink(const char *file, int line, const char *path1, const char *path2) { if (is_hardlink(file, line, path1, path2)) return (1); failure_start(file, line, "Files %s and %s are not hardlinked", path1, path2); failure_finish(NULL); return (0); } int assertion_is_not_hardlink(const char *file, int line, const char *path1, const char *path2) { if (!is_hardlink(file, line, path1, path2)) return (1); failure_start(file, line, "Files %s and %s should not be hardlinked", path1, path2); failure_finish(NULL); return (0); } /* Verify a/b/mtime of 'pathname'. */ /* If 'recent', verify that it's within last 10 seconds. */ static int assertion_file_time(const char *file, int line, const char *pathname, long t, long nsec, char type, int recent) { long long filet, filet_nsec; int r; #if defined(_WIN32) && !defined(__CYGWIN__) #define EPOC_TIME (116444736000000000ULL) FILETIME fxtime, fbirthtime, fatime, fmtime; ULARGE_INTEGER wintm; HANDLE h; fxtime.dwLowDateTime = 0; fxtime.dwHighDateTime = 0; assertion_count(file, line); /* Note: FILE_FLAG_BACKUP_SEMANTICS applies to open * a directory file. If not, CreateFile() will fail when * the pathname is a directory. */ h = CreateFile(pathname, FILE_READ_ATTRIBUTES, 0, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); if (h == INVALID_HANDLE_VALUE) { failure_start(file, line, "Can't access %s\n", pathname); failure_finish(NULL); return (0); } r = GetFileTime(h, &fbirthtime, &fatime, &fmtime); switch (type) { case 'a': fxtime = fatime; break; case 'b': fxtime = fbirthtime; break; case 'm': fxtime = fmtime; break; } CloseHandle(h); if (r == 0) { failure_start(file, line, "Can't GetFileTime %s\n", pathname); failure_finish(NULL); return (0); } wintm.LowPart = fxtime.dwLowDateTime; wintm.HighPart = fxtime.dwHighDateTime; filet = (wintm.QuadPart - EPOC_TIME) / 10000000; filet_nsec = ((wintm.QuadPart - EPOC_TIME) % 10000000) * 100; nsec = (nsec / 100) * 100; /* Round the request */ #else struct stat st; assertion_count(file, line); r = lstat(pathname, &st); if (r != 0) { failure_start(file, line, "Can't stat %s\n", pathname); failure_finish(NULL); return (0); } switch (type) { case 'a': filet = st.st_atime; break; case 'm': filet = st.st_mtime; break; case 'b': filet = 0; break; default: fprintf(stderr, "INTERNAL: Bad type %c for file time", type); exit(1); } #if defined(__FreeBSD__) switch (type) { case 'a': filet_nsec = st.st_atimespec.tv_nsec; break; case 'b': filet = st.st_birthtime; /* FreeBSD filesystems that don't support birthtime * (e.g., UFS1) always return -1 here. */ if (filet == -1) { return (1); } filet_nsec = st.st_birthtimespec.tv_nsec; break; case 'm': filet_nsec = st.st_mtimespec.tv_nsec; break; default: fprintf(stderr, "INTERNAL: Bad type %c for file time", type); exit(1); } /* FreeBSD generally only stores to microsecond res, so round. */ filet_nsec = (filet_nsec / 1000) * 1000; nsec = (nsec / 1000) * 1000; #else filet_nsec = nsec = 0; /* Generic POSIX only has whole seconds. */ if (type == 'b') return (1); /* Generic POSIX doesn't have birthtime */ #if defined(__HAIKU__) if (type == 'a') return (1); /* Haiku doesn't have atime. */ #endif #endif #endif if (recent) { /* Check that requested time is up-to-date. */ time_t now = time(NULL); if (filet < now - 10 || filet > now + 1) { failure_start(file, line, "File %s has %ctime %lld, %lld seconds ago\n", pathname, type, filet, now - filet); failure_finish(NULL); return (0); } } else if (filet != t || filet_nsec != nsec) { failure_start(file, line, "File %s has %ctime %lld.%09lld, expected %lld.%09lld", pathname, type, filet, filet_nsec, t, nsec); failure_finish(NULL); return (0); } return (1); } /* Verify atime of 'pathname'. */ int assertion_file_atime(const char *file, int line, const char *pathname, long t, long nsec) { return assertion_file_time(file, line, pathname, t, nsec, 'a', 0); } /* Verify atime of 'pathname' is up-to-date. */ int assertion_file_atime_recent(const char *file, int line, const char *pathname) { return assertion_file_time(file, line, pathname, 0, 0, 'a', 1); } /* Verify birthtime of 'pathname'. */ int assertion_file_birthtime(const char *file, int line, const char *pathname, long t, long nsec) { return assertion_file_time(file, line, pathname, t, nsec, 'b', 0); } /* Verify birthtime of 'pathname' is up-to-date. */ int assertion_file_birthtime_recent(const char *file, int line, const char *pathname) { return assertion_file_time(file, line, pathname, 0, 0, 'b', 1); } /* Verify mode of 'pathname'. */ int assertion_file_mode(const char *file, int line, const char *pathname, int expected_mode) { int mode; int r; assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) failure_start(file, line, "assertFileMode not yet implemented for Windows"); (void)mode; /* UNUSED */ (void)r; /* UNUSED */ #else { struct stat st; r = lstat(pathname, &st); mode = (int)(st.st_mode & 0777); } if (r == 0 && mode == expected_mode) return (1); failure_start(file, line, "File %s has mode %o, expected %o", pathname, mode, expected_mode); #endif failure_finish(NULL); return (0); } /* Verify mtime of 'pathname'. */ int assertion_file_mtime(const char *file, int line, const char *pathname, long t, long nsec) { return assertion_file_time(file, line, pathname, t, nsec, 'm', 0); } /* Verify mtime of 'pathname' is up-to-date. */ int assertion_file_mtime_recent(const char *file, int line, const char *pathname) { return assertion_file_time(file, line, pathname, 0, 0, 'm', 1); } /* Verify number of links to 'pathname'. */ int assertion_file_nlinks(const char *file, int line, const char *pathname, int nlinks) { #if defined(_WIN32) && !defined(__CYGWIN__) BY_HANDLE_FILE_INFORMATION bhfi; int r; assertion_count(file, line); r = my_GetFileInformationByName(pathname, &bhfi); if (r != 0 && bhfi.nNumberOfLinks == (DWORD)nlinks) return (1); failure_start(file, line, "File %s has %d links, expected %d", pathname, bhfi.nNumberOfLinks, nlinks); failure_finish(NULL); return (0); #else struct stat st; int r; assertion_count(file, line); r = lstat(pathname, &st); if (r == 0 && (int)st.st_nlink == nlinks) return (1); failure_start(file, line, "File %s has %d links, expected %d", pathname, st.st_nlink, nlinks); failure_finish(NULL); return (0); #endif } /* Verify size of 'pathname'. */ int assertion_file_size(const char *file, int line, const char *pathname, long size) { int64_t filesize; int r; assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) { BY_HANDLE_FILE_INFORMATION bhfi; r = !my_GetFileInformationByName(pathname, &bhfi); filesize = ((int64_t)bhfi.nFileSizeHigh << 32) + bhfi.nFileSizeLow; } #else { struct stat st; r = lstat(pathname, &st); filesize = st.st_size; } #endif if (r == 0 && filesize == size) return (1); failure_start(file, line, "File %s has size %ld, expected %ld", pathname, (long)filesize, (long)size); failure_finish(NULL); return (0); } /* Assert that 'pathname' is a dir. If mode >= 0, verify that too. */ int assertion_is_dir(const char *file, int line, const char *pathname, int mode) { struct stat st; int r; #if defined(_WIN32) && !defined(__CYGWIN__) (void)mode; /* UNUSED */ #endif assertion_count(file, line); r = lstat(pathname, &st); if (r != 0) { failure_start(file, line, "Dir should exist: %s", pathname); failure_finish(NULL); return (0); } if (!S_ISDIR(st.st_mode)) { failure_start(file, line, "%s is not a dir", pathname); failure_finish(NULL); return (0); } #if !defined(_WIN32) || defined(__CYGWIN__) /* Windows doesn't handle permissions the same way as POSIX, * so just ignore the mode tests. */ /* TODO: Can we do better here? */ if (mode >= 0 && (mode_t)mode != (st.st_mode & 07777)) { failure_start(file, line, "Dir %s has wrong mode", pathname); logprintf(" Expected: 0%3o\n", mode); logprintf(" Found: 0%3o\n", st.st_mode & 07777); failure_finish(NULL); return (0); } #endif return (1); } /* Verify that 'pathname' is a regular file. If 'mode' is >= 0, * verify that too. */ int assertion_is_reg(const char *file, int line, const char *pathname, int mode) { struct stat st; int r; #if defined(_WIN32) && !defined(__CYGWIN__) (void)mode; /* UNUSED */ #endif assertion_count(file, line); r = lstat(pathname, &st); if (r != 0 || !S_ISREG(st.st_mode)) { failure_start(file, line, "File should exist: %s", pathname); failure_finish(NULL); return (0); } #if !defined(_WIN32) || defined(__CYGWIN__) /* Windows doesn't handle permissions the same way as POSIX, * so just ignore the mode tests. */ /* TODO: Can we do better here? */ if (mode >= 0 && (mode_t)mode != (st.st_mode & 07777)) { failure_start(file, line, "File %s has wrong mode", pathname); logprintf(" Expected: 0%3o\n", mode); logprintf(" Found: 0%3o\n", st.st_mode & 07777); failure_finish(NULL); return (0); } #endif return (1); } /* Check whether 'pathname' is a symbolic link. If 'contents' is * non-NULL, verify that the symlink has those contents. */ static int is_symlink(const char *file, int line, const char *pathname, const char *contents) { #if defined(_WIN32) && !defined(__CYGWIN__) (void)pathname; /* UNUSED */ (void)contents; /* UNUSED */ assertion_count(file, line); /* Windows sort-of has real symlinks, but they're only usable * by privileged users and are crippled even then, so there's * really not much point in bothering with this. */ return (0); #else char buff[300]; struct stat st; ssize_t linklen; int r; assertion_count(file, line); r = lstat(pathname, &st); if (r != 0) { failure_start(file, line, "Symlink should exist: %s", pathname); failure_finish(NULL); return (0); } if (!S_ISLNK(st.st_mode)) return (0); if (contents == NULL) return (1); linklen = readlink(pathname, buff, sizeof(buff)); if (linklen < 0) { failure_start(file, line, "Can't read symlink %s", pathname); failure_finish(NULL); return (0); } buff[linklen] = '\0'; if (strcmp(buff, contents) != 0) return (0); return (1); #endif } /* Assert that path is a symlink that (optionally) contains contents. */ int assertion_is_symlink(const char *file, int line, const char *path, const char *contents) { if (is_symlink(file, line, path, contents)) return (1); if (contents) failure_start(file, line, "File %s is not a symlink to %s", path, contents); else failure_start(file, line, "File %s is not a symlink", path); failure_finish(NULL); return (0); } /* Create a directory and report any errors. */ int assertion_make_dir(const char *file, int line, const char *dirname, int mode) { assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) (void)mode; /* UNUSED */ if (0 == _mkdir(dirname)) return (1); #else if (0 == mkdir(dirname, mode)) { if (0 == chmod(dirname, mode)) { assertion_file_mode(file, line, dirname, mode); return (1); } } #endif failure_start(file, line, "Could not create directory %s", dirname); failure_finish(NULL); return(0); } /* Create a file with the specified contents and report any failures. */ int assertion_make_file(const char *file, int line, const char *path, int mode, int csize, const void *contents) { #if defined(_WIN32) && !defined(__CYGWIN__) /* TODO: Rework this to set file mode as well. */ FILE *f; (void)mode; /* UNUSED */ assertion_count(file, line); f = fopen(path, "wb"); if (f == NULL) { failure_start(file, line, "Could not create file %s", path); failure_finish(NULL); return (0); } if (contents != NULL) { size_t wsize; if (csize < 0) wsize = strlen(contents); else wsize = (size_t)csize; if (wsize != fwrite(contents, 1, wsize, f)) { fclose(f); failure_start(file, line, "Could not write file %s", path); failure_finish(NULL); return (0); } } fclose(f); return (1); #else int fd; assertion_count(file, line); fd = open(path, O_CREAT | O_WRONLY, mode >= 0 ? mode : 0644); if (fd < 0) { failure_start(file, line, "Could not create %s", path); failure_finish(NULL); return (0); } if (0 != chmod(path, mode)) { failure_start(file, line, "Could not chmod %s", path); failure_finish(NULL); close(fd); return (0); } if (contents != NULL) { ssize_t wsize; if (csize < 0) wsize = (ssize_t)strlen(contents); else wsize = (ssize_t)csize; if (wsize != write(fd, contents, wsize)) { close(fd); failure_start(file, line, "Could not write to %s", path); failure_finish(NULL); close(fd); return (0); } } close(fd); assertion_file_mode(file, line, path, mode); return (1); #endif } /* Create a hardlink and report any failures. */ int assertion_make_hardlink(const char *file, int line, const char *newpath, const char *linkto) { int succeeded; assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) succeeded = my_CreateHardLinkA(newpath, linkto); #elif HAVE_LINK succeeded = !link(linkto, newpath); #else succeeded = 0; #endif if (succeeded) return (1); failure_start(file, line, "Could not create hardlink"); logprintf(" New link: %s\n", newpath); logprintf(" Old name: %s\n", linkto); failure_finish(NULL); return(0); } /* Create a symlink and report any failures. */ int assertion_make_symlink(const char *file, int line, const char *newpath, const char *linkto) { #if defined(_WIN32) && !defined(__CYGWIN__) int targetIsDir = 0; /* TODO: Fix this */ assertion_count(file, line); if (my_CreateSymbolicLinkA(newpath, linkto, targetIsDir)) return (1); #elif HAVE_SYMLINK assertion_count(file, line); if (0 == symlink(linkto, newpath)) return (1); #endif failure_start(file, line, "Could not create symlink"); logprintf(" New link: %s\n", newpath); logprintf(" Old name: %s\n", linkto); failure_finish(NULL); return(0); } /* Set umask, report failures. */ int assertion_umask(const char *file, int line, int mask) { assertion_count(file, line); (void)file; /* UNUSED */ (void)line; /* UNUSED */ umask(mask); return (1); } /* Set times, report failures. */ int assertion_utimes(const char *file, int line, const char *pathname, long at, long at_nsec, long mt, long mt_nsec) { int r; #if defined(_WIN32) && !defined(__CYGWIN__) #define WINTIME(sec, nsec) ((Int32x32To64(sec, 10000000) + EPOC_TIME)\ + (((nsec)/1000)*10)) HANDLE h; ULARGE_INTEGER wintm; FILETIME fatime, fmtime; FILETIME *pat, *pmt; assertion_count(file, line); h = CreateFileA(pathname,GENERIC_READ | GENERIC_WRITE, FILE_SHARE_READ | FILE_SHARE_WRITE, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); if (h == INVALID_HANDLE_VALUE) { failure_start(file, line, "Can't access %s\n", pathname); failure_finish(NULL); return (0); } if (at > 0 || at_nsec > 0) { wintm.QuadPart = WINTIME(at, at_nsec); fatime.dwLowDateTime = wintm.LowPart; fatime.dwHighDateTime = wintm.HighPart; pat = &fatime; } else pat = NULL; if (mt > 0 || mt_nsec > 0) { wintm.QuadPart = WINTIME(mt, mt_nsec); fmtime.dwLowDateTime = wintm.LowPart; fmtime.dwHighDateTime = wintm.HighPart; pmt = &fmtime; } else pmt = NULL; if (pat != NULL || pmt != NULL) r = SetFileTime(h, NULL, pat, pmt); else r = 1; CloseHandle(h); if (r == 0) { failure_start(file, line, "Can't SetFileTime %s\n", pathname); failure_finish(NULL); return (0); } return (1); #else /* defined(_WIN32) && !defined(__CYGWIN__) */ struct stat st; struct timeval times[2]; #if !defined(__FreeBSD__) mt_nsec = at_nsec = 0; /* Generic POSIX only has whole seconds. */ #endif if (mt == 0 && mt_nsec == 0 && at == 0 && at_nsec == 0) return (1); r = lstat(pathname, &st); if (r < 0) { failure_start(file, line, "Can't stat %s\n", pathname); failure_finish(NULL); return (0); } if (mt == 0 && mt_nsec == 0) { mt = st.st_mtime; #if defined(__FreeBSD__) mt_nsec = st.st_mtimespec.tv_nsec; /* FreeBSD generally only stores to microsecond res, so round. */ mt_nsec = (mt_nsec / 1000) * 1000; #endif } if (at == 0 && at_nsec == 0) { at = st.st_atime; #if defined(__FreeBSD__) at_nsec = st.st_atimespec.tv_nsec; /* FreeBSD generally only stores to microsecond res, so round. */ at_nsec = (at_nsec / 1000) * 1000; #endif } times[1].tv_sec = mt; times[1].tv_usec = mt_nsec / 1000; times[0].tv_sec = at; times[0].tv_usec = at_nsec / 1000; #ifdef HAVE_LUTIMES r = lutimes(pathname, times); #else r = utimes(pathname, times); #endif if (r < 0) { failure_start(file, line, "Can't utimes %s\n", pathname); failure_finish(NULL); return (0); } return (1); #endif /* defined(_WIN32) && !defined(__CYGWIN__) */ } /* Set nodump, report failures. */ int assertion_nodump(const char *file, int line, const char *pathname) { #if defined(HAVE_STRUCT_STAT_ST_FLAGS) && defined(UF_NODUMP) int r; assertion_count(file, line); r = chflags(pathname, UF_NODUMP); if (r < 0) { failure_start(file, line, "Can't set nodump %s\n", pathname); failure_finish(NULL); return (0); } #elif defined(EXT2_IOC_GETFLAGS) && defined(HAVE_WORKING_EXT2_IOC_GETFLAGS)\ && defined(EXT2_NODUMP_FL) int fd, r, flags; assertion_count(file, line); fd = open(pathname, O_RDONLY | O_NONBLOCK); if (fd < 0) { failure_start(file, line, "Can't open %s\n", pathname); failure_finish(NULL); return (0); } r = ioctl(fd, EXT2_IOC_GETFLAGS, &flags); if (r < 0) { failure_start(file, line, "Can't get flags %s\n", pathname); failure_finish(NULL); return (0); } flags |= EXT2_NODUMP_FL; r = ioctl(fd, EXT2_IOC_SETFLAGS, &flags); if (r < 0) { failure_start(file, line, "Can't set nodump %s\n", pathname); failure_finish(NULL); return (0); } close(fd); #else (void)pathname; /* UNUSED */ assertion_count(file, line); #endif return (1); } /* * * UTILITIES for use by tests. * */ /* * Check whether platform supports symlinks. This is intended * for tests to use in deciding whether to bother testing symlink * support; if the platform doesn't support symlinks, there's no point * in checking whether the program being tested can create them. * * Note that the first time this test is called, we actually go out to * disk to create and verify a symlink. This is necessary because * symlink support is actually a property of a particular filesystem * and can thus vary between directories on a single system. After * the first call, this returns the cached result from memory, so it's * safe to call it as often as you wish. */ int canSymlink(void) { /* Remember the test result */ static int value = 0, tested = 0; if (tested) return (value); ++tested; assertion_make_file(__FILE__, __LINE__, "canSymlink.0", 0644, 1, "a"); /* Note: Cygwin has its own symlink() emulation that does not * use the Win32 CreateSymbolicLink() function. */ #if defined(_WIN32) && !defined(__CYGWIN__) value = my_CreateSymbolicLinkA("canSymlink.1", "canSymlink.0", 0) && is_symlink(__FILE__, __LINE__, "canSymlink.1", "canSymlink.0"); #elif HAVE_SYMLINK value = (0 == symlink("canSymlink.0", "canSymlink.1")) && is_symlink(__FILE__, __LINE__, "canSymlink.1","canSymlink.0"); #endif return (value); } /* Platform-dependent options for hiding the output of a subcommand. */ #if defined(_WIN32) && !defined(__CYGWIN__) static const char *redirectArgs = ">NUL 2>NUL"; /* Win32 cmd.exe */ #else static const char *redirectArgs = ">/dev/null 2>/dev/null"; /* POSIX 'sh' */ #endif /* * Can this platform run the bzip2 program? */ int canBzip2(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("bzip2 -d -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the grzip program? */ int canGrzip(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("grzip -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the gzip program? */ int canGzip(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("gzip -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lrzip program? */ int canRunCommand(const char *cmd) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("%s %s", cmd, redirectArgs) == 0) value = 1; } return (value); } int canLrzip(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lrzip -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lz4 program? */ int canLz4(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lz4 -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lzip program? */ int canLzip(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lzip -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lzma program? */ int canLzma(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lzma -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lzop program? */ int canLzop(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lzop -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the xz program? */ int canXz(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("xz -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this filesystem handle nodump flags. */ #if defined(HAVE_STRUCT_STAT_ST_FLAGS) && defined(UF_NODUMP) int canNodump(void) { const char *path = "cannodumptest"; struct stat sb; assertion_make_file(__FILE__, __LINE__, path, 0644, 0, NULL); if (chflags(path, UF_NODUMP) < 0) return (0); if (stat(path, &sb) < 0) return (0); if (sb.st_flags & UF_NODUMP) return (1); return (0); } #elif defined(EXT2_IOC_GETFLAGS) && defined(HAVE_WORKING_EXT2_IOC_GETFLAGS)\ && defined(EXT2_NODUMP_FL) int canNodump(void) { const char *path = "cannodumptest"; int fd, r, flags; assertion_make_file(__FILE__, __LINE__, path, 0644, 0, NULL); fd = open(path, O_RDONLY | O_NONBLOCK); if (fd < 0) return (0); r = ioctl(fd, EXT2_IOC_GETFLAGS, &flags); if (r < 0) return (0); flags |= EXT2_NODUMP_FL; r = ioctl(fd, EXT2_IOC_SETFLAGS, &flags); if (r < 0) return (0); close(fd); fd = open(path, O_RDONLY | O_NONBLOCK); if (fd < 0) return (0); r = ioctl(fd, EXT2_IOC_GETFLAGS, &flags); if (r < 0) return (0); close(fd); if (flags & EXT2_NODUMP_FL) return (1); return (0); } #else int canNodump() { return (0); } #endif /* * Sleep as needed; useful for verifying disk timestamp changes by * ensuring that the wall-clock time has actually changed before we * go back to re-read something from disk. */ void sleepUntilAfter(time_t t) { while (t >= time(NULL)) #if defined(_WIN32) && !defined(__CYGWIN__) Sleep(500); #else sleep(1); #endif } /* * Call standard system() call, but build up the command line using * sprintf() conventions. */ int systemf(const char *fmt, ...) { char buff[8192]; va_list ap; int r; va_start(ap, fmt); vsprintf(buff, fmt, ap); if (verbosity > VERBOSITY_FULL) logprintf("Cmd: %s\n", buff); r = system(buff); va_end(ap); return (r); } /* * Slurp a file into memory for ease of comparison and testing. * Returns size of file in 'sizep' if non-NULL, null-terminates * data in memory for ease of use. */ char * slurpfile(size_t * sizep, const char *fmt, ...) { char filename[8192]; struct stat st; va_list ap; char *p; ssize_t bytes_read; FILE *f; int r; va_start(ap, fmt); vsprintf(filename, fmt, ap); va_end(ap); f = fopen(filename, "rb"); if (f == NULL) { /* Note: No error; non-existent file is okay here. */ return (NULL); } r = fstat(fileno(f), &st); if (r != 0) { logprintf("Can't stat file %s\n", filename); fclose(f); return (NULL); } p = malloc((size_t)st.st_size + 1); if (p == NULL) { logprintf("Can't allocate %ld bytes of memory to read file %s\n", (long int)st.st_size, filename); fclose(f); return (NULL); } bytes_read = fread(p, 1, (size_t)st.st_size, f); if (bytes_read < st.st_size) { logprintf("Can't read file %s\n", filename); fclose(f); free(p); return (NULL); } p[st.st_size] = '\0'; if (sizep != NULL) *sizep = (size_t)st.st_size; fclose(f); return (p); } /* * Slurp a file into memory for ease of comparison and testing. * Returns size of file in 'sizep' if non-NULL, null-terminates * data in memory for ease of use. */ void dumpfile(const char *filename, void *data, size_t len) { ssize_t bytes_written; FILE *f; f = fopen(filename, "wb"); if (f == NULL) { logprintf("Can't open file %s for writing\n", filename); return; } bytes_written = fwrite(data, 1, len, f); if (bytes_written < (ssize_t)len) logprintf("Can't write file %s\n", filename); fclose(f); } /* Read a uuencoded file from the reference directory, decode, and * write the result into the current directory. */ #define VALID_UUDECODE(c) (c >= 32 && c <= 96) #define UUDECODE(c) (((c) - 0x20) & 0x3f) void extract_reference_file(const char *name) { char buff[1024]; FILE *in, *out; sprintf(buff, "%s/%s.uu", refdir, name); in = fopen(buff, "r"); failure("Couldn't open reference file %s", buff); assert(in != NULL); if (in == NULL) return; /* Read up to and including the 'begin' line. */ for (;;) { if (fgets(buff, sizeof(buff), in) == NULL) { /* TODO: This is a failure. */ return; } if (memcmp(buff, "begin ", 6) == 0) break; } /* Now, decode the rest and write it. */ out = fopen(name, "wb"); while (fgets(buff, sizeof(buff), in) != NULL) { char *p = buff; int bytes; if (memcmp(buff, "end", 3) == 0) break; bytes = UUDECODE(*p++); while (bytes > 0) { int n = 0; /* Write out 1-3 bytes from that. */ if (bytes > 0) { assert(VALID_UUDECODE(p[0])); assert(VALID_UUDECODE(p[1])); n = UUDECODE(*p++) << 18; n |= UUDECODE(*p++) << 12; fputc(n >> 16, out); --bytes; } if (bytes > 0) { assert(VALID_UUDECODE(p[0])); n |= UUDECODE(*p++) << 6; fputc((n >> 8) & 0xFF, out); --bytes; } if (bytes > 0) { assert(VALID_UUDECODE(p[0])); n |= UUDECODE(*p++); fputc(n & 0xFF, out); --bytes; } } } fclose(out); fclose(in); } void copy_reference_file(const char *name) { char buff[1024]; FILE *in, *out; size_t rbytes; sprintf(buff, "%s/%s", refdir, name); in = fopen(buff, "rb"); failure("Couldn't open reference file %s", buff); assert(in != NULL); if (in == NULL) return; /* Now, decode the rest and write it. */ /* Not a lot of error checking here; the input better be right. */ out = fopen(name, "wb"); while ((rbytes = fread(buff, 1, sizeof(buff), in)) > 0) { if (fwrite(buff, 1, rbytes, out) != rbytes) { logprintf("Error: fwrite\n"); break; } } fclose(out); fclose(in); } int is_LargeInode(const char *file) { #if defined(_WIN32) && !defined(__CYGWIN__) BY_HANDLE_FILE_INFORMATION bhfi; int r; r = my_GetFileInformationByName(file, &bhfi); if (r != 0) return (0); return (bhfi.nFileIndexHigh & 0x0000FFFFUL); #else struct stat st; int64_t ino; if (stat(file, &st) < 0) return (0); ino = (int64_t)st.st_ino; return (ino > 0xffffffff); #endif } void extract_reference_files(const char **names) { while (names && *names) extract_reference_file(*names++); } /* * * TEST management * */ /* * "list.h" is simply created by "grep DEFINE_TEST test_*.c"; it has * a line like * DEFINE_TEST(test_function) * for each test. */ /* Use "list.h" to declare all of the test functions. */ #undef DEFINE_TEST #define DEFINE_TEST(name) void name(void); #include "list.h" /* Use "list.h" to create a list of all tests (functions and names). */ #undef DEFINE_TEST #define DEFINE_TEST(n) { n, #n, 0 }, struct test_list_t tests[] = { #include "list.h" }; /* * Summarize repeated failures in the just-completed test. */ static void test_summarize(int failed, int skips_num) { unsigned int i; switch (verbosity) { case VERBOSITY_SUMMARY_ONLY: printf(failed ? "E" : "."); fflush(stdout); break; case VERBOSITY_PASSFAIL: printf(failed ? "FAIL\n" : skips_num ? "ok (S)\n" : "ok\n"); break; } log_console = (verbosity == VERBOSITY_LIGHT_REPORT); for (i = 0; i < sizeof(failed_lines)/sizeof(failed_lines[0]); i++) { if (failed_lines[i].count > 1 && !failed_lines[i].skip) logprintf("%s:%d: Summary: Failed %d times\n", failed_filename, i, failed_lines[i].count); } /* Clear the failure history for the next file. */ failed_filename = NULL; memset(failed_lines, 0, sizeof(failed_lines)); } /* * Actually run a single test, with appropriate setup and cleanup. */ static int test_run(int i, const char *tmpdir) { char workdir[1024]; char logfilename[64]; int failures_before = failures; int skips_before = skips; int oldumask; switch (verbosity) { case VERBOSITY_SUMMARY_ONLY: /* No per-test reports at all */ break; case VERBOSITY_PASSFAIL: /* rest of line will include ok/FAIL marker */ printf("%3d: %-64s", i, tests[i].name); fflush(stdout); break; default: /* Title of test, details will follow */ printf("%3d: %s\n", i, tests[i].name); } /* Chdir to the top-level work directory. */ if (!assertChdir(tmpdir)) { fprintf(stderr, "ERROR: Can't chdir to top work dir %s\n", tmpdir); exit(1); } /* Create a log file for this test. */ sprintf(logfilename, "%s.log", tests[i].name); logfile = fopen(logfilename, "w"); fprintf(logfile, "%s\n\n", tests[i].name); /* Chdir() to a work dir for this specific test. */ snprintf(workdir, sizeof(workdir), "%s/%s", tmpdir, tests[i].name); testworkdir = workdir; if (!assertMakeDir(testworkdir, 0755) || !assertChdir(testworkdir)) { fprintf(stderr, "ERROR: Can't chdir to work dir %s\n", testworkdir); exit(1); } /* Explicitly reset the locale before each test. */ setlocale(LC_ALL, "C"); /* Record the umask before we run the test. */ umask(oldumask = umask(0)); /* * Run the actual test. */ (*tests[i].func)(); /* * Clean up and report afterwards. */ testworkdir = NULL; /* Restore umask */ umask(oldumask); /* Reset locale. */ setlocale(LC_ALL, "C"); /* Reset directory. */ if (!assertChdir(tmpdir)) { fprintf(stderr, "ERROR: Couldn't chdir to temp dir %s\n", tmpdir); exit(1); } /* Report per-test summaries. */ tests[i].failures = failures - failures_before; test_summarize(tests[i].failures, skips - skips_before); /* Close the per-test log file. */ fclose(logfile); logfile = NULL; /* If there were no failures, we can remove the work dir and logfile. */ if (tests[i].failures == 0) { if (!keep_temp_files && assertChdir(tmpdir)) { #if defined(_WIN32) && !defined(__CYGWIN__) /* Make sure not to leave empty directories. * Sometimes a processing of closing files used by tests * is not done, then rmdir will be failed and it will * leave a empty test directory. So we should wait a few * seconds and retry rmdir. */ int r, t; for (t = 0; t < 10; t++) { if (t > 0) Sleep(1000); r = systemf("rmdir /S /Q %s", tests[i].name); if (r == 0) break; } systemf("del %s", logfilename); #else systemf("rm -rf %s", tests[i].name); systemf("rm %s", logfilename); #endif } } /* Return appropriate status. */ return (tests[i].failures); } /* * * * MAIN and support routines. * * */ static void usage(const char *program) { static const int limit = sizeof(tests) / sizeof(tests[0]); int i; printf("Usage: %s [options] ...\n", program); printf("Default is to run all tests.\n"); printf("Otherwise, specify the numbers of the tests you wish to run.\n"); printf("Options:\n"); printf(" -d Dump core after any failure, for debugging.\n"); printf(" -k Keep all temp files.\n"); printf(" Default: temp files for successful tests deleted.\n"); #ifdef PROGRAM printf(" -p Path to executable to be tested.\n"); printf(" Default: path taken from " ENVBASE " environment variable.\n"); #endif printf(" -q Quiet.\n"); printf(" -r Path to dir containing reference files.\n"); printf(" Default: Current directory.\n"); printf(" -u Keep running specifies tests until one fails.\n"); printf(" -v Verbose.\n"); printf("Available tests:\n"); for (i = 0; i < limit; i++) printf(" %d: %s\n", i, tests[i].name); exit(1); } static char * get_refdir(const char *d) { size_t tried_size, buff_size; char *buff, *tried, *pwd = NULL, *p = NULL; #ifdef PATH_MAX buff_size = PATH_MAX; #else buff_size = 8192; #endif buff = calloc(buff_size, 1); if (buff == NULL) { fprintf(stderr, "Unable to allocate memory\n"); exit(1); } /* Allocate a buffer to hold the various directories we checked. */ tried_size = buff_size * 2; tried = calloc(tried_size, 1); if (tried == NULL) { fprintf(stderr, "Unable to allocate memory\n"); exit(1); } /* If a dir was specified, try that */ if (d != NULL) { pwd = NULL; snprintf(buff, buff_size, "%s", d); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); goto failure; } /* Get the current dir. */ #ifdef PATH_MAX pwd = getcwd(NULL, PATH_MAX);/* Solaris getcwd needs the size. */ #else pwd = getcwd(NULL, 0); #endif while (pwd[strlen(pwd) - 1] == '\n') pwd[strlen(pwd) - 1] = '\0'; /* Look for a known file. */ snprintf(buff, buff_size, "%s", pwd); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); snprintf(buff, buff_size, "%s/test", pwd); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); #if defined(LIBRARY) snprintf(buff, buff_size, "%s/%s/test", pwd, LIBRARY); #else snprintf(buff, buff_size, "%s/%s/test", pwd, PROGRAM); #endif p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); #if defined(PROGRAM_ALIAS) snprintf(buff, buff_size, "%s/%s/test", pwd, PROGRAM_ALIAS); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); #endif if (memcmp(pwd, "/usr/obj", 8) == 0) { snprintf(buff, buff_size, "%s", pwd + 8); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); snprintf(buff, buff_size, "%s/test", pwd + 8); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); } failure: printf("Unable to locate known reference file %s\n", KNOWNREF); printf(" Checked following directories:\n%s\n", tried); printf("Use -r option to specify full path to reference directory\n"); #if defined(_WIN32) && !defined(__CYGWIN__) && defined(_DEBUG) DebugBreak(); #endif exit(1); success: free(p); free(pwd); free(tried); /* Copy result into a fresh buffer to reduce memory usage. */ p = strdup(buff); free(buff); return p; } int main(int argc, char **argv) { static const int limit = sizeof(tests) / sizeof(tests[0]); int test_set[sizeof(tests) / sizeof(tests[0])]; int i = 0, j = 0, tests_run = 0, tests_failed = 0, option; time_t now; char *refdir_alloc = NULL; const char *progname; char **saved_argv; const char *tmp, *option_arg, *p; char tmpdir[256], *pwd, *testprogdir, *tmp2 = NULL, *vlevel = NULL; char tmpdir_timestamp[256]; (void)argc; /* UNUSED */ /* Get the current dir. */ #ifdef PATH_MAX pwd = getcwd(NULL, PATH_MAX);/* Solaris getcwd needs the size. */ #else pwd = getcwd(NULL, 0); #endif while (pwd[strlen(pwd) - 1] == '\n') pwd[strlen(pwd) - 1] = '\0'; #if defined(HAVE__CrtSetReportMode) && !defined(__WATCOMC__) /* To stop to run the default invalid parameter handler. */ _set_invalid_parameter_handler(invalid_parameter_handler); /* Disable annoying assertion message box. */ _CrtSetReportMode(_CRT_ASSERT, 0); #endif /* * Name of this program, used to build root of our temp directory * tree. */ progname = p = argv[0]; if ((testprogdir = (char *)malloc(strlen(progname) + 1)) == NULL) { fprintf(stderr, "ERROR: Out of memory."); exit(1); } strcpy(testprogdir, progname); while (*p != '\0') { /* Support \ or / dir separators for Windows compat. */ if (*p == '/' || *p == '\\') { progname = p + 1; i = j; } ++p; j++; } testprogdir[i] = '\0'; #if defined(_WIN32) && !defined(__CYGWIN__) if (testprogdir[0] != '/' && testprogdir[0] != '\\' && !(((testprogdir[0] >= 'a' && testprogdir[0] <= 'z') || (testprogdir[0] >= 'A' && testprogdir[0] <= 'Z')) && testprogdir[1] == ':' && (testprogdir[2] == '/' || testprogdir[2] == '\\'))) #else if (testprogdir[0] != '/') #endif { /* Fixup path for relative directories. */ if ((testprogdir = (char *)realloc(testprogdir, strlen(pwd) + 1 + strlen(testprogdir) + 1)) == NULL) { fprintf(stderr, "ERROR: Out of memory."); exit(1); } memmove(testprogdir + strlen(pwd) + 1, testprogdir, strlen(testprogdir) + 1); memcpy(testprogdir, pwd, strlen(pwd)); testprogdir[strlen(pwd)] = '/'; } #ifdef PROGRAM /* Get the target program from environment, if available. */ testprogfile = getenv(ENVBASE); #endif if (getenv("TMPDIR") != NULL) tmp = getenv("TMPDIR"); else if (getenv("TMP") != NULL) tmp = getenv("TMP"); else if (getenv("TEMP") != NULL) tmp = getenv("TEMP"); else if (getenv("TEMPDIR") != NULL) tmp = getenv("TEMPDIR"); else tmp = "/tmp"; /* Allow -d to be controlled through the environment. */ if (getenv(ENVBASE "_DEBUG") != NULL) dump_on_failure = 1; /* Allow -v to be controlled through the environment. */ if (getenv("_VERBOSITY_LEVEL") != NULL) { vlevel = getenv("_VERBOSITY_LEVEL"); verbosity = atoi(vlevel); if (verbosity < VERBOSITY_SUMMARY_ONLY || verbosity > VERBOSITY_FULL) { /* Unsupported verbosity levels are silently ignored */ vlevel = NULL; verbosity = VERBOSITY_PASSFAIL; } } /* Get the directory holding test files from environment. */ refdir = getenv(ENVBASE "_TEST_FILES"); /* * Parse options, without using getopt(), which isn't available * on all platforms. */ ++argv; /* Skip program name */ while (*argv != NULL) { if (**argv != '-') break; p = *argv++; ++p; /* Skip '-' */ while (*p != '\0') { option = *p++; option_arg = NULL; /* If 'opt' takes an argument, parse that. */ if (option == 'p' || option == 'r') { if (*p != '\0') option_arg = p; else if (*argv == NULL) { fprintf(stderr, "Option -%c requires argument.\n", option); usage(progname); } else option_arg = *argv++; p = ""; /* End of this option word. */ } /* Now, handle the option. */ switch (option) { case 'd': dump_on_failure = 1; break; case 'k': keep_temp_files = 1; break; case 'p': #ifdef PROGRAM testprogfile = option_arg; #else fprintf(stderr, "-p option not permitted\n"); usage(progname); #endif break; case 'q': if (!vlevel) verbosity--; break; case 'r': refdir = option_arg; break; case 'u': until_failure++; break; case 'v': if (!vlevel) verbosity++; break; default: fprintf(stderr, "Unrecognized option '%c'\n", option); usage(progname); } } } /* * Sanity-check that our options make sense. */ #ifdef PROGRAM if (testprogfile == NULL) { if ((tmp2 = (char *)malloc(strlen(testprogdir) + 1 + strlen(PROGRAM) + 1)) == NULL) { fprintf(stderr, "ERROR: Out of memory."); exit(1); } strcpy(tmp2, testprogdir); strcat(tmp2, "/"); strcat(tmp2, PROGRAM); testprogfile = tmp2; } { char *testprg; #if defined(_WIN32) && !defined(__CYGWIN__) /* Command.com sometimes rejects '/' separators. */ testprg = strdup(testprogfile); for (i = 0; testprg[i] != '\0'; i++) { if (testprg[i] == '/') testprg[i] = '\\'; } testprogfile = testprg; #endif /* Quote the name that gets put into shell command lines. */ testprg = malloc(strlen(testprogfile) + 3); strcpy(testprg, "\""); strcat(testprg, testprogfile); strcat(testprg, "\""); testprog = testprg; } #endif #if !defined(_WIN32) && defined(SIGPIPE) { /* Ignore SIGPIPE signals */ struct sigaction sa; sa.sa_handler = SIG_IGN; sigemptyset(&sa.sa_mask); sa.sa_flags = 0; sigaction(SIGPIPE, &sa, NULL); } #endif /* * Create a temp directory for the following tests. * Include the time the tests started as part of the name, * to make it easier to track the results of multiple tests. */ now = time(NULL); for (i = 0; ; i++) { strftime(tmpdir_timestamp, sizeof(tmpdir_timestamp), "%Y-%m-%dT%H.%M.%S", localtime(&now)); sprintf(tmpdir, "%s/%s.%s-%03d", tmp, progname, tmpdir_timestamp, i); if (assertMakeDir(tmpdir,0755)) break; if (i >= 999) { fprintf(stderr, "ERROR: Unable to create temp directory %s\n", tmpdir); exit(1); } } /* * If the user didn't specify a directory for locating * reference files, try to find the reference files in * the "usual places." */ refdir = refdir_alloc = get_refdir(refdir); /* * Banner with basic information. */ printf("\n"); printf("If tests fail or crash, details will be in:\n"); printf(" %s\n", tmpdir); printf("\n"); if (verbosity > VERBOSITY_SUMMARY_ONLY) { printf("Reference files will be read from: %s\n", refdir); #ifdef PROGRAM printf("Running tests on: %s\n", testprog); #endif printf("Exercising: "); fflush(stdout); printf("%s\n", EXTRA_VERSION); } else { printf("Running "); fflush(stdout); } /* * Run some or all of the individual tests. */ saved_argv = argv; do { argv = saved_argv; do { int test_num; test_num = get_test_set(test_set, limit, *argv, tests); if (test_num < 0) { printf("*** INVALID Test %s\n", *argv); free(refdir_alloc); free(testprogdir); usage(progname); return (1); } for (i = 0; i < test_num; i++) { tests_run++; if (test_run(test_set[i], tmpdir)) { tests_failed++; if (until_failure) goto finish; } } if (*argv != NULL) argv++; } while (*argv != NULL); } while (until_failure); finish: /* Must be freed after all tests run */ free(tmp2); free(testprogdir); free(pwd); /* * Report summary statistics. */ if (verbosity > VERBOSITY_SUMMARY_ONLY) { printf("\n"); printf("Totals:\n"); printf(" Tests run: %8d\n", tests_run); printf(" Tests failed: %8d\n", tests_failed); printf(" Assertions checked:%8d\n", assertions); printf(" Assertions failed: %8d\n", failures); printf(" Skips reported: %8d\n", skips); } if (failures) { printf("\n"); printf("Failing tests:\n"); for (i = 0; i < limit; ++i) { if (tests[i].failures) printf(" %d: %s (%d failures)\n", i, tests[i].name, tests[i].failures); } printf("\n"); printf("Details for failing tests: %s\n", tmpdir); printf("\n"); } else { if (verbosity == VERBOSITY_SUMMARY_ONLY) printf("\n"); printf("%d tests passed, no failures\n", tests_run); } free(refdir_alloc); /* If the final tmpdir is empty, we can remove it. */ /* This should be the usual case when all tests succeed. */ assertChdir(".."); rmdir(tmpdir); return (tests_failed ? 1 : 0); } Index: head/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c (revision 309590) @@ -1,1393 +1,1393 @@ /*- * Copyright (c) 2003-2009 Tim Kientzle * Copyright (c) 2010-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 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 /* Mac OSX requires sys/types.h before sys/acl.h. */ #include #endif #ifdef HAVE_SYS_ACL_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_ACL_LIBACL_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 /* * Linux and FreeBSD plug this obvious hole in POSIX.1e in * different ways. */ #if HAVE_ACL_GET_PERM #define ACL_GET_PERM acl_get_perm #elif HAVE_ACL_GET_PERM_NP #define ACL_GET_PERM acl_get_perm_np #endif static int setup_acls(struct archive_read_disk *, struct archive_entry *, int *fd); 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 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_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 (st->st_flags != 0) archive_entry_set_fflags(entry, st->st_flags, 0); #endif #if 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 ((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, EXT2_IOC_GETFLAGS, &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 = setup_acls(a, entry, &fd); if (!a->suppress_xattr) { r1 = setup_xattrs(a, entry, &fd); if (r1 < r) r = r1; } if (a->enable_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_entry_sourcepath(entry); if (name == NULL) name = archive_entry_pathname(entry); if (name == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't open file to read extended attributes: No name"); return (ARCHIVE_WARN); } if (a->tree != NULL) { if (a->tree_enter_working_dir(a->tree) != 0) { archive_set_error(&a->archive, errno, "Couldn't change dir"); return (ARCHIVE_FAILED); } } /* 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 #ifdef HAVE_POSIX_ACL static int translate_acl(struct archive_read_disk *a, struct archive_entry *entry, acl_t acl, int archive_entry_acl_type); static int setup_acls(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { const char *accpath; acl_t acl; int r; accpath = archive_entry_sourcepath(entry); if (accpath == NULL) accpath = archive_entry_pathname(entry); if (*fd < 0 && a->tree != NULL) { if (a->follow_symlinks || archive_entry_filetype(entry) != AE_IFLNK) *fd = a->open_on_current_dir(a->tree, accpath, O_RDONLY | O_NONBLOCK); if (*fd < 0) { if (a->tree_enter_working_dir(a->tree) != 0) { archive_set_error(&a->archive, errno, "Couldn't access %s", accpath); return (ARCHIVE_FAILED); } } } archive_entry_acl_clear(entry); acl = NULL; #ifdef ACL_TYPE_NFS4 /* Try NFS4 ACL first. */ if (*fd >= 0) #if HAVE_ACL_GET_FD_NP acl = acl_get_fd_np(*fd, ACL_TYPE_NFS4); #else acl = acl_get_fd(*fd); #endif #if HAVE_ACL_GET_LINK_NP else if (!a->follow_symlinks) acl = acl_get_link_np(accpath, ACL_TYPE_NFS4); #else else if ((!a->follow_symlinks) && (archive_entry_filetype(entry) == AE_IFLNK)) /* We can't get the ACL of a symlink, so we assume it can't have one. */ acl = NULL; #endif else acl = acl_get_file(accpath, ACL_TYPE_NFS4); #if HAVE_ACL_IS_TRIVIAL_NP if (acl != NULL && acl_is_trivial_np(acl, &r) == 0) { /* Ignore "trivial" ACLs that just mirror the file mode. */ if (r) { acl_free(acl); acl = NULL; /* * Simultaneous NFSv4 and POSIX.1e ACLs for the same * entry are not allowed, so we should return here */ return (ARCHIVE_OK); } } #endif if (acl != NULL) { r = translate_acl(a, entry, acl, ARCHIVE_ENTRY_ACL_TYPE_NFS4); acl_free(acl); if (r != ARCHIVE_OK) { archive_set_error(&a->archive, errno, "Couldn't translate NFSv4 ACLs: %s", accpath); } return (r); } #endif /* ACL_TYPE_NFS4 */ /* Retrieve access ACL from file. */ if (*fd >= 0) acl = acl_get_fd(*fd); #if HAVE_ACL_GET_LINK_NP else if (!a->follow_symlinks) acl = acl_get_link_np(accpath, ACL_TYPE_ACCESS); #else else if ((!a->follow_symlinks) && (archive_entry_filetype(entry) == AE_IFLNK)) /* We can't get the ACL of a symlink, so we assume it can't have one. */ acl = NULL; #endif else acl = acl_get_file(accpath, ACL_TYPE_ACCESS); #if HAVE_ACL_IS_TRIVIAL_NP /* Ignore "trivial" ACLs that just mirror the file mode. */ if (acl != NULL && acl_is_trivial_np(acl, &r) == 0) { if (r) { acl_free(acl); acl = NULL; } } #endif if (acl != NULL) { r = translate_acl(a, entry, acl, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); acl_free(acl); acl = NULL; if (r != ARCHIVE_OK) { archive_set_error(&a->archive, errno, "Couldn't translate access ACLs: %s", accpath); return (r); } } /* Only directories can have default ACLs. */ if (S_ISDIR(archive_entry_mode(entry))) { acl = acl_get_file(accpath, ACL_TYPE_DEFAULT); if (acl != NULL) { r = translate_acl(a, entry, acl, ARCHIVE_ENTRY_ACL_TYPE_DEFAULT); acl_free(acl); if (r != ARCHIVE_OK) { archive_set_error(&a->archive, errno, "Couldn't translate default ACLs: %s", accpath); return (r); } } } return (ARCHIVE_OK); } /* * Translate system ACL into libarchive internal structure. */ static struct { int archive_perm; int platform_perm; } acl_perm_map[] = { {ARCHIVE_ENTRY_ACL_EXECUTE, ACL_EXECUTE}, {ARCHIVE_ENTRY_ACL_WRITE, ACL_WRITE}, {ARCHIVE_ENTRY_ACL_READ, ACL_READ}, #ifdef ACL_TYPE_NFS4 {ARCHIVE_ENTRY_ACL_READ_DATA, ACL_READ_DATA}, {ARCHIVE_ENTRY_ACL_LIST_DIRECTORY, ACL_LIST_DIRECTORY}, {ARCHIVE_ENTRY_ACL_WRITE_DATA, ACL_WRITE_DATA}, {ARCHIVE_ENTRY_ACL_ADD_FILE, ACL_ADD_FILE}, {ARCHIVE_ENTRY_ACL_APPEND_DATA, ACL_APPEND_DATA}, {ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY, ACL_ADD_SUBDIRECTORY}, {ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS, ACL_READ_NAMED_ATTRS}, {ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS, ACL_WRITE_NAMED_ATTRS}, {ARCHIVE_ENTRY_ACL_DELETE_CHILD, ACL_DELETE_CHILD}, {ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES, ACL_READ_ATTRIBUTES}, {ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES, ACL_WRITE_ATTRIBUTES}, {ARCHIVE_ENTRY_ACL_DELETE, ACL_DELETE}, {ARCHIVE_ENTRY_ACL_READ_ACL, ACL_READ_ACL}, {ARCHIVE_ENTRY_ACL_WRITE_ACL, ACL_WRITE_ACL}, {ARCHIVE_ENTRY_ACL_WRITE_OWNER, ACL_WRITE_OWNER}, {ARCHIVE_ENTRY_ACL_SYNCHRONIZE, ACL_SYNCHRONIZE} #endif }; #ifdef ACL_TYPE_NFS4 static struct { int archive_inherit; int platform_inherit; } acl_inherit_map[] = { {ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, ACL_ENTRY_FILE_INHERIT}, {ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT, ACL_ENTRY_DIRECTORY_INHERIT}, {ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT, ACL_ENTRY_NO_PROPAGATE_INHERIT}, {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ACL_ENTRY_INHERIT_ONLY} }; #endif static int translate_acl(struct archive_read_disk *a, struct archive_entry *entry, acl_t acl, int default_entry_acl_type) { acl_tag_t acl_tag; #ifdef ACL_TYPE_NFS4 acl_entry_type_t acl_type; acl_flagset_t acl_flagset; int brand; #endif acl_entry_t acl_entry; acl_permset_t acl_permset; int i, entry_acl_type; int r, s, ae_id, ae_tag, ae_perm; const char *ae_name; #ifdef ACL_TYPE_NFS4 // FreeBSD "brands" ACLs as POSIX.1e or NFSv4 // Make sure the "brand" on this ACL is consistent // with the default_entry_acl_type bits provided. if (acl_get_brand_np(acl, &brand) != 0) { archive_set_error(&a->archive, errno, "Failed to read ACL brand"); return (ARCHIVE_WARN); } switch (brand) { case ACL_BRAND_POSIX: switch (default_entry_acl_type) { case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid ACL entry type for POSIX.1e ACL"); return (ARCHIVE_WARN); } break; case ACL_BRAND_NFS4: if (default_entry_acl_type & ~ARCHIVE_ENTRY_ACL_TYPE_NFS4) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid ACL entry type for NFSv4 ACL"); return (ARCHIVE_WARN); } break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unknown ACL brand"); return (ARCHIVE_WARN); } #endif s = acl_get_entry(acl, ACL_FIRST_ENTRY, &acl_entry); if (s == -1) { archive_set_error(&a->archive, errno, "Failed to get first ACL entry"); return (ARCHIVE_WARN); } while (s == 1) { ae_id = -1; ae_name = NULL; ae_perm = 0; if (acl_get_tag_type(acl_entry, &acl_tag) != 0) { archive_set_error(&a->archive, errno, "Failed to get ACL tag type"); return (ARCHIVE_WARN); } switch (acl_tag) { case ACL_USER: ae_id = (int)*(uid_t *)acl_get_qualifier(acl_entry); ae_name = archive_read_disk_uname(&a->archive, ae_id); ae_tag = ARCHIVE_ENTRY_ACL_USER; break; case ACL_GROUP: ae_id = (int)*(gid_t *)acl_get_qualifier(acl_entry); ae_name = archive_read_disk_gname(&a->archive, ae_id); ae_tag = ARCHIVE_ENTRY_ACL_GROUP; break; case ACL_MASK: ae_tag = ARCHIVE_ENTRY_ACL_MASK; break; case ACL_USER_OBJ: ae_tag = ARCHIVE_ENTRY_ACL_USER_OBJ; break; case ACL_GROUP_OBJ: ae_tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; break; case ACL_OTHER: ae_tag = ARCHIVE_ENTRY_ACL_OTHER; break; #ifdef ACL_TYPE_NFS4 case ACL_EVERYONE: ae_tag = ARCHIVE_ENTRY_ACL_EVERYONE; break; #endif default: /* Skip types that libarchive can't support. */ s = acl_get_entry(acl, ACL_NEXT_ENTRY, &acl_entry); continue; } // XXX acl_type maps to allow/deny/audit/YYYY bits entry_acl_type = default_entry_acl_type; #ifdef ACL_TYPE_NFS4 if (default_entry_acl_type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) { /* - * acl_get_entry_type_np() falis with non-NFSv4 ACLs + * acl_get_entry_type_np() fails with non-NFSv4 ACLs */ if (acl_get_entry_type_np(acl_entry, &acl_type) != 0) { archive_set_error(&a->archive, errno, "Failed " "to get ACL type from a NFSv4 ACL entry"); return (ARCHIVE_WARN); } switch (acl_type) { case ACL_ENTRY_TYPE_ALLOW: entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_ALLOW; break; case ACL_ENTRY_TYPE_DENY: entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_DENY; break; case ACL_ENTRY_TYPE_AUDIT: entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_AUDIT; break; case ACL_ENTRY_TYPE_ALARM: entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_ALARM; break; default: archive_set_error(&a->archive, errno, "Invalid NFSv4 ACL entry type"); return (ARCHIVE_WARN); } /* * Libarchive stores "flag" (NFSv4 inheritance bits) * in the ae_perm bitmap. * * acl_get_flagset_np() fails with non-NFSv4 ACLs */ if (acl_get_flagset_np(acl_entry, &acl_flagset) != 0) { archive_set_error(&a->archive, errno, "Failed to get flagset from a NFSv4 ACL entry"); return (ARCHIVE_WARN); } for (i = 0; i < (int)(sizeof(acl_inherit_map) / sizeof(acl_inherit_map[0])); ++i) { r = acl_get_flag_np(acl_flagset, acl_inherit_map[i].platform_inherit); if (r == -1) { archive_set_error(&a->archive, errno, "Failed to check flag in a NFSv4 " "ACL flagset"); return (ARCHIVE_WARN); } else if (r) ae_perm |= acl_inherit_map[i].archive_inherit; } } #endif if (acl_get_permset(acl_entry, &acl_permset) != 0) { archive_set_error(&a->archive, errno, "Failed to get ACL permission set"); return (ARCHIVE_WARN); } for (i = 0; i < (int)(sizeof(acl_perm_map) / sizeof(acl_perm_map[0])); ++i) { /* * acl_get_perm() is spelled differently on different * platforms; see above. */ r = ACL_GET_PERM(acl_permset, acl_perm_map[i].platform_perm); if (r == -1) { archive_set_error(&a->archive, errno, "Failed to check permission in an ACL permission set"); return (ARCHIVE_WARN); } else if (r) ae_perm |= acl_perm_map[i].archive_perm; } archive_entry_acl_add_entry(entry, entry_acl_type, ae_perm, ae_tag, ae_id, ae_name); s = acl_get_entry(acl, ACL_NEXT_ENTRY, &acl_entry); if (s == -1) { archive_set_error(&a->archive, errno, "Failed to get next ACL entry"); return (ARCHIVE_WARN); } } return (ARCHIVE_OK); } #else static int 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 #if (HAVE_FGETXATTR && HAVE_FLISTXATTR && HAVE_LISTXATTR && \ HAVE_LLISTXATTR && HAVE_GETXATTR && HAVE_LGETXATTR) || \ (HAVE_FGETEA && HAVE_FLISTEA && HAVE_LISTEA) /* * Linux 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) { ssize_t size; void *value = NULL; const char *accpath; accpath = archive_entry_sourcepath(entry); if (accpath == NULL) accpath = archive_entry_pathname(entry); #if HAVE_FGETXATTR if (fd >= 0) size = fgetxattr(fd, name, NULL, 0); else if (!a->follow_symlinks) size = lgetxattr(accpath, name, NULL, 0); else size = getxattr(accpath, name, NULL, 0); #elif HAVE_FGETEA if (fd >= 0) size = fgetea(fd, name, NULL, 0); else if (!a->follow_symlinks) size = lgetea(accpath, name, NULL, 0); else 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 HAVE_FGETXATTR if (fd >= 0) size = fgetxattr(fd, name, value, size); else if (!a->follow_symlinks) size = lgetxattr(accpath, name, value, size); else size = getxattr(accpath, name, value, size); #elif HAVE_FGETEA if (fd >= 0) size = fgetea(fd, name, value, size); else if (!a->follow_symlinks) size = lgetea(accpath, name, value, size); else 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 = archive_entry_sourcepath(entry); if (path == NULL) path = archive_entry_pathname(entry); if (*fd < 0 && a->tree != NULL) { if (a->follow_symlinks || archive_entry_filetype(entry) != AE_IFLNK) *fd = a->open_on_current_dir(a->tree, path, O_RDONLY | O_NONBLOCK); if (*fd < 0) { if (a->tree_enter_working_dir(a->tree) != 0) { archive_set_error(&a->archive, errno, "Couldn't access %s", path); return (ARCHIVE_FAILED); } } } #if HAVE_FLISTXATTR if (*fd >= 0) list_size = flistxattr(*fd, NULL, 0); else if (!a->follow_symlinks) list_size = llistxattr(path, NULL, 0); else list_size = listxattr(path, NULL, 0); #elif HAVE_FLISTEA if (*fd >= 0) list_size = flistea(*fd, NULL, 0); else if (!a->follow_symlinks) list_size = llistea(path, NULL, 0); else 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 HAVE_FLISTXATTR if (*fd >= 0) list_size = flistxattr(*fd, list, list_size); else if (!a->follow_symlinks) list_size = llistxattr(path, list, list_size); else list_size = listxattr(path, list, list_size); #elif HAVE_FLISTEA if (*fd >= 0) list_size = flistea(*fd, list, list_size); else if (!a->follow_symlinks) list_size = llistea(path, list, list_size); else 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 (strncmp(p, "system.", 7) == 0 || strncmp(p, "xfsroot.", 8) == 0) continue; setup_xattr(a, entry, p, *fd); } free(list); return (ARCHIVE_OK); } #elif HAVE_EXTATTR_GET_FILE && HAVE_EXTATTR_LIST_FILE && \ HAVE_DECL_EXTATTR_NAMESPACE_USER /* * 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); static int setup_xattr(struct archive_read_disk *a, struct archive_entry *entry, int namespace, const char *name, const char *fullname, int fd) { ssize_t size; void *value = NULL; const char *accpath; accpath = archive_entry_sourcepath(entry); if (accpath == NULL) accpath = archive_entry_pathname(entry); 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 = archive_entry_sourcepath(entry); if (path == NULL) path = archive_entry_pathname(entry); if (*fd < 0 && a->tree != NULL) { if (a->follow_symlinks || archive_entry_filetype(entry) != AE_IFLNK) *fd = a->open_on_current_dir(a->tree, path, O_RDONLY | O_NONBLOCK); if (*fd < 0) { if (a->tree_enter_working_dir(a->tree) != 0) { archive_set_error(&a->archive, errno, "Couldn't access %s", path); return (ARCHIVE_FAILED); } } } 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); 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; if (archive_entry_filetype(entry) != AE_IFREG || archive_entry_size(entry) <= 0 || archive_entry_hardlink(entry) != NULL) return (ARCHIVE_OK); if (*fd < 0) { const char *path; path = archive_entry_sourcepath(entry); if (path == NULL) path = archive_entry_pathname(entry); 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 perfom FS_IOC_FIEMAP. */ + * 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; 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 && a->tree != NULL) { const char *path; path = archive_entry_sourcepath(entry); if (path == NULL) path = archive_entry_pathname(entry); *fd = a->open_on_current_dir(a->tree, path, O_RDONLY | O_NONBLOCK); if (*fd < 0) { archive_set_error(&a->archive, errno, "Can't open `%s'", path); return (ARCHIVE_FAILED); } } 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 { const char *path; path = archive_entry_sourcepath(entry); if (path == NULL) path = archive_entry_pathname(entry); #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 spase. */ + 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_filter_uu.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_filter_uu.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_read_support_filter_uu.c (revision 309590) @@ -1,696 +1,703 @@ /*- * Copyright (c) 2009-2011 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_private.h" #include "archive_read_private.h" /* Maximum lookahead during bid phase */ #define UUENCODE_BID_MAX_READ 128*1024 /* in bytes */ struct uudecode { int64_t total; unsigned char *in_buff; #define IN_BUFF_SIZE (1024) int in_cnt; size_t in_allocated; unsigned char *out_buff; #define OUT_BUFF_SIZE (64 * 1024) int state; #define ST_FIND_HEAD 0 #define ST_READ_UU 1 #define ST_UUEND 2 #define ST_READ_BASE64 3 #define ST_IGNORE 4 }; static int uudecode_bidder_bid(struct archive_read_filter_bidder *, struct archive_read_filter *filter); static int uudecode_bidder_init(struct archive_read_filter *); static ssize_t uudecode_filter_read(struct archive_read_filter *, const void **); static int uudecode_filter_close(struct archive_read_filter *); #if ARCHIVE_VERSION_NUMBER < 4000000 /* Deprecated; remove in libarchive 4.0 */ int archive_read_support_compression_uu(struct archive *a) { return archive_read_support_filter_uu(a); } #endif int archive_read_support_filter_uu(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct archive_read_filter_bidder *bidder; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_filter_uu"); if (__archive_read_get_bidder(a, &bidder) != ARCHIVE_OK) return (ARCHIVE_FATAL); bidder->data = NULL; bidder->name = "uu"; bidder->bid = uudecode_bidder_bid; bidder->init = uudecode_bidder_init; bidder->options = NULL; bidder->free = NULL; return (ARCHIVE_OK); } static const unsigned char ascii[256] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, '\n', 0, 0, '\r', 0, 0, /* 00 - 0F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 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, 0, /* 70 - 7F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ }; static const unsigned char uuchar[256] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ 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, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 60 - 6F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 70 - 7F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ }; static const unsigned char base64[256] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, /* 20 - 2F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, /* 30 - 3F */ 0, 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, 0, 0, 0, 0, 0, /* 50 - 5F */ 0, 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, 0, 0, 0, 0, 0, /* 70 - 7F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ }; static const int base64num[128] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 62, 0, 0, 0, 63, /* 20 - 2F */ 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 0, 0, 0, 0, 0, 0, /* 30 - 3F */ 0, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, /* 40 - 4F */ 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 0, 0, 0, 0, 0, /* 50 - 5F */ 0, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, /* 60 - 6F */ 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 0, 0, 0, 0, 0, /* 70 - 7F */ }; static ssize_t get_line(const unsigned char *b, ssize_t avail, ssize_t *nlsize) { ssize_t len; len = 0; while (len < avail) { switch (ascii[*b]) { case 0: /* Non-ascii character or control character. */ if (nlsize != NULL) *nlsize = 0; return (-1); case '\r': if (avail-len > 1 && b[1] == '\n') { if (nlsize != NULL) *nlsize = 2; return (len+2); } /* FALL THROUGH */ case '\n': if (nlsize != NULL) *nlsize = 1; return (len+1); case 1: b++; len++; break; } } if (nlsize != NULL) *nlsize = 0; return (avail); } static ssize_t bid_get_line(struct archive_read_filter *filter, const unsigned char **b, ssize_t *avail, ssize_t *ravail, ssize_t *nl, size_t* nbytes_read) { ssize_t len; int quit; quit = 0; if (*avail == 0) { *nl = 0; len = 0; } else len = get_line(*b, *avail, nl); /* * Read bytes more while it does not reach the end of line. */ while (*nl == 0 && len == *avail && !quit && *nbytes_read < UUENCODE_BID_MAX_READ) { ssize_t diff = *ravail - *avail; size_t nbytes_req = (*ravail+1023) & ~1023U; ssize_t tested; /* Increase reading bytes if it is not enough to at least * new two lines. */ if (nbytes_req < (size_t)*ravail + 160) nbytes_req <<= 1; *b = __archive_read_filter_ahead(filter, nbytes_req, avail); if (*b == NULL) { if (*ravail >= *avail) return (0); /* Reading bytes reaches the end of a stream. */ *b = __archive_read_filter_ahead(filter, *avail, avail); quit = 1; } *nbytes_read = *avail; *ravail = *avail; *b += diff; *avail -= diff; tested = len;/* Skip some bytes we already determinated. */ len = get_line(*b + tested, *avail - tested, nl); if (len >= 0) len += tested; } return (len); } #define UUDECODE(c) (((c) - 0x20) & 0x3f) static int uudecode_bidder_bid(struct archive_read_filter_bidder *self, struct archive_read_filter *filter) { const unsigned char *b; ssize_t avail, ravail; ssize_t len, nl; int l; int firstline; size_t nbytes_read; (void)self; /* UNUSED */ b = __archive_read_filter_ahead(filter, 1, &avail); if (b == NULL) return (0); firstline = 20; ravail = avail; nbytes_read = avail; for (;;) { len = bid_get_line(filter, &b, &avail, &ravail, &nl, &nbytes_read); if (len < 0 || nl == 0) return (0); /* No match found. */ if (len - nl >= 11 && memcmp(b, "begin ", 6) == 0) l = 6; else if (len -nl >= 18 && memcmp(b, "begin-base64 ", 13) == 0) l = 13; else l = 0; if (l > 0 && (b[l] < '0' || b[l] > '7' || b[l+1] < '0' || b[l+1] > '7' || b[l+2] < '0' || b[l+2] > '7' || b[l+3] != ' ')) l = 0; b += len; avail -= len; if (l) break; firstline = 0; /* Do not read more than UUENCODE_BID_MAX_READ bytes */ if (nbytes_read >= UUENCODE_BID_MAX_READ) return (0); } if (!avail) return (0); len = bid_get_line(filter, &b, &avail, &ravail, &nl, &nbytes_read); if (len < 0 || nl == 0) return (0);/* There are non-ascii characters. */ avail -= len; if (l == 6) { /* "begin " */ if (!uuchar[*b]) return (0); /* Get a length of decoded bytes. */ l = UUDECODE(*b++); len--; if (l > 45) /* Normally, maximum length is 45(character 'M'). */ return (0); while (l && len-nl > 0) { if (l > 0) { if (!uuchar[*b++]) return (0); if (!uuchar[*b++]) return (0); len -= 2; --l; } if (l > 0) { if (!uuchar[*b++]) return (0); --len; --l; } if (l > 0) { if (!uuchar[*b++]) return (0); --len; --l; } } if (len-nl < 0) return (0); if (len-nl == 1 && (uuchar[*b] || /* Check sum. */ (*b >= 'a' && *b <= 'z'))) {/* Padding data(MINIX). */ ++b; --len; } b += nl; if (avail && uuchar[*b]) return (firstline+30); } else if (l == 13) { /* "begin-base64 " */ while (len-nl > 0) { if (!base64[*b++]) return (0); --len; } b += nl; if (avail >= 5 && memcmp(b, "====\n", 5) == 0) return (firstline+40); if (avail >= 6 && memcmp(b, "====\r\n", 6) == 0) return (firstline+40); if (avail > 0 && base64[*b]) return (firstline+30); } return (0); } static int uudecode_bidder_init(struct archive_read_filter *self) { struct uudecode *uudecode; void *out_buff; void *in_buff; self->code = ARCHIVE_FILTER_UU; self->name = "uu"; self->read = uudecode_filter_read; self->skip = NULL; /* not supported */ self->close = uudecode_filter_close; uudecode = (struct uudecode *)calloc(sizeof(*uudecode), 1); out_buff = malloc(OUT_BUFF_SIZE); in_buff = malloc(IN_BUFF_SIZE); if (uudecode == NULL || out_buff == NULL || in_buff == NULL) { archive_set_error(&self->archive->archive, ENOMEM, "Can't allocate data for uudecode"); free(uudecode); free(out_buff); free(in_buff); return (ARCHIVE_FATAL); } self->data = uudecode; uudecode->in_buff = in_buff; uudecode->in_cnt = 0; uudecode->in_allocated = IN_BUFF_SIZE; uudecode->out_buff = out_buff; uudecode->state = ST_FIND_HEAD; return (ARCHIVE_OK); } static int ensure_in_buff_size(struct archive_read_filter *self, struct uudecode *uudecode, size_t size) { if (size > uudecode->in_allocated) { unsigned char *ptr; size_t newsize; /* * Calculate a new buffer size for in_buff. * Increase its value until it has enough size we need. */ newsize = uudecode->in_allocated; do { if (newsize < IN_BUFF_SIZE*32) newsize <<= 1; else newsize += IN_BUFF_SIZE; } while (size > newsize); /* Allocate the new buffer. */ ptr = malloc(newsize); if (ptr == NULL) { free(ptr); archive_set_error(&self->archive->archive, ENOMEM, "Can't allocate data for uudecode"); return (ARCHIVE_FATAL); } /* Move the remaining data in in_buff into the new buffer. */ if (uudecode->in_cnt) memmove(ptr, uudecode->in_buff, uudecode->in_cnt); /* Replace in_buff with the new buffer. */ free(uudecode->in_buff); uudecode->in_buff = ptr; uudecode->in_allocated = newsize; } return (ARCHIVE_OK); } static ssize_t uudecode_filter_read(struct archive_read_filter *self, const void **buff) { struct uudecode *uudecode; const unsigned char *b, *d; unsigned char *out; ssize_t avail_in, ravail; ssize_t used; ssize_t total; ssize_t len, llen, nl; uudecode = (struct uudecode *)self->data; read_more: d = __archive_read_filter_ahead(self->upstream, 1, &avail_in); if (d == NULL && avail_in < 0) return (ARCHIVE_FATAL); /* Quiet a code analyzer; make sure avail_in must be zero * when d is NULL. */ if (d == NULL) avail_in = 0; used = 0; total = 0; out = uudecode->out_buff; ravail = avail_in; if (uudecode->state == ST_IGNORE) { used = avail_in; goto finish; } if (uudecode->in_cnt) { /* * If there is remaining data which is saved by * previous calling, use it first. */ if (ensure_in_buff_size(self, uudecode, avail_in + uudecode->in_cnt) != ARCHIVE_OK) return (ARCHIVE_FATAL); memcpy(uudecode->in_buff + uudecode->in_cnt, d, avail_in); d = uudecode->in_buff; avail_in += uudecode->in_cnt; uudecode->in_cnt = 0; } for (;used < avail_in; d += llen, used += llen) { int64_t l, body; b = d; len = get_line(b, avail_in - used, &nl); if (len < 0) { /* Non-ascii character is found. */ if (uudecode->state == ST_FIND_HEAD && (uudecode->total > 0 || total > 0)) { uudecode->state = ST_IGNORE; used = avail_in; goto finish; } archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Insufficient compressed data"); return (ARCHIVE_FATAL); } llen = len; if ((nl == 0) && (uudecode->state != ST_UUEND)) { + if (total == 0 && ravail <= 0) { + /* There is nothing more to read, fail */ + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Missing format data"); + return (ARCHIVE_FATAL); + } /* * Save remaining data which does not contain * NL('\n','\r'). */ if (ensure_in_buff_size(self, uudecode, len) != ARCHIVE_OK) return (ARCHIVE_FATAL); if (uudecode->in_buff != b) memmove(uudecode->in_buff, b, len); uudecode->in_cnt = (int)len; if (total == 0) { /* Do not return 0; it means end-of-file. * We should try to read bytes more. */ __archive_read_filter_consume( self->upstream, ravail); goto read_more; } used += len; break; } switch (uudecode->state) { default: case ST_FIND_HEAD: /* Do not read more than UUENCODE_BID_MAX_READ bytes */ if (total + len >= UUENCODE_BID_MAX_READ) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid format data"); return (ARCHIVE_FATAL); } if (len - nl >= 11 && memcmp(b, "begin ", 6) == 0) l = 6; else if (len - nl >= 18 && memcmp(b, "begin-base64 ", 13) == 0) l = 13; else l = 0; if (l != 0 && b[l] >= '0' && b[l] <= '7' && b[l+1] >= '0' && b[l+1] <= '7' && b[l+2] >= '0' && b[l+2] <= '7' && b[l+3] == ' ') { if (l == 6) uudecode->state = ST_READ_UU; else uudecode->state = ST_READ_BASE64; } break; case ST_READ_UU: if (total + len * 2 > OUT_BUFF_SIZE) goto finish; body = len - nl; if (!uuchar[*b] || body <= 0) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Insufficient compressed data"); return (ARCHIVE_FATAL); } - /* Get length of undecoded bytes of curent line. */ + /* Get length of undecoded bytes of current line. */ l = UUDECODE(*b++); body--; if (l > body) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Insufficient compressed data"); return (ARCHIVE_FATAL); } if (l == 0) { uudecode->state = ST_UUEND; break; } while (l > 0) { int n = 0; if (l > 0) { if (!uuchar[b[0]] || !uuchar[b[1]]) break; n = UUDECODE(*b++) << 18; n |= UUDECODE(*b++) << 12; *out++ = n >> 16; total++; --l; } if (l > 0) { if (!uuchar[b[0]]) break; n |= UUDECODE(*b++) << 6; *out++ = (n >> 8) & 0xFF; total++; --l; } if (l > 0) { if (!uuchar[b[0]]) break; n |= UUDECODE(*b++); *out++ = n & 0xFF; total++; --l; } } if (l) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Insufficient compressed data"); return (ARCHIVE_FATAL); } break; case ST_UUEND: if (len - nl == 3 && memcmp(b, "end ", 3) == 0) uudecode->state = ST_FIND_HEAD; else { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Insufficient compressed data"); return (ARCHIVE_FATAL); } break; case ST_READ_BASE64: if (total + len * 2 > OUT_BUFF_SIZE) goto finish; l = len - nl; if (l >= 3 && b[0] == '=' && b[1] == '=' && b[2] == '=') { uudecode->state = ST_FIND_HEAD; break; } while (l > 0) { int n = 0; if (l > 0) { if (!base64[b[0]] || !base64[b[1]]) break; n = base64num[*b++] << 18; n |= base64num[*b++] << 12; *out++ = n >> 16; total++; l -= 2; } if (l > 0) { if (*b == '=') break; if (!base64[*b]) break; n |= base64num[*b++] << 6; *out++ = (n >> 8) & 0xFF; total++; --l; } if (l > 0) { if (*b == '=') break; if (!base64[*b]) break; n |= base64num[*b++]; *out++ = n & 0xFF; total++; --l; } } if (l && *b != '=') { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Insufficient compressed data"); return (ARCHIVE_FATAL); } break; } } finish: if (ravail < avail_in) used -= avail_in - ravail; __archive_read_filter_consume(self->upstream, used); *buff = uudecode->out_buff; uudecode->total += total; return (total); } static int uudecode_filter_close(struct archive_read_filter *self) { struct uudecode *uudecode; uudecode = (struct uudecode *)self->data; free(uudecode->in_buff); free(uudecode->out_buff); free(uudecode); return (ARCHIVE_OK); } Index: head/contrib/libarchive/libarchive/archive_read_support_format_ar.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_format_ar.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_read_support_format_ar.c (revision 309590) @@ -1,635 +1,638 @@ /*- * Copyright (c) 2007 Kai Wang * Copyright (c) 2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_read_private.h" struct ar { int64_t entry_bytes_remaining; /* unconsumed is purely to track data we've gotten from readahead, * but haven't yet marked as consumed. Must be paired with * entry_bytes_remaining usage/modification. */ size_t entry_bytes_unconsumed; int64_t entry_offset; int64_t entry_padding; char *strtab; size_t strtab_size; char read_global_header; }; /* * Define structure of the "ar" header. */ #define AR_name_offset 0 #define AR_name_size 16 #define AR_date_offset 16 #define AR_date_size 12 #define AR_uid_offset 28 #define AR_uid_size 6 #define AR_gid_offset 34 #define AR_gid_size 6 #define AR_mode_offset 40 #define AR_mode_size 8 #define AR_size_offset 48 #define AR_size_size 10 #define AR_fmag_offset 58 #define AR_fmag_size 2 static int archive_read_format_ar_bid(struct archive_read *a, int); static int archive_read_format_ar_cleanup(struct archive_read *a); static int archive_read_format_ar_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset); static int archive_read_format_ar_skip(struct archive_read *a); static int archive_read_format_ar_read_header(struct archive_read *a, struct archive_entry *e); static uint64_t ar_atol8(const char *p, unsigned char_cnt); static uint64_t ar_atol10(const char *p, unsigned char_cnt); static int ar_parse_gnu_filename_table(struct archive_read *a); static int ar_parse_common_header(struct ar *ar, struct archive_entry *, const char *h); int archive_read_support_format_ar(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct ar *ar; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_ar"); ar = (struct ar *)malloc(sizeof(*ar)); if (ar == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate ar data"); return (ARCHIVE_FATAL); } memset(ar, 0, sizeof(*ar)); ar->strtab = NULL; r = __archive_read_register_format(a, ar, "ar", archive_read_format_ar_bid, NULL, archive_read_format_ar_read_header, archive_read_format_ar_read_data, archive_read_format_ar_skip, NULL, archive_read_format_ar_cleanup, NULL, NULL); if (r != ARCHIVE_OK) { free(ar); return (r); } return (ARCHIVE_OK); } static int archive_read_format_ar_cleanup(struct archive_read *a) { struct ar *ar; ar = (struct ar *)(a->format->data); if (ar->strtab) free(ar->strtab); free(ar); (a->format->data) = NULL; return (ARCHIVE_OK); } static int archive_read_format_ar_bid(struct archive_read *a, int best_bid) { const void *h; (void)best_bid; /* UNUSED */ /* * Verify the 8-byte file signature. * TODO: Do we need to check more than this? */ if ((h = __archive_read_ahead(a, 8, NULL)) == NULL) return (-1); if (memcmp(h, "!\n", 8) == 0) { return (64); } return (-1); } static int _ar_read_header(struct archive_read *a, struct archive_entry *entry, struct ar *ar, const char *h, size_t *unconsumed) { char filename[AR_name_size + 1]; uint64_t number; /* Used to hold parsed numbers before validation. */ size_t bsd_name_length, entry_size; char *p, *st; const void *b; int r; /* Verify the magic signature on the file header. */ if (strncmp(h + AR_fmag_offset, "`\n", 2) != 0) { archive_set_error(&a->archive, EINVAL, "Incorrect file header signature"); return (ARCHIVE_FATAL); } /* Copy filename into work buffer. */ strncpy(filename, h + AR_name_offset, AR_name_size); filename[AR_name_size] = '\0'; /* * Guess the format variant based on the filename. */ if (a->archive.archive_format == ARCHIVE_FORMAT_AR) { /* We don't already know the variant, so let's guess. */ /* * Biggest clue is presence of '/': GNU starts special * filenames with '/', appends '/' as terminator to * non-special names, so anything with '/' should be * GNU except for BSD long filenames. */ if (strncmp(filename, "#1/", 3) == 0) a->archive.archive_format = ARCHIVE_FORMAT_AR_BSD; else if (strchr(filename, '/') != NULL) a->archive.archive_format = ARCHIVE_FORMAT_AR_GNU; else if (strncmp(filename, "__.SYMDEF", 9) == 0) a->archive.archive_format = ARCHIVE_FORMAT_AR_BSD; /* * XXX Do GNU/SVR4 'ar' programs ever omit trailing '/' * if name exactly fills 16-byte field? If so, we * can't assume entries without '/' are BSD. XXX */ } /* Update format name from the code. */ if (a->archive.archive_format == ARCHIVE_FORMAT_AR_GNU) a->archive.archive_format_name = "ar (GNU/SVR4)"; else if (a->archive.archive_format == ARCHIVE_FORMAT_AR_BSD) a->archive.archive_format_name = "ar (BSD)"; else a->archive.archive_format_name = "ar"; /* * Remove trailing spaces from the filename. GNU and BSD * variants both pad filename area out with spaces. * This will only be wrong if GNU/SVR4 'ar' implementations * omit trailing '/' for 16-char filenames and we have * a 16-char filename that ends in ' '. */ p = filename + AR_name_size - 1; while (p >= filename && *p == ' ') { *p = '\0'; p--; } /* * Remove trailing slash unless first character is '/'. * (BSD entries never end in '/', so this will only trim * GNU-format entries. GNU special entries start with '/' * and are not terminated in '/', so we don't trim anything * that starts with '/'.) */ if (filename[0] != '/' && p > filename && *p == '/') { *p = '\0'; } if (p < filename) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Found entry with empty filename"); return (ARCHIVE_FATAL); } /* * '//' is the GNU filename table. * Later entries can refer to names in this table. */ if (strcmp(filename, "//") == 0) { /* This must come before any call to _read_ahead. */ ar_parse_common_header(ar, entry, h); archive_entry_copy_pathname(entry, filename); archive_entry_set_filetype(entry, AE_IFREG); /* Get the size of the filename table. */ number = ar_atol10(h + AR_size_offset, AR_size_size); - if (number > SIZE_MAX) { + if (number > SIZE_MAX || number > 1024 * 1024 * 1024) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Filename table too large"); return (ARCHIVE_FATAL); } entry_size = (size_t)number; if (entry_size == 0) { archive_set_error(&a->archive, EINVAL, "Invalid string table"); return (ARCHIVE_FATAL); } if (ar->strtab != NULL) { archive_set_error(&a->archive, EINVAL, "More than one string tables exist"); return (ARCHIVE_FATAL); } /* Read the filename table into memory. */ st = malloc(entry_size); if (st == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate filename table buffer"); return (ARCHIVE_FATAL); } ar->strtab = st; ar->strtab_size = entry_size; if (*unconsumed) { __archive_read_consume(a, *unconsumed); *unconsumed = 0; } if ((b = __archive_read_ahead(a, entry_size, NULL)) == NULL) return (ARCHIVE_FATAL); memcpy(st, b, entry_size); __archive_read_consume(a, entry_size); /* All contents are consumed. */ ar->entry_bytes_remaining = 0; archive_entry_set_size(entry, ar->entry_bytes_remaining); /* Parse the filename table. */ return (ar_parse_gnu_filename_table(a)); } /* * GNU variant handles long filenames by storing / * to indicate a name stored in the filename table. * XXX TODO: Verify that it's all digits... Don't be fooled * by "/9xyz" XXX */ if (filename[0] == '/' && filename[1] >= '0' && filename[1] <= '9') { number = ar_atol10(h + AR_name_offset + 1, AR_name_size - 1); /* * If we can't look up the real name, warn and return * the entry with the wrong name. */ if (ar->strtab == NULL || number > ar->strtab_size) { archive_set_error(&a->archive, EINVAL, "Can't find long filename for GNU/SVR4 archive entry"); archive_entry_copy_pathname(entry, filename); /* Parse the time, owner, mode, size fields. */ ar_parse_common_header(ar, entry, h); return (ARCHIVE_FATAL); } archive_entry_copy_pathname(entry, &ar->strtab[(size_t)number]); /* Parse the time, owner, mode, size fields. */ return (ar_parse_common_header(ar, entry, h)); } /* * BSD handles long filenames by storing "#1/" followed by the * length of filename as a decimal number, then prepends the * the filename to the file contents. */ if (strncmp(filename, "#1/", 3) == 0) { /* Parse the time, owner, mode, size fields. */ /* This must occur before _read_ahead is called again. */ ar_parse_common_header(ar, entry, h); /* Parse the size of the name, adjust the file size. */ number = ar_atol10(h + AR_name_offset + 3, AR_name_size - 3); - bsd_name_length = (size_t)number; - /* Guard against the filename + trailing NUL - * overflowing a size_t and against the filename size - * being larger than the entire entry. */ - if (number > (uint64_t)(bsd_name_length + 1) - || (int64_t)bsd_name_length > ar->entry_bytes_remaining) { + /* Sanity check the filename length: + * = Must be <= SIZE_MAX - 1 + * = Must be <= 1MB + * = Cannot be bigger than the entire entry + */ + if (number > SIZE_MAX - 1 + || number > 1024 * 1024 + || (int64_t)number > ar->entry_bytes_remaining) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Bad input file size"); return (ARCHIVE_FATAL); } + bsd_name_length = (size_t)number; ar->entry_bytes_remaining -= bsd_name_length; /* Adjust file size reported to client. */ archive_entry_set_size(entry, ar->entry_bytes_remaining); if (*unconsumed) { __archive_read_consume(a, *unconsumed); *unconsumed = 0; } /* Read the long name into memory. */ if ((b = __archive_read_ahead(a, bsd_name_length, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Truncated input file"); return (ARCHIVE_FATAL); } /* Store it in the entry. */ p = (char *)malloc(bsd_name_length + 1); if (p == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate fname buffer"); return (ARCHIVE_FATAL); } strncpy(p, b, bsd_name_length); p[bsd_name_length] = '\0'; __archive_read_consume(a, bsd_name_length); archive_entry_copy_pathname(entry, p); free(p); return (ARCHIVE_OK); } /* * "/" is the SVR4/GNU archive symbol table. */ if (strcmp(filename, "/") == 0) { archive_entry_copy_pathname(entry, "/"); /* Parse the time, owner, mode, size fields. */ r = ar_parse_common_header(ar, entry, h); /* Force the file type to a regular file. */ archive_entry_set_filetype(entry, AE_IFREG); return (r); } /* * "__.SYMDEF" is a BSD archive symbol table. */ if (strcmp(filename, "__.SYMDEF") == 0) { archive_entry_copy_pathname(entry, filename); /* Parse the time, owner, mode, size fields. */ return (ar_parse_common_header(ar, entry, h)); } /* * Otherwise, this is a standard entry. The filename * has already been trimmed as much as possible, based * on our current knowledge of the format. */ archive_entry_copy_pathname(entry, filename); return (ar_parse_common_header(ar, entry, h)); } static int archive_read_format_ar_read_header(struct archive_read *a, struct archive_entry *entry) { struct ar *ar = (struct ar*)(a->format->data); size_t unconsumed; const void *header_data; int ret; if (!ar->read_global_header) { /* * We are now at the beginning of the archive, * so we need first consume the ar global header. */ __archive_read_consume(a, 8); ar->read_global_header = 1; /* Set a default format code for now. */ a->archive.archive_format = ARCHIVE_FORMAT_AR; } /* Read the header for the next file entry. */ if ((header_data = __archive_read_ahead(a, 60, NULL)) == NULL) /* Broken header. */ return (ARCHIVE_EOF); unconsumed = 60; ret = _ar_read_header(a, entry, ar, (const char *)header_data, &unconsumed); if (unconsumed) __archive_read_consume(a, unconsumed); return ret; } static int ar_parse_common_header(struct ar *ar, struct archive_entry *entry, const char *h) { uint64_t n; /* Copy remaining header */ archive_entry_set_mtime(entry, (time_t)ar_atol10(h + AR_date_offset, AR_date_size), 0L); archive_entry_set_uid(entry, (uid_t)ar_atol10(h + AR_uid_offset, AR_uid_size)); archive_entry_set_gid(entry, (gid_t)ar_atol10(h + AR_gid_offset, AR_gid_size)); archive_entry_set_mode(entry, (mode_t)ar_atol8(h + AR_mode_offset, AR_mode_size)); n = ar_atol10(h + AR_size_offset, AR_size_size); ar->entry_offset = 0; ar->entry_padding = n % 2; archive_entry_set_size(entry, n); ar->entry_bytes_remaining = n; return (ARCHIVE_OK); } static int archive_read_format_ar_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { ssize_t bytes_read; struct ar *ar; ar = (struct ar *)(a->format->data); if (ar->entry_bytes_unconsumed) { __archive_read_consume(a, ar->entry_bytes_unconsumed); ar->entry_bytes_unconsumed = 0; } if (ar->entry_bytes_remaining > 0) { *buff = __archive_read_ahead(a, 1, &bytes_read); if (bytes_read == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Truncated ar archive"); return (ARCHIVE_FATAL); } if (bytes_read < 0) return (ARCHIVE_FATAL); if (bytes_read > ar->entry_bytes_remaining) bytes_read = (ssize_t)ar->entry_bytes_remaining; *size = bytes_read; ar->entry_bytes_unconsumed = bytes_read; *offset = ar->entry_offset; ar->entry_offset += bytes_read; ar->entry_bytes_remaining -= bytes_read; return (ARCHIVE_OK); } else { int64_t skipped = __archive_read_consume(a, ar->entry_padding); if (skipped >= 0) { ar->entry_padding -= skipped; } if (ar->entry_padding) { if (skipped >= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Truncated ar archive- failed consuming padding"); } return (ARCHIVE_FATAL); } *buff = NULL; *size = 0; *offset = ar->entry_offset; return (ARCHIVE_EOF); } } static int archive_read_format_ar_skip(struct archive_read *a) { int64_t bytes_skipped; struct ar* ar; ar = (struct ar *)(a->format->data); bytes_skipped = __archive_read_consume(a, ar->entry_bytes_remaining + ar->entry_padding + ar->entry_bytes_unconsumed); if (bytes_skipped < 0) return (ARCHIVE_FATAL); ar->entry_bytes_remaining = 0; ar->entry_bytes_unconsumed = 0; ar->entry_padding = 0; return (ARCHIVE_OK); } static int ar_parse_gnu_filename_table(struct archive_read *a) { struct ar *ar; char *p; size_t size; ar = (struct ar*)(a->format->data); size = ar->strtab_size; for (p = ar->strtab; p < ar->strtab + size - 1; ++p) { if (*p == '/') { *p++ = '\0'; if (*p != '\n') goto bad_string_table; *p = '\0'; } } /* * GNU ar always pads the table to an even size. * The pad character is either '\n' or '`'. */ if (p != ar->strtab + size && *p != '\n' && *p != '`') goto bad_string_table; /* Enforce zero termination. */ ar->strtab[size - 1] = '\0'; return (ARCHIVE_OK); bad_string_table: archive_set_error(&a->archive, EINVAL, "Invalid string table"); free(ar->strtab); ar->strtab = NULL; return (ARCHIVE_FATAL); } static uint64_t ar_atol8(const char *p, unsigned char_cnt) { uint64_t l, limit, last_digit_limit; unsigned int digit, base; base = 8; limit = UINT64_MAX / base; last_digit_limit = UINT64_MAX % base; while ((*p == ' ' || *p == '\t') && char_cnt-- > 0) p++; l = 0; digit = *p - '0'; while (*p >= '0' && digit < base && char_cnt-- > 0) { if (l>limit || (l == limit && digit > last_digit_limit)) { l = UINT64_MAX; /* Truncate on overflow. */ break; } l = (l * base) + digit; digit = *++p - '0'; } return (l); } static uint64_t ar_atol10(const char *p, unsigned char_cnt) { uint64_t l, limit, last_digit_limit; unsigned int base, digit; base = 10; limit = UINT64_MAX / base; last_digit_limit = UINT64_MAX % base; while ((*p == ' ' || *p == '\t') && char_cnt-- > 0) p++; l = 0; digit = *p - '0'; while (*p >= '0' && digit < base && char_cnt-- > 0) { if (l > limit || (l == limit && digit > last_digit_limit)) { l = UINT64_MAX; /* Truncate on overflow. */ break; } l = (l * base) + digit; digit = *++p - '0'; } return (l); } Index: head/contrib/libarchive/libarchive/archive_read_support_format_cab.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_format_cab.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_read_support_format_cab.c (revision 309590) @@ -1,3351 +1,3351 @@ /*- * Copyright (c) 2010-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_read_private.h" #include "archive_endian.h" struct lzx_dec { /* Decoding status. */ int state; /* * Window to see last decoded data, from 32KBi to 2MBi. */ int w_size; int w_mask; /* Window buffer, which is a loop buffer. */ unsigned char *w_buff; /* The insert position to the window. */ int w_pos; /* The position where we can copy decoded code from the window. */ int copy_pos; /* The length how many bytes we can copy decoded code from * the window. */ int copy_len; - /* Translation reversal for x86 proccessor CALL byte sequence(E8). + /* Translation reversal for x86 processor CALL byte sequence(E8). * This is used for LZX only. */ uint32_t translation_size; char translation; char block_type; #define VERBATIM_BLOCK 1 #define ALIGNED_OFFSET_BLOCK 2 #define UNCOMPRESSED_BLOCK 3 size_t block_size; size_t block_bytes_avail; /* Repeated offset. */ int r0, r1, r2; unsigned char rbytes[4]; int rbytes_avail; int length_header; int position_slot; int offset_bits; struct lzx_pos_tbl { int base; int footer_bits; } *pos_tbl; /* * Bit stream reader. */ struct lzx_br { #define CACHE_TYPE uint64_t #define CACHE_BITS (8 * sizeof(CACHE_TYPE)) /* Cache buffer. */ CACHE_TYPE cache_buffer; /* Indicates how many bits avail in cache_buffer. */ int cache_avail; unsigned char odd; char have_odd; } br; /* * Huffman coding. */ struct huffman { int len_size; int freq[17]; unsigned char *bitlen; /* * Use a index table. It's faster than searching a huffman * coding tree, which is a binary tree. But a use of a large * index table causes L1 cache read miss many times. */ #define HTBL_BITS 10 int max_bits; int shift_bits; int tbl_bits; int tree_used; int tree_avail; /* Direct access table. */ uint16_t *tbl; /* Binary tree table for extra bits over the direct access. */ struct htree_t { uint16_t left; uint16_t right; } *tree; } at, lt, mt, pt; int loop; int error; }; static const int slots[] = { 30, 32, 34, 36, 38, 42, 50, 66, 98, 162, 290 }; #define SLOT_BASE 15 #define SLOT_MAX 21/*->25*/ struct lzx_stream { const unsigned char *next_in; int64_t avail_in; int64_t total_in; unsigned char *next_out; int64_t avail_out; int64_t total_out; struct lzx_dec *ds; }; /* * Cabinet file definitions. */ /* CFHEADER offset */ #define CFHEADER_signature 0 #define CFHEADER_cbCabinet 8 #define CFHEADER_coffFiles 16 #define CFHEADER_versionMinor 24 #define CFHEADER_versionMajor 25 #define CFHEADER_cFolders 26 #define CFHEADER_cFiles 28 #define CFHEADER_flags 30 #define CFHEADER_setID 32 #define CFHEADER_iCabinet 34 #define CFHEADER_cbCFHeader 36 #define CFHEADER_cbCFFolder 38 #define CFHEADER_cbCFData 39 /* CFFOLDER offset */ #define CFFOLDER_coffCabStart 0 #define CFFOLDER_cCFData 4 #define CFFOLDER_typeCompress 6 #define CFFOLDER_abReserve 8 /* CFFILE offset */ #define CFFILE_cbFile 0 #define CFFILE_uoffFolderStart 4 #define CFFILE_iFolder 8 #define CFFILE_date_time 10 #define CFFILE_attribs 14 /* CFDATA offset */ #define CFDATA_csum 0 #define CFDATA_cbData 4 #define CFDATA_cbUncomp 6 static const char *compression_name[] = { "NONE", "MSZIP", "Quantum", "LZX", }; struct cfdata { /* Sum value of this CFDATA. */ uint32_t sum; uint16_t compressed_size; uint16_t compressed_bytes_remaining; uint16_t uncompressed_size; uint16_t uncompressed_bytes_remaining; /* To know how many bytes we have decompressed. */ uint16_t uncompressed_avail; /* Offset from the beginning of compressed data of this CFDATA */ uint16_t read_offset; int64_t unconsumed; /* To keep memory image of this CFDATA to compute the sum. */ size_t memimage_size; unsigned char *memimage; /* Result of calculation of sum. */ uint32_t sum_calculated; unsigned char sum_extra[4]; int sum_extra_avail; const void *sum_ptr; }; struct cffolder { uint32_t cfdata_offset_in_cab; uint16_t cfdata_count; uint16_t comptype; #define COMPTYPE_NONE 0x0000 #define COMPTYPE_MSZIP 0x0001 #define COMPTYPE_QUANTUM 0x0002 #define COMPTYPE_LZX 0x0003 uint16_t compdata; const char *compname; /* At the time reading CFDATA */ struct cfdata cfdata; int cfdata_index; /* Flags to mark progress of decompression. */ char decompress_init; }; struct cffile { uint32_t uncompressed_size; uint32_t offset; time_t mtime; uint16_t folder; #define iFoldCONTINUED_FROM_PREV 0xFFFD #define iFoldCONTINUED_TO_NEXT 0xFFFE #define iFoldCONTINUED_PREV_AND_NEXT 0xFFFF unsigned char attr; #define ATTR_RDONLY 0x01 #define ATTR_NAME_IS_UTF 0x80 struct archive_string pathname; }; struct cfheader { /* Total bytes of all file size in a Cabinet. */ uint32_t total_bytes; uint32_t files_offset; uint16_t folder_count; uint16_t file_count; uint16_t flags; #define PREV_CABINET 0x0001 #define NEXT_CABINET 0x0002 #define RESERVE_PRESENT 0x0004 uint16_t setid; uint16_t cabinet; /* Version number. */ unsigned char major; unsigned char minor; unsigned char cffolder; unsigned char cfdata; /* All folders in a cabinet. */ struct cffolder *folder_array; /* All files in a cabinet. */ struct cffile *file_array; int file_index; }; struct cab { /* entry_bytes_remaining is the number of bytes we expect. */ int64_t entry_offset; int64_t entry_bytes_remaining; int64_t entry_unconsumed; int64_t entry_compressed_bytes_read; int64_t entry_uncompressed_bytes_read; struct cffolder *entry_cffolder; struct cffile *entry_cffile; struct cfdata *entry_cfdata; /* Offset from beginning of a cabinet file. */ int64_t cab_offset; struct cfheader cfheader; struct archive_wstring ws; /* Flag to mark progress that an archive was read their first header.*/ char found_header; char end_of_archive; char end_of_entry; char end_of_entry_cleanup; char read_data_invoked; int64_t bytes_skipped; unsigned char *uncompressed_buffer; size_t uncompressed_buffer_size; int init_default_conversion; struct archive_string_conv *sconv; struct archive_string_conv *sconv_default; struct archive_string_conv *sconv_utf8; char format_name[64]; #ifdef HAVE_ZLIB_H z_stream stream; char stream_valid; #endif struct lzx_stream xstrm; }; static int archive_read_format_cab_bid(struct archive_read *, int); static int archive_read_format_cab_options(struct archive_read *, const char *, const char *); static int archive_read_format_cab_read_header(struct archive_read *, struct archive_entry *); static int archive_read_format_cab_read_data(struct archive_read *, const void **, size_t *, int64_t *); static int archive_read_format_cab_read_data_skip(struct archive_read *); static int archive_read_format_cab_cleanup(struct archive_read *); static int cab_skip_sfx(struct archive_read *); static time_t cab_dos_time(const unsigned char *); static int cab_read_data(struct archive_read *, const void **, size_t *, int64_t *); static int cab_read_header(struct archive_read *); static uint32_t cab_checksum_cfdata_4(const void *, size_t bytes, uint32_t); static uint32_t cab_checksum_cfdata(const void *, size_t bytes, uint32_t); static void cab_checksum_update(struct archive_read *, size_t); static int cab_checksum_finish(struct archive_read *); static int cab_next_cfdata(struct archive_read *); static const void *cab_read_ahead_cfdata(struct archive_read *, ssize_t *); static const void *cab_read_ahead_cfdata_none(struct archive_read *, ssize_t *); static const void *cab_read_ahead_cfdata_deflate(struct archive_read *, ssize_t *); static const void *cab_read_ahead_cfdata_lzx(struct archive_read *, ssize_t *); static int64_t cab_consume_cfdata(struct archive_read *, int64_t); static int64_t cab_minimum_consume_cfdata(struct archive_read *, int64_t); static int lzx_decode_init(struct lzx_stream *, int); static int lzx_read_blocks(struct lzx_stream *, int); static int lzx_decode_blocks(struct lzx_stream *, int); static void lzx_decode_free(struct lzx_stream *); static void lzx_translation(struct lzx_stream *, void *, size_t, uint32_t); static void lzx_cleanup_bitstream(struct lzx_stream *); static int lzx_decode(struct lzx_stream *, int); static int lzx_read_pre_tree(struct lzx_stream *); static int lzx_read_bitlen(struct lzx_stream *, struct huffman *, int); static int lzx_huffman_init(struct huffman *, size_t, int); static void lzx_huffman_free(struct huffman *); static int lzx_make_huffman_table(struct huffman *); static inline int lzx_decode_huffman(struct huffman *, unsigned); static int lzx_decode_huffman_tree(struct huffman *, unsigned, int); int archive_read_support_format_cab(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct cab *cab; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_cab"); cab = (struct cab *)calloc(1, sizeof(*cab)); if (cab == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate CAB data"); return (ARCHIVE_FATAL); } archive_string_init(&cab->ws); archive_wstring_ensure(&cab->ws, 256); r = __archive_read_register_format(a, cab, "cab", archive_read_format_cab_bid, archive_read_format_cab_options, archive_read_format_cab_read_header, archive_read_format_cab_read_data, archive_read_format_cab_read_data_skip, NULL, archive_read_format_cab_cleanup, NULL, NULL); if (r != ARCHIVE_OK) free(cab); return (ARCHIVE_OK); } static int find_cab_magic(const char *p) { switch (p[4]) { case 0: /* * Note: Self-Extraction program has 'MSCF' string in their * program. If we were finding 'MSCF' string only, we got * wrong place for Cabinet header, thus, we have to check * following four bytes which are reserved and must be set * to zero. */ if (memcmp(p, "MSCF\0\0\0\0", 8) == 0) return 0; return 5; case 'F': return 1; case 'C': return 2; case 'S': return 3; case 'M': return 4; default: return 5; } } static int archive_read_format_cab_bid(struct archive_read *a, int best_bid) { const char *p; ssize_t bytes_avail, offset, window; /* If there's already a better bid than we can ever make, don't bother testing. */ if (best_bid > 64) return (-1); if ((p = __archive_read_ahead(a, 8, NULL)) == NULL) return (-1); if (memcmp(p, "MSCF\0\0\0\0", 8) == 0) return (64); /* * Attempt to handle self-extracting archives * by noting a PE header and searching forward * up to 128k for a 'MSCF' marker. */ if (p[0] == 'M' && p[1] == 'Z') { offset = 0; window = 4096; while (offset < (1024 * 128)) { const char *h = __archive_read_ahead(a, offset + window, &bytes_avail); if (h == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < 128) return (0); continue; } p = h + offset; while (p + 8 < h + bytes_avail) { int next; if ((next = find_cab_magic(p)) == 0) return (64); p += next; } offset = p - h; } } return (0); } static int archive_read_format_cab_options(struct archive_read *a, const char *key, const char *val) { struct cab *cab; int ret = ARCHIVE_FAILED; cab = (struct cab *)(a->format->data); if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "cab: hdrcharset option needs a character-set name"); else { cab->sconv = archive_string_conversion_from_charset( &a->archive, val, 0); if (cab->sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int cab_skip_sfx(struct archive_read *a) { const char *p, *q; size_t skip; ssize_t bytes, window; window = 4096; for (;;) { const char *h = __archive_read_ahead(a, window, &bytes); if (h == NULL) { /* Remaining size are less than window. */ window >>= 1; if (window < 128) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Couldn't find out CAB header"); return (ARCHIVE_FATAL); } continue; } p = h; q = p + bytes; /* * Scan ahead until we find something that looks * like the cab header. */ while (p + 8 < q) { int next; if ((next = find_cab_magic(p)) == 0) { skip = p - h; __archive_read_consume(a, skip); return (ARCHIVE_OK); } p += next; } skip = p - h; __archive_read_consume(a, skip); } } static int truncated_error(struct archive_read *a) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated CAB header"); return (ARCHIVE_FATAL); } static ssize_t cab_strnlen(const unsigned char *p, size_t maxlen) { size_t i; for (i = 0; i <= maxlen; i++) { if (p[i] == 0) break; } if (i > maxlen) return (-1);/* invalid */ return ((ssize_t)i); } /* Read bytes as much as remaining. */ static const void * cab_read_ahead_remaining(struct archive_read *a, size_t min, ssize_t *avail) { const void *p; while (min > 0) { p = __archive_read_ahead(a, min, avail); if (p != NULL) return (p); min--; } return (NULL); } /* Convert a path separator '\' -> '/' */ static int cab_convert_path_separator_1(struct archive_string *fn, unsigned char attr) { size_t i; int mb; /* Easy check if we have '\' in multi-byte string. */ mb = 0; for (i = 0; i < archive_strlen(fn); i++) { if (fn->s[i] == '\\') { if (mb) { /* This may be second byte of multi-byte * character. */ break; } fn->s[i] = '/'; mb = 0; } else if ((fn->s[i] & 0x80) && !(attr & ATTR_NAME_IS_UTF)) mb = 1; else mb = 0; } if (i == archive_strlen(fn)) return (0); return (-1); } /* * Replace a character '\' with '/' in wide character. */ static void cab_convert_path_separator_2(struct cab *cab, struct archive_entry *entry) { const wchar_t *wp; size_t i; /* If a conversion to wide character failed, force the replacement. */ if ((wp = archive_entry_pathname_w(entry)) != NULL) { archive_wstrcpy(&(cab->ws), wp); for (i = 0; i < archive_strlen(&(cab->ws)); i++) { if (cab->ws.s[i] == L'\\') cab->ws.s[i] = L'/'; } archive_entry_copy_pathname_w(entry, cab->ws.s); } } /* * Read CFHEADER, CFFOLDER and CFFILE. */ static int cab_read_header(struct archive_read *a) { const unsigned char *p; struct cab *cab; struct cfheader *hd; size_t bytes, used; ssize_t len; int64_t skip; int err, i; int cur_folder, prev_folder; uint32_t offset32; a->archive.archive_format = ARCHIVE_FORMAT_CAB; if (a->archive.archive_format_name == NULL) a->archive.archive_format_name = "CAB"; if ((p = __archive_read_ahead(a, 42, NULL)) == NULL) return (truncated_error(a)); cab = (struct cab *)(a->format->data); if (cab->found_header == 0 && p[0] == 'M' && p[1] == 'Z') { /* This is an executable? Must be self-extracting... */ err = cab_skip_sfx(a); if (err < ARCHIVE_WARN) return (err); /* Re-read header after processing the SFX. */ if ((p = __archive_read_ahead(a, 42, NULL)) == NULL) return (truncated_error(a)); } cab->cab_offset = 0; /* * Read CFHEADER. */ hd = &cab->cfheader; if (p[CFHEADER_signature+0] != 'M' || p[CFHEADER_signature+1] != 'S' || p[CFHEADER_signature+2] != 'C' || p[CFHEADER_signature+3] != 'F') { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Couldn't find out CAB header"); return (ARCHIVE_FATAL); } hd->total_bytes = archive_le32dec(p + CFHEADER_cbCabinet); hd->files_offset = archive_le32dec(p + CFHEADER_coffFiles); hd->minor = p[CFHEADER_versionMinor]; hd->major = p[CFHEADER_versionMajor]; hd->folder_count = archive_le16dec(p + CFHEADER_cFolders); if (hd->folder_count == 0) goto invalid; hd->file_count = archive_le16dec(p + CFHEADER_cFiles); if (hd->file_count == 0) goto invalid; hd->flags = archive_le16dec(p + CFHEADER_flags); hd->setid = archive_le16dec(p + CFHEADER_setID); hd->cabinet = archive_le16dec(p + CFHEADER_iCabinet); used = CFHEADER_iCabinet + 2; if (hd->flags & RESERVE_PRESENT) { uint16_t cfheader; cfheader = archive_le16dec(p + CFHEADER_cbCFHeader); if (cfheader > 60000U) goto invalid; hd->cffolder = p[CFHEADER_cbCFFolder]; hd->cfdata = p[CFHEADER_cbCFData]; used += 4;/* cbCFHeader, cbCFFolder and cbCFData */ used += cfheader;/* abReserve */ } else hd->cffolder = 0;/* Avoid compiling warning. */ if (hd->flags & PREV_CABINET) { /* How many bytes are used for szCabinetPrev. */ if ((p = __archive_read_ahead(a, used+256, NULL)) == NULL) return (truncated_error(a)); if ((len = cab_strnlen(p + used, 255)) <= 0) goto invalid; used += len + 1; /* How many bytes are used for szDiskPrev. */ if ((p = __archive_read_ahead(a, used+256, NULL)) == NULL) return (truncated_error(a)); if ((len = cab_strnlen(p + used, 255)) <= 0) goto invalid; used += len + 1; } if (hd->flags & NEXT_CABINET) { /* How many bytes are used for szCabinetNext. */ if ((p = __archive_read_ahead(a, used+256, NULL)) == NULL) return (truncated_error(a)); if ((len = cab_strnlen(p + used, 255)) <= 0) goto invalid; used += len + 1; /* How many bytes are used for szDiskNext. */ if ((p = __archive_read_ahead(a, used+256, NULL)) == NULL) return (truncated_error(a)); if ((len = cab_strnlen(p + used, 255)) <= 0) goto invalid; used += len + 1; } __archive_read_consume(a, used); cab->cab_offset += used; used = 0; /* * Read CFFOLDER. */ hd->folder_array = (struct cffolder *)calloc( hd->folder_count, sizeof(struct cffolder)); if (hd->folder_array == NULL) goto nomem; bytes = 8; if (hd->flags & RESERVE_PRESENT) bytes += hd->cffolder; bytes *= hd->folder_count; if ((p = __archive_read_ahead(a, bytes, NULL)) == NULL) return (truncated_error(a)); offset32 = 0; for (i = 0; i < hd->folder_count; i++) { struct cffolder *folder = &(hd->folder_array[i]); folder->cfdata_offset_in_cab = archive_le32dec(p + CFFOLDER_coffCabStart); folder->cfdata_count = archive_le16dec(p+CFFOLDER_cCFData); folder->comptype = archive_le16dec(p+CFFOLDER_typeCompress) & 0x0F; folder->compdata = archive_le16dec(p+CFFOLDER_typeCompress) >> 8; /* Get a compression name. */ if (folder->comptype < sizeof(compression_name) / sizeof(compression_name[0])) folder->compname = compression_name[folder->comptype]; else folder->compname = "UNKNOWN"; p += 8; used += 8; if (hd->flags & RESERVE_PRESENT) { p += hd->cffolder;/* abReserve */ used += hd->cffolder; } /* * Sanity check if each data is acceptable. */ if (offset32 >= folder->cfdata_offset_in_cab) goto invalid; offset32 = folder->cfdata_offset_in_cab; /* Set a request to initialize zlib for the CFDATA of * this folder. */ folder->decompress_init = 0; } __archive_read_consume(a, used); cab->cab_offset += used; /* * Read CFFILE. */ /* Seek read pointer to the offset of CFFILE if needed. */ skip = (int64_t)hd->files_offset - cab->cab_offset; if (skip < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid offset of CFFILE %jd < %jd", (intmax_t)hd->files_offset, (intmax_t)cab->cab_offset); return (ARCHIVE_FATAL); } if (skip) { __archive_read_consume(a, skip); cab->cab_offset += skip; } /* Allocate memory for CFDATA */ hd->file_array = (struct cffile *)calloc( hd->file_count, sizeof(struct cffile)); if (hd->file_array == NULL) goto nomem; prev_folder = -1; for (i = 0; i < hd->file_count; i++) { struct cffile *file = &(hd->file_array[i]); ssize_t avail; if ((p = __archive_read_ahead(a, 16, NULL)) == NULL) return (truncated_error(a)); file->uncompressed_size = archive_le32dec(p + CFFILE_cbFile); file->offset = archive_le32dec(p + CFFILE_uoffFolderStart); file->folder = archive_le16dec(p + CFFILE_iFolder); file->mtime = cab_dos_time(p + CFFILE_date_time); file->attr = (uint8_t)archive_le16dec(p + CFFILE_attribs); __archive_read_consume(a, 16); cab->cab_offset += 16; if ((p = cab_read_ahead_remaining(a, 256, &avail)) == NULL) return (truncated_error(a)); if ((len = cab_strnlen(p, avail-1)) <= 0) goto invalid; /* Copy a pathname. */ archive_string_init(&(file->pathname)); archive_strncpy(&(file->pathname), p, len); __archive_read_consume(a, len + 1); cab->cab_offset += len + 1; /* * Sanity check if each data is acceptable. */ if (file->uncompressed_size > 0x7FFF8000) goto invalid;/* Too large */ if ((int64_t)file->offset + (int64_t)file->uncompressed_size > ARCHIVE_LITERAL_LL(0x7FFF8000)) goto invalid;/* Too large */ switch (file->folder) { case iFoldCONTINUED_TO_NEXT: /* This must be last file in a folder. */ if (i != hd->file_count -1) goto invalid; cur_folder = hd->folder_count -1; break; case iFoldCONTINUED_PREV_AND_NEXT: /* This must be only one file in a folder. */ if (hd->file_count != 1) goto invalid; /* FALL THROUGH */ case iFoldCONTINUED_FROM_PREV: /* This must be first file in a folder. */ if (i != 0) goto invalid; prev_folder = cur_folder = 0; offset32 = file->offset; break; default: if (file->folder >= hd->folder_count) goto invalid; cur_folder = file->folder; break; } /* Dot not back track. */ if (cur_folder < prev_folder) goto invalid; if (cur_folder != prev_folder) offset32 = 0; prev_folder = cur_folder; /* Make sure there are not any blanks from last file * contents. */ if (offset32 != file->offset) goto invalid; offset32 += file->uncompressed_size; /* CFDATA is available for file contents. */ if (file->uncompressed_size > 0 && hd->folder_array[cur_folder].cfdata_count == 0) goto invalid; } if (hd->cabinet != 0 || hd->flags & (PREV_CABINET | NEXT_CABINET)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Multivolume cabinet file is unsupported"); return (ARCHIVE_WARN); } return (ARCHIVE_OK); invalid: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid CAB header"); return (ARCHIVE_FATAL); nomem: archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for CAB data"); return (ARCHIVE_FATAL); } static int archive_read_format_cab_read_header(struct archive_read *a, struct archive_entry *entry) { struct cab *cab; struct cfheader *hd; struct cffolder *prev_folder; struct cffile *file; struct archive_string_conv *sconv; int err = ARCHIVE_OK, r; cab = (struct cab *)(a->format->data); if (cab->found_header == 0) { err = cab_read_header(a); if (err < ARCHIVE_WARN) return (err); /* We've found the header. */ cab->found_header = 1; } hd = &cab->cfheader; if (hd->file_index >= hd->file_count) { cab->end_of_archive = 1; return (ARCHIVE_EOF); } file = &hd->file_array[hd->file_index++]; cab->end_of_entry = 0; cab->end_of_entry_cleanup = 0; cab->entry_compressed_bytes_read = 0; cab->entry_uncompressed_bytes_read = 0; cab->entry_unconsumed = 0; cab->entry_cffile = file; /* * Choose a proper folder. */ prev_folder = cab->entry_cffolder; switch (file->folder) { case iFoldCONTINUED_FROM_PREV: case iFoldCONTINUED_PREV_AND_NEXT: cab->entry_cffolder = &hd->folder_array[0]; break; case iFoldCONTINUED_TO_NEXT: cab->entry_cffolder = &hd->folder_array[hd->folder_count-1]; break; default: cab->entry_cffolder = &hd->folder_array[file->folder]; break; } /* If a cffolder of this file is changed, reset a cfdata to read * file contents from next cfdata. */ if (prev_folder != cab->entry_cffolder) cab->entry_cfdata = NULL; /* If a pathname is UTF-8, prepare a string conversion object * for UTF-8 and use it. */ if (file->attr & ATTR_NAME_IS_UTF) { if (cab->sconv_utf8 == NULL) { cab->sconv_utf8 = archive_string_conversion_from_charset( &(a->archive), "UTF-8", 1); if (cab->sconv_utf8 == NULL) return (ARCHIVE_FATAL); } sconv = cab->sconv_utf8; } else if (cab->sconv != NULL) { /* Choose the conversion specified by the option. */ sconv = cab->sconv; } else { /* Choose the default conversion. */ if (!cab->init_default_conversion) { cab->sconv_default = archive_string_default_conversion_for_read( &(a->archive)); cab->init_default_conversion = 1; } sconv = cab->sconv_default; } /* * Set a default value and common data */ r = cab_convert_path_separator_1(&(file->pathname), file->attr); if (archive_entry_copy_pathname_l(entry, file->pathname.s, archive_strlen(&(file->pathname)), 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)); err = ARCHIVE_WARN; } if (r < 0) { /* Convert a path separator '\' -> '/' */ cab_convert_path_separator_2(cab, entry); } archive_entry_set_size(entry, file->uncompressed_size); if (file->attr & ATTR_RDONLY) archive_entry_set_mode(entry, AE_IFREG | 0555); else archive_entry_set_mode(entry, AE_IFREG | 0666); archive_entry_set_mtime(entry, file->mtime, 0); cab->entry_bytes_remaining = file->uncompressed_size; cab->entry_offset = 0; /* We don't need compress data. */ if (file->uncompressed_size == 0) cab->end_of_entry_cleanup = cab->end_of_entry = 1; /* Set up a more descriptive format name. */ sprintf(cab->format_name, "CAB %d.%d (%s)", hd->major, hd->minor, cab->entry_cffolder->compname); a->archive.archive_format_name = cab->format_name; return (err); } static int archive_read_format_cab_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct cab *cab = (struct cab *)(a->format->data); int r; switch (cab->entry_cffile->folder) { case iFoldCONTINUED_FROM_PREV: case iFoldCONTINUED_TO_NEXT: case iFoldCONTINUED_PREV_AND_NEXT: *buff = NULL; *size = 0; *offset = 0; archive_clear_error(&a->archive); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Cannot restore this file split in multivolume."); return (ARCHIVE_FAILED); default: break; } if (cab->read_data_invoked == 0) { if (cab->bytes_skipped) { if (cab->entry_cfdata == NULL) { r = cab_next_cfdata(a); if (r < 0) return (r); } if (cab_consume_cfdata(a, cab->bytes_skipped) < 0) return (ARCHIVE_FATAL); cab->bytes_skipped = 0; } cab->read_data_invoked = 1; } if (cab->entry_unconsumed) { /* Consume as much as the compressor actually used. */ r = (int)cab_consume_cfdata(a, cab->entry_unconsumed); cab->entry_unconsumed = 0; if (r < 0) return (r); } if (cab->end_of_archive || cab->end_of_entry) { if (!cab->end_of_entry_cleanup) { /* End-of-entry cleanup done. */ cab->end_of_entry_cleanup = 1; } *offset = cab->entry_offset; *size = 0; *buff = NULL; return (ARCHIVE_EOF); } return (cab_read_data(a, buff, size, offset)); } static uint32_t cab_checksum_cfdata_4(const void *p, size_t bytes, uint32_t seed) { const unsigned char *b; unsigned u32num; uint32_t sum; u32num = (unsigned)bytes / 4; sum = seed; b = p; for (;u32num > 0; --u32num) { sum ^= archive_le32dec(b); b += 4; } return (sum); } static uint32_t cab_checksum_cfdata(const void *p, size_t bytes, uint32_t seed) { const unsigned char *b; uint32_t sum; uint32_t t; sum = cab_checksum_cfdata_4(p, bytes, seed); b = p; b += bytes & ~3; t = 0; switch (bytes & 3) { case 3: t |= ((uint32_t)(*b++)) << 16; /* FALL THROUGH */ case 2: t |= ((uint32_t)(*b++)) << 8; /* FALL THROUGH */ case 1: t |= *b; /* FALL THROUGH */ default: break; } sum ^= t; return (sum); } static void cab_checksum_update(struct archive_read *a, size_t bytes) { struct cab *cab = (struct cab *)(a->format->data); struct cfdata *cfdata = cab->entry_cfdata; const unsigned char *p; size_t sumbytes; if (cfdata->sum == 0 || cfdata->sum_ptr == NULL) return; /* * Calculate the sum of this CFDATA. * Make sure CFDATA must be calculated in four bytes. */ p = cfdata->sum_ptr; sumbytes = bytes; if (cfdata->sum_extra_avail) { while (cfdata->sum_extra_avail < 4 && sumbytes > 0) { cfdata->sum_extra[ cfdata->sum_extra_avail++] = *p++; sumbytes--; } if (cfdata->sum_extra_avail == 4) { cfdata->sum_calculated = cab_checksum_cfdata_4( cfdata->sum_extra, 4, cfdata->sum_calculated); cfdata->sum_extra_avail = 0; } } if (sumbytes) { int odd = sumbytes & 3; if (sumbytes - odd > 0) cfdata->sum_calculated = cab_checksum_cfdata_4( p, sumbytes - odd, cfdata->sum_calculated); if (odd) memcpy(cfdata->sum_extra, p + sumbytes - odd, odd); cfdata->sum_extra_avail = odd; } cfdata->sum_ptr = NULL; } static int cab_checksum_finish(struct archive_read *a) { struct cab *cab = (struct cab *)(a->format->data); struct cfdata *cfdata = cab->entry_cfdata; int l; /* Do not need to compute a sum. */ if (cfdata->sum == 0) return (ARCHIVE_OK); /* * Calculate the sum of remaining CFDATA. */ if (cfdata->sum_extra_avail) { cfdata->sum_calculated = cab_checksum_cfdata(cfdata->sum_extra, cfdata->sum_extra_avail, cfdata->sum_calculated); cfdata->sum_extra_avail = 0; } l = 4; if (cab->cfheader.flags & RESERVE_PRESENT) l += cab->cfheader.cfdata; cfdata->sum_calculated = cab_checksum_cfdata( cfdata->memimage + CFDATA_cbData, l, cfdata->sum_calculated); if (cfdata->sum_calculated != cfdata->sum) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Checksum error CFDATA[%d] %x:%x in %d bytes", cab->entry_cffolder->cfdata_index -1, cfdata->sum, cfdata->sum_calculated, cfdata->compressed_size); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } /* * Read CFDATA if needed. */ static int cab_next_cfdata(struct archive_read *a) { struct cab *cab = (struct cab *)(a->format->data); struct cfdata *cfdata = cab->entry_cfdata; /* There are remaining bytes in current CFDATA, use it first. */ if (cfdata != NULL && cfdata->uncompressed_bytes_remaining > 0) return (ARCHIVE_OK); if (cfdata == NULL) { int64_t skip; cab->entry_cffolder->cfdata_index = 0; /* Seek read pointer to the offset of CFDATA if needed. */ skip = cab->entry_cffolder->cfdata_offset_in_cab - cab->cab_offset; if (skip < 0) { int folder_index; switch (cab->entry_cffile->folder) { case iFoldCONTINUED_FROM_PREV: case iFoldCONTINUED_PREV_AND_NEXT: folder_index = 0; break; case iFoldCONTINUED_TO_NEXT: folder_index = cab->cfheader.folder_count-1; break; default: folder_index = cab->entry_cffile->folder; break; } archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid offset of CFDATA in folder(%d) %jd < %jd", folder_index, (intmax_t)cab->entry_cffolder->cfdata_offset_in_cab, (intmax_t)cab->cab_offset); return (ARCHIVE_FATAL); } if (skip > 0) { if (__archive_read_consume(a, skip) < 0) return (ARCHIVE_FATAL); cab->cab_offset = cab->entry_cffolder->cfdata_offset_in_cab; } } /* * Read a CFDATA. */ if (cab->entry_cffolder->cfdata_index < cab->entry_cffolder->cfdata_count) { const unsigned char *p; int l; cfdata = &(cab->entry_cffolder->cfdata); cab->entry_cffolder->cfdata_index++; cab->entry_cfdata = cfdata; cfdata->sum_calculated = 0; cfdata->sum_extra_avail = 0; cfdata->sum_ptr = NULL; l = 8; if (cab->cfheader.flags & RESERVE_PRESENT) l += cab->cfheader.cfdata; if ((p = __archive_read_ahead(a, l, NULL)) == NULL) return (truncated_error(a)); cfdata->sum = archive_le32dec(p + CFDATA_csum); cfdata->compressed_size = archive_le16dec(p + CFDATA_cbData); cfdata->compressed_bytes_remaining = cfdata->compressed_size; cfdata->uncompressed_size = archive_le16dec(p + CFDATA_cbUncomp); cfdata->uncompressed_bytes_remaining = cfdata->uncompressed_size; cfdata->uncompressed_avail = 0; cfdata->read_offset = 0; cfdata->unconsumed = 0; /* * Sanity check if data size is acceptable. */ if (cfdata->compressed_size == 0 || cfdata->compressed_size > (0x8000+6144)) goto invalid; if (cfdata->uncompressed_size > 0x8000) goto invalid; if (cfdata->uncompressed_size == 0) { switch (cab->entry_cffile->folder) { case iFoldCONTINUED_PREV_AND_NEXT: case iFoldCONTINUED_TO_NEXT: break; case iFoldCONTINUED_FROM_PREV: default: goto invalid; } } /* If CFDATA is not last in a folder, an uncompressed * size must be 0x8000(32KBi) */ if ((cab->entry_cffolder->cfdata_index < cab->entry_cffolder->cfdata_count) && cfdata->uncompressed_size != 0x8000) goto invalid; /* A compressed data size and an uncompressed data size must * be the same in no compression mode. */ if (cab->entry_cffolder->comptype == COMPTYPE_NONE && cfdata->compressed_size != cfdata->uncompressed_size) goto invalid; /* * Save CFDATA image for sum check. */ if (cfdata->memimage_size < (size_t)l) { free(cfdata->memimage); cfdata->memimage = malloc(l); if (cfdata->memimage == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for CAB data"); return (ARCHIVE_FATAL); } cfdata->memimage_size = l; } memcpy(cfdata->memimage, p, l); /* Consume bytes as much as we used. */ __archive_read_consume(a, l); cab->cab_offset += l; } else if (cab->entry_cffolder->cfdata_count > 0) { /* Run out of all CFDATA in a folder. */ cfdata->compressed_size = 0; cfdata->uncompressed_size = 0; cfdata->compressed_bytes_remaining = 0; cfdata->uncompressed_bytes_remaining = 0; } else { /* Current folder does not have any CFDATA. */ cfdata = &(cab->entry_cffolder->cfdata); cab->entry_cfdata = cfdata; memset(cfdata, 0, sizeof(*cfdata)); } return (ARCHIVE_OK); invalid: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid CFDATA"); return (ARCHIVE_FATAL); } /* * Read ahead CFDATA. */ static const void * cab_read_ahead_cfdata(struct archive_read *a, ssize_t *avail) { struct cab *cab = (struct cab *)(a->format->data); int err; err = cab_next_cfdata(a); if (err < ARCHIVE_OK) { *avail = err; return (NULL); } switch (cab->entry_cffolder->comptype) { case COMPTYPE_NONE: return (cab_read_ahead_cfdata_none(a, avail)); case COMPTYPE_MSZIP: return (cab_read_ahead_cfdata_deflate(a, avail)); case COMPTYPE_LZX: return (cab_read_ahead_cfdata_lzx(a, avail)); default: /* Unsupported compression. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported CAB compression : %s", cab->entry_cffolder->compname); *avail = ARCHIVE_FAILED; return (NULL); } } /* * Read ahead CFDATA as uncompressed data. */ static const void * cab_read_ahead_cfdata_none(struct archive_read *a, ssize_t *avail) { struct cab *cab = (struct cab *)(a->format->data); struct cfdata *cfdata; const void *d; cfdata = cab->entry_cfdata; /* * 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. */ d = __archive_read_ahead(a, 1, avail); if (*avail <= 0) { *avail = truncated_error(a); return (NULL); } if (*avail > cfdata->uncompressed_bytes_remaining) *avail = cfdata->uncompressed_bytes_remaining; cfdata->uncompressed_avail = cfdata->uncompressed_size; cfdata->unconsumed = *avail; cfdata->sum_ptr = d; return (d); } /* * Read ahead CFDATA as deflate data. */ #ifdef HAVE_ZLIB_H static const void * cab_read_ahead_cfdata_deflate(struct archive_read *a, ssize_t *avail) { struct cab *cab = (struct cab *)(a->format->data); struct cfdata *cfdata; const void *d; int r, mszip; uint16_t uavail; char eod = 0; cfdata = cab->entry_cfdata; /* If the buffer hasn't been allocated, allocate it now. */ if (cab->uncompressed_buffer == NULL) { cab->uncompressed_buffer_size = 0x8000; cab->uncompressed_buffer = (unsigned char *)malloc(cab->uncompressed_buffer_size); if (cab->uncompressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for CAB reader"); *avail = ARCHIVE_FATAL; return (NULL); } } uavail = cfdata->uncompressed_avail; if (uavail == cfdata->uncompressed_size) { d = cab->uncompressed_buffer + cfdata->read_offset; *avail = uavail - cfdata->read_offset; return (d); } if (!cab->entry_cffolder->decompress_init) { cab->stream.next_in = NULL; cab->stream.avail_in = 0; cab->stream.total_in = 0; cab->stream.next_out = NULL; cab->stream.avail_out = 0; cab->stream.total_out = 0; if (cab->stream_valid) r = inflateReset(&cab->stream); else r = inflateInit2(&cab->stream, -15 /* Don't check for zlib header */); if (r != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't initialize deflate decompression."); *avail = ARCHIVE_FATAL; return (NULL); } /* Stream structure has been set up. */ cab->stream_valid = 1; /* We've initialized decompression for this stream. */ cab->entry_cffolder->decompress_init = 1; } if (cfdata->compressed_bytes_remaining == cfdata->compressed_size) mszip = 2; else mszip = 0; eod = 0; cab->stream.total_out = uavail; /* * We always uncompress all data in current CFDATA. */ while (!eod && cab->stream.total_out < cfdata->uncompressed_size) { ssize_t bytes_avail; cab->stream.next_out = cab->uncompressed_buffer + cab->stream.total_out; cab->stream.avail_out = cfdata->uncompressed_size - cab->stream.total_out; d = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { *avail = truncated_error(a); return (NULL); } if (bytes_avail > cfdata->compressed_bytes_remaining) bytes_avail = cfdata->compressed_bytes_remaining; /* * 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'. */ cab->stream.next_in = (Bytef *)(uintptr_t)d; cab->stream.avail_in = (uInt)bytes_avail; cab->stream.total_in = 0; /* Cut out a tow-byte MSZIP signature(0x43, 0x4b). */ if (mszip > 0) { if (bytes_avail <= mszip) { if (mszip == 2) { if (cab->stream.next_in[0] != 0x43) goto nomszip; if (bytes_avail > 1 && cab->stream.next_in[1] != 0x4b) goto nomszip; } else if (cab->stream.next_in[0] != 0x4b) goto nomszip; cfdata->unconsumed = bytes_avail; cfdata->sum_ptr = d; if (cab_minimum_consume_cfdata( a, cfdata->unconsumed) < 0) { *avail = ARCHIVE_FATAL; return (NULL); } mszip -= (int)bytes_avail; continue; } if (mszip == 1 && cab->stream.next_in[0] != 0x4b) goto nomszip; else if (cab->stream.next_in[0] != 0x43 || cab->stream.next_in[1] != 0x4b) goto nomszip; cab->stream.next_in += mszip; cab->stream.avail_in -= mszip; cab->stream.total_in += mszip; mszip = 0; } r = inflate(&cab->stream, 0); switch (r) { case Z_OK: break; case Z_STREAM_END: eod = 1; break; default: goto zlibfailed; } cfdata->unconsumed = cab->stream.total_in; cfdata->sum_ptr = d; if (cab_minimum_consume_cfdata(a, cfdata->unconsumed) < 0) { *avail = ARCHIVE_FATAL; return (NULL); } } uavail = (uint16_t)cab->stream.total_out; if (uavail < cfdata->uncompressed_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid uncompressed size (%d < %d)", uavail, cfdata->uncompressed_size); *avail = ARCHIVE_FATAL; return (NULL); } /* * Note: I suspect there is a bug in makecab.exe because, in rare * case, compressed bytes are still remaining regardless we have - * gotten all uncompressed bytes, which size is recoded in CFDATA, + * gotten all uncompressed bytes, which size is recorded in CFDATA, * as much as we need, and we have to use the garbage so as to * correctly compute the sum of CFDATA accordingly. */ if (cfdata->compressed_bytes_remaining > 0) { ssize_t bytes_avail; d = __archive_read_ahead(a, cfdata->compressed_bytes_remaining, &bytes_avail); if (bytes_avail <= 0) { *avail = truncated_error(a); return (NULL); } cfdata->unconsumed = cfdata->compressed_bytes_remaining; cfdata->sum_ptr = d; if (cab_minimum_consume_cfdata(a, cfdata->unconsumed) < 0) { *avail = ARCHIVE_FATAL; return (NULL); } } /* * Set dictionary data for decompressing of next CFDATA, which * in the same folder. This is why we always do decompress CFDATA * even if beginning CFDATA or some of CFDATA are not used in * skipping file data. */ if (cab->entry_cffolder->cfdata_index < cab->entry_cffolder->cfdata_count) { r = inflateReset(&cab->stream); if (r != Z_OK) goto zlibfailed; r = inflateSetDictionary(&cab->stream, cab->uncompressed_buffer, cfdata->uncompressed_size); if (r != Z_OK) goto zlibfailed; } d = cab->uncompressed_buffer + cfdata->read_offset; *avail = uavail - cfdata->read_offset; cfdata->uncompressed_avail = uavail; return (d); zlibfailed: switch (r) { case Z_MEM_ERROR: archive_set_error(&a->archive, ENOMEM, "Out of memory for deflate decompression"); break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Deflate decompression failed (%d)", r); break; } *avail = ARCHIVE_FATAL; return (NULL); nomszip: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "CFDATA incorrect(no MSZIP signature)"); *avail = ARCHIVE_FATAL; return (NULL); } #else /* HAVE_ZLIB_H */ static const void * cab_read_ahead_cfdata_deflate(struct archive_read *a, ssize_t *avail) { *avail = ARCHIVE_FATAL; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "libarchive compiled without deflate support (no libz)"); return (NULL); } #endif /* HAVE_ZLIB_H */ static const void * cab_read_ahead_cfdata_lzx(struct archive_read *a, ssize_t *avail) { struct cab *cab = (struct cab *)(a->format->data); struct cfdata *cfdata; const void *d; int r; uint16_t uavail; cfdata = cab->entry_cfdata; /* If the buffer hasn't been allocated, allocate it now. */ if (cab->uncompressed_buffer == NULL) { cab->uncompressed_buffer_size = 0x8000; cab->uncompressed_buffer = (unsigned char *)malloc(cab->uncompressed_buffer_size); if (cab->uncompressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for CAB reader"); *avail = ARCHIVE_FATAL; return (NULL); } } uavail = cfdata->uncompressed_avail; if (uavail == cfdata->uncompressed_size) { d = cab->uncompressed_buffer + cfdata->read_offset; *avail = uavail - cfdata->read_offset; return (d); } if (!cab->entry_cffolder->decompress_init) { r = lzx_decode_init(&cab->xstrm, cab->entry_cffolder->compdata); if (r != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't initialize LZX decompression."); *avail = ARCHIVE_FATAL; return (NULL); } /* We've initialized decompression for this stream. */ cab->entry_cffolder->decompress_init = 1; } /* Clean up remaining bits of previous CFDATA. */ lzx_cleanup_bitstream(&cab->xstrm); cab->xstrm.total_out = uavail; while (cab->xstrm.total_out < cfdata->uncompressed_size) { ssize_t bytes_avail; cab->xstrm.next_out = cab->uncompressed_buffer + cab->xstrm.total_out; cab->xstrm.avail_out = cfdata->uncompressed_size - cab->xstrm.total_out; d = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated CAB file data"); *avail = ARCHIVE_FATAL; return (NULL); } if (bytes_avail > cfdata->compressed_bytes_remaining) bytes_avail = cfdata->compressed_bytes_remaining; cab->xstrm.next_in = d; cab->xstrm.avail_in = bytes_avail; cab->xstrm.total_in = 0; r = lzx_decode(&cab->xstrm, cfdata->compressed_bytes_remaining == bytes_avail); switch (r) { case ARCHIVE_OK: case ARCHIVE_EOF: break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "LZX decompression failed (%d)", r); *avail = ARCHIVE_FATAL; return (NULL); } cfdata->unconsumed = cab->xstrm.total_in; cfdata->sum_ptr = d; if (cab_minimum_consume_cfdata(a, cfdata->unconsumed) < 0) { *avail = ARCHIVE_FATAL; return (NULL); } } uavail = (uint16_t)cab->xstrm.total_out; /* * Make sure a read pointer advances to next CFDATA. */ if (cfdata->compressed_bytes_remaining > 0) { ssize_t bytes_avail; d = __archive_read_ahead(a, cfdata->compressed_bytes_remaining, &bytes_avail); if (bytes_avail <= 0) { *avail = truncated_error(a); return (NULL); } cfdata->unconsumed = cfdata->compressed_bytes_remaining; cfdata->sum_ptr = d; if (cab_minimum_consume_cfdata(a, cfdata->unconsumed) < 0) { *avail = ARCHIVE_FATAL; return (NULL); } } /* - * Translation reversal of x86 proccessor CALL byte sequence(E8). + * Translation reversal of x86 processor CALL byte sequence(E8). */ lzx_translation(&cab->xstrm, cab->uncompressed_buffer, cfdata->uncompressed_size, (cab->entry_cffolder->cfdata_index-1) * 0x8000); d = cab->uncompressed_buffer + cfdata->read_offset; *avail = uavail - cfdata->read_offset; cfdata->uncompressed_avail = uavail; return (d); } /* * Consume CFDATA. * We always decompress CFDATA to consume CFDATA as much as we need * in uncompressed bytes because all CFDATA in a folder are related * so we do not skip any CFDATA without decompressing. * Note: If the folder of a CFFILE is iFoldCONTINUED_PREV_AND_NEXT or * iFoldCONTINUED_FROM_PREV, we won't decompress because a CFDATA for * the CFFILE is remaining bytes of previous Multivolume CAB file. */ static int64_t cab_consume_cfdata(struct archive_read *a, int64_t consumed_bytes) { struct cab *cab = (struct cab *)(a->format->data); struct cfdata *cfdata; int64_t cbytes, rbytes; int err; rbytes = cab_minimum_consume_cfdata(a, consumed_bytes); if (rbytes < 0) return (ARCHIVE_FATAL); cfdata = cab->entry_cfdata; while (rbytes > 0) { ssize_t avail; if (cfdata->compressed_size == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid CFDATA"); return (ARCHIVE_FATAL); } cbytes = cfdata->uncompressed_bytes_remaining; if (cbytes > rbytes) cbytes = rbytes; rbytes -= cbytes; if (cfdata->uncompressed_avail == 0 && (cab->entry_cffile->folder == iFoldCONTINUED_PREV_AND_NEXT || cab->entry_cffile->folder == iFoldCONTINUED_FROM_PREV)) { /* We have not read any data yet. */ if (cbytes == cfdata->uncompressed_bytes_remaining) { /* Skip whole current CFDATA. */ __archive_read_consume(a, cfdata->compressed_size); cab->cab_offset += cfdata->compressed_size; cfdata->compressed_bytes_remaining = 0; cfdata->uncompressed_bytes_remaining = 0; err = cab_next_cfdata(a); if (err < 0) return (err); cfdata = cab->entry_cfdata; if (cfdata->uncompressed_size == 0) { switch (cab->entry_cffile->folder) { case iFoldCONTINUED_PREV_AND_NEXT: case iFoldCONTINUED_TO_NEXT: case iFoldCONTINUED_FROM_PREV: rbytes = 0; break; default: break; } } continue; } cfdata->read_offset += (uint16_t)cbytes; cfdata->uncompressed_bytes_remaining -= (uint16_t)cbytes; break; } else if (cbytes == 0) { err = cab_next_cfdata(a); if (err < 0) return (err); cfdata = cab->entry_cfdata; if (cfdata->uncompressed_size == 0) { switch (cab->entry_cffile->folder) { case iFoldCONTINUED_PREV_AND_NEXT: case iFoldCONTINUED_TO_NEXT: case iFoldCONTINUED_FROM_PREV: return (ARCHIVE_FATAL); default: break; } } continue; } while (cbytes > 0) { (void)cab_read_ahead_cfdata(a, &avail); if (avail <= 0) return (ARCHIVE_FATAL); if (avail > cbytes) avail = (ssize_t)cbytes; if (cab_minimum_consume_cfdata(a, avail) < 0) return (ARCHIVE_FATAL); cbytes -= avail; } } return (consumed_bytes); } /* * Consume CFDATA as much as we have already gotten and * compute the sum of CFDATA. */ static int64_t cab_minimum_consume_cfdata(struct archive_read *a, int64_t consumed_bytes) { struct cab *cab = (struct cab *)(a->format->data); struct cfdata *cfdata; int64_t cbytes, rbytes; int err; cfdata = cab->entry_cfdata; rbytes = consumed_bytes; if (cab->entry_cffolder->comptype == COMPTYPE_NONE) { if (consumed_bytes < cfdata->unconsumed) cbytes = consumed_bytes; else cbytes = cfdata->unconsumed; rbytes -= cbytes; cfdata->read_offset += (uint16_t)cbytes; cfdata->uncompressed_bytes_remaining -= (uint16_t)cbytes; cfdata->unconsumed -= cbytes; } else { cbytes = cfdata->uncompressed_avail - cfdata->read_offset; if (cbytes > 0) { if (consumed_bytes < cbytes) cbytes = consumed_bytes; rbytes -= cbytes; cfdata->read_offset += (uint16_t)cbytes; cfdata->uncompressed_bytes_remaining -= (uint16_t)cbytes; } if (cfdata->unconsumed) { cbytes = cfdata->unconsumed; cfdata->unconsumed = 0; } else cbytes = 0; } if (cbytes) { /* Compute the sum. */ cab_checksum_update(a, (size_t)cbytes); /* Consume as much as the compressor actually used. */ __archive_read_consume(a, cbytes); cab->cab_offset += cbytes; cfdata->compressed_bytes_remaining -= (uint16_t)cbytes; if (cfdata->compressed_bytes_remaining == 0) { err = cab_checksum_finish(a); if (err < 0) return (err); } } return (rbytes); } /* * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets * cab->end_of_entry if it consumes all of the data. */ static int cab_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct cab *cab = (struct cab *)(a->format->data); ssize_t bytes_avail; if (cab->entry_bytes_remaining == 0) { *buff = NULL; *size = 0; *offset = cab->entry_offset; cab->end_of_entry = 1; return (ARCHIVE_OK); } *buff = cab_read_ahead_cfdata(a, &bytes_avail); if (bytes_avail <= 0) { *buff = NULL; *size = 0; *offset = 0; if (bytes_avail == 0 && cab->entry_cfdata->uncompressed_size == 0) { /* All of CFDATA in a folder has been handled. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid CFDATA"); return (ARCHIVE_FATAL); } else return ((int)bytes_avail); } if (bytes_avail > cab->entry_bytes_remaining) bytes_avail = (ssize_t)cab->entry_bytes_remaining; *size = bytes_avail; *offset = cab->entry_offset; cab->entry_offset += bytes_avail; cab->entry_bytes_remaining -= bytes_avail; if (cab->entry_bytes_remaining == 0) cab->end_of_entry = 1; cab->entry_unconsumed = bytes_avail; if (cab->entry_cffolder->comptype == COMPTYPE_NONE) { /* Don't consume more than current entry used. */ if (cab->entry_cfdata->unconsumed > cab->entry_unconsumed) cab->entry_cfdata->unconsumed = cab->entry_unconsumed; } return (ARCHIVE_OK); } static int archive_read_format_cab_read_data_skip(struct archive_read *a) { struct cab *cab; int64_t bytes_skipped; int r; cab = (struct cab *)(a->format->data); if (cab->end_of_archive) return (ARCHIVE_EOF); if (!cab->read_data_invoked) { cab->bytes_skipped += cab->entry_bytes_remaining; cab->entry_bytes_remaining = 0; /* This entry is finished and done. */ cab->end_of_entry_cleanup = cab->end_of_entry = 1; return (ARCHIVE_OK); } if (cab->entry_unconsumed) { /* Consume as much as the compressor actually used. */ r = (int)cab_consume_cfdata(a, cab->entry_unconsumed); cab->entry_unconsumed = 0; if (r < 0) return (r); } else if (cab->entry_cfdata == NULL) { r = cab_next_cfdata(a); if (r < 0) return (r); } /* if we've already read to end of data, we're done. */ if (cab->end_of_entry_cleanup) return (ARCHIVE_OK); /* * If the length is at the beginning, we can skip the * compressed data much more quickly. */ bytes_skipped = cab_consume_cfdata(a, cab->entry_bytes_remaining); if (bytes_skipped < 0) return (ARCHIVE_FATAL); /* If the compression type is none(uncompressed), we've already * consumed data as much as the current entry size. */ if (cab->entry_cffolder->comptype == COMPTYPE_NONE && cab->entry_cfdata != NULL) cab->entry_cfdata->unconsumed = 0; /* This entry is finished and done. */ cab->end_of_entry_cleanup = cab->end_of_entry = 1; return (ARCHIVE_OK); } static int archive_read_format_cab_cleanup(struct archive_read *a) { struct cab *cab = (struct cab *)(a->format->data); struct cfheader *hd = &cab->cfheader; int i; if (hd->folder_array != NULL) { for (i = 0; i < hd->folder_count; i++) free(hd->folder_array[i].cfdata.memimage); free(hd->folder_array); } if (hd->file_array != NULL) { for (i = 0; i < cab->cfheader.file_count; i++) archive_string_free(&(hd->file_array[i].pathname)); free(hd->file_array); } #ifdef HAVE_ZLIB_H if (cab->stream_valid) inflateEnd(&cab->stream); #endif lzx_decode_free(&cab->xstrm); archive_wstring_free(&cab->ws); free(cab->uncompressed_buffer); free(cab); (a->format->data) = NULL; return (ARCHIVE_OK); } /* Convert an MSDOS-style date/time into Unix-style time. */ static time_t cab_dos_time(const unsigned char *p) { int msTime, msDate; struct tm ts; msDate = archive_le16dec(p); msTime = archive_le16dec(p+2); 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)); } /***************************************************************** * * LZX decompression code. * *****************************************************************/ /* * Initialize LZX decoder. * * Returns ARCHIVE_OK if initialization was successful. * Returns ARCHIVE_FAILED if w_bits has unsupported value. * Returns ARCHIVE_FATAL if initialization failed; memory allocation * error occurred. */ static int lzx_decode_init(struct lzx_stream *strm, int w_bits) { struct lzx_dec *ds; int slot, w_size, w_slot; int base, footer; int base_inc[18]; if (strm->ds == NULL) { strm->ds = calloc(1, sizeof(*strm->ds)); if (strm->ds == NULL) return (ARCHIVE_FATAL); } ds = strm->ds; ds->error = ARCHIVE_FAILED; /* Allow bits from 15(32KBi) up to 21(2MBi) */ if (w_bits < SLOT_BASE || w_bits > SLOT_MAX) return (ARCHIVE_FAILED); ds->error = ARCHIVE_FATAL; /* * Alloc window */ w_size = ds->w_size; w_slot = slots[w_bits - SLOT_BASE]; ds->w_size = 1U << w_bits; ds->w_mask = ds->w_size -1; if (ds->w_buff == NULL || w_size != ds->w_size) { free(ds->w_buff); ds->w_buff = malloc(ds->w_size); if (ds->w_buff == NULL) return (ARCHIVE_FATAL); free(ds->pos_tbl); ds->pos_tbl = malloc(sizeof(ds->pos_tbl[0]) * w_slot); if (ds->pos_tbl == NULL) return (ARCHIVE_FATAL); lzx_huffman_free(&(ds->mt)); } for (footer = 0; footer < 18; footer++) base_inc[footer] = 1 << footer; base = footer = 0; for (slot = 0; slot < w_slot; slot++) { int n; if (footer == 0) base = slot; else base += base_inc[footer]; if (footer < 17) { footer = -2; for (n = base; n; n >>= 1) footer++; if (footer <= 0) footer = 0; } ds->pos_tbl[slot].base = base; ds->pos_tbl[slot].footer_bits = footer; } ds->w_pos = 0; ds->state = 0; ds->br.cache_buffer = 0; ds->br.cache_avail = 0; ds->r0 = ds->r1 = ds->r2 = 1; /* Initialize aligned offset tree. */ if (lzx_huffman_init(&(ds->at), 8, 8) != ARCHIVE_OK) return (ARCHIVE_FATAL); /* Initialize pre-tree. */ if (lzx_huffman_init(&(ds->pt), 20, 10) != ARCHIVE_OK) return (ARCHIVE_FATAL); /* Initialize Main tree. */ if (lzx_huffman_init(&(ds->mt), 256+(w_slot<<3), 16) != ARCHIVE_OK) return (ARCHIVE_FATAL); /* Initialize Length tree. */ if (lzx_huffman_init(&(ds->lt), 249, 16) != ARCHIVE_OK) return (ARCHIVE_FATAL); ds->error = 0; return (ARCHIVE_OK); } /* * Release LZX decoder. */ static void lzx_decode_free(struct lzx_stream *strm) { if (strm->ds == NULL) return; free(strm->ds->w_buff); free(strm->ds->pos_tbl); lzx_huffman_free(&(strm->ds->at)); lzx_huffman_free(&(strm->ds->pt)); lzx_huffman_free(&(strm->ds->mt)); lzx_huffman_free(&(strm->ds->lt)); free(strm->ds); strm->ds = NULL; } /* * E8 Call Translation reversal. */ static void lzx_translation(struct lzx_stream *strm, void *p, size_t size, uint32_t offset) { struct lzx_dec *ds = strm->ds; unsigned char *b, *end; if (!ds->translation || size <= 10) return; b = p; end = b + size - 10; while (b < end && (b = memchr(b, 0xE8, end - b)) != NULL) { size_t i = b - (unsigned char *)p; int32_t cp, displacement, value; cp = (int32_t)(offset + (uint32_t)i); value = archive_le32dec(&b[1]); if (value >= -cp && value < (int32_t)ds->translation_size) { if (value >= 0) displacement = value - cp; else displacement = value + ds->translation_size; archive_le32enc(&b[1], (uint32_t)displacement); } b += 5; } } /* * Bit stream reader. */ /* Check that the cache buffer has enough bits. */ #define lzx_br_has(br, n) ((br)->cache_avail >= n) /* Get compressed data by bit. */ #define lzx_br_bits(br, n) \ (((uint32_t)((br)->cache_buffer >> \ ((br)->cache_avail - (n)))) & cache_masks[n]) #define lzx_br_bits_forced(br, n) \ (((uint32_t)((br)->cache_buffer << \ ((n) - (br)->cache_avail))) & cache_masks[n]) /* Read ahead to make sure the cache buffer has enough compressed data we * will use. * True : completed, there is enough data in the cache buffer. * False : we met that strm->next_in is empty, we have to get following * bytes. */ #define lzx_br_read_ahead_0(strm, br, n) \ (lzx_br_has((br), (n)) || lzx_br_fillup(strm, br)) /* True : the cache buffer has some bits as much as we need. * False : there are no enough bits in the cache buffer to be used, * we have to get following bytes if we could. */ #define lzx_br_read_ahead(strm, br, n) \ (lzx_br_read_ahead_0((strm), (br), (n)) || lzx_br_has((br), (n))) /* Notify how many bits we consumed. */ #define lzx_br_consume(br, n) ((br)->cache_avail -= (n)) #define lzx_br_consume_unaligned_bits(br) ((br)->cache_avail &= ~0x0f) #define lzx_br_is_unaligned(br) ((br)->cache_avail & 0x0f) static const uint32_t cache_masks[] = { 0x00000000, 0x00000001, 0x00000003, 0x00000007, 0x0000000F, 0x0000001F, 0x0000003F, 0x0000007F, 0x000000FF, 0x000001FF, 0x000003FF, 0x000007FF, 0x00000FFF, 0x00001FFF, 0x00003FFF, 0x00007FFF, 0x0000FFFF, 0x0001FFFF, 0x0003FFFF, 0x0007FFFF, 0x000FFFFF, 0x001FFFFF, 0x003FFFFF, 0x007FFFFF, 0x00FFFFFF, 0x01FFFFFF, 0x03FFFFFF, 0x07FFFFFF, 0x0FFFFFFF, 0x1FFFFFFF, 0x3FFFFFFF, 0x7FFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF }; /* * Shift away used bits in the cache data and fill it up with following bits. * Call this when cache buffer does not have enough bits you need. * * Returns 1 if the cache buffer is full. * Returns 0 if the cache buffer is not full; input buffer is empty. */ static int lzx_br_fillup(struct lzx_stream *strm, struct lzx_br *br) { /* - * x86 proccessor family can read misaligned data without an access error. + * x86 processor family can read misaligned data without an access error. */ int n = CACHE_BITS - br->cache_avail; for (;;) { switch (n >> 4) { case 4: if (strm->avail_in >= 8) { br->cache_buffer = ((uint64_t)strm->next_in[1]) << 56 | ((uint64_t)strm->next_in[0]) << 48 | ((uint64_t)strm->next_in[3]) << 40 | ((uint64_t)strm->next_in[2]) << 32 | ((uint32_t)strm->next_in[5]) << 24 | ((uint32_t)strm->next_in[4]) << 16 | ((uint32_t)strm->next_in[7]) << 8 | (uint32_t)strm->next_in[6]; strm->next_in += 8; strm->avail_in -= 8; br->cache_avail += 8 * 8; return (1); } break; case 3: if (strm->avail_in >= 6) { br->cache_buffer = (br->cache_buffer << 48) | ((uint64_t)strm->next_in[1]) << 40 | ((uint64_t)strm->next_in[0]) << 32 | ((uint32_t)strm->next_in[3]) << 24 | ((uint32_t)strm->next_in[2]) << 16 | ((uint32_t)strm->next_in[5]) << 8 | (uint32_t)strm->next_in[4]; strm->next_in += 6; strm->avail_in -= 6; br->cache_avail += 6 * 8; return (1); } break; case 0: /* We have enough compressed data in * the cache buffer.*/ return (1); default: break; } if (strm->avail_in < 2) { /* There is not enough compressed data to * fill up the cache buffer. */ if (strm->avail_in == 1) { br->odd = *strm->next_in++; strm->avail_in--; br->have_odd = 1; } return (0); } br->cache_buffer = (br->cache_buffer << 16) | archive_le16dec(strm->next_in); strm->next_in += 2; strm->avail_in -= 2; br->cache_avail += 16; n -= 16; } } static void lzx_br_fixup(struct lzx_stream *strm, struct lzx_br *br) { int n = CACHE_BITS - br->cache_avail; if (br->have_odd && n >= 16 && strm->avail_in > 0) { br->cache_buffer = (br->cache_buffer << 16) | ((uint16_t)(*strm->next_in)) << 8 | br->odd; strm->next_in++; strm->avail_in--; br->cache_avail += 16; br->have_odd = 0; } } static void lzx_cleanup_bitstream(struct lzx_stream *strm) { strm->ds->br.cache_avail = 0; strm->ds->br.have_odd = 0; } /* * Decode LZX. * * 1. Returns ARCHIVE_OK if output buffer or input buffer are empty. * Please set available buffer and call this function again. * 2. Returns ARCHIVE_EOF if decompression has been completed. * 3. Returns ARCHIVE_FAILED if an error occurred; compressed data * is broken or you do not set 'last' flag properly. */ #define ST_RD_TRANSLATION 0 #define ST_RD_TRANSLATION_SIZE 1 #define ST_RD_BLOCK_TYPE 2 #define ST_RD_BLOCK_SIZE 3 #define ST_RD_ALIGNMENT 4 #define ST_RD_R0 5 #define ST_RD_R1 6 #define ST_RD_R2 7 #define ST_COPY_UNCOMP1 8 #define ST_COPY_UNCOMP2 9 #define ST_RD_ALIGNED_OFFSET 10 #define ST_RD_VERBATIM 11 #define ST_RD_PRE_MAIN_TREE_256 12 #define ST_MAIN_TREE_256 13 #define ST_RD_PRE_MAIN_TREE_REM 14 #define ST_MAIN_TREE_REM 15 #define ST_RD_PRE_LENGTH_TREE 16 #define ST_LENGTH_TREE 17 #define ST_MAIN 18 #define ST_LENGTH 19 #define ST_OFFSET 20 #define ST_REAL_POS 21 #define ST_COPY 22 static int lzx_decode(struct lzx_stream *strm, int last) { struct lzx_dec *ds = strm->ds; int64_t avail_in; int r; if (ds->error) return (ds->error); avail_in = strm->avail_in; lzx_br_fixup(strm, &(ds->br)); do { if (ds->state < ST_MAIN) r = lzx_read_blocks(strm, last); else { int64_t bytes_written = strm->avail_out; r = lzx_decode_blocks(strm, last); bytes_written -= strm->avail_out; strm->next_out += bytes_written; strm->total_out += bytes_written; } } while (r == 100); strm->total_in += avail_in - strm->avail_in; return (r); } static int lzx_read_blocks(struct lzx_stream *strm, int last) { struct lzx_dec *ds = strm->ds; struct lzx_br *br = &(ds->br); int i, r; for (;;) { switch (ds->state) { case ST_RD_TRANSLATION: if (!lzx_br_read_ahead(strm, br, 1)) { ds->state = ST_RD_TRANSLATION; if (last) goto failed; return (ARCHIVE_OK); } ds->translation = lzx_br_bits(br, 1); lzx_br_consume(br, 1); /* FALL THROUGH */ case ST_RD_TRANSLATION_SIZE: if (ds->translation) { if (!lzx_br_read_ahead(strm, br, 32)) { ds->state = ST_RD_TRANSLATION_SIZE; if (last) goto failed; return (ARCHIVE_OK); } ds->translation_size = lzx_br_bits(br, 16); lzx_br_consume(br, 16); ds->translation_size <<= 16; ds->translation_size |= lzx_br_bits(br, 16); lzx_br_consume(br, 16); } /* FALL THROUGH */ case ST_RD_BLOCK_TYPE: if (!lzx_br_read_ahead(strm, br, 3)) { ds->state = ST_RD_BLOCK_TYPE; if (last) goto failed; return (ARCHIVE_OK); } ds->block_type = lzx_br_bits(br, 3); lzx_br_consume(br, 3); /* Check a block type. */ switch (ds->block_type) { case VERBATIM_BLOCK: case ALIGNED_OFFSET_BLOCK: case UNCOMPRESSED_BLOCK: break; default: goto failed;/* Invalid */ } /* FALL THROUGH */ case ST_RD_BLOCK_SIZE: if (!lzx_br_read_ahead(strm, br, 24)) { ds->state = ST_RD_BLOCK_SIZE; if (last) goto failed; return (ARCHIVE_OK); } ds->block_size = lzx_br_bits(br, 8); lzx_br_consume(br, 8); ds->block_size <<= 16; ds->block_size |= lzx_br_bits(br, 16); lzx_br_consume(br, 16); if (ds->block_size == 0) goto failed; ds->block_bytes_avail = ds->block_size; if (ds->block_type != UNCOMPRESSED_BLOCK) { if (ds->block_type == VERBATIM_BLOCK) ds->state = ST_RD_VERBATIM; else ds->state = ST_RD_ALIGNED_OFFSET; break; } /* FALL THROUGH */ case ST_RD_ALIGNMENT: /* * Handle an Uncompressed Block. */ /* Skip padding to align following field on * 16-bit boundary. */ if (lzx_br_is_unaligned(br)) lzx_br_consume_unaligned_bits(br); else { if (lzx_br_read_ahead(strm, br, 16)) lzx_br_consume(br, 16); else { ds->state = ST_RD_ALIGNMENT; if (last) goto failed; return (ARCHIVE_OK); } } /* Preparation to read repeated offsets R0,R1 and R2. */ ds->rbytes_avail = 0; ds->state = ST_RD_R0; /* FALL THROUGH */ case ST_RD_R0: case ST_RD_R1: case ST_RD_R2: do { uint16_t u16; /* Drain bits in the cache buffer of * bit-stream. */ if (lzx_br_has(br, 32)) { u16 = lzx_br_bits(br, 16); lzx_br_consume(br, 16); archive_le16enc(ds->rbytes, u16); u16 = lzx_br_bits(br, 16); lzx_br_consume(br, 16); archive_le16enc(ds->rbytes+2, u16); ds->rbytes_avail = 4; } else if (lzx_br_has(br, 16)) { u16 = lzx_br_bits(br, 16); lzx_br_consume(br, 16); archive_le16enc(ds->rbytes, u16); ds->rbytes_avail = 2; } if (ds->rbytes_avail < 4 && ds->br.have_odd) { ds->rbytes[ds->rbytes_avail++] = ds->br.odd; ds->br.have_odd = 0; } while (ds->rbytes_avail < 4) { if (strm->avail_in <= 0) { if (last) goto failed; return (ARCHIVE_OK); } ds->rbytes[ds->rbytes_avail++] = *strm->next_in++; strm->avail_in--; } ds->rbytes_avail = 0; if (ds->state == ST_RD_R0) { ds->r0 = archive_le32dec(ds->rbytes); if (ds->r0 < 0) goto failed; ds->state = ST_RD_R1; } else if (ds->state == ST_RD_R1) { ds->r1 = archive_le32dec(ds->rbytes); if (ds->r1 < 0) goto failed; ds->state = ST_RD_R2; } else if (ds->state == ST_RD_R2) { ds->r2 = archive_le32dec(ds->rbytes); if (ds->r2 < 0) goto failed; /* We've gotten all repeated offsets. */ ds->state = ST_COPY_UNCOMP1; } } while (ds->state != ST_COPY_UNCOMP1); /* FALL THROUGH */ case ST_COPY_UNCOMP1: /* * Copy bytes form next_in to next_out directly. */ while (ds->block_bytes_avail) { int l; if (strm->avail_out <= 0) /* Output buffer is empty. */ return (ARCHIVE_OK); if (strm->avail_in <= 0) { /* Input buffer is empty. */ if (last) goto failed; return (ARCHIVE_OK); } l = (int)ds->block_bytes_avail; if (l > ds->w_size - ds->w_pos) l = ds->w_size - ds->w_pos; if (l > strm->avail_out) l = (int)strm->avail_out; if (l > strm->avail_in) l = (int)strm->avail_in; memcpy(strm->next_out, strm->next_in, l); memcpy(&(ds->w_buff[ds->w_pos]), strm->next_in, l); strm->next_in += l; strm->avail_in -= l; strm->next_out += l; strm->avail_out -= l; strm->total_out += l; ds->w_pos = (ds->w_pos + l) & ds->w_mask; ds->block_bytes_avail -= l; } /* FALL THROUGH */ case ST_COPY_UNCOMP2: /* Re-align; skip padding byte. */ if (ds->block_size & 1) { if (strm->avail_in <= 0) { /* Input buffer is empty. */ ds->state = ST_COPY_UNCOMP2; if (last) goto failed; return (ARCHIVE_OK); } strm->next_in++; strm->avail_in --; } /* This block ended. */ ds->state = ST_RD_BLOCK_TYPE; return (ARCHIVE_EOF); /********************/ case ST_RD_ALIGNED_OFFSET: /* * Read Aligned offset tree. */ if (!lzx_br_read_ahead(strm, br, 3 * ds->at.len_size)) { ds->state = ST_RD_ALIGNED_OFFSET; if (last) goto failed; return (ARCHIVE_OK); } memset(ds->at.freq, 0, sizeof(ds->at.freq)); for (i = 0; i < ds->at.len_size; i++) { ds->at.bitlen[i] = lzx_br_bits(br, 3); ds->at.freq[ds->at.bitlen[i]]++; lzx_br_consume(br, 3); } if (!lzx_make_huffman_table(&ds->at)) goto failed; /* FALL THROUGH */ case ST_RD_VERBATIM: ds->loop = 0; /* FALL THROUGH */ case ST_RD_PRE_MAIN_TREE_256: /* * Read Pre-tree for first 256 elements of main tree. */ if (!lzx_read_pre_tree(strm)) { ds->state = ST_RD_PRE_MAIN_TREE_256; if (last) goto failed; return (ARCHIVE_OK); } if (!lzx_make_huffman_table(&ds->pt)) goto failed; ds->loop = 0; /* FALL THROUGH */ case ST_MAIN_TREE_256: /* * Get path lengths of first 256 elements of main tree. */ r = lzx_read_bitlen(strm, &ds->mt, 256); if (r < 0) goto failed; else if (!r) { ds->state = ST_MAIN_TREE_256; if (last) goto failed; return (ARCHIVE_OK); } ds->loop = 0; /* FALL THROUGH */ case ST_RD_PRE_MAIN_TREE_REM: /* * Read Pre-tree for remaining elements of main tree. */ if (!lzx_read_pre_tree(strm)) { ds->state = ST_RD_PRE_MAIN_TREE_REM; if (last) goto failed; return (ARCHIVE_OK); } if (!lzx_make_huffman_table(&ds->pt)) goto failed; ds->loop = 256; /* FALL THROUGH */ case ST_MAIN_TREE_REM: /* * Get path lengths of remaining elements of main tree. */ r = lzx_read_bitlen(strm, &ds->mt, -1); if (r < 0) goto failed; else if (!r) { ds->state = ST_MAIN_TREE_REM; if (last) goto failed; return (ARCHIVE_OK); } if (!lzx_make_huffman_table(&ds->mt)) goto failed; ds->loop = 0; /* FALL THROUGH */ case ST_RD_PRE_LENGTH_TREE: /* * Read Pre-tree for remaining elements of main tree. */ if (!lzx_read_pre_tree(strm)) { ds->state = ST_RD_PRE_LENGTH_TREE; if (last) goto failed; return (ARCHIVE_OK); } if (!lzx_make_huffman_table(&ds->pt)) goto failed; ds->loop = 0; /* FALL THROUGH */ case ST_LENGTH_TREE: /* * Get path lengths of remaining elements of main tree. */ r = lzx_read_bitlen(strm, &ds->lt, -1); if (r < 0) goto failed; else if (!r) { ds->state = ST_LENGTH_TREE; if (last) goto failed; return (ARCHIVE_OK); } if (!lzx_make_huffman_table(&ds->lt)) goto failed; ds->state = ST_MAIN; return (100); } } failed: return (ds->error = ARCHIVE_FAILED); } static int lzx_decode_blocks(struct lzx_stream *strm, int last) { struct lzx_dec *ds = strm->ds; struct lzx_br bre = ds->br; struct huffman *at = &(ds->at), *lt = &(ds->lt), *mt = &(ds->mt); const struct lzx_pos_tbl *pos_tbl = ds->pos_tbl; unsigned char *noutp = strm->next_out; unsigned char *endp = noutp + strm->avail_out; unsigned char *w_buff = ds->w_buff; unsigned char *at_bitlen = at->bitlen; unsigned char *lt_bitlen = lt->bitlen; unsigned char *mt_bitlen = mt->bitlen; size_t block_bytes_avail = ds->block_bytes_avail; int at_max_bits = at->max_bits; int lt_max_bits = lt->max_bits; int mt_max_bits = mt->max_bits; int c, copy_len = ds->copy_len, copy_pos = ds->copy_pos; int w_pos = ds->w_pos, w_mask = ds->w_mask, w_size = ds->w_size; int length_header = ds->length_header; int offset_bits = ds->offset_bits; int position_slot = ds->position_slot; int r0 = ds->r0, r1 = ds->r1, r2 = ds->r2; int state = ds->state; char block_type = ds->block_type; for (;;) { switch (state) { case ST_MAIN: for (;;) { if (block_bytes_avail == 0) { /* This block ended. */ ds->state = ST_RD_BLOCK_TYPE; ds->br = bre; ds->block_bytes_avail = block_bytes_avail; ds->copy_len = copy_len; ds->copy_pos = copy_pos; ds->length_header = length_header; ds->position_slot = position_slot; ds->r0 = r0; ds->r1 = r1; ds->r2 = r2; ds->w_pos = w_pos; strm->avail_out = endp - noutp; return (ARCHIVE_EOF); } if (noutp >= endp) /* Output buffer is empty. */ goto next_data; if (!lzx_br_read_ahead(strm, &bre, mt_max_bits)) { if (!last) goto next_data; /* Remaining bits are less than * maximum bits(mt.max_bits) but maybe * it still remains as much as we need, * so we should try to use it with * dummy bits. */ c = lzx_decode_huffman(mt, lzx_br_bits_forced( &bre, mt_max_bits)); lzx_br_consume(&bre, mt_bitlen[c]); if (!lzx_br_has(&bre, 0)) goto failed;/* Over read. */ } else { c = lzx_decode_huffman(mt, lzx_br_bits(&bre, mt_max_bits)); lzx_br_consume(&bre, mt_bitlen[c]); } if (c > UCHAR_MAX) break; /* * 'c' is exactly literal code. */ /* Save a decoded code to reference it * afterward. */ w_buff[w_pos] = c; w_pos = (w_pos + 1) & w_mask; /* Store the decoded code to output buffer. */ *noutp++ = c; block_bytes_avail--; } /* * Get a match code, its length and offset. */ c -= UCHAR_MAX + 1; length_header = c & 7; position_slot = c >> 3; /* FALL THROUGH */ case ST_LENGTH: /* * Get a length. */ if (length_header == 7) { if (!lzx_br_read_ahead(strm, &bre, lt_max_bits)) { if (!last) { state = ST_LENGTH; goto next_data; } c = lzx_decode_huffman(lt, lzx_br_bits_forced( &bre, lt_max_bits)); lzx_br_consume(&bre, lt_bitlen[c]); if (!lzx_br_has(&bre, 0)) goto failed;/* Over read. */ } else { c = lzx_decode_huffman(lt, lzx_br_bits(&bre, lt_max_bits)); lzx_br_consume(&bre, lt_bitlen[c]); } copy_len = c + 7 + 2; } else copy_len = length_header + 2; if ((size_t)copy_len > block_bytes_avail) goto failed; /* * Get an offset. */ switch (position_slot) { case 0: /* Use repeated offset 0. */ copy_pos = r0; state = ST_REAL_POS; continue; case 1: /* Use repeated offset 1. */ copy_pos = r1; /* Swap repeated offset. */ r1 = r0; r0 = copy_pos; state = ST_REAL_POS; continue; case 2: /* Use repeated offset 2. */ copy_pos = r2; /* Swap repeated offset. */ r2 = r0; r0 = copy_pos; state = ST_REAL_POS; continue; default: offset_bits = pos_tbl[position_slot].footer_bits; break; } /* FALL THROUGH */ case ST_OFFSET: /* * Get the offset, which is a distance from * current window position. */ if (block_type == ALIGNED_OFFSET_BLOCK && offset_bits >= 3) { int offbits = offset_bits - 3; if (!lzx_br_read_ahead(strm, &bre, offbits)) { state = ST_OFFSET; if (last) goto failed; goto next_data; } copy_pos = lzx_br_bits(&bre, offbits) << 3; /* Get an aligned number. */ if (!lzx_br_read_ahead(strm, &bre, offbits + at_max_bits)) { if (!last) { state = ST_OFFSET; goto next_data; } lzx_br_consume(&bre, offbits); c = lzx_decode_huffman(at, lzx_br_bits_forced(&bre, at_max_bits)); lzx_br_consume(&bre, at_bitlen[c]); if (!lzx_br_has(&bre, 0)) goto failed;/* Over read. */ } else { lzx_br_consume(&bre, offbits); c = lzx_decode_huffman(at, lzx_br_bits(&bre, at_max_bits)); lzx_br_consume(&bre, at_bitlen[c]); } /* Add an aligned number. */ copy_pos += c; } else { if (!lzx_br_read_ahead(strm, &bre, offset_bits)) { state = ST_OFFSET; if (last) goto failed; goto next_data; } copy_pos = lzx_br_bits(&bre, offset_bits); lzx_br_consume(&bre, offset_bits); } copy_pos += pos_tbl[position_slot].base -2; /* Update repeated offset LRU queue. */ r2 = r1; r1 = r0; r0 = copy_pos; /* FALL THROUGH */ case ST_REAL_POS: /* * Compute a real position in window. */ copy_pos = (w_pos - copy_pos) & w_mask; /* FALL THROUGH */ case ST_COPY: /* * Copy several bytes as extracted data from the window * into the output buffer. */ for (;;) { const unsigned char *s; int l; l = copy_len; if (copy_pos > w_pos) { if (l > w_size - copy_pos) l = w_size - copy_pos; } else { if (l > w_size - w_pos) l = w_size - w_pos; } if (noutp + l >= endp) l = (int)(endp - noutp); s = w_buff + copy_pos; if (l >= 8 && ((copy_pos + l < w_pos) || (w_pos + l < copy_pos))) { memcpy(w_buff + w_pos, s, l); memcpy(noutp, s, l); } else { unsigned char *d; int li; d = w_buff + w_pos; for (li = 0; li < l; li++) noutp[li] = d[li] = s[li]; } noutp += l; copy_pos = (copy_pos + l) & w_mask; w_pos = (w_pos + l) & w_mask; block_bytes_avail -= l; if (copy_len <= l) /* A copy of current pattern ended. */ break; copy_len -= l; if (noutp >= endp) { /* Output buffer is empty. */ state = ST_COPY; goto next_data; } } state = ST_MAIN; break; } } failed: return (ds->error = ARCHIVE_FAILED); next_data: ds->br = bre; ds->block_bytes_avail = block_bytes_avail; ds->copy_len = copy_len; ds->copy_pos = copy_pos; ds->length_header = length_header; ds->offset_bits = offset_bits; ds->position_slot = position_slot; ds->r0 = r0; ds->r1 = r1; ds->r2 = r2; ds->state = state; ds->w_pos = w_pos; strm->avail_out = endp - noutp; return (ARCHIVE_OK); } static int lzx_read_pre_tree(struct lzx_stream *strm) { struct lzx_dec *ds = strm->ds; struct lzx_br *br = &(ds->br); int i; if (ds->loop == 0) memset(ds->pt.freq, 0, sizeof(ds->pt.freq)); for (i = ds->loop; i < ds->pt.len_size; i++) { if (!lzx_br_read_ahead(strm, br, 4)) { ds->loop = i; return (0); } ds->pt.bitlen[i] = lzx_br_bits(br, 4); ds->pt.freq[ds->pt.bitlen[i]]++; lzx_br_consume(br, 4); } ds->loop = i; return (1); } /* * Read a bunch of bit-lengths from pre-tree. */ static int lzx_read_bitlen(struct lzx_stream *strm, struct huffman *d, int end) { struct lzx_dec *ds = strm->ds; struct lzx_br *br = &(ds->br); int c, i, j, ret, same; unsigned rbits; i = ds->loop; if (i == 0) memset(d->freq, 0, sizeof(d->freq)); ret = 0; if (end < 0) end = d->len_size; while (i < end) { ds->loop = i; if (!lzx_br_read_ahead(strm, br, ds->pt.max_bits)) goto getdata; rbits = lzx_br_bits(br, ds->pt.max_bits); c = lzx_decode_huffman(&(ds->pt), rbits); switch (c) { case 17:/* several zero lengths, from 4 to 19. */ if (!lzx_br_read_ahead(strm, br, ds->pt.bitlen[c]+4)) goto getdata; lzx_br_consume(br, ds->pt.bitlen[c]); same = lzx_br_bits(br, 4) + 4; if (i + same > end) return (-1);/* Invalid */ lzx_br_consume(br, 4); for (j = 0; j < same; j++) d->bitlen[i++] = 0; break; case 18:/* many zero lengths, from 20 to 51. */ if (!lzx_br_read_ahead(strm, br, ds->pt.bitlen[c]+5)) goto getdata; lzx_br_consume(br, ds->pt.bitlen[c]); same = lzx_br_bits(br, 5) + 20; if (i + same > end) return (-1);/* Invalid */ lzx_br_consume(br, 5); memset(d->bitlen + i, 0, same); i += same; break; case 19:/* a few same lengths. */ if (!lzx_br_read_ahead(strm, br, ds->pt.bitlen[c]+1+ds->pt.max_bits)) goto getdata; lzx_br_consume(br, ds->pt.bitlen[c]); same = lzx_br_bits(br, 1) + 4; if (i + same > end) return (-1); lzx_br_consume(br, 1); rbits = lzx_br_bits(br, ds->pt.max_bits); c = lzx_decode_huffman(&(ds->pt), rbits); lzx_br_consume(br, ds->pt.bitlen[c]); c = (d->bitlen[i] - c + 17) % 17; if (c < 0) return (-1);/* Invalid */ for (j = 0; j < same; j++) d->bitlen[i++] = c; d->freq[c] += same; break; default: lzx_br_consume(br, ds->pt.bitlen[c]); c = (d->bitlen[i] - c + 17) % 17; if (c < 0) return (-1);/* Invalid */ d->freq[c]++; d->bitlen[i++] = c; break; } } ret = 1; getdata: ds->loop = i; return (ret); } static int lzx_huffman_init(struct huffman *hf, size_t len_size, int tbl_bits) { int bits; if (hf->bitlen == NULL || hf->len_size != (int)len_size) { free(hf->bitlen); hf->bitlen = calloc(len_size, sizeof(hf->bitlen[0])); if (hf->bitlen == NULL) return (ARCHIVE_FATAL); hf->len_size = (int)len_size; } else memset(hf->bitlen, 0, len_size * sizeof(hf->bitlen[0])); if (hf->tbl == NULL) { if (tbl_bits < HTBL_BITS) bits = tbl_bits; else bits = HTBL_BITS; hf->tbl = malloc(((size_t)1 << bits) * sizeof(hf->tbl[0])); if (hf->tbl == NULL) return (ARCHIVE_FATAL); hf->tbl_bits = tbl_bits; } if (hf->tree == NULL && tbl_bits > HTBL_BITS) { hf->tree_avail = 1 << (tbl_bits - HTBL_BITS + 4); hf->tree = malloc(hf->tree_avail * sizeof(hf->tree[0])); if (hf->tree == NULL) return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } static void lzx_huffman_free(struct huffman *hf) { free(hf->bitlen); free(hf->tbl); free(hf->tree); } /* * Make a huffman coding table. */ static int lzx_make_huffman_table(struct huffman *hf) { uint16_t *tbl; const unsigned char *bitlen; int bitptn[17], weight[17]; int i, maxbits = 0, ptn, tbl_size, w; int diffbits, len_avail; /* * Initialize bit patterns. */ ptn = 0; for (i = 1, w = 1 << 15; i <= 16; i++, w >>= 1) { bitptn[i] = ptn; weight[i] = w; if (hf->freq[i]) { ptn += hf->freq[i] * w; maxbits = i; } } if ((ptn & 0xffff) != 0 || maxbits > hf->tbl_bits) return (0);/* Invalid */ hf->max_bits = maxbits; /* * Cut out extra bits which we won't house in the table. * This preparation reduces the same calculation in the for-loop * making the table. */ if (maxbits < 16) { int ebits = 16 - maxbits; for (i = 1; i <= maxbits; i++) { bitptn[i] >>= ebits; weight[i] >>= ebits; } } if (maxbits > HTBL_BITS) { int htbl_max; uint16_t *p; diffbits = maxbits - HTBL_BITS; for (i = 1; i <= HTBL_BITS; i++) { bitptn[i] >>= diffbits; weight[i] >>= diffbits; } htbl_max = bitptn[HTBL_BITS] + weight[HTBL_BITS] * hf->freq[HTBL_BITS]; p = &(hf->tbl[htbl_max]); while (p < &hf->tbl[1U<shift_bits = diffbits; /* * Make the table. */ tbl_size = 1 << HTBL_BITS; tbl = hf->tbl; bitlen = hf->bitlen; len_avail = hf->len_size; hf->tree_used = 0; for (i = 0; i < len_avail; i++) { uint16_t *p; int len, cnt; uint16_t bit; int extlen; struct htree_t *ht; if (bitlen[i] == 0) continue; /* Get a bit pattern */ len = bitlen[i]; ptn = bitptn[len]; cnt = weight[len]; if (len <= HTBL_BITS) { /* Calculate next bit pattern */ if ((bitptn[len] = ptn + cnt) > tbl_size) return (0);/* Invalid */ /* Update the table */ p = &(tbl[ptn]); while (--cnt >= 0) p[cnt] = (uint16_t)i; continue; } /* * A bit length is too big to be housed to a direct table, * so we use a tree model for its extra bits. */ bitptn[len] = ptn + cnt; bit = 1U << (diffbits -1); extlen = len - HTBL_BITS; p = &(tbl[ptn >> diffbits]); if (*p == 0) { *p = len_avail + hf->tree_used; ht = &(hf->tree[hf->tree_used++]); if (hf->tree_used > hf->tree_avail) return (0);/* Invalid */ ht->left = 0; ht->right = 0; } else { if (*p < len_avail || *p >= (len_avail + hf->tree_used)) return (0);/* Invalid */ ht = &(hf->tree[*p - len_avail]); } while (--extlen > 0) { if (ptn & bit) { if (ht->left < len_avail) { ht->left = len_avail + hf->tree_used; ht = &(hf->tree[hf->tree_used++]); if (hf->tree_used > hf->tree_avail) return (0);/* Invalid */ ht->left = 0; ht->right = 0; } else { ht = &(hf->tree[ht->left - len_avail]); } } else { if (ht->right < len_avail) { ht->right = len_avail + hf->tree_used; ht = &(hf->tree[hf->tree_used++]); if (hf->tree_used > hf->tree_avail) return (0);/* Invalid */ ht->left = 0; ht->right = 0; } else { ht = &(hf->tree[ht->right - len_avail]); } } bit >>= 1; } if (ptn & bit) { if (ht->left != 0) return (0);/* Invalid */ ht->left = (uint16_t)i; } else { if (ht->right != 0) return (0);/* Invalid */ ht->right = (uint16_t)i; } } return (1); } static int lzx_decode_huffman_tree(struct huffman *hf, unsigned rbits, int c) { struct htree_t *ht; int extlen; ht = hf->tree; extlen = hf->shift_bits; while (c >= hf->len_size) { c -= hf->len_size; if (extlen-- <= 0 || c >= hf->tree_used) return (0); if (rbits & (1U << extlen)) c = ht[c].left; else c = ht[c].right; } return (c); } static inline int lzx_decode_huffman(struct huffman *hf, unsigned rbits) { int c; /* * At first search an index table for a bit pattern. * If it fails, search a huffman tree for. */ c = hf->tbl[rbits >> hf->shift_bits]; if (c < hf->len_size) return (c); /* This bit pattern needs to be found out at a huffman tree. */ return (lzx_decode_huffman_tree(hf, rbits, c)); } Index: head/contrib/libarchive/libarchive/archive_read_support_format_tar.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_format_tar.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_read_support_format_tar.c (revision 309590) @@ -1,2832 +1,2839 @@ /*- * Copyright (c) 2003-2007 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("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_acl_private.h" /* For ACL parsing routines. */ #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_read_private.h" #define tar_min(a,b) ((a) < (b) ? (a) : (b)) /* * Layout of POSIX 'ustar' tar header. */ struct archive_entry_header_ustar { char name[100]; char mode[8]; char uid[8]; char gid[8]; char size[12]; char mtime[12]; char checksum[8]; char typeflag[1]; char linkname[100]; /* "old format" header ends here */ char magic[6]; /* For POSIX: "ustar\0" */ char version[2]; /* For POSIX: "00" */ char uname[32]; char gname[32]; char rdevmajor[8]; char rdevminor[8]; char prefix[155]; }; /* * Structure of GNU tar header */ struct gnu_sparse { char offset[12]; char numbytes[12]; }; struct archive_entry_header_gnutar { char name[100]; char mode[8]; char uid[8]; char gid[8]; char size[12]; char mtime[12]; char checksum[8]; char typeflag[1]; char linkname[100]; char magic[8]; /* "ustar \0" (note blank/blank/null at end) */ char uname[32]; char gname[32]; char rdevmajor[8]; char rdevminor[8]; char atime[12]; char ctime[12]; char offset[12]; char longnames[4]; char unused[1]; struct gnu_sparse sparse[4]; char isextended[1]; char realsize[12]; /* * Old GNU format doesn't use POSIX 'prefix' field; they use * the 'L' (longname) entry instead. */ }; /* * Data specific to this format. */ struct sparse_block { struct sparse_block *next; int64_t offset; int64_t remaining; int hole; }; struct tar { struct archive_string acl_text; struct archive_string entry_pathname; /* For "GNU.sparse.name" and other similar path extensions. */ struct archive_string entry_pathname_override; struct archive_string entry_linkpath; struct archive_string entry_uname; struct archive_string entry_gname; struct archive_string longlink; struct archive_string longname; struct archive_string pax_header; struct archive_string pax_global; struct archive_string line; int pax_hdrcharset_binary; int header_recursion_depth; int64_t entry_bytes_remaining; int64_t entry_offset; int64_t entry_padding; int64_t entry_bytes_unconsumed; int64_t realsize; int sparse_allowed; struct sparse_block *sparse_list; struct sparse_block *sparse_last; int64_t sparse_offset; int64_t sparse_numbytes; int sparse_gnu_major; int sparse_gnu_minor; char sparse_gnu_pending; struct archive_string localname; struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv; struct archive_string_conv *sconv_acl; struct archive_string_conv *sconv_default; int init_default_conversion; int compat_2x; int process_mac_extensions; int read_concatenated_archives; }; static int archive_block_is_null(const char *p); static char *base64_decode(const char *, size_t, size_t *); static int gnu_add_sparse_entry(struct archive_read *, struct tar *, int64_t offset, int64_t remaining); static void gnu_clear_sparse_list(struct tar *); static int gnu_sparse_old_read(struct archive_read *, struct tar *, const struct archive_entry_header_gnutar *header, size_t *); static int gnu_sparse_old_parse(struct archive_read *, struct tar *, const struct gnu_sparse *sparse, int length); static int gnu_sparse_01_parse(struct archive_read *, struct tar *, const char *); static ssize_t gnu_sparse_10_read(struct archive_read *, struct tar *, size_t *); static int header_Solaris_ACL(struct archive_read *, struct tar *, struct archive_entry *, const void *, size_t *); static int header_common(struct archive_read *, struct tar *, struct archive_entry *, const void *); static int header_old_tar(struct archive_read *, struct tar *, struct archive_entry *, const void *); static int header_pax_extensions(struct archive_read *, struct tar *, struct archive_entry *, const void *, size_t *); static int header_pax_global(struct archive_read *, struct tar *, struct archive_entry *, const void *h, size_t *); static int header_longlink(struct archive_read *, struct tar *, struct archive_entry *, const void *h, size_t *); static int header_longname(struct archive_read *, struct tar *, struct archive_entry *, const void *h, size_t *); static int read_mac_metadata_blob(struct archive_read *, struct tar *, struct archive_entry *, const void *h, size_t *); static int header_volume(struct archive_read *, struct tar *, struct archive_entry *, const void *h, size_t *); static int header_ustar(struct archive_read *, struct tar *, struct archive_entry *, const void *h); static int header_gnutar(struct archive_read *, struct tar *, struct archive_entry *, const void *h, size_t *); static int archive_read_format_tar_bid(struct archive_read *, int); static int archive_read_format_tar_options(struct archive_read *, const char *, const char *); static int archive_read_format_tar_cleanup(struct archive_read *); static int archive_read_format_tar_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset); static int archive_read_format_tar_skip(struct archive_read *a); static int archive_read_format_tar_read_header(struct archive_read *, struct archive_entry *); static int checksum(struct archive_read *, const void *); static int pax_attribute(struct archive_read *, struct tar *, struct archive_entry *, const char *key, const char *value); static int pax_header(struct archive_read *, struct tar *, struct archive_entry *, char *attr); static void pax_time(const char *, int64_t *sec, long *nanos); static ssize_t readline(struct archive_read *, struct tar *, const char **, ssize_t limit, size_t *); static int read_body_to_string(struct archive_read *, struct tar *, struct archive_string *, const void *h, size_t *); static int solaris_sparse_parse(struct archive_read *, struct tar *, struct archive_entry *, const char *); static int64_t tar_atol(const char *, size_t); static int64_t tar_atol10(const char *, size_t); static int64_t tar_atol256(const char *, size_t); static int64_t tar_atol8(const char *, size_t); static int tar_read_header(struct archive_read *, struct tar *, struct archive_entry *, size_t *); static int tohex(int c); static char *url_decode(const char *); static void tar_flush_unconsumed(struct archive_read *, size_t *); int archive_read_support_format_gnutar(struct archive *a) { archive_check_magic(a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_gnutar"); return (archive_read_support_format_tar(a)); } int archive_read_support_format_tar(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct tar *tar; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_tar"); tar = (struct tar *)calloc(1, sizeof(*tar)); #ifdef HAVE_COPYFILE_H /* Set this by default on Mac OS. */ tar->process_mac_extensions = 1; #endif if (tar == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate tar data"); return (ARCHIVE_FATAL); } r = __archive_read_register_format(a, tar, "tar", archive_read_format_tar_bid, archive_read_format_tar_options, archive_read_format_tar_read_header, archive_read_format_tar_read_data, archive_read_format_tar_skip, NULL, archive_read_format_tar_cleanup, NULL, NULL); if (r != ARCHIVE_OK) free(tar); return (ARCHIVE_OK); } static int archive_read_format_tar_cleanup(struct archive_read *a) { struct tar *tar; tar = (struct tar *)(a->format->data); gnu_clear_sparse_list(tar); archive_string_free(&tar->acl_text); archive_string_free(&tar->entry_pathname); archive_string_free(&tar->entry_pathname_override); archive_string_free(&tar->entry_linkpath); archive_string_free(&tar->entry_uname); archive_string_free(&tar->entry_gname); archive_string_free(&tar->line); archive_string_free(&tar->pax_global); archive_string_free(&tar->pax_header); archive_string_free(&tar->longname); archive_string_free(&tar->longlink); archive_string_free(&tar->localname); free(tar); (a->format->data) = NULL; return (ARCHIVE_OK); } /* * Validate number field * * This has to be pretty lenient in order to accomodate the enormous * variety of tar writers in the world: - * = POSIX ustar requires octal values with leading zeros and - * specific termination on fields + * = POSIX (IEEE Std 1003.1-1988) ustar requires octal values with leading + * zeros and allows fields to be terminated with space or null characters * = Many writers use different termination (in particular, libarchive * omits terminator bytes to squeeze one or two more digits) * = Many writers pad with space and omit leading zeros * = GNU tar and star write base-256 values if numbers are too * big to be represented in octal + * + * Examples of specific tar headers that we should support: + * = Perl Archive::Tar terminates uid, gid, devminor and devmajor with two + * null bytes, pads size with spaces and other numeric fields with zeroes + * = plexus-archiver prior to 2.6.3 (before switching to commons-compress) + * may have uid and gid fields filled with spaces without any octal digits + * at all and pads all numeric fields with spaces * * This should tolerate all variants in use. It will reject a field * where the writer just left garbage after a trailing NUL. */ static int validate_number_field(const char* p_field, size_t i_size) { unsigned char marker = (unsigned char)p_field[0]; if (marker == 128 || marker == 255 || marker == 0) { /* Base-256 marker, there's nothing we can check. */ return 1; } else { /* Must be octal */ size_t i = 0; /* Skip any leading spaces */ while (i < i_size && p_field[i] == ' ') { ++i; } /* Skip octal digits. */ while (i < i_size && p_field[i] >= '0' && p_field[i] <= '7') { ++i; } /* Any remaining characters must be space or NUL padding. */ while (i < i_size) { if (p_field[i] != ' ' && p_field[i] != 0) { return 0; } ++i; } return 1; } } static int archive_read_format_tar_bid(struct archive_read *a, int best_bid) { int bid; const char *h; const struct archive_entry_header_ustar *header; (void)best_bid; /* UNUSED */ bid = 0; /* Now let's look at the actual header and see if it matches. */ h = __archive_read_ahead(a, 512, NULL); if (h == NULL) return (-1); /* If it's an end-of-archive mark, we can handle it. */ if (h[0] == 0 && archive_block_is_null(h)) { /* * Usually, I bid the number of bits verified, but * in this case, 4096 seems excessive so I picked 10 as * an arbitrary but reasonable-seeming value. */ return (10); } /* If it's not an end-of-archive mark, it must have a valid checksum.*/ if (!checksum(a, h)) return (0); bid += 48; /* Checksum is usually 6 octal digits. */ header = (const struct archive_entry_header_ustar *)h; /* Recognize POSIX formats. */ if ((memcmp(header->magic, "ustar\0", 6) == 0) && (memcmp(header->version, "00", 2) == 0)) bid += 56; /* Recognize GNU tar format. */ if ((memcmp(header->magic, "ustar ", 6) == 0) && (memcmp(header->version, " \0", 2) == 0)) bid += 56; /* Type flag must be null, digit or A-Z, a-z. */ if (header->typeflag[0] != 0 && !( header->typeflag[0] >= '0' && header->typeflag[0] <= '9') && !( header->typeflag[0] >= 'A' && header->typeflag[0] <= 'Z') && !( header->typeflag[0] >= 'a' && header->typeflag[0] <= 'z') ) return (0); bid += 2; /* 6 bits of variation in an 8-bit field leaves 2 bits. */ /* * Check format of mode/uid/gid/mtime/size/rdevmajor/rdevminor fields. */ if (bid > 0 && ( validate_number_field(header->mode, sizeof(header->mode)) == 0 || validate_number_field(header->uid, sizeof(header->uid)) == 0 || validate_number_field(header->gid, sizeof(header->gid)) == 0 || validate_number_field(header->mtime, sizeof(header->mtime)) == 0 || validate_number_field(header->size, sizeof(header->size)) == 0 || validate_number_field(header->rdevmajor, sizeof(header->rdevmajor)) == 0 || validate_number_field(header->rdevminor, sizeof(header->rdevminor)) == 0)) { bid = 0; } return (bid); } static int archive_read_format_tar_options(struct archive_read *a, const char *key, const char *val) { struct tar *tar; int ret = ARCHIVE_FAILED; tar = (struct tar *)(a->format->data); if (strcmp(key, "compat-2x") == 0) { /* Handle UTF-8 filnames as libarchive 2.x */ tar->compat_2x = (val != NULL && val[0] != 0); tar->init_default_conversion = tar->compat_2x; return (ARCHIVE_OK); } else if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "tar: hdrcharset option needs a character-set name"); else { tar->opt_sconv = archive_string_conversion_from_charset( &a->archive, val, 0); if (tar->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } else if (strcmp(key, "mac-ext") == 0) { tar->process_mac_extensions = (val != NULL && val[0] != 0); return (ARCHIVE_OK); } else if (strcmp(key, "read_concatenated_archives") == 0) { tar->read_concatenated_archives = (val != NULL && val[0] != 0); return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } /* utility function- this exists to centralize the logic of tracking * how much unconsumed data we have floating around, and to consume * anything outstanding since we're going to do read_aheads */ static void tar_flush_unconsumed(struct archive_read *a, size_t *unconsumed) { if (*unconsumed) { /* void *data = (void *)__archive_read_ahead(a, *unconsumed, NULL); * this block of code is to poison claimed unconsumed space, ensuring * things break if it is in use still. * currently it WILL break things, so enable it only for debugging this issue if (data) { memset(data, 0xff, *unconsumed); } */ __archive_read_consume(a, *unconsumed); *unconsumed = 0; } } /* * The function invoked by archive_read_next_header(). This * just sets up a few things and then calls the internal * tar_read_header() function below. */ static int archive_read_format_tar_read_header(struct archive_read *a, struct archive_entry *entry) { /* * When converting tar archives to cpio archives, it is * essential that each distinct file have a distinct inode * number. To simplify this, we keep a static count here to * assign fake dev/inode numbers to each tar entry. Note that * pax format archives may overwrite this with something more * useful. * * Ideally, we would track every file read from the archive so * that we could assign the same dev/ino pair to hardlinks, * but the memory required to store a complete lookup table is * probably not worthwhile just to support the relatively * obscure tar->cpio conversion case. */ static int default_inode; static int default_dev; struct tar *tar; const char *p; const wchar_t *wp; int r; size_t l, unconsumed = 0; /* Assign default device/inode values. */ archive_entry_set_dev(entry, 1 + default_dev); /* Don't use zero. */ archive_entry_set_ino(entry, ++default_inode); /* Don't use zero. */ /* Limit generated st_ino number to 16 bits. */ if (default_inode >= 0xffff) { ++default_dev; default_inode = 0; } tar = (struct tar *)(a->format->data); tar->entry_offset = 0; gnu_clear_sparse_list(tar); tar->realsize = -1; /* Mark this as "unset" */ /* Setup default string conversion. */ tar->sconv = tar->opt_sconv; if (tar->sconv == NULL) { if (!tar->init_default_conversion) { tar->sconv_default = archive_string_default_conversion_for_read(&(a->archive)); tar->init_default_conversion = 1; } tar->sconv = tar->sconv_default; } r = tar_read_header(a, tar, entry, &unconsumed); tar_flush_unconsumed(a, &unconsumed); /* * "non-sparse" files are really just sparse files with * a single block. */ if (tar->sparse_list == NULL) { if (gnu_add_sparse_entry(a, tar, 0, tar->entry_bytes_remaining) != ARCHIVE_OK) return (ARCHIVE_FATAL); } else { struct sparse_block *sb; for (sb = tar->sparse_list; sb != NULL; sb = sb->next) { if (!sb->hole) archive_entry_sparse_add_entry(entry, sb->offset, sb->remaining); } } if (r == ARCHIVE_OK && archive_entry_filetype(entry) == AE_IFREG) { /* * "Regular" entry with trailing '/' is really * directory: This is needed for certain old tar * variants and even for some broken newer ones. */ if ((wp = archive_entry_pathname_w(entry)) != NULL) { l = wcslen(wp); if (l > 0 && wp[l - 1] == L'/') { archive_entry_set_filetype(entry, AE_IFDIR); } } else if ((p = archive_entry_pathname(entry)) != NULL) { l = strlen(p); if (l > 0 && p[l - 1] == '/') { archive_entry_set_filetype(entry, AE_IFDIR); } } } return (r); } static int archive_read_format_tar_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { ssize_t bytes_read; struct tar *tar; struct sparse_block *p; tar = (struct tar *)(a->format->data); for (;;) { /* Remove exhausted entries from sparse list. */ while (tar->sparse_list != NULL && tar->sparse_list->remaining == 0) { p = tar->sparse_list; tar->sparse_list = p->next; free(p); } if (tar->entry_bytes_unconsumed) { __archive_read_consume(a, tar->entry_bytes_unconsumed); tar->entry_bytes_unconsumed = 0; } /* If we're at end of file, return EOF. */ if (tar->sparse_list == NULL || tar->entry_bytes_remaining == 0) { if (__archive_read_consume(a, tar->entry_padding) < 0) return (ARCHIVE_FATAL); tar->entry_padding = 0; *buff = NULL; *size = 0; *offset = tar->realsize; return (ARCHIVE_EOF); } *buff = __archive_read_ahead(a, 1, &bytes_read); if (bytes_read < 0) return (ARCHIVE_FATAL); if (*buff == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Truncated tar archive"); return (ARCHIVE_FATAL); } if (bytes_read > tar->entry_bytes_remaining) bytes_read = (ssize_t)tar->entry_bytes_remaining; /* Don't read more than is available in the * current sparse block. */ if (tar->sparse_list->remaining < bytes_read) bytes_read = (ssize_t)tar->sparse_list->remaining; *size = bytes_read; *offset = tar->sparse_list->offset; tar->sparse_list->remaining -= bytes_read; tar->sparse_list->offset += bytes_read; tar->entry_bytes_remaining -= bytes_read; tar->entry_bytes_unconsumed = bytes_read; if (!tar->sparse_list->hole) return (ARCHIVE_OK); /* Current is hole data and skip this. */ } } static int archive_read_format_tar_skip(struct archive_read *a) { int64_t bytes_skipped; int64_t request; struct sparse_block *p; struct tar* tar; tar = (struct tar *)(a->format->data); /* Do not consume the hole of a sparse file. */ request = 0; for (p = tar->sparse_list; p != NULL; p = p->next) { if (!p->hole) { if (p->remaining >= INT64_MAX - request) { return ARCHIVE_FATAL; } request += p->remaining; } } if (request > tar->entry_bytes_remaining) request = tar->entry_bytes_remaining; request += tar->entry_padding + tar->entry_bytes_unconsumed; bytes_skipped = __archive_read_consume(a, request); if (bytes_skipped < 0) return (ARCHIVE_FATAL); tar->entry_bytes_remaining = 0; tar->entry_bytes_unconsumed = 0; tar->entry_padding = 0; /* Free the sparse list. */ gnu_clear_sparse_list(tar); return (ARCHIVE_OK); } /* * This function recursively interprets all of the headers associated * with a single entry. */ static int tar_read_header(struct archive_read *a, struct tar *tar, struct archive_entry *entry, size_t *unconsumed) { ssize_t bytes; int err; const char *h; const struct archive_entry_header_ustar *header; const struct archive_entry_header_gnutar *gnuheader; /* Loop until we find a workable header record. */ for (;;) { tar_flush_unconsumed(a, unconsumed); /* Read 512-byte header record */ h = __archive_read_ahead(a, 512, &bytes); if (bytes < 0) return ((int)bytes); if (bytes == 0) { /* EOF at a block boundary. */ /* Some writers do omit the block of nulls. */ return (ARCHIVE_EOF); } if (bytes < 512) { /* Short block at EOF; this is bad. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated tar archive"); return (ARCHIVE_FATAL); } *unconsumed = 512; /* Header is workable if it's not an end-of-archive mark. */ if (h[0] != 0 || !archive_block_is_null(h)) break; /* Ensure format is set for archives with only null blocks. */ if (a->archive.archive_format_name == NULL) { a->archive.archive_format = ARCHIVE_FORMAT_TAR; a->archive.archive_format_name = "tar"; } if (!tar->read_concatenated_archives) { /* Try to consume a second all-null record, as well. */ tar_flush_unconsumed(a, unconsumed); h = __archive_read_ahead(a, 512, NULL); if (h != NULL && h[0] == 0 && archive_block_is_null(h)) __archive_read_consume(a, 512); archive_clear_error(&a->archive); return (ARCHIVE_EOF); } /* * We're reading concatenated archives, ignore this block and * loop to get the next. */ } /* * Note: If the checksum fails and we return ARCHIVE_RETRY, * then the client is likely to just retry. This is a very * crude way to search for the next valid header! * * TODO: Improve this by implementing a real header scan. */ if (!checksum(a, h)) { tar_flush_unconsumed(a, unconsumed); archive_set_error(&a->archive, EINVAL, "Damaged tar archive"); return (ARCHIVE_RETRY); /* Retryable: Invalid header */ } if (++tar->header_recursion_depth > 32) { tar_flush_unconsumed(a, unconsumed); archive_set_error(&a->archive, EINVAL, "Too many special headers"); return (ARCHIVE_WARN); } /* Determine the format variant. */ header = (const struct archive_entry_header_ustar *)h; switch(header->typeflag[0]) { case 'A': /* Solaris tar ACL */ a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "Solaris tar"; err = header_Solaris_ACL(a, tar, entry, h, unconsumed); break; case 'g': /* POSIX-standard 'g' header. */ a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "POSIX pax interchange format"; err = header_pax_global(a, tar, entry, h, unconsumed); if (err == ARCHIVE_EOF) return (err); break; case 'K': /* Long link name (GNU tar, others) */ err = header_longlink(a, tar, entry, h, unconsumed); break; case 'L': /* Long filename (GNU tar, others) */ err = header_longname(a, tar, entry, h, unconsumed); break; case 'V': /* GNU volume header */ err = header_volume(a, tar, entry, h, unconsumed); break; case 'X': /* Used by SUN tar; same as 'x'. */ a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "POSIX pax interchange format (Sun variant)"; err = header_pax_extensions(a, tar, entry, h, unconsumed); break; case 'x': /* POSIX-standard 'x' header. */ a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "POSIX pax interchange format"; err = header_pax_extensions(a, tar, entry, h, unconsumed); break; default: gnuheader = (const struct archive_entry_header_gnutar *)h; if (memcmp(gnuheader->magic, "ustar \0", 8) == 0) { a->archive.archive_format = ARCHIVE_FORMAT_TAR_GNUTAR; a->archive.archive_format_name = "GNU tar format"; err = header_gnutar(a, tar, entry, h, unconsumed); } else if (memcmp(header->magic, "ustar", 5) == 0) { if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE) { a->archive.archive_format = ARCHIVE_FORMAT_TAR_USTAR; a->archive.archive_format_name = "POSIX ustar format"; } err = header_ustar(a, tar, entry, h); } else { a->archive.archive_format = ARCHIVE_FORMAT_TAR; a->archive.archive_format_name = "tar (non-POSIX)"; err = header_old_tar(a, tar, entry, h); } } if (err == ARCHIVE_FATAL) return (err); tar_flush_unconsumed(a, unconsumed); h = NULL; header = NULL; --tar->header_recursion_depth; /* Yuck. Apple's design here ends up storing long pathname * extensions for both the AppleDouble extension entry and the * regular entry. */ if ((err == ARCHIVE_WARN || err == ARCHIVE_OK) && tar->header_recursion_depth == 0 && tar->process_mac_extensions) { int err2 = read_mac_metadata_blob(a, tar, entry, h, unconsumed); if (err2 < err) err = err2; } /* We return warnings or success as-is. Anything else is fatal. */ if (err == ARCHIVE_WARN || err == ARCHIVE_OK) { if (tar->sparse_gnu_pending) { if (tar->sparse_gnu_major == 1 && tar->sparse_gnu_minor == 0) { ssize_t bytes_read; tar->sparse_gnu_pending = 0; /* Read initial sparse map. */ bytes_read = gnu_sparse_10_read(a, tar, unconsumed); tar->entry_bytes_remaining -= bytes_read; if (bytes_read < 0) return ((int)bytes_read); } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unrecognized GNU sparse file format"); return (ARCHIVE_WARN); } tar->sparse_gnu_pending = 0; } return (err); } if (err == ARCHIVE_EOF) /* EOF when recursively reading a header is bad. */ archive_set_error(&a->archive, EINVAL, "Damaged tar archive"); return (ARCHIVE_FATAL); } /* * Return true if block checksum is correct. */ static int checksum(struct archive_read *a, const void *h) { const unsigned char *bytes; const struct archive_entry_header_ustar *header; int check, sum; size_t i; (void)a; /* UNUSED */ bytes = (const unsigned char *)h; header = (const struct archive_entry_header_ustar *)h; /* Checksum field must hold an octal number */ for (i = 0; i < sizeof(header->checksum); ++i) { char c = header->checksum[i]; if (c != ' ' && c != '\0' && (c < '0' || c > '7')) return 0; } /* * Test the checksum. Note that POSIX specifies _unsigned_ * bytes for this calculation. */ sum = (int)tar_atol(header->checksum, sizeof(header->checksum)); check = 0; for (i = 0; i < 148; i++) check += (unsigned char)bytes[i]; for (; i < 156; i++) check += 32; for (; i < 512; i++) check += (unsigned char)bytes[i]; if (sum == check) return (1); /* * Repeat test with _signed_ bytes, just in case this archive * was created by an old BSD, Solaris, or HP-UX tar with a * broken checksum calculation. */ check = 0; for (i = 0; i < 148; i++) check += (signed char)bytes[i]; for (; i < 156; i++) check += 32; for (; i < 512; i++) check += (signed char)bytes[i]; if (sum == check) return (1); return (0); } /* * Return true if this block contains only nulls. */ static int archive_block_is_null(const char *p) { unsigned i; for (i = 0; i < 512; i++) if (*p++) return (0); return (1); } /* * Interpret 'A' Solaris ACL header */ static int header_Solaris_ACL(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { const struct archive_entry_header_ustar *header; size_t size; int err; int64_t type; char *acl, *p; /* * read_body_to_string adds a NUL terminator, but we need a little * more to make sure that we don't overrun acl_text later. */ header = (const struct archive_entry_header_ustar *)h; size = (size_t)tar_atol(header->size, sizeof(header->size)); err = read_body_to_string(a, tar, &(tar->acl_text), h, unconsumed); if (err != ARCHIVE_OK) return (err); /* Recursively read next header */ err = tar_read_header(a, tar, entry, unconsumed); if ((err != ARCHIVE_OK) && (err != ARCHIVE_WARN)) return (err); /* TODO: Examine the first characters to see if this * is an AIX ACL descriptor. We'll likely never support * them, but it would be polite to recognize and warn when * we do see them. */ /* Leading octal number indicates ACL type and number of entries. */ p = acl = tar->acl_text.s; type = 0; while (*p != '\0' && p < acl + size) { if (*p < '0' || *p > '7') { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (invalid digit)"); return(ARCHIVE_WARN); } type <<= 3; type += *p - '0'; if (type > 077777777) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (count too large)"); return (ARCHIVE_WARN); } p++; } switch ((int)type & ~0777777) { case 01000000: /* POSIX.1e ACL */ break; case 03000000: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Solaris NFSv4 ACLs not supported"); return (ARCHIVE_WARN); default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (unsupported type %o)", (int)type); return (ARCHIVE_WARN); } p++; if (p >= acl + size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (body overflow)"); return(ARCHIVE_WARN); } /* ACL text is null-terminated; find the end. */ size -= (p - acl); acl = p; while (*p != '\0' && p < acl + size) p++; if (tar->sconv_acl == NULL) { tar->sconv_acl = archive_string_conversion_from_charset( &(a->archive), "UTF-8", 1); if (tar->sconv_acl == NULL) return (ARCHIVE_FATAL); } archive_strncpy(&(tar->localname), acl, p - acl); err = archive_acl_parse_l(archive_entry_acl(entry), tar->localname.s, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, tar->sconv_acl); if (err != ARCHIVE_OK) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for ACL"); } else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (unparsable)"); } return (err); } /* * Interpret 'K' long linkname header. */ static int header_longlink(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { int err; err = read_body_to_string(a, tar, &(tar->longlink), h, unconsumed); if (err != ARCHIVE_OK) return (err); err = tar_read_header(a, tar, entry, unconsumed); if ((err != ARCHIVE_OK) && (err != ARCHIVE_WARN)) return (err); /* Set symlink if symlink already set, else hardlink. */ archive_entry_copy_link(entry, tar->longlink.s); return (ARCHIVE_OK); } static int set_conversion_failed_error(struct archive_read *a, struct archive_string_conv *sconv, const char *name) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for %s", name); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "%s can't be converted from %s to current locale.", name, archive_string_conversion_charset_name(sconv)); return (ARCHIVE_WARN); } /* * Interpret 'L' long filename header. */ static int header_longname(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { int err; err = read_body_to_string(a, tar, &(tar->longname), h, unconsumed); if (err != ARCHIVE_OK) return (err); /* Read and parse "real" header, then override name. */ err = tar_read_header(a, tar, entry, unconsumed); if ((err != ARCHIVE_OK) && (err != ARCHIVE_WARN)) return (err); if (archive_entry_copy_pathname_l(entry, tar->longname.s, archive_strlen(&(tar->longname)), tar->sconv) != 0) err = set_conversion_failed_error(a, tar->sconv, "Pathname"); return (err); } /* * Interpret 'V' GNU tar volume header. */ static int header_volume(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { (void)h; /* Just skip this and read the next header. */ return (tar_read_header(a, tar, entry, unconsumed)); } /* * Read body of an archive entry into an archive_string object. */ static int read_body_to_string(struct archive_read *a, struct tar *tar, struct archive_string *as, const void *h, size_t *unconsumed) { int64_t size; const struct archive_entry_header_ustar *header; const void *src; (void)tar; /* UNUSED */ header = (const struct archive_entry_header_ustar *)h; size = tar_atol(header->size, sizeof(header->size)); if ((size > 1048576) || (size < 0)) { archive_set_error(&a->archive, EINVAL, "Special header too large"); return (ARCHIVE_FATAL); } /* Fail if we can't make our buffer big enough. */ if (archive_string_ensure(as, (size_t)size+1) == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory"); return (ARCHIVE_FATAL); } tar_flush_unconsumed(a, unconsumed); /* Read the body into the string. */ *unconsumed = (size_t)((size + 511) & ~ 511); src = __archive_read_ahead(a, *unconsumed, NULL); if (src == NULL) { *unconsumed = 0; return (ARCHIVE_FATAL); } memcpy(as->s, src, (size_t)size); as->s[size] = '\0'; as->length = (size_t)size; return (ARCHIVE_OK); } /* * Parse out common header elements. * * This would be the same as header_old_tar, except that the * filename is handled slightly differently for old and POSIX * entries (POSIX entries support a 'prefix'). This factoring * allows header_old_tar and header_ustar * to handle filenames differently, while still putting most of the * common parsing into one place. */ static int header_common(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h) { const struct archive_entry_header_ustar *header; char tartype; int err = ARCHIVE_OK; header = (const struct archive_entry_header_ustar *)h; if (header->linkname[0]) archive_strncpy(&(tar->entry_linkpath), header->linkname, sizeof(header->linkname)); else archive_string_empty(&(tar->entry_linkpath)); /* Parse out the numeric fields (all are octal) */ archive_entry_set_mode(entry, (mode_t)tar_atol(header->mode, sizeof(header->mode))); archive_entry_set_uid(entry, tar_atol(header->uid, sizeof(header->uid))); archive_entry_set_gid(entry, tar_atol(header->gid, sizeof(header->gid))); tar->entry_bytes_remaining = tar_atol(header->size, sizeof(header->size)); if (tar->entry_bytes_remaining < 0) { tar->entry_bytes_remaining = 0; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Tar entry has negative size"); return (ARCHIVE_FATAL); } if (tar->entry_bytes_remaining == INT64_MAX) { /* Note: tar_atol returns INT64_MAX on overflow */ tar->entry_bytes_remaining = 0; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Tar entry size overflow"); return (ARCHIVE_FATAL); } tar->realsize = tar->entry_bytes_remaining; archive_entry_set_size(entry, tar->entry_bytes_remaining); archive_entry_set_mtime(entry, tar_atol(header->mtime, sizeof(header->mtime)), 0); /* Handle the tar type flag appropriately. */ tartype = header->typeflag[0]; switch (tartype) { case '1': /* Hard link */ if (archive_entry_copy_hardlink_l(entry, tar->entry_linkpath.s, archive_strlen(&(tar->entry_linkpath)), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Linkname"); if (err == ARCHIVE_FATAL) return (err); } /* * The following may seem odd, but: Technically, tar * does not store the file type for a "hard link" * entry, only the fact that it is a hard link. So, I * leave the type zero normally. But, pax interchange * format allows hard links to have data, which * implies that the underlying entry is a regular * file. */ if (archive_entry_size(entry) > 0) archive_entry_set_filetype(entry, AE_IFREG); /* * A tricky point: Traditionally, tar readers have * ignored the size field when reading hardlink * entries, and some writers put non-zero sizes even * though the body is empty. POSIX blessed this * convention in the 1988 standard, but broke with * this tradition in 2001 by permitting hardlink * entries to store valid bodies in pax interchange * format, but not in ustar format. Since there is no * hard and fast way to distinguish pax interchange * from earlier archives (the 'x' and 'g' entries are * optional, after all), we need a heuristic. */ if (archive_entry_size(entry) == 0) { /* If the size is already zero, we're done. */ } else if (a->archive.archive_format == ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE) { /* Definitely pax extended; must obey hardlink size. */ } else if (a->archive.archive_format == ARCHIVE_FORMAT_TAR || a->archive.archive_format == ARCHIVE_FORMAT_TAR_GNUTAR) { /* Old-style or GNU tar: we must ignore the size. */ archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; } else if (archive_read_format_tar_bid(a, 50) > 50) { /* * We don't know if it's pax: If the bid * function sees a valid ustar header * immediately following, then let's ignore * the hardlink size. */ archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; } /* * TODO: There are still two cases I'd like to handle: * = a ustar non-pax archive with a hardlink entry at * end-of-archive. (Look for block of nulls following?) * = a pax archive that has not seen any pax headers * and has an entry which is a hardlink entry storing * a body containing an uncompressed tar archive. * The first is worth addressing; I don't see any reliable * way to deal with the second possibility. */ break; case '2': /* Symlink */ archive_entry_set_filetype(entry, AE_IFLNK); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; if (archive_entry_copy_symlink_l(entry, tar->entry_linkpath.s, archive_strlen(&(tar->entry_linkpath)), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Linkname"); if (err == ARCHIVE_FATAL) return (err); } break; case '3': /* Character device */ archive_entry_set_filetype(entry, AE_IFCHR); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case '4': /* Block device */ archive_entry_set_filetype(entry, AE_IFBLK); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case '5': /* Dir */ archive_entry_set_filetype(entry, AE_IFDIR); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case '6': /* FIFO device */ archive_entry_set_filetype(entry, AE_IFIFO); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case 'D': /* GNU incremental directory type */ /* * No special handling is actually required here. * It might be nice someday to preprocess the file list and * provide it to the client, though. */ archive_entry_set_filetype(entry, AE_IFDIR); break; case 'M': /* GNU "Multi-volume" (remainder of file from last archive)*/ /* * As far as I can tell, this is just like a regular file * entry, except that the contents should be _appended_ to * the indicated file at the indicated offset. This may * require some API work to fully support. */ break; case 'N': /* Old GNU "long filename" entry. */ /* The body of this entry is a script for renaming * previously-extracted entries. Ugh. It will never * be supported by libarchive. */ archive_entry_set_filetype(entry, AE_IFREG); break; case 'S': /* GNU sparse files */ /* * Sparse files are really just regular files with * sparse information in the extended area. */ /* FALLTHROUGH */ case '0': /* * Enable sparse file "read" support only for regular * files and explicit GNU sparse files. However, we * don't allow non-standard file types to be sparse. */ tar->sparse_allowed = 1; /* FALLTHROUGH */ default: /* Regular file and non-standard types */ /* * Per POSIX: non-recognized types should always be * treated as regular files. */ archive_entry_set_filetype(entry, AE_IFREG); break; } return (err); } /* * Parse out header elements for "old-style" tar archives. */ static int header_old_tar(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h) { const struct archive_entry_header_ustar *header; int err = ARCHIVE_OK, err2; /* Copy filename over (to ensure null termination). */ header = (const struct archive_entry_header_ustar *)h; if (archive_entry_copy_pathname_l(entry, header->name, sizeof(header->name), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Pathname"); if (err == ARCHIVE_FATAL) return (err); } /* Grab rest of common fields */ err2 = header_common(a, tar, entry, h); if (err > err2) err = err2; tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); return (err); } /* * Read a Mac AppleDouble-encoded blob of file metadata, * if there is one. */ static int read_mac_metadata_blob(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { int64_t size; const void *data; const char *p, *name; const wchar_t *wp, *wname; (void)h; /* UNUSED */ wname = wp = archive_entry_pathname_w(entry); if (wp != NULL) { /* Find the last path element. */ for (; *wp != L'\0'; ++wp) { if (wp[0] == '/' && wp[1] != L'\0') wname = wp + 1; } /* * If last path element starts with "._", then * this is a Mac extension. */ if (wname[0] != L'.' || wname[1] != L'_' || wname[2] == L'\0') return ARCHIVE_OK; } else { /* Find the last path element. */ name = p = archive_entry_pathname(entry); if (p == NULL) return (ARCHIVE_FAILED); for (; *p != '\0'; ++p) { if (p[0] == '/' && p[1] != '\0') name = p + 1; } /* * If last path element starts with "._", then * this is a Mac extension. */ if (name[0] != '.' || name[1] != '_' || name[2] == '\0') return ARCHIVE_OK; } /* Read the body as a Mac OS metadata blob. */ size = archive_entry_size(entry); /* * TODO: Look beyond the body here to peek at the next header. * If it's a regular header (not an extension header) * that has the wrong name, just return the current * entry as-is, without consuming the body here. * That would reduce the risk of us mis-identifying * an ordinary file that just happened to have * a name starting with "._". * * Q: Is the above idea really possible? Even * when there are GNU or pax extension entries? */ data = __archive_read_ahead(a, (size_t)size, NULL); if (data == NULL) { *unconsumed = 0; return (ARCHIVE_FATAL); } archive_entry_copy_mac_metadata(entry, data, (size_t)size); *unconsumed = (size_t)((size + 511) & ~ 511); tar_flush_unconsumed(a, unconsumed); return (tar_read_header(a, tar, entry, unconsumed)); } /* * Parse a file header for a pax extended archive entry. */ static int header_pax_global(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { int err; err = read_body_to_string(a, tar, &(tar->pax_global), h, unconsumed); if (err != ARCHIVE_OK) return (err); err = tar_read_header(a, tar, entry, unconsumed); return (err); } static int header_pax_extensions(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { int err, err2; err = read_body_to_string(a, tar, &(tar->pax_header), h, unconsumed); if (err != ARCHIVE_OK) return (err); /* Parse the next header. */ err = tar_read_header(a, tar, entry, unconsumed); if ((err != ARCHIVE_OK) && (err != ARCHIVE_WARN)) return (err); /* * TODO: Parse global/default options into 'entry' struct here * before handling file-specific options. * * This design (parse standard header, then overwrite with pax * extended attribute data) usually works well, but isn't ideal; * it would be better to parse the pax extended attributes first * and then skip any fields in the standard header that were * defined in the pax header. */ err2 = pax_header(a, tar, entry, tar->pax_header.s); err = err_combine(err, err2); tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); return (err); } /* * Parse a file header for a Posix "ustar" archive entry. This also * handles "pax" or "extended ustar" entries. */ static int header_ustar(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h) { const struct archive_entry_header_ustar *header; struct archive_string *as; int err = ARCHIVE_OK, r; header = (const struct archive_entry_header_ustar *)h; /* Copy name into an internal buffer to ensure null-termination. */ as = &(tar->entry_pathname); if (header->prefix[0]) { archive_strncpy(as, header->prefix, sizeof(header->prefix)); if (as->s[archive_strlen(as) - 1] != '/') archive_strappend_char(as, '/'); archive_strncat(as, header->name, sizeof(header->name)); } else { archive_strncpy(as, header->name, sizeof(header->name)); } if (archive_entry_copy_pathname_l(entry, as->s, archive_strlen(as), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Pathname"); if (err == ARCHIVE_FATAL) return (err); } /* Handle rest of common fields. */ r = header_common(a, tar, entry, h); if (r == ARCHIVE_FATAL) return (r); if (r < err) err = r; /* Handle POSIX ustar fields. */ if (archive_entry_copy_uname_l(entry, header->uname, sizeof(header->uname), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Uname"); if (err == ARCHIVE_FATAL) return (err); } if (archive_entry_copy_gname_l(entry, header->gname, sizeof(header->gname), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Gname"); if (err == ARCHIVE_FATAL) return (err); } /* Parse out device numbers only for char and block specials. */ if (header->typeflag[0] == '3' || header->typeflag[0] == '4') { archive_entry_set_rdevmajor(entry, (dev_t) tar_atol(header->rdevmajor, sizeof(header->rdevmajor))); archive_entry_set_rdevminor(entry, (dev_t) tar_atol(header->rdevminor, sizeof(header->rdevminor))); } tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); return (err); } /* * Parse the pax extended attributes record. * * Returns non-zero if there's an error in the data. */ static int pax_header(struct archive_read *a, struct tar *tar, struct archive_entry *entry, char *attr) { size_t attr_length, l, line_length; char *p; char *key, *value; struct archive_string *as; struct archive_string_conv *sconv; int err, err2; attr_length = strlen(attr); tar->pax_hdrcharset_binary = 0; archive_string_empty(&(tar->entry_gname)); archive_string_empty(&(tar->entry_linkpath)); archive_string_empty(&(tar->entry_pathname)); archive_string_empty(&(tar->entry_pathname_override)); archive_string_empty(&(tar->entry_uname)); err = ARCHIVE_OK; while (attr_length > 0) { /* Parse decimal length field at start of line. */ line_length = 0; l = attr_length; p = attr; /* Record start of line. */ while (l>0) { if (*p == ' ') { p++; l--; break; } if (*p < '0' || *p > '9') { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring malformed pax extended attributes"); return (ARCHIVE_WARN); } line_length *= 10; line_length += *p - '0'; if (line_length > 999999) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Rejecting pax extended attribute > 1MB"); return (ARCHIVE_WARN); } p++; l--; } /* * Parsed length must be no bigger than available data, * at least 1, and the last character of the line must * be '\n'. */ if (line_length > attr_length || line_length < 1 || attr[line_length - 1] != '\n') { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring malformed pax extended attribute"); return (ARCHIVE_WARN); } /* Null-terminate the line. */ attr[line_length - 1] = '\0'; /* Find end of key and null terminate it. */ key = p; if (key[0] == '=') return (-1); while (*p && *p != '=') ++p; if (*p == '\0') { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid pax extended attributes"); return (ARCHIVE_WARN); } *p = '\0'; /* Identify null-terminated 'value' portion. */ value = p + 1; /* Identify this attribute and set it in the entry. */ err2 = pax_attribute(a, tar, entry, key, value); if (err2 == ARCHIVE_FATAL) return (err2); err = err_combine(err, err2); /* Skip to next line */ attr += line_length; attr_length -= line_length; } /* * PAX format uses UTF-8 as default charset for its metadata * unless hdrcharset=BINARY is present in its header. * We apply the charset specified by the hdrcharset option only * when the hdrcharset attribute(in PAX header) is BINARY because * we respect the charset described in PAX header and BINARY also * means that metadata(filename,uname and gname) character-set * is unknown. */ if (tar->pax_hdrcharset_binary) sconv = tar->opt_sconv; else { sconv = archive_string_conversion_from_charset( &(a->archive), "UTF-8", 1); if (sconv == NULL) return (ARCHIVE_FATAL); if (tar->compat_2x) archive_string_conversion_set_opt(sconv, SCONV_SET_OPT_UTF8_LIBARCHIVE2X); } if (archive_strlen(&(tar->entry_gname)) > 0) { if (archive_entry_copy_gname_l(entry, tar->entry_gname.s, archive_strlen(&(tar->entry_gname)), sconv) != 0) { err = set_conversion_failed_error(a, sconv, "Gname"); if (err == ARCHIVE_FATAL) return (err); /* Use a converted an original name. */ archive_entry_copy_gname(entry, tar->entry_gname.s); } } if (archive_strlen(&(tar->entry_linkpath)) > 0) { if (archive_entry_copy_link_l(entry, tar->entry_linkpath.s, archive_strlen(&(tar->entry_linkpath)), sconv) != 0) { err = set_conversion_failed_error(a, sconv, "Linkname"); if (err == ARCHIVE_FATAL) return (err); /* Use a converted an original name. */ archive_entry_copy_link(entry, tar->entry_linkpath.s); } } /* * Some extensions (such as the GNU sparse file extensions) * deliberately store a synthetic name under the regular 'path' * attribute and the real file name under a different attribute. * Since we're supposed to not care about the order, we * have no choice but to store all of the various filenames * we find and figure it all out afterwards. This is the * figuring out part. */ as = NULL; if (archive_strlen(&(tar->entry_pathname_override)) > 0) as = &(tar->entry_pathname_override); else if (archive_strlen(&(tar->entry_pathname)) > 0) as = &(tar->entry_pathname); if (as != NULL) { if (archive_entry_copy_pathname_l(entry, as->s, archive_strlen(as), sconv) != 0) { err = set_conversion_failed_error(a, sconv, "Pathname"); if (err == ARCHIVE_FATAL) return (err); /* Use a converted an original name. */ archive_entry_copy_pathname(entry, as->s); } } if (archive_strlen(&(tar->entry_uname)) > 0) { if (archive_entry_copy_uname_l(entry, tar->entry_uname.s, archive_strlen(&(tar->entry_uname)), sconv) != 0) { err = set_conversion_failed_error(a, sconv, "Uname"); if (err == ARCHIVE_FATAL) return (err); /* Use a converted an original name. */ archive_entry_copy_uname(entry, tar->entry_uname.s); } } return (err); } static int pax_attribute_xattr(struct archive_entry *entry, const char *name, const char *value) { char *name_decoded; void *value_decoded; size_t value_len; if (strlen(name) < 18 || (memcmp(name, "LIBARCHIVE.xattr.", 17)) != 0) return 3; name += 17; /* URL-decode name */ name_decoded = url_decode(name); if (name_decoded == NULL) return 2; /* Base-64 decode value */ value_decoded = base64_decode(value, strlen(value), &value_len); if (value_decoded == NULL) { free(name_decoded); return 1; } archive_entry_xattr_add_entry(entry, name_decoded, value_decoded, value_len); free(name_decoded); free(value_decoded); return 0; } /* * Parse a single key=value attribute. key/value pointers are * assumed to point into reasonably long-lived storage. * * Note that POSIX reserves all-lowercase keywords. Vendor-specific * extensions should always have keywords of the form "VENDOR.attribute" * In particular, it's quite feasible to support many different * vendor extensions here. I'm using "LIBARCHIVE" for extensions * unique to this library. * * Investigate other vendor-specific extensions and see if * any of them look useful. */ static int pax_attribute(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const char *key, const char *value) { int64_t s; long n; int err = ARCHIVE_OK, r; #ifndef __FreeBSD__ if (value == NULL) value = ""; /* Disable compiler warning; do not pass * NULL pointer to strlen(). */ #endif switch (key[0]) { case 'G': /* Reject GNU.sparse.* headers on non-regular files. */ if (strncmp(key, "GNU.sparse", 10) == 0 && !tar->sparse_allowed) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Non-regular file cannot be sparse"); return (ARCHIVE_FATAL); } /* GNU "0.0" sparse pax format. */ if (strcmp(key, "GNU.sparse.numblocks") == 0) { tar->sparse_offset = -1; tar->sparse_numbytes = -1; tar->sparse_gnu_major = 0; tar->sparse_gnu_minor = 0; } if (strcmp(key, "GNU.sparse.offset") == 0) { tar->sparse_offset = tar_atol10(value, strlen(value)); if (tar->sparse_numbytes != -1) { if (gnu_add_sparse_entry(a, tar, tar->sparse_offset, tar->sparse_numbytes) != ARCHIVE_OK) return (ARCHIVE_FATAL); tar->sparse_offset = -1; tar->sparse_numbytes = -1; } } if (strcmp(key, "GNU.sparse.numbytes") == 0) { tar->sparse_numbytes = tar_atol10(value, strlen(value)); if (tar->sparse_numbytes != -1) { if (gnu_add_sparse_entry(a, tar, tar->sparse_offset, tar->sparse_numbytes) != ARCHIVE_OK) return (ARCHIVE_FATAL); tar->sparse_offset = -1; tar->sparse_numbytes = -1; } } if (strcmp(key, "GNU.sparse.size") == 0) { tar->realsize = tar_atol10(value, strlen(value)); archive_entry_set_size(entry, tar->realsize); } /* GNU "0.1" sparse pax format. */ if (strcmp(key, "GNU.sparse.map") == 0) { tar->sparse_gnu_major = 0; tar->sparse_gnu_minor = 1; if (gnu_sparse_01_parse(a, tar, value) != ARCHIVE_OK) return (ARCHIVE_WARN); } /* GNU "1.0" sparse pax format */ if (strcmp(key, "GNU.sparse.major") == 0) { tar->sparse_gnu_major = (int)tar_atol10(value, strlen(value)); tar->sparse_gnu_pending = 1; } if (strcmp(key, "GNU.sparse.minor") == 0) { tar->sparse_gnu_minor = (int)tar_atol10(value, strlen(value)); tar->sparse_gnu_pending = 1; } if (strcmp(key, "GNU.sparse.name") == 0) { /* * The real filename; when storing sparse * files, GNU tar puts a synthesized name into * the regular 'path' attribute in an attempt * to limit confusion. ;-) */ archive_strcpy(&(tar->entry_pathname_override), value); } if (strcmp(key, "GNU.sparse.realsize") == 0) { tar->realsize = tar_atol10(value, strlen(value)); archive_entry_set_size(entry, tar->realsize); } break; case 'L': /* Our extensions */ /* TODO: Handle arbitrary extended attributes... */ /* if (strcmp(key, "LIBARCHIVE.xxxxxxx") == 0) archive_entry_set_xxxxxx(entry, value); */ if (strcmp(key, "LIBARCHIVE.creationtime") == 0) { pax_time(value, &s, &n); archive_entry_set_birthtime(entry, s, n); } if (memcmp(key, "LIBARCHIVE.xattr.", 17) == 0) pax_attribute_xattr(entry, key, value); break; case 'S': /* We support some keys used by the "star" archiver */ if (strcmp(key, "SCHILY.acl.access") == 0) { if (tar->sconv_acl == NULL) { tar->sconv_acl = archive_string_conversion_from_charset( &(a->archive), "UTF-8", 1); if (tar->sconv_acl == NULL) return (ARCHIVE_FATAL); } r = archive_acl_parse_l(archive_entry_acl(entry), value, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, tar->sconv_acl); if (r != ARCHIVE_OK) { err = r; if (err == ARCHIVE_FATAL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for " "SCHILY.acl.access"); return (err); } archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Parse error: SCHILY.acl.access"); } } else if (strcmp(key, "SCHILY.acl.default") == 0) { if (tar->sconv_acl == NULL) { tar->sconv_acl = archive_string_conversion_from_charset( &(a->archive), "UTF-8", 1); if (tar->sconv_acl == NULL) return (ARCHIVE_FATAL); } r = archive_acl_parse_l(archive_entry_acl(entry), value, ARCHIVE_ENTRY_ACL_TYPE_DEFAULT, tar->sconv_acl); if (r != ARCHIVE_OK) { err = r; if (err == ARCHIVE_FATAL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for " "SCHILY.acl.default"); return (err); } archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Parse error: SCHILY.acl.default"); } } else if (strcmp(key, "SCHILY.devmajor") == 0) { archive_entry_set_rdevmajor(entry, (dev_t)tar_atol10(value, strlen(value))); } else if (strcmp(key, "SCHILY.devminor") == 0) { archive_entry_set_rdevminor(entry, (dev_t)tar_atol10(value, strlen(value))); } else if (strcmp(key, "SCHILY.fflags") == 0) { archive_entry_copy_fflags_text(entry, value); } else if (strcmp(key, "SCHILY.dev") == 0) { archive_entry_set_dev(entry, (dev_t)tar_atol10(value, strlen(value))); } else if (strcmp(key, "SCHILY.ino") == 0) { archive_entry_set_ino(entry, tar_atol10(value, strlen(value))); } else if (strcmp(key, "SCHILY.nlink") == 0) { archive_entry_set_nlink(entry, (unsigned) tar_atol10(value, strlen(value))); } else if (strcmp(key, "SCHILY.realsize") == 0) { tar->realsize = tar_atol10(value, strlen(value)); archive_entry_set_size(entry, tar->realsize); } else if (strcmp(key, "SUN.holesdata") == 0) { /* A Solaris extension for sparse. */ r = solaris_sparse_parse(a, tar, entry, value); if (r < err) { if (r == ARCHIVE_FATAL) return (r); err = r; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Parse error: SUN.holesdata"); } } break; case 'a': if (strcmp(key, "atime") == 0) { pax_time(value, &s, &n); archive_entry_set_atime(entry, s, n); } break; case 'c': if (strcmp(key, "ctime") == 0) { pax_time(value, &s, &n); archive_entry_set_ctime(entry, s, n); } else if (strcmp(key, "charset") == 0) { /* TODO: Publish charset information in entry. */ } else if (strcmp(key, "comment") == 0) { /* TODO: Publish comment in entry. */ } break; case 'g': if (strcmp(key, "gid") == 0) { archive_entry_set_gid(entry, tar_atol10(value, strlen(value))); } else if (strcmp(key, "gname") == 0) { archive_strcpy(&(tar->entry_gname), value); } break; case 'h': if (strcmp(key, "hdrcharset") == 0) { if (strcmp(value, "BINARY") == 0) /* Binary mode. */ tar->pax_hdrcharset_binary = 1; else if (strcmp(value, "ISO-IR 10646 2000 UTF-8") == 0) tar->pax_hdrcharset_binary = 0; } break; case 'l': /* pax interchange doesn't distinguish hardlink vs. symlink. */ if (strcmp(key, "linkpath") == 0) { archive_strcpy(&(tar->entry_linkpath), value); } break; case 'm': if (strcmp(key, "mtime") == 0) { pax_time(value, &s, &n); archive_entry_set_mtime(entry, s, n); } break; case 'p': if (strcmp(key, "path") == 0) { archive_strcpy(&(tar->entry_pathname), value); } break; case 'r': /* POSIX has reserved 'realtime.*' */ break; case 's': /* POSIX has reserved 'security.*' */ /* Someday: if (strcmp(key, "security.acl") == 0) { ... } */ if (strcmp(key, "size") == 0) { /* "size" is the size of the data in the entry. */ tar->entry_bytes_remaining = tar_atol10(value, strlen(value)); /* * But, "size" is not necessarily the size of * the file on disk; if this is a sparse file, * the disk size may have already been set from * GNU.sparse.realsize or GNU.sparse.size or * an old GNU header field or SCHILY.realsize * or .... */ if (tar->realsize < 0) { archive_entry_set_size(entry, tar->entry_bytes_remaining); tar->realsize = tar->entry_bytes_remaining; } } break; case 'u': if (strcmp(key, "uid") == 0) { archive_entry_set_uid(entry, tar_atol10(value, strlen(value))); } else if (strcmp(key, "uname") == 0) { archive_strcpy(&(tar->entry_uname), value); } break; } return (err); } /* * parse a decimal time value, which may include a fractional portion */ static void pax_time(const char *p, int64_t *ps, long *pn) { char digit; int64_t s; unsigned long l; int sign; int64_t limit, last_digit_limit; limit = INT64_MAX / 10; last_digit_limit = INT64_MAX % 10; s = 0; sign = 1; if (*p == '-') { sign = -1; p++; } while (*p >= '0' && *p <= '9') { digit = *p - '0'; if (s > limit || (s == limit && digit > last_digit_limit)) { s = INT64_MAX; break; } s = (s * 10) + digit; ++p; } *ps = s * sign; /* Calculate nanoseconds. */ *pn = 0; if (*p != '.') return; l = 100000000UL; do { ++p; if (*p >= '0' && *p <= '9') *pn += (*p - '0') * l; else break; } while (l /= 10); } /* * Parse GNU tar header */ static int header_gnutar(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { const struct archive_entry_header_gnutar *header; int64_t t; int err = ARCHIVE_OK; /* * GNU header is like POSIX ustar, except 'prefix' is * replaced with some other fields. This also means the * filename is stored as in old-style archives. */ /* Grab fields common to all tar variants. */ err = header_common(a, tar, entry, h); if (err == ARCHIVE_FATAL) return (err); /* Copy filename over (to ensure null termination). */ header = (const struct archive_entry_header_gnutar *)h; if (archive_entry_copy_pathname_l(entry, header->name, sizeof(header->name), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Pathname"); if (err == ARCHIVE_FATAL) return (err); } /* Fields common to ustar and GNU */ /* XXX Can the following be factored out since it's common * to ustar and gnu tar? Is it okay to move it down into * header_common, perhaps? */ if (archive_entry_copy_uname_l(entry, header->uname, sizeof(header->uname), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Uname"); if (err == ARCHIVE_FATAL) return (err); } if (archive_entry_copy_gname_l(entry, header->gname, sizeof(header->gname), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Gname"); if (err == ARCHIVE_FATAL) return (err); } /* Parse out device numbers only for char and block specials */ if (header->typeflag[0] == '3' || header->typeflag[0] == '4') { archive_entry_set_rdevmajor(entry, (dev_t) tar_atol(header->rdevmajor, sizeof(header->rdevmajor))); archive_entry_set_rdevminor(entry, (dev_t) tar_atol(header->rdevminor, sizeof(header->rdevminor))); } else archive_entry_set_rdev(entry, 0); tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); /* Grab GNU-specific fields. */ t = tar_atol(header->atime, sizeof(header->atime)); if (t > 0) archive_entry_set_atime(entry, t, 0); t = tar_atol(header->ctime, sizeof(header->ctime)); if (t > 0) archive_entry_set_ctime(entry, t, 0); if (header->realsize[0] != 0) { tar->realsize = tar_atol(header->realsize, sizeof(header->realsize)); archive_entry_set_size(entry, tar->realsize); } if (header->sparse[0].offset[0] != 0) { if (gnu_sparse_old_read(a, tar, header, unconsumed) != ARCHIVE_OK) return (ARCHIVE_FATAL); } else { if (header->isextended[0] != 0) { /* XXX WTF? XXX */ } } return (err); } static int gnu_add_sparse_entry(struct archive_read *a, struct tar *tar, int64_t offset, int64_t remaining) { struct sparse_block *p; p = (struct sparse_block *)malloc(sizeof(*p)); if (p == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } memset(p, 0, sizeof(*p)); if (tar->sparse_last != NULL) tar->sparse_last->next = p; else tar->sparse_list = p; tar->sparse_last = p; if (remaining < 0 || offset < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed sparse map data"); return (ARCHIVE_FATAL); } p->offset = offset; p->remaining = remaining; return (ARCHIVE_OK); } static void gnu_clear_sparse_list(struct tar *tar) { struct sparse_block *p; while (tar->sparse_list != NULL) { p = tar->sparse_list; tar->sparse_list = p->next; free(p); } tar->sparse_last = NULL; } /* * GNU tar old-format sparse data. * * GNU old-format sparse data is stored in a fixed-field * format. Offset/size values are 11-byte octal fields (same * format as 'size' field in ustart header). These are * stored in the header, allocating subsequent header blocks * as needed. Extending the header in this way is a pretty * severe POSIX violation; this design has earned GNU tar a * lot of criticism. */ static int gnu_sparse_old_read(struct archive_read *a, struct tar *tar, const struct archive_entry_header_gnutar *header, size_t *unconsumed) { ssize_t bytes_read; const void *data; struct extended { struct gnu_sparse sparse[21]; char isextended[1]; char padding[7]; }; const struct extended *ext; if (gnu_sparse_old_parse(a, tar, header->sparse, 4) != ARCHIVE_OK) return (ARCHIVE_FATAL); if (header->isextended[0] == 0) return (ARCHIVE_OK); do { tar_flush_unconsumed(a, unconsumed); data = __archive_read_ahead(a, 512, &bytes_read); if (bytes_read < 0) return (ARCHIVE_FATAL); if (bytes_read < 512) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated tar archive " "detected while reading sparse file data"); return (ARCHIVE_FATAL); } *unconsumed = 512; ext = (const struct extended *)data; if (gnu_sparse_old_parse(a, tar, ext->sparse, 21) != ARCHIVE_OK) return (ARCHIVE_FATAL); } while (ext->isextended[0] != 0); if (tar->sparse_list != NULL) tar->entry_offset = tar->sparse_list->offset; return (ARCHIVE_OK); } static int gnu_sparse_old_parse(struct archive_read *a, struct tar *tar, const struct gnu_sparse *sparse, int length) { while (length > 0 && sparse->offset[0] != 0) { if (gnu_add_sparse_entry(a, tar, tar_atol(sparse->offset, sizeof(sparse->offset)), tar_atol(sparse->numbytes, sizeof(sparse->numbytes))) != ARCHIVE_OK) return (ARCHIVE_FATAL); sparse++; length--; } return (ARCHIVE_OK); } /* * GNU tar sparse format 0.0 * * Beginning with GNU tar 1.15, sparse files are stored using * information in the pax extended header. The GNU tar maintainers * have gone through a number of variations in the process of working * out this scheme; fortunately, they're all numbered. * * Sparse format 0.0 uses attribute GNU.sparse.numblocks to store the * number of blocks, and GNU.sparse.offset/GNU.sparse.numbytes to * store offset/size for each block. The repeated instances of these * latter fields violate the pax specification (which frowns on * duplicate keys), so this format was quickly replaced. */ /* * GNU tar sparse format 0.1 * * This version replaced the offset/numbytes attributes with * a single "map" attribute that stored a list of integers. This * format had two problems: First, the "map" attribute could be very * long, which caused problems for some implementations. More * importantly, the sparse data was lost when extracted by archivers * that didn't recognize this extension. */ static int gnu_sparse_01_parse(struct archive_read *a, struct tar *tar, const char *p) { const char *e; int64_t offset = -1, size = -1; for (;;) { e = p; while (*e != '\0' && *e != ',') { if (*e < '0' || *e > '9') return (ARCHIVE_WARN); e++; } if (offset < 0) { offset = tar_atol10(p, e - p); if (offset < 0) return (ARCHIVE_WARN); } else { size = tar_atol10(p, e - p); if (size < 0) return (ARCHIVE_WARN); if (gnu_add_sparse_entry(a, tar, offset, size) != ARCHIVE_OK) return (ARCHIVE_FATAL); offset = -1; } if (*e == '\0') return (ARCHIVE_OK); p = e + 1; } } /* * GNU tar sparse format 1.0 * * The idea: The offset/size data is stored as a series of base-10 * ASCII numbers prepended to the file data, so that dearchivers that * don't support this format will extract the block map along with the * data and a separate post-process can restore the sparseness. * * Unfortunately, GNU tar 1.16 had a bug that added unnecessary * padding to the body of the file when using this format. GNU tar * 1.17 corrected this bug without bumping the version number, so * it's not possible to support both variants. This code supports * the later variant at the expense of not supporting the former. * * This variant also replaced GNU.sparse.size with GNU.sparse.realsize * and introduced the GNU.sparse.major/GNU.sparse.minor attributes. */ /* * Read the next line from the input, and parse it as a decimal * integer followed by '\n'. Returns positive integer value or * negative on error. */ static int64_t gnu_sparse_10_atol(struct archive_read *a, struct tar *tar, int64_t *remaining, size_t *unconsumed) { int64_t l, limit, last_digit_limit; const char *p; ssize_t bytes_read; int base, digit; base = 10; limit = INT64_MAX / base; last_digit_limit = INT64_MAX % base; /* * Skip any lines starting with '#'; GNU tar specs * don't require this, but they should. */ do { bytes_read = readline(a, tar, &p, (ssize_t)tar_min(*remaining, 100), unconsumed); if (bytes_read <= 0) return (ARCHIVE_FATAL); *remaining -= bytes_read; } while (p[0] == '#'); l = 0; while (bytes_read > 0) { if (*p == '\n') return (l); if (*p < '0' || *p >= '0' + base) return (ARCHIVE_WARN); digit = *p - '0'; if (l > limit || (l == limit && digit > last_digit_limit)) l = INT64_MAX; /* Truncate on overflow. */ else l = (l * base) + digit; p++; bytes_read--; } /* TODO: Error message. */ return (ARCHIVE_WARN); } /* * Returns length (in bytes) of the sparse data description * that was read. */ static ssize_t gnu_sparse_10_read(struct archive_read *a, struct tar *tar, size_t *unconsumed) { ssize_t bytes_read; int entries; int64_t offset, size, to_skip, remaining; /* Clear out the existing sparse list. */ gnu_clear_sparse_list(tar); remaining = tar->entry_bytes_remaining; /* Parse entries. */ entries = (int)gnu_sparse_10_atol(a, tar, &remaining, unconsumed); if (entries < 0) return (ARCHIVE_FATAL); /* Parse the individual entries. */ while (entries-- > 0) { /* Parse offset/size */ offset = gnu_sparse_10_atol(a, tar, &remaining, unconsumed); if (offset < 0) return (ARCHIVE_FATAL); size = gnu_sparse_10_atol(a, tar, &remaining, unconsumed); if (size < 0) return (ARCHIVE_FATAL); /* Add a new sparse entry. */ if (gnu_add_sparse_entry(a, tar, offset, size) != ARCHIVE_OK) return (ARCHIVE_FATAL); } /* Skip rest of block... */ tar_flush_unconsumed(a, unconsumed); bytes_read = (ssize_t)(tar->entry_bytes_remaining - remaining); to_skip = 0x1ff & -bytes_read; if (to_skip != __archive_read_consume(a, to_skip)) return (ARCHIVE_FATAL); return ((ssize_t)(bytes_read + to_skip)); } /* * Solaris pax extension for a sparse file. This is recorded with the * data and hole pairs. The way recording sparse information by Solaris' * pax simply indicates where data and sparse are, so the stored contents * consist of both data and hole. */ static int solaris_sparse_parse(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const char *p) { const char *e; int64_t start, end; int hole = 1; (void)entry; /* UNUSED */ end = 0; if (*p == ' ') p++; else return (ARCHIVE_WARN); for (;;) { e = p; while (*e != '\0' && *e != ' ') { if (*e < '0' || *e > '9') return (ARCHIVE_WARN); e++; } start = end; end = tar_atol10(p, e - p); if (end < 0) return (ARCHIVE_WARN); if (start < end) { if (gnu_add_sparse_entry(a, tar, start, end - start) != ARCHIVE_OK) return (ARCHIVE_FATAL); tar->sparse_last->hole = hole; } if (*e == '\0') return (ARCHIVE_OK); p = e + 1; hole = hole == 0; } } /*- * Convert text->integer. * * Traditional tar formats (including POSIX) specify base-8 for * all of the standard numeric fields. This is a significant limitation * in practice: * = file size is limited to 8GB * = rdevmajor and rdevminor are limited to 21 bits * = uid/gid are limited to 21 bits * * There are two workarounds for this: * = pax extended headers, which use variable-length string fields * = GNU tar and STAR both allow either base-8 or base-256 in * most fields. The high bit is set to indicate base-256. * * On read, this implementation supports both extensions. */ static int64_t tar_atol(const char *p, size_t char_cnt) { /* * Technically, GNU tar considers a field to be in base-256 * only if the first byte is 0xff or 0x80. */ if (*p & 0x80) return (tar_atol256(p, char_cnt)); return (tar_atol8(p, char_cnt)); } /* * Note that this implementation does not (and should not!) obey * locale settings; you cannot simply substitute strtol here, since * it does obey locale. */ static int64_t tar_atol_base_n(const char *p, size_t char_cnt, int base) { int64_t l, maxval, limit, last_digit_limit; int digit, sign; maxval = INT64_MAX; limit = INT64_MAX / base; last_digit_limit = INT64_MAX % base; /* the pointer will not be dereferenced if char_cnt is zero * due to the way the && operator is evaulated. */ while (char_cnt != 0 && (*p == ' ' || *p == '\t')) { p++; char_cnt--; } sign = 1; if (char_cnt != 0 && *p == '-') { sign = -1; p++; char_cnt--; maxval = INT64_MIN; limit = -(INT64_MIN / base); last_digit_limit = INT64_MIN % base; } l = 0; if (char_cnt != 0) { digit = *p - '0'; while (digit >= 0 && digit < base && char_cnt != 0) { if (l>limit || (l == limit && digit > last_digit_limit)) { return maxval; /* Truncate on overflow. */ } l = (l * base) + digit; digit = *++p - '0'; char_cnt--; } } return (sign < 0) ? -l : l; } static int64_t tar_atol8(const char *p, size_t char_cnt) { return tar_atol_base_n(p, char_cnt, 8); } static int64_t tar_atol10(const char *p, size_t char_cnt) { return tar_atol_base_n(p, char_cnt, 10); } /* * Parse a base-256 integer. This is just a variable-length * twos-complement signed binary value in big-endian order, except * that the high-order bit is ignored. The values here can be up to * 12 bytes, so we need to be careful about overflowing 64-bit * (8-byte) integers. * * This code unashamedly assumes that the local machine uses 8-bit * bytes and twos-complement arithmetic. */ static int64_t tar_atol256(const char *_p, size_t char_cnt) { uint64_t l; const unsigned char *p = (const unsigned char *)_p; unsigned char c, neg; /* Extend 7-bit 2s-comp to 8-bit 2s-comp, decide sign. */ c = *p; if (c & 0x40) { neg = 0xff; c |= 0x80; l = ~ARCHIVE_LITERAL_ULL(0); } else { neg = 0; c &= 0x7f; l = 0; } /* If more than 8 bytes, check that we can ignore * high-order bits without overflow. */ while (char_cnt > sizeof(int64_t)) { --char_cnt; if (c != neg) return neg ? INT64_MIN : INT64_MAX; c = *++p; } /* c is first byte that fits; if sign mismatch, return overflow */ if ((c ^ neg) & 0x80) { return neg ? INT64_MIN : INT64_MAX; } /* Accumulate remaining bytes. */ while (--char_cnt > 0) { l = (l << 8) | c; c = *++p; } l = (l << 8) | c; /* Return signed twos-complement value. */ return (int64_t)(l); } /* * Returns length of line (including trailing newline) * or negative on error. 'start' argument is updated to * point to first character of line. This avoids copying * when possible. */ static ssize_t readline(struct archive_read *a, struct tar *tar, const char **start, ssize_t limit, size_t *unconsumed) { ssize_t bytes_read; ssize_t total_size = 0; const void *t; const char *s; void *p; tar_flush_unconsumed(a, unconsumed); t = __archive_read_ahead(a, 1, &bytes_read); if (bytes_read <= 0) return (ARCHIVE_FATAL); s = t; /* Start of line? */ p = memchr(t, '\n', bytes_read); /* If we found '\n' in the read buffer, return pointer to that. */ if (p != NULL) { bytes_read = 1 + ((const char *)p) - s; if (bytes_read > limit) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Line too long"); return (ARCHIVE_FATAL); } *unconsumed = bytes_read; *start = s; return (bytes_read); } *unconsumed = bytes_read; /* Otherwise, we need to accumulate in a line buffer. */ for (;;) { if (total_size + bytes_read > limit) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Line too long"); return (ARCHIVE_FATAL); } if (archive_string_ensure(&tar->line, total_size + bytes_read) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate working buffer"); return (ARCHIVE_FATAL); } memcpy(tar->line.s + total_size, t, bytes_read); tar_flush_unconsumed(a, unconsumed); total_size += bytes_read; /* If we found '\n', clean up and return. */ if (p != NULL) { *start = tar->line.s; return (total_size); } /* Read some more. */ t = __archive_read_ahead(a, 1, &bytes_read); if (bytes_read <= 0) return (ARCHIVE_FATAL); s = t; /* Start of line? */ p = memchr(t, '\n', bytes_read); /* If we found '\n', trim the read. */ if (p != NULL) { bytes_read = 1 + ((const char *)p) - s; } *unconsumed = bytes_read; } } /* * base64_decode - Base64 decode * * This accepts most variations of base-64 encoding, including: * * with or without line breaks * * with or without the final group padded with '=' or '_' characters * (The most economical Base-64 variant does not pad the last group and * omits line breaks; RFC1341 used for MIME requires both.) */ static char * base64_decode(const char *s, size_t len, size_t *out_len) { static const unsigned char digits[64] = { 'A','B','C','D','E','F','G','H','I','J','K','L','M','N', 'O','P','Q','R','S','T','U','V','W','X','Y','Z','a','b', 'c','d','e','f','g','h','i','j','k','l','m','n','o','p', 'q','r','s','t','u','v','w','x','y','z','0','1','2','3', '4','5','6','7','8','9','+','/' }; static unsigned char decode_table[128]; char *out, *d; const unsigned char *src = (const unsigned char *)s; /* If the decode table is not yet initialized, prepare it. */ if (decode_table[digits[1]] != 1) { unsigned i; memset(decode_table, 0xff, sizeof(decode_table)); for (i = 0; i < sizeof(digits); i++) decode_table[digits[i]] = i; } /* Allocate enough space to hold the entire output. */ /* Note that we may not use all of this... */ out = (char *)malloc(len - len / 4 + 1); if (out == NULL) { *out_len = 0; return (NULL); } d = out; while (len > 0) { /* Collect the next group of (up to) four characters. */ int v = 0; int group_size = 0; while (group_size < 4 && len > 0) { /* '=' or '_' padding indicates final group. */ if (*src == '=' || *src == '_') { len = 0; break; } /* Skip illegal characters (including line breaks) */ if (*src > 127 || *src < 32 || decode_table[*src] == 0xff) { len--; src++; continue; } v <<= 6; v |= decode_table[*src++]; len --; group_size++; } /* Align a short group properly. */ v <<= 6 * (4 - group_size); /* Unpack the group we just collected. */ switch (group_size) { case 4: d[2] = v & 0xff; /* FALLTHROUGH */ case 3: d[1] = (v >> 8) & 0xff; /* FALLTHROUGH */ case 2: d[0] = (v >> 16) & 0xff; break; case 1: /* this is invalid! */ break; } d += group_size * 3 / 4; } *out_len = d - out; return (out); } static char * url_decode(const char *in) { char *out, *d; const char *s; out = (char *)malloc(strlen(in) + 1); if (out == NULL) return (NULL); for (s = in, d = out; *s != '\0'; ) { if (s[0] == '%' && s[1] != '\0' && s[2] != '\0') { /* Try to convert % escape */ int digit1 = tohex(s[1]); int digit2 = tohex(s[2]); if (digit1 >= 0 && digit2 >= 0) { /* Looks good, consume three chars */ s += 3; /* Convert output */ *d++ = ((digit1 << 4) | digit2); continue; } /* Else fall through and treat '%' as normal char */ } *d++ = *s++; } *d = '\0'; return (out); } static int tohex(int c) { if (c >= '0' && c <= '9') return (c - '0'); else if (c >= 'A' && c <= 'F') return (c - 'A' + 10); else if (c >= 'a' && c <= 'f') return (c - 'a' + 10); else return (-1); } Index: head/contrib/libarchive/libarchive/archive_string.c =================================================================== --- head/contrib/libarchive/libarchive/archive_string.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_string.c (revision 309590) @@ -1,4198 +1,4198 @@ /*- * 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); 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); wmemmove(as->s + as->length, p, s); as->length += s; as->s[as->length] = 0; return (as); } 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 consis 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)) { /* * Unfortunaly, 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 programing. 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()); cp = my_atoi(p+1); if (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 compatibillty. */ #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 filname 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; int i, r = 0, r2; /* We must allocate memory even if there is no data for conversion * or copy. This simulates archive_string_append behavior. */ if (_p == NULL || n == 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) { length = mbsnbytes(_p, n); if (archive_string_append(as, _p, length) == NULL) return (-1);/* No memory */ return (0); } if (sc->flag & SCONV_FROM_UTF16) length = utf16nbytes(_p, n); else length = mbsnbytes(_p, n); 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 failes. */ 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 failes. */ 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 '?' charater 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]; - /* Invalide sequence or there are not plenty bytes. */ + /* 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 leagal + /* 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 leagal Unicode values. */ + /* 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 leagal or not. + * 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 culculation, the code * must not be surrogate values, and Unicode has no codes - * larger than 0x10ffff. Thus, those are not leagal Unicode + * 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 libarchie 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 failes. */ 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 failes. */ 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 failes. */ 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 failes. */ 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 programing 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; } 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; } 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 programing 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_string.h =================================================================== --- head/contrib/libarchive/libarchive/archive_string.h (revision 309589) +++ head/contrib/libarchive/libarchive/archive_string.h (revision 309590) @@ -1,239 +1,239 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * $FreeBSD$ * */ #ifndef __LIBARCHIVE_BUILD #ifndef __LIBARCHIVE_TEST #error This header is only to be used internally to libarchive. #endif #endif #ifndef ARCHIVE_STRING_H_INCLUDED #define ARCHIVE_STRING_H_INCLUDED #include #ifdef HAVE_STDLIB_H #include /* required for wchar_t on some systems */ #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_WCHAR_H #include #endif #include "archive.h" /* * Basic resizable/reusable string support similar to Java's "StringBuffer." * * Unlike sbuf(9), the buffers here are fully reusable and track the * length throughout. */ struct archive_string { char *s; /* Pointer to the storage */ size_t length; /* Length of 's' in characters */ size_t buffer_length; /* Length of malloc-ed storage in bytes. */ }; struct archive_wstring { wchar_t *s; /* Pointer to the storage */ size_t length; /* Length of 's' in characters */ size_t buffer_length; /* Length of malloc-ed storage in bytes. */ }; struct archive_string_conv; /* Initialize an archive_string object on the stack or elsewhere. */ #define archive_string_init(a) \ do { (a)->s = NULL; (a)->length = 0; (a)->buffer_length = 0; } while(0) /* Append a C char to an archive_string, resizing as necessary. */ struct archive_string * archive_strappend_char(struct archive_string *, char); /* Ditto for a wchar_t and an archive_wstring. */ struct archive_wstring * archive_wstrappend_wchar(struct archive_wstring *, wchar_t); /* Convert a Unicode string to current locale and append the result. */ /* Returns -1 if conversion fails. */ int archive_string_append_from_wcs(struct archive_string *, const wchar_t *, size_t); /* Create a string conversion object. * Return NULL and set a error message if the conversion is not supported * on the platform. */ struct archive_string_conv * archive_string_conversion_to_charset(struct archive *, const char *, int); struct archive_string_conv * archive_string_conversion_from_charset(struct archive *, const char *, int); /* Create the default string conversion object for reading/writing an archive. * Return NULL if the conversion is unneeded. * Note: On non Windows platform this always returns NULL. */ struct archive_string_conv * archive_string_default_conversion_for_read(struct archive *); struct archive_string_conv * archive_string_default_conversion_for_write(struct archive *); /* Dispose of a string conversion object. */ void archive_string_conversion_free(struct archive *); const char * archive_string_conversion_charset_name(struct archive_string_conv *); void archive_string_conversion_set_opt(struct archive_string_conv *, int); #define SCONV_SET_OPT_UTF8_LIBARCHIVE2X 1 #define SCONV_SET_OPT_NORMALIZATION_C 2 #define SCONV_SET_OPT_NORMALIZATION_D 4 /* Copy one archive_string to another in locale conversion. - * Return -1 if conversion failes. */ + * Return -1 if conversion fails. */ int archive_strncpy_l(struct archive_string *, const void *, size_t, struct archive_string_conv *); /* Copy one archive_string to another in locale conversion. - * Return -1 if conversion failes. */ + * Return -1 if conversion fails. */ int archive_strncat_l(struct archive_string *, const void *, size_t, struct archive_string_conv *); /* Copy one archive_string to another */ #define archive_string_copy(dest, src) \ ((dest)->length = 0, archive_string_concat((dest), (src))) #define archive_wstring_copy(dest, src) \ ((dest)->length = 0, archive_wstring_concat((dest), (src))) /* Concatenate one archive_string to another */ void archive_string_concat(struct archive_string *dest, struct archive_string *src); void archive_wstring_concat(struct archive_wstring *dest, struct archive_wstring *src); /* Ensure that the underlying buffer is at least as large as the request. */ struct archive_string * archive_string_ensure(struct archive_string *, size_t); struct archive_wstring * archive_wstring_ensure(struct archive_wstring *, size_t); /* Append C string, which may lack trailing \0. */ /* The source is declared void * here because this gets used with * "signed char *", "unsigned char *" and "char *" arguments. * Declaring it "char *" as with some of the other functions just * leads to a lot of extra casts. */ struct archive_string * archive_strncat(struct archive_string *, const void *, size_t); struct archive_wstring * archive_wstrncat(struct archive_wstring *, const wchar_t *, size_t); /* Append a C string to an archive_string, resizing as necessary. */ struct archive_string * archive_strcat(struct archive_string *, const void *); struct archive_wstring * archive_wstrcat(struct archive_wstring *, const wchar_t *); /* Copy a C string to an archive_string, resizing as necessary. */ #define archive_strcpy(as,p) \ archive_strncpy((as), (p), ((p) == NULL ? 0 : strlen(p))) #define archive_wstrcpy(as,p) \ archive_wstrncpy((as), (p), ((p) == NULL ? 0 : wcslen(p))) #define archive_strcpy_l(as,p,lo) \ archive_strncpy_l((as), (p), ((p) == NULL ? 0 : strlen(p)), (lo)) /* Copy a C string to an archive_string with limit, resizing as necessary. */ #define archive_strncpy(as,p,l) \ ((as)->length=0, archive_strncat((as), (p), (l))) #define archive_wstrncpy(as,p,l) \ ((as)->length = 0, archive_wstrncat((as), (p), (l))) /* Return length of string. */ #define archive_strlen(a) ((a)->length) /* Set string length to zero. */ #define archive_string_empty(a) ((a)->length = 0) #define archive_wstring_empty(a) ((a)->length = 0) /* Release any allocated storage resources. */ void archive_string_free(struct archive_string *); void archive_wstring_free(struct archive_wstring *); /* Like 'vsprintf', but resizes the underlying string as necessary. */ /* Note: This only implements a small subset of standard printf functionality. */ void archive_string_vsprintf(struct archive_string *, const char *, va_list) __LA_PRINTF(2, 0); void archive_string_sprintf(struct archive_string *, const char *, ...) __LA_PRINTF(2, 3); /* Translates from MBS to Unicode. */ /* Returns non-zero if conversion failed in any way. */ int archive_wstring_append_from_mbs(struct archive_wstring *dest, const char *, size_t); /* A "multistring" can hold Unicode, UTF8, or MBS versions of * the string. If you set and read the same version, no translation * is done. If you set and read different versions, the library * will attempt to transparently convert. */ struct archive_mstring { struct archive_string aes_mbs; struct archive_string aes_utf8; struct archive_wstring aes_wcs; struct archive_string aes_mbs_in_locale; /* Bitmap of which of the above are valid. Because we're lazy * about malloc-ing and reusing the underlying storage, we * can't rely on NULL pointers to indicate whether a string * has been set. */ int aes_set; #define AES_SET_MBS 1 #define AES_SET_UTF8 2 #define AES_SET_WCS 4 }; void archive_mstring_clean(struct archive_mstring *); void archive_mstring_copy(struct archive_mstring *dest, struct archive_mstring *src); int archive_mstring_get_mbs(struct archive *, struct archive_mstring *, const char **); int archive_mstring_get_utf8(struct archive *, struct archive_mstring *, const char **); int archive_mstring_get_wcs(struct archive *, struct archive_mstring *, const wchar_t **); int archive_mstring_get_mbs_l(struct archive_mstring *, const char **, size_t *, struct archive_string_conv *); int archive_mstring_copy_mbs(struct archive_mstring *, const char *mbs); int archive_mstring_copy_mbs_len(struct archive_mstring *, const char *mbs, size_t); int archive_mstring_copy_utf8(struct archive_mstring *, const char *utf8); int archive_mstring_copy_wcs(struct archive_mstring *, const wchar_t *wcs); int archive_mstring_copy_wcs_len(struct archive_mstring *, const wchar_t *wcs, size_t); int archive_mstring_copy_mbs_len_l(struct archive_mstring *, const char *mbs, size_t, struct archive_string_conv *); int archive_mstring_update_utf8(struct archive *, struct archive_mstring *aes, const char *utf8); #endif Index: head/contrib/libarchive/libarchive/archive_write_set_format_7zip.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_7zip.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_write_set_format_7zip.c (revision 309590) @@ -1,2328 +1,2328 @@ /*- * 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 menber of struct file to chain. + * 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); 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 Nume 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_alloc(void *p, size_t size) { (void)p; return malloc(size); } static void ppmd_free(void *p, void *address) { (void)p; free(address); } static ISzAlloc g_szalloc = { ppmd_alloc, ppmd_free }; 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, &g_szalloc); 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, &g_szalloc); 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/archive_write_set_format_cpio.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_cpio.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_write_set_format_cpio.c (revision 309590) @@ -1,500 +1,500 @@ /*- * Copyright (c) 2003-2007 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("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_write_private.h" static ssize_t archive_write_cpio_data(struct archive_write *, const void *buff, size_t s); static int archive_write_cpio_close(struct archive_write *); static int archive_write_cpio_free(struct archive_write *); static int archive_write_cpio_finish_entry(struct archive_write *); static int archive_write_cpio_header(struct archive_write *, struct archive_entry *); static int archive_write_cpio_options(struct archive_write *, const char *, const char *); static int format_octal(int64_t, void *, int); static int64_t format_octal_recursive(int64_t, char *, int); static int write_header(struct archive_write *, struct archive_entry *); struct cpio { uint64_t entry_bytes_remaining; int64_t ino_next; struct { int64_t old; int new;} *ino_list; size_t ino_list_size; size_t ino_list_next; struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv_default; int init_default_conversion; }; #define c_magic_offset 0 #define c_magic_size 6 #define c_dev_offset 6 #define c_dev_size 6 #define c_ino_offset 12 #define c_ino_size 6 #define c_mode_offset 18 #define c_mode_size 6 #define c_uid_offset 24 #define c_uid_size 6 #define c_gid_offset 30 #define c_gid_size 6 #define c_nlink_offset 36 #define c_nlink_size 6 #define c_rdev_offset 42 #define c_rdev_size 6 #define c_mtime_offset 48 #define c_mtime_size 11 #define c_namesize_offset 59 #define c_namesize_size 6 #define c_filesize_offset 65 #define c_filesize_size 11 /* * Set output format to 'cpio' format. */ int archive_write_set_format_cpio(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct cpio *cpio; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_cpio"); /* If someone else was already registered, unregister them. */ if (a->format_free != NULL) (a->format_free)(a); cpio = (struct cpio *)calloc(1, sizeof(*cpio)); if (cpio == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate cpio data"); return (ARCHIVE_FATAL); } a->format_data = cpio; a->format_name = "cpio"; a->format_options = archive_write_cpio_options; a->format_write_header = archive_write_cpio_header; a->format_write_data = archive_write_cpio_data; a->format_finish_entry = archive_write_cpio_finish_entry; a->format_close = archive_write_cpio_close; a->format_free = archive_write_cpio_free; a->archive.archive_format = ARCHIVE_FORMAT_CPIO_POSIX; a->archive.archive_format_name = "POSIX cpio"; return (ARCHIVE_OK); } static int archive_write_cpio_options(struct archive_write *a, const char *key, const char *val) { struct cpio *cpio = (struct cpio *)a->format_data; int ret = ARCHIVE_FAILED; if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: hdrcharset option needs a character-set name", a->format_name); else { cpio->opt_sconv = archive_string_conversion_to_charset( &a->archive, val, 0); if (cpio->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } /* * Ino values are as long as 64 bits on some systems; cpio format * only allows 18 bits and relies on the ino values to identify hardlinked * files. So, we can't merely "hash" the ino numbers since collisions * would corrupt the archive. Instead, we generate synthetic ino values * to store in the archive and maintain a map of original ino values to * synthetic ones so we can preserve hardlink information. * * TODO: Make this more efficient. It's not as bad as it looks (most * files don't have any hardlinks and we don't do any work here for those), * but it wouldn't be hard to do better. * * TODO: Work with dev/ino pairs here instead of just ino values. */ static int synthesize_ino_value(struct cpio *cpio, struct archive_entry *entry) { int64_t ino = archive_entry_ino64(entry); int ino_new; size_t i; /* * If no index number was given, don't assign one. In * particular, this handles the end-of-archive marker * correctly by giving it a zero index value. (This is also * why we start our synthetic index numbers with one below.) */ if (ino == 0) return (0); /* Don't store a mapping if we don't need to. */ if (archive_entry_nlink(entry) < 2) { return (int)(++cpio->ino_next); } /* Look up old ino; if we have it, this is a hardlink * and we reuse the same value. */ for (i = 0; i < cpio->ino_list_next; ++i) { if (cpio->ino_list[i].old == ino) return (cpio->ino_list[i].new); } /* Assign a new index number. */ ino_new = (int)(++cpio->ino_next); /* Ensure space for the new mapping. */ if (cpio->ino_list_size <= cpio->ino_list_next) { size_t newsize = cpio->ino_list_size < 512 ? 512 : cpio->ino_list_size * 2; void *newlist = realloc(cpio->ino_list, sizeof(cpio->ino_list[0]) * newsize); if (newlist == NULL) return (-1); cpio->ino_list_size = newsize; cpio->ino_list = newlist; } /* Record and return the new value. */ cpio->ino_list[cpio->ino_list_next].old = ino; cpio->ino_list[cpio->ino_list_next].new = ino_new; ++cpio->ino_list_next; return (ino_new); } static struct archive_string_conv * get_sconv(struct archive_write *a) { struct cpio *cpio; struct archive_string_conv *sconv; cpio = (struct cpio *)a->format_data; sconv = cpio->opt_sconv; if (sconv == NULL) { if (!cpio->init_default_conversion) { cpio->sconv_default = archive_string_default_conversion_for_write( &(a->archive)); cpio->init_default_conversion = 1; } sconv = cpio->sconv_default; } return (sconv); } static int archive_write_cpio_header(struct archive_write *a, struct archive_entry *entry) { const char *path; size_t len; if (archive_entry_filetype(entry) == 0) { archive_set_error(&a->archive, -1, "Filetype required"); return (ARCHIVE_FAILED); } if (archive_entry_pathname_l(entry, &path, &len, get_sconv(a)) != 0 && errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } if (len == 0 || path == NULL || path[0] == '\0') { archive_set_error(&a->archive, -1, "Pathname required"); return (ARCHIVE_FAILED); } if (!archive_entry_size_is_set(entry) || archive_entry_size(entry) < 0) { archive_set_error(&a->archive, -1, "Size required"); return (ARCHIVE_FAILED); } return write_header(a, entry); } static int write_header(struct archive_write *a, struct archive_entry *entry) { struct cpio *cpio; const char *p, *path; int pathlength, ret, ret_final; int64_t ino; char h[76]; struct archive_string_conv *sconv; struct archive_entry *entry_main; size_t len; cpio = (struct cpio *)a->format_data; ret_final = ARCHIVE_OK; sconv = get_sconv(a); #if defined(_WIN32) && !defined(__CYGWIN__) - /* Make sure the path separators in pahtname, hardlink and symlink + /* Make sure the path separators in pathname, hardlink and symlink * are all slash '/', not the Windows path separator '\'. */ entry_main = __la_win_entry_in_posix_pathseparator(entry); if (entry_main == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate ustar data"); return(ARCHIVE_FATAL); } if (entry != entry_main) entry = entry_main; else entry_main = NULL; #else entry_main = NULL; #endif ret = archive_entry_pathname_l(entry, &path, &len, sconv); if (ret != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); ret_final = ARCHIVE_FATAL; goto exit_write_header; } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate pathname '%s' to %s", archive_entry_pathname(entry), archive_string_conversion_charset_name(sconv)); ret_final = ARCHIVE_WARN; } /* Include trailing null. */ pathlength = (int)len + 1; memset(h, 0, sizeof(h)); format_octal(070707, h + c_magic_offset, c_magic_size); format_octal(archive_entry_dev(entry), h + c_dev_offset, c_dev_size); ino = synthesize_ino_value(cpio, entry); if (ino < 0) { archive_set_error(&a->archive, ENOMEM, "No memory for ino translation table"); ret_final = ARCHIVE_FATAL; goto exit_write_header; } else if (ino > 0777777) { archive_set_error(&a->archive, ERANGE, "Too many files for this cpio format"); ret_final = ARCHIVE_FATAL; goto exit_write_header; } format_octal(ino & 0777777, h + c_ino_offset, c_ino_size); /* TODO: Set ret_final to ARCHIVE_WARN if any of these overflow. */ format_octal(archive_entry_mode(entry), h + c_mode_offset, c_mode_size); format_octal(archive_entry_uid(entry), h + c_uid_offset, c_uid_size); format_octal(archive_entry_gid(entry), h + c_gid_offset, c_gid_size); format_octal(archive_entry_nlink(entry), h + c_nlink_offset, c_nlink_size); if (archive_entry_filetype(entry) == AE_IFBLK || archive_entry_filetype(entry) == AE_IFCHR) format_octal(archive_entry_dev(entry), h + c_rdev_offset, c_rdev_size); else format_octal(0, h + c_rdev_offset, c_rdev_size); format_octal(archive_entry_mtime(entry), h + c_mtime_offset, c_mtime_size); format_octal(pathlength, h + c_namesize_offset, c_namesize_size); /* Non-regular files don't store bodies. */ if (archive_entry_filetype(entry) != AE_IFREG) archive_entry_set_size(entry, 0); /* Symlinks get the link written as the body of the entry. */ ret = archive_entry_symlink_l(entry, &p, &len, sconv); if (ret != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); ret_final = ARCHIVE_FATAL; goto exit_write_header; } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate linkname '%s' to %s", archive_entry_symlink(entry), archive_string_conversion_charset_name(sconv)); ret_final = ARCHIVE_WARN; } if (len > 0 && p != NULL && *p != '\0') ret = format_octal(strlen(p), h + c_filesize_offset, c_filesize_size); else ret = format_octal(archive_entry_size(entry), h + c_filesize_offset, c_filesize_size); if (ret) { archive_set_error(&a->archive, ERANGE, "File is too large for cpio format."); ret_final = ARCHIVE_FAILED; goto exit_write_header; } ret = __archive_write_output(a, h, sizeof(h)); if (ret != ARCHIVE_OK) { ret_final = ARCHIVE_FATAL; goto exit_write_header; } ret = __archive_write_output(a, path, pathlength); if (ret != ARCHIVE_OK) { ret_final = ARCHIVE_FATAL; goto exit_write_header; } cpio->entry_bytes_remaining = archive_entry_size(entry); /* Write the symlink now. */ if (p != NULL && *p != '\0') { ret = __archive_write_output(a, p, strlen(p)); if (ret != ARCHIVE_OK) { ret_final = ARCHIVE_FATAL; goto exit_write_header; } } exit_write_header: if (entry_main) archive_entry_free(entry_main); return (ret_final); } static ssize_t archive_write_cpio_data(struct archive_write *a, const void *buff, size_t s) { struct cpio *cpio; int ret; cpio = (struct cpio *)a->format_data; if (s > cpio->entry_bytes_remaining) s = (size_t)cpio->entry_bytes_remaining; ret = __archive_write_output(a, buff, s); cpio->entry_bytes_remaining -= s; if (ret >= 0) return (s); else return (ret); } /* * Format a number into the specified field. */ static int format_octal(int64_t v, void *p, int digits) { int64_t max; int ret; max = (((int64_t)1) << (digits * 3)) - 1; if (v >= 0 && v <= max) { format_octal_recursive(v, (char *)p, digits); ret = 0; } else { format_octal_recursive(max, (char *)p, digits); ret = -1; } return (ret); } static int64_t format_octal_recursive(int64_t v, char *p, int s) { if (s == 0) return (v); v = format_octal_recursive(v, p+1, s-1); *p = '0' + ((char)v & 7); return (v >> 3); } static int archive_write_cpio_close(struct archive_write *a) { int er; struct archive_entry *trailer; trailer = archive_entry_new2(NULL); /* nlink = 1 here for GNU cpio compat. */ archive_entry_set_nlink(trailer, 1); archive_entry_set_size(trailer, 0); archive_entry_set_pathname(trailer, "TRAILER!!!"); er = write_header(a, trailer); archive_entry_free(trailer); return (er); } static int archive_write_cpio_free(struct archive_write *a) { struct cpio *cpio; cpio = (struct cpio *)a->format_data; free(cpio->ino_list); free(cpio); a->format_data = NULL; return (ARCHIVE_OK); } static int archive_write_cpio_finish_entry(struct archive_write *a) { struct cpio *cpio; cpio = (struct cpio *)a->format_data; return (__archive_write_nulls(a, (size_t)cpio->entry_bytes_remaining)); } Index: head/contrib/libarchive/libarchive/archive_write_set_format_cpio_newc.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_cpio_newc.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_write_set_format_cpio_newc.c (revision 309590) @@ -1,458 +1,458 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * Copyright (c) 2006 Rudolf Marek SYSGO s.r.o. * 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_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_write_private.h" static ssize_t archive_write_newc_data(struct archive_write *, const void *buff, size_t s); static int archive_write_newc_close(struct archive_write *); static int archive_write_newc_free(struct archive_write *); static int archive_write_newc_finish_entry(struct archive_write *); static int archive_write_newc_header(struct archive_write *, struct archive_entry *); static int archive_write_newc_options(struct archive_write *, const char *, const char *); static int format_hex(int64_t, void *, int); static int64_t format_hex_recursive(int64_t, char *, int); static int write_header(struct archive_write *, struct archive_entry *); struct cpio { uint64_t entry_bytes_remaining; int padding; struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv_default; int init_default_conversion; }; #define c_magic_offset 0 #define c_magic_size 6 #define c_ino_offset 6 #define c_ino_size 8 #define c_mode_offset 14 #define c_mode_size 8 #define c_uid_offset 22 #define c_uid_size 8 #define c_gid_offset 30 #define c_gid_size 8 #define c_nlink_offset 38 #define c_nlink_size 8 #define c_mtime_offset 46 #define c_mtime_size 8 #define c_filesize_offset 54 #define c_filesize_size 8 #define c_devmajor_offset 62 #define c_devmajor_size 8 #define c_devminor_offset 70 #define c_devminor_size 8 #define c_rdevmajor_offset 78 #define c_rdevmajor_size 8 #define c_rdevminor_offset 86 #define c_rdevminor_size 8 #define c_namesize_offset 94 #define c_namesize_size 8 #define c_checksum_offset 102 #define c_checksum_size 8 #define c_header_size 110 /* Logic trick: difference between 'n' and next multiple of 4 */ #define PAD4(n) (3 & (1 + ~(n))) /* * Set output format to 'cpio' format. */ int archive_write_set_format_cpio_newc(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct cpio *cpio; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_cpio_newc"); /* If someone else was already registered, unregister them. */ if (a->format_free != NULL) (a->format_free)(a); cpio = (struct cpio *)malloc(sizeof(*cpio)); if (cpio == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate cpio data"); return (ARCHIVE_FATAL); } memset(cpio, 0, sizeof(*cpio)); a->format_data = cpio; a->format_name = "cpio"; a->format_options = archive_write_newc_options; a->format_write_header = archive_write_newc_header; a->format_write_data = archive_write_newc_data; a->format_finish_entry = archive_write_newc_finish_entry; a->format_close = archive_write_newc_close; a->format_free = archive_write_newc_free; a->archive.archive_format = ARCHIVE_FORMAT_CPIO_SVR4_NOCRC; a->archive.archive_format_name = "SVR4 cpio nocrc"; return (ARCHIVE_OK); } static int archive_write_newc_options(struct archive_write *a, const char *key, const char *val) { struct cpio *cpio = (struct cpio *)a->format_data; int ret = ARCHIVE_FAILED; if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: hdrcharset option needs a character-set name", a->format_name); else { cpio->opt_sconv = archive_string_conversion_to_charset( &a->archive, val, 0); if (cpio->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static struct archive_string_conv * get_sconv(struct archive_write *a) { struct cpio *cpio; struct archive_string_conv *sconv; cpio = (struct cpio *)a->format_data; sconv = cpio->opt_sconv; if (sconv == NULL) { if (!cpio->init_default_conversion) { cpio->sconv_default = archive_string_default_conversion_for_write( &(a->archive)); cpio->init_default_conversion = 1; } sconv = cpio->sconv_default; } return (sconv); } static int archive_write_newc_header(struct archive_write *a, struct archive_entry *entry) { const char *path; size_t len; if (archive_entry_filetype(entry) == 0) { archive_set_error(&a->archive, -1, "Filetype required"); return (ARCHIVE_FAILED); } if (archive_entry_pathname_l(entry, &path, &len, get_sconv(a)) != 0 && errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } if (len == 0 || path == NULL || path[0] == '\0') { archive_set_error(&a->archive, -1, "Pathname required"); return (ARCHIVE_FAILED); } if (archive_entry_hardlink(entry) == NULL && (!archive_entry_size_is_set(entry) || archive_entry_size(entry) < 0)) { archive_set_error(&a->archive, -1, "Size required"); return (ARCHIVE_FAILED); } return write_header(a, entry); } static int write_header(struct archive_write *a, struct archive_entry *entry) { int64_t ino; struct cpio *cpio; const char *p, *path; int pathlength, ret, ret_final; char h[c_header_size]; struct archive_string_conv *sconv; struct archive_entry *entry_main; size_t len; int pad; cpio = (struct cpio *)a->format_data; ret_final = ARCHIVE_OK; sconv = get_sconv(a); #if defined(_WIN32) && !defined(__CYGWIN__) - /* Make sure the path separators in pahtname, hardlink and symlink + /* Make sure the path separators in pathname, hardlink and symlink * are all slash '/', not the Windows path separator '\'. */ entry_main = __la_win_entry_in_posix_pathseparator(entry); if (entry_main == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate ustar data"); return(ARCHIVE_FATAL); } if (entry != entry_main) entry = entry_main; else entry_main = NULL; #else entry_main = NULL; #endif ret = archive_entry_pathname_l(entry, &path, &len, sconv); if (ret != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); ret_final = ARCHIVE_FATAL; goto exit_write_header; } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate pathname '%s' to %s", archive_entry_pathname(entry), archive_string_conversion_charset_name(sconv)); ret_final = ARCHIVE_WARN; } pathlength = (int)len + 1; /* Include trailing null. */ memset(h, 0, c_header_size); format_hex(0x070701, h + c_magic_offset, c_magic_size); format_hex(archive_entry_devmajor(entry), h + c_devmajor_offset, c_devmajor_size); format_hex(archive_entry_devminor(entry), h + c_devminor_offset, c_devminor_size); ino = archive_entry_ino64(entry); if (ino > 0xffffffff) { archive_set_error(&a->archive, ERANGE, "large inode number truncated"); ret_final = ARCHIVE_WARN; } /* TODO: Set ret_final to ARCHIVE_WARN if any of these overflow. */ format_hex(ino & 0xffffffff, h + c_ino_offset, c_ino_size); format_hex(archive_entry_mode(entry), h + c_mode_offset, c_mode_size); format_hex(archive_entry_uid(entry), h + c_uid_offset, c_uid_size); format_hex(archive_entry_gid(entry), h + c_gid_offset, c_gid_size); format_hex(archive_entry_nlink(entry), h + c_nlink_offset, c_nlink_size); if (archive_entry_filetype(entry) == AE_IFBLK || archive_entry_filetype(entry) == AE_IFCHR) { format_hex(archive_entry_rdevmajor(entry), h + c_rdevmajor_offset, c_rdevmajor_size); format_hex(archive_entry_rdevminor(entry), h + c_rdevminor_offset, c_rdevminor_size); } else { format_hex(0, h + c_rdevmajor_offset, c_rdevmajor_size); format_hex(0, h + c_rdevminor_offset, c_rdevminor_size); } format_hex(archive_entry_mtime(entry), h + c_mtime_offset, c_mtime_size); format_hex(pathlength, h + c_namesize_offset, c_namesize_size); format_hex(0, h + c_checksum_offset, c_checksum_size); /* Non-regular files don't store bodies. */ if (archive_entry_filetype(entry) != AE_IFREG) archive_entry_set_size(entry, 0); /* Symlinks get the link written as the body of the entry. */ ret = archive_entry_symlink_l(entry, &p, &len, sconv); if (ret != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Likname"); ret_final = ARCHIVE_FATAL; goto exit_write_header; } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate linkname '%s' to %s", archive_entry_symlink(entry), archive_string_conversion_charset_name(sconv)); ret_final = ARCHIVE_WARN; } if (len > 0 && p != NULL && *p != '\0') ret = format_hex(strlen(p), h + c_filesize_offset, c_filesize_size); else ret = format_hex(archive_entry_size(entry), h + c_filesize_offset, c_filesize_size); if (ret) { archive_set_error(&a->archive, ERANGE, "File is too large for this format."); ret_final = ARCHIVE_FAILED; goto exit_write_header; } ret = __archive_write_output(a, h, c_header_size); if (ret != ARCHIVE_OK) { ret_final = ARCHIVE_FATAL; goto exit_write_header; } /* Pad pathname to even length. */ ret = __archive_write_output(a, path, pathlength); if (ret != ARCHIVE_OK) { ret_final = ARCHIVE_FATAL; goto exit_write_header; } pad = PAD4(pathlength + c_header_size); if (pad) { ret = __archive_write_output(a, "\0\0\0", pad); if (ret != ARCHIVE_OK) { ret_final = ARCHIVE_FATAL; goto exit_write_header; } } cpio->entry_bytes_remaining = archive_entry_size(entry); cpio->padding = (int)PAD4(cpio->entry_bytes_remaining); /* Write the symlink now. */ if (p != NULL && *p != '\0') { ret = __archive_write_output(a, p, strlen(p)); if (ret != ARCHIVE_OK) { ret_final = ARCHIVE_FATAL; goto exit_write_header; } pad = PAD4(strlen(p)); ret = __archive_write_output(a, "\0\0\0", pad); if (ret != ARCHIVE_OK) { ret_final = ARCHIVE_FATAL; goto exit_write_header; } } exit_write_header: if (entry_main) archive_entry_free(entry_main); return (ret_final); } static ssize_t archive_write_newc_data(struct archive_write *a, const void *buff, size_t s) { struct cpio *cpio; int ret; cpio = (struct cpio *)a->format_data; if (s > cpio->entry_bytes_remaining) s = (size_t)cpio->entry_bytes_remaining; ret = __archive_write_output(a, buff, s); cpio->entry_bytes_remaining -= s; if (ret >= 0) return (s); else return (ret); } /* * Format a number into the specified field. */ static int format_hex(int64_t v, void *p, int digits) { int64_t max; int ret; max = (((int64_t)1) << (digits * 4)) - 1; if (v >= 0 && v <= max) { format_hex_recursive(v, (char *)p, digits); ret = 0; } else { format_hex_recursive(max, (char *)p, digits); ret = -1; } return (ret); } static int64_t format_hex_recursive(int64_t v, char *p, int s) { if (s == 0) return (v); v = format_hex_recursive(v, p+1, s-1); *p = "0123456789abcdef"[v & 0xf]; return (v >> 4); } static int archive_write_newc_close(struct archive_write *a) { int er; struct archive_entry *trailer; trailer = archive_entry_new(); archive_entry_set_nlink(trailer, 1); archive_entry_set_size(trailer, 0); archive_entry_set_pathname(trailer, "TRAILER!!!"); /* Bypass the required data checks. */ er = write_header(a, trailer); archive_entry_free(trailer); return (er); } static int archive_write_newc_free(struct archive_write *a) { struct cpio *cpio; cpio = (struct cpio *)a->format_data; free(cpio); a->format_data = NULL; return (ARCHIVE_OK); } static int archive_write_newc_finish_entry(struct archive_write *a) { struct cpio *cpio; cpio = (struct cpio *)a->format_data; return (__archive_write_nulls(a, (size_t)cpio->entry_bytes_remaining + cpio->padding)); } Index: head/contrib/libarchive/libarchive/archive_write_set_format_gnutar.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_gnutar.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_write_set_format_gnutar.c (revision 309590) @@ -1,763 +1,763 @@ /*- * Copyright (c) 2010 Nokia Corporation and/or its subsidiary(-ies). * Author: Jonas Gastal * 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: head/lib/libarchive/archive_write_set_format_gnu_tar.c 191579 2009-04-27 18:35:03Z gastal $"); #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_write_private.h" struct gnutar { uint64_t entry_bytes_remaining; uint64_t entry_padding; const char * linkname; size_t linkname_length; const char * pathname; size_t pathname_length; const char * uname; size_t uname_length; const char * gname; size_t gname_length; struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv_default; int init_default_conversion; }; /* * Define structure of GNU tar header. */ #define GNUTAR_name_offset 0 #define GNUTAR_name_size 100 #define GNUTAR_mode_offset 100 #define GNUTAR_mode_size 7 #define GNUTAR_mode_max_size 8 #define GNUTAR_uid_offset 108 #define GNUTAR_uid_size 7 #define GNUTAR_uid_max_size 8 #define GNUTAR_gid_offset 116 #define GNUTAR_gid_size 7 #define GNUTAR_gid_max_size 8 #define GNUTAR_size_offset 124 #define GNUTAR_size_size 11 #define GNUTAR_size_max_size 12 #define GNUTAR_mtime_offset 136 #define GNUTAR_mtime_size 11 #define GNUTAR_mtime_max_size 11 #define GNUTAR_checksum_offset 148 #define GNUTAR_checksum_size 8 #define GNUTAR_typeflag_offset 156 #define GNUTAR_typeflag_size 1 #define GNUTAR_linkname_offset 157 #define GNUTAR_linkname_size 100 #define GNUTAR_magic_offset 257 #define GNUTAR_magic_size 6 #define GNUTAR_version_offset 263 #define GNUTAR_version_size 2 #define GNUTAR_uname_offset 265 #define GNUTAR_uname_size 32 #define GNUTAR_gname_offset 297 #define GNUTAR_gname_size 32 #define GNUTAR_rdevmajor_offset 329 #define GNUTAR_rdevmajor_size 6 #define GNUTAR_rdevmajor_max_size 8 #define GNUTAR_rdevminor_offset 337 #define GNUTAR_rdevminor_size 6 #define GNUTAR_rdevminor_max_size 8 /* * A filled-in copy of the header for initialization. */ static const char template_header[] = { /* name: 100 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0, /* Mode, null termination: 8 bytes */ '0','0','0','0','0','0', '0','\0', /* uid, null termination: 8 bytes */ '0','0','0','0','0','0', '0','\0', /* gid, null termination: 8 bytes */ '0','0','0','0','0','0', '0','\0', - /* size, space termation: 12 bytes */ + /* size, space termination: 12 bytes */ '0','0','0','0','0','0','0','0','0','0','0', '\0', - /* mtime, space termation: 12 bytes */ + /* mtime, space termination: 12 bytes */ '0','0','0','0','0','0','0','0','0','0','0', '\0', /* Initial checksum value: 8 spaces */ ' ',' ',' ',' ',' ',' ',' ',' ', /* Typeflag: 1 byte */ '0', /* '0' = regular file */ /* Linkname: 100 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0, /* Magic: 8 bytes */ 'u','s','t','a','r',' ', ' ','\0', /* Uname: 32 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, /* Gname: 32 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, /* rdevmajor + null padding: 8 bytes */ '\0','\0','\0','\0','\0','\0', '\0','\0', /* rdevminor + null padding: 8 bytes */ '\0','\0','\0','\0','\0','\0', '\0','\0', /* Padding: 167 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0 }; static int archive_write_gnutar_options(struct archive_write *, const char *, const char *); static int archive_format_gnutar_header(struct archive_write *, char h[512], struct archive_entry *, int tartype); static int archive_write_gnutar_header(struct archive_write *, struct archive_entry *entry); static ssize_t archive_write_gnutar_data(struct archive_write *a, const void *buff, size_t s); static int archive_write_gnutar_free(struct archive_write *); static int archive_write_gnutar_close(struct archive_write *); static int archive_write_gnutar_finish_entry(struct archive_write *); static int format_256(int64_t, char *, int); static int format_number(int64_t, char *, int size, int maxsize); static int format_octal(int64_t, char *, int); /* * Set output format to 'GNU tar' format. */ int archive_write_set_format_gnutar(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct gnutar *gnutar; gnutar = (struct gnutar *)calloc(1, sizeof(*gnutar)); if (gnutar == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate gnutar data"); return (ARCHIVE_FATAL); } a->format_data = gnutar; a->format_name = "gnutar"; a->format_options = archive_write_gnutar_options; a->format_write_header = archive_write_gnutar_header; a->format_write_data = archive_write_gnutar_data; a->format_close = archive_write_gnutar_close; a->format_free = archive_write_gnutar_free; a->format_finish_entry = archive_write_gnutar_finish_entry; a->archive.archive_format = ARCHIVE_FORMAT_TAR_GNUTAR; a->archive.archive_format_name = "GNU tar"; return (ARCHIVE_OK); } static int archive_write_gnutar_options(struct archive_write *a, const char *key, const char *val) { struct gnutar *gnutar = (struct gnutar *)a->format_data; int ret = ARCHIVE_FAILED; if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: hdrcharset option needs a character-set name", a->format_name); else { gnutar->opt_sconv = archive_string_conversion_to_charset( &a->archive, val, 0); if (gnutar->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int archive_write_gnutar_close(struct archive_write *a) { return (__archive_write_nulls(a, 512*2)); } static int archive_write_gnutar_free(struct archive_write *a) { struct gnutar *gnutar; gnutar = (struct gnutar *)a->format_data; free(gnutar); a->format_data = NULL; return (ARCHIVE_OK); } static int archive_write_gnutar_finish_entry(struct archive_write *a) { struct gnutar *gnutar; int ret; gnutar = (struct gnutar *)a->format_data; ret = __archive_write_nulls(a, (size_t) (gnutar->entry_bytes_remaining + gnutar->entry_padding)); gnutar->entry_bytes_remaining = gnutar->entry_padding = 0; return (ret); } static ssize_t archive_write_gnutar_data(struct archive_write *a, const void *buff, size_t s) { struct gnutar *gnutar; int ret; gnutar = (struct gnutar *)a->format_data; if (s > gnutar->entry_bytes_remaining) s = (size_t)gnutar->entry_bytes_remaining; ret = __archive_write_output(a, buff, s); gnutar->entry_bytes_remaining -= s; if (ret != ARCHIVE_OK) return (ret); return (s); } static int archive_write_gnutar_header(struct archive_write *a, struct archive_entry *entry) { char buff[512]; int r, ret, ret2 = ARCHIVE_OK; int tartype; struct gnutar *gnutar; struct archive_string_conv *sconv; struct archive_entry *entry_main; gnutar = (struct gnutar *)a->format_data; /* Setup default string conversion. */ if (gnutar->opt_sconv == NULL) { if (!gnutar->init_default_conversion) { gnutar->sconv_default = archive_string_default_conversion_for_write( &(a->archive)); gnutar->init_default_conversion = 1; } sconv = gnutar->sconv_default; } else sconv = gnutar->opt_sconv; /* Only regular files (not hardlinks) have data. */ if (archive_entry_hardlink(entry) != NULL || archive_entry_symlink(entry) != NULL || !(archive_entry_filetype(entry) == AE_IFREG)) archive_entry_set_size(entry, 0); if (AE_IFDIR == archive_entry_filetype(entry)) { const char *p; size_t path_length; /* * Ensure a trailing '/'. Modify the entry so * the client sees the change. */ #if defined(_WIN32) && !defined(__CYGWIN__) const wchar_t *wp; wp = archive_entry_pathname_w(entry); if (wp != NULL && wp[wcslen(wp) -1] != L'/') { struct archive_wstring ws; archive_string_init(&ws); path_length = wcslen(wp); if (archive_wstring_ensure(&ws, path_length + 2) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate ustar data"); archive_wstring_free(&ws); return(ARCHIVE_FATAL); } /* Should we keep '\' ? */ if (wp[path_length -1] == L'\\') path_length--; archive_wstrncpy(&ws, wp, path_length); archive_wstrappend_wchar(&ws, L'/'); archive_entry_copy_pathname_w(entry, ws.s); archive_wstring_free(&ws); p = NULL; } else #endif p = archive_entry_pathname(entry); /* * On Windows, this is a backup operation just in * case getting WCS failed. On POSIX, this is a * normal operation. */ if (p != NULL && p[strlen(p) - 1] != '/') { struct archive_string as; archive_string_init(&as); path_length = strlen(p); if (archive_string_ensure(&as, path_length + 2) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate ustar data"); archive_string_free(&as); return(ARCHIVE_FATAL); } #if defined(_WIN32) && !defined(__CYGWIN__) /* NOTE: This might break the pathname * if the current code page is CP932 and * the pathname includes a character '\' * as a part of its multibyte pathname. */ if (p[strlen(p) -1] == '\\') path_length--; else #endif archive_strncpy(&as, p, path_length); archive_strappend_char(&as, '/'); archive_entry_copy_pathname(entry, as.s); archive_string_free(&as); } } #if defined(_WIN32) && !defined(__CYGWIN__) - /* Make sure the path separators in pahtname, hardlink and symlink + /* Make sure the path separators in pathname, hardlink and symlink * are all slash '/', not the Windows path separator '\'. */ entry_main = __la_win_entry_in_posix_pathseparator(entry); if (entry_main == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate ustar data"); return(ARCHIVE_FATAL); } if (entry != entry_main) entry = entry_main; else entry_main = NULL; #else entry_main = NULL; #endif r = archive_entry_pathname_l(entry, &(gnutar->pathname), &(gnutar->pathname_length), sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathame"); ret = ARCHIVE_FATAL; goto exit_write_header; } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate pathname '%s' to %s", archive_entry_pathname(entry), archive_string_conversion_charset_name(sconv)); ret2 = ARCHIVE_WARN; } r = archive_entry_uname_l(entry, &(gnutar->uname), &(gnutar->uname_length), sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Uname"); ret = ARCHIVE_FATAL; goto exit_write_header; } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate uname '%s' to %s", archive_entry_uname(entry), archive_string_conversion_charset_name(sconv)); ret2 = ARCHIVE_WARN; } r = archive_entry_gname_l(entry, &(gnutar->gname), &(gnutar->gname_length), sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Gname"); ret = ARCHIVE_FATAL; goto exit_write_header; } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate gname '%s' to %s", archive_entry_gname(entry), archive_string_conversion_charset_name(sconv)); ret2 = ARCHIVE_WARN; } /* If linkname is longer than 100 chars we need to add a 'K' header. */ r = archive_entry_hardlink_l(entry, &(gnutar->linkname), &(gnutar->linkname_length), sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); ret = ARCHIVE_FATAL; goto exit_write_header; } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate linkname '%s' to %s", archive_entry_hardlink(entry), archive_string_conversion_charset_name(sconv)); ret2 = ARCHIVE_WARN; } if (gnutar->linkname_length == 0) { r = archive_entry_symlink_l(entry, &(gnutar->linkname), &(gnutar->linkname_length), sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); ret = ARCHIVE_FATAL; goto exit_write_header; } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate linkname '%s' to %s", archive_entry_hardlink(entry), archive_string_conversion_charset_name(sconv)); ret2 = ARCHIVE_WARN; } } if (gnutar->linkname_length > GNUTAR_linkname_size) { size_t length = gnutar->linkname_length + 1; struct archive_entry *temp = archive_entry_new2(&a->archive); /* Uname/gname here don't really matter since no one reads them; * these are the values that GNU tar happens to use on FreeBSD. */ archive_entry_set_uname(temp, "root"); archive_entry_set_gname(temp, "wheel"); archive_entry_set_pathname(temp, "././@LongLink"); archive_entry_set_size(temp, length); ret = archive_format_gnutar_header(a, buff, temp, 'K'); if (ret < ARCHIVE_WARN) goto exit_write_header; ret = __archive_write_output(a, buff, 512); if(ret < ARCHIVE_WARN) goto exit_write_header; archive_entry_free(temp); /* Write name and trailing null byte. */ ret = __archive_write_output(a, gnutar->linkname, length); if(ret < ARCHIVE_WARN) goto exit_write_header; /* Pad to 512 bytes */ ret = __archive_write_nulls(a, 0x1ff & (-(ssize_t)length)); if (ret < ARCHIVE_WARN) goto exit_write_header; } /* If pathname is longer than 100 chars we need to add an 'L' header. */ if (gnutar->pathname_length > GNUTAR_name_size) { const char *pathname = gnutar->pathname; size_t length = gnutar->pathname_length + 1; struct archive_entry *temp = archive_entry_new2(&a->archive); /* Uname/gname here don't really matter since no one reads them; * these are the values that GNU tar happens to use on FreeBSD. */ archive_entry_set_uname(temp, "root"); archive_entry_set_gname(temp, "wheel"); archive_entry_set_pathname(temp, "././@LongLink"); archive_entry_set_size(temp, length); ret = archive_format_gnutar_header(a, buff, temp, 'L'); if (ret < ARCHIVE_WARN) goto exit_write_header; ret = __archive_write_output(a, buff, 512); if(ret < ARCHIVE_WARN) goto exit_write_header; archive_entry_free(temp); /* Write pathname + trailing null byte. */ ret = __archive_write_output(a, pathname, length); if(ret < ARCHIVE_WARN) goto exit_write_header; /* Pad to multiple of 512 bytes. */ ret = __archive_write_nulls(a, 0x1ff & (-(ssize_t)length)); if (ret < ARCHIVE_WARN) goto exit_write_header; } if (archive_entry_hardlink(entry) != NULL) { tartype = '1'; } else switch (archive_entry_filetype(entry)) { case AE_IFREG: tartype = '0' ; break; case AE_IFLNK: tartype = '2' ; break; case AE_IFCHR: tartype = '3' ; break; case AE_IFBLK: tartype = '4' ; break; case AE_IFDIR: tartype = '5' ; break; case AE_IFIFO: tartype = '6' ; break; case AE_IFSOCK: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive socket"); ret = ARCHIVE_FAILED; goto exit_write_header; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive this (mode=0%lo)", (unsigned long)archive_entry_mode(entry)); ret = ARCHIVE_FAILED; goto exit_write_header; } ret = archive_format_gnutar_header(a, buff, entry, tartype); if (ret < ARCHIVE_WARN) goto exit_write_header; if (ret2 < ret) ret = ret2; ret2 = __archive_write_output(a, buff, 512); if (ret2 < ARCHIVE_WARN) { ret = ret2; goto exit_write_header; } if (ret2 < ret) ret = ret2; gnutar->entry_bytes_remaining = archive_entry_size(entry); gnutar->entry_padding = 0x1ff & (-(int64_t)gnutar->entry_bytes_remaining); exit_write_header: if (entry_main) archive_entry_free(entry_main); return (ret); } static int archive_format_gnutar_header(struct archive_write *a, char h[512], struct archive_entry *entry, int tartype) { unsigned int checksum; int i, ret; size_t copy_length; const char *p; struct gnutar *gnutar; gnutar = (struct gnutar *)a->format_data; ret = 0; /* * The "template header" already includes the signature, * various end-of-field markers, and other required elements. */ memcpy(h, &template_header, 512); /* * Because the block is already null-filled, and strings * are allowed to exactly fill their destination (without null), * I use memcpy(dest, src, strlen()) here a lot to copy strings. */ if (tartype == 'K' || tartype == 'L') { p = archive_entry_pathname(entry); copy_length = strlen(p); } else { p = gnutar->pathname; copy_length = gnutar->pathname_length; } if (copy_length > GNUTAR_name_size) copy_length = GNUTAR_name_size; memcpy(h + GNUTAR_name_offset, p, copy_length); if ((copy_length = gnutar->linkname_length) > 0) { if (copy_length > GNUTAR_linkname_size) copy_length = GNUTAR_linkname_size; memcpy(h + GNUTAR_linkname_offset, gnutar->linkname, copy_length); } /* TODO: How does GNU tar handle unames longer than GNUTAR_uname_size? */ if (tartype == 'K' || tartype == 'L') { p = archive_entry_uname(entry); copy_length = strlen(p); } else { p = gnutar->uname; copy_length = gnutar->uname_length; } if (copy_length > 0) { if (copy_length > GNUTAR_uname_size) copy_length = GNUTAR_uname_size; memcpy(h + GNUTAR_uname_offset, p, copy_length); } /* TODO: How does GNU tar handle gnames longer than GNUTAR_gname_size? */ if (tartype == 'K' || tartype == 'L') { p = archive_entry_gname(entry); copy_length = strlen(p); } else { p = gnutar->gname; copy_length = gnutar->gname_length; } if (copy_length > 0) { if (strlen(p) > GNUTAR_gname_size) copy_length = GNUTAR_gname_size; memcpy(h + GNUTAR_gname_offset, p, copy_length); } /* By truncating the mode here, we ensure it always fits. */ format_octal(archive_entry_mode(entry) & 07777, h + GNUTAR_mode_offset, GNUTAR_mode_size); /* GNU tar supports base-256 here, so should never overflow. */ if (format_number(archive_entry_uid(entry), h + GNUTAR_uid_offset, GNUTAR_uid_size, GNUTAR_uid_max_size)) { archive_set_error(&a->archive, ERANGE, "Numeric user ID %jd too large", (intmax_t)archive_entry_uid(entry)); ret = ARCHIVE_FAILED; } /* GNU tar supports base-256 here, so should never overflow. */ if (format_number(archive_entry_gid(entry), h + GNUTAR_gid_offset, GNUTAR_gid_size, GNUTAR_gid_max_size)) { archive_set_error(&a->archive, ERANGE, "Numeric group ID %jd too large", (intmax_t)archive_entry_gid(entry)); ret = ARCHIVE_FAILED; } /* GNU tar supports base-256 here, so should never overflow. */ if (format_number(archive_entry_size(entry), h + GNUTAR_size_offset, GNUTAR_size_size, GNUTAR_size_max_size)) { archive_set_error(&a->archive, ERANGE, "File size out of range"); ret = ARCHIVE_FAILED; } /* Shouldn't overflow before 2106, since mtime field is 33 bits. */ format_octal(archive_entry_mtime(entry), h + GNUTAR_mtime_offset, GNUTAR_mtime_size); if (archive_entry_filetype(entry) == AE_IFBLK || archive_entry_filetype(entry) == AE_IFCHR) { if (format_octal(archive_entry_rdevmajor(entry), h + GNUTAR_rdevmajor_offset, GNUTAR_rdevmajor_size)) { archive_set_error(&a->archive, ERANGE, "Major device number too large"); ret = ARCHIVE_FAILED; } if (format_octal(archive_entry_rdevminor(entry), h + GNUTAR_rdevminor_offset, GNUTAR_rdevminor_size)) { archive_set_error(&a->archive, ERANGE, "Minor device number too large"); ret = ARCHIVE_FAILED; } } h[GNUTAR_typeflag_offset] = tartype; checksum = 0; for (i = 0; i < 512; i++) checksum += 255 & (unsigned int)h[i]; h[GNUTAR_checksum_offset + 6] = '\0'; /* Can't be pre-set in the template. */ /* h[GNUTAR_checksum_offset + 7] = ' '; */ /* This is pre-set in the template. */ format_octal(checksum, h + GNUTAR_checksum_offset, 6); return (ret); } /* * Format a number into a field, falling back to base-256 if necessary. */ static int format_number(int64_t v, char *p, int s, int maxsize) { int64_t limit = ((int64_t)1 << (s*3)); if (v < limit) return (format_octal(v, p, s)); return (format_256(v, p, maxsize)); } /* * Format a number into the specified field using base-256. */ static int format_256(int64_t v, char *p, int s) { p += s; while (s-- > 0) { *--p = (char)(v & 0xff); v >>= 8; } *p |= 0x80; /* Set the base-256 marker bit. */ return (0); } /* * Format a number into the specified field using octal. */ static int format_octal(int64_t v, char *p, int s) { int len = s; /* Octal values can't be negative, so use 0. */ if (v < 0) v = 0; p += s; /* Start at the end and work backwards. */ while (s-- > 0) { *--p = (char)('0' + (v & 7)); v >>= 3; } if (v == 0) return (0); /* If it overflowed, fill field with max value. */ while (len-- > 0) *p++ = '7'; return (-1); } Index: head/contrib/libarchive/libarchive/archive_write_set_format_mtree.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_mtree.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_write_set_format_mtree.c (revision 309590) @@ -1,2228 +1,2228 @@ /*- * Copyright (c) 2008 Joerg Sonnenberger * Copyright (c) 2009-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_SYS_TYPES_H #include #endif #include #include #include #include "archive.h" #include "archive_digest_private.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_rb.h" #include "archive_string.h" #include "archive_write_private.h" #define INDENTNAMELEN 15 #define MAXLINELEN 80 #define SET_KEYS \ (F_FLAGS | F_GID | F_GNAME | F_MODE | F_TYPE | F_UID | F_UNAME) struct attr_counter { struct attr_counter *prev; struct attr_counter *next; struct mtree_entry *m_entry; int count; }; struct att_counter_set { struct attr_counter *uid_list; struct attr_counter *gid_list; struct attr_counter *mode_list; struct attr_counter *flags_list; }; struct mtree_chain { struct mtree_entry *first; struct mtree_entry **last; }; /* * The Data only for a directory file. */ struct dir_info { struct archive_rb_tree rbtree; struct mtree_chain children; struct mtree_entry *chnext; int virtual; }; /* * The Data only for a regular file. */ struct reg_info { int compute_sum; uint32_t crc; #ifdef ARCHIVE_HAS_MD5 unsigned char buf_md5[16]; #endif #ifdef ARCHIVE_HAS_RMD160 unsigned char buf_rmd160[20]; #endif #ifdef ARCHIVE_HAS_SHA1 unsigned char buf_sha1[20]; #endif #ifdef ARCHIVE_HAS_SHA256 unsigned char buf_sha256[32]; #endif #ifdef ARCHIVE_HAS_SHA384 unsigned char buf_sha384[48]; #endif #ifdef ARCHIVE_HAS_SHA512 unsigned char buf_sha512[64]; #endif }; struct mtree_entry { struct archive_rb_node rbnode; struct mtree_entry *next; struct mtree_entry *parent; struct dir_info *dir_info; struct reg_info *reg_info; struct archive_string parentdir; struct archive_string basename; struct archive_string pathname; struct archive_string symlink; struct archive_string uname; struct archive_string gname; struct archive_string fflags_text; unsigned int nlink; mode_t filetype; mode_t mode; int64_t size; int64_t uid; int64_t gid; time_t mtime; long mtime_nsec; unsigned long fflags_set; unsigned long fflags_clear; dev_t rdevmajor; dev_t rdevminor; dev_t devmajor; dev_t devminor; int64_t ino; }; struct mtree_writer { struct mtree_entry *mtree_entry; struct mtree_entry *root; struct mtree_entry *cur_dirent; struct archive_string cur_dirstr; struct mtree_chain file_list; struct archive_string ebuf; struct archive_string buf; int first; uint64_t entry_bytes_remaining; /* * Set global value. */ struct { int processing; mode_t type; int keys; int64_t uid; int64_t gid; mode_t mode; unsigned long fflags_set; unsigned long fflags_clear; } set; struct att_counter_set acs; int classic; int depth; /* check sum */ int compute_sum; uint32_t crc; uint64_t crc_len; #ifdef ARCHIVE_HAS_MD5 archive_md5_ctx md5ctx; #endif #ifdef ARCHIVE_HAS_RMD160 archive_rmd160_ctx rmd160ctx; #endif #ifdef ARCHIVE_HAS_SHA1 archive_sha1_ctx sha1ctx; #endif #ifdef ARCHIVE_HAS_SHA256 archive_sha256_ctx sha256ctx; #endif #ifdef ARCHIVE_HAS_SHA384 archive_sha384_ctx sha384ctx; #endif #ifdef ARCHIVE_HAS_SHA512 archive_sha512_ctx sha512ctx; #endif /* Keyword options */ int keys; #define F_CKSUM 0x00000001 /* check sum */ #define F_DEV 0x00000002 /* device type */ #define F_DONE 0x00000004 /* directory done */ #define F_FLAGS 0x00000008 /* file flags */ #define F_GID 0x00000010 /* gid */ #define F_GNAME 0x00000020 /* group name */ #define F_IGN 0x00000040 /* ignore */ #define F_MAGIC 0x00000080 /* name has magic chars */ #define F_MD5 0x00000100 /* MD5 digest */ #define F_MODE 0x00000200 /* mode */ #define F_NLINK 0x00000400 /* number of links */ #define F_NOCHANGE 0x00000800 /* If owner/mode "wrong", do * not change */ #define F_OPT 0x00001000 /* existence optional */ #define F_RMD160 0x00002000 /* RIPEMD160 digest */ #define F_SHA1 0x00004000 /* SHA-1 digest */ #define F_SIZE 0x00008000 /* size */ #define F_SLINK 0x00010000 /* symbolic link */ #define F_TAGS 0x00020000 /* tags */ #define F_TIME 0x00040000 /* modification time */ #define F_TYPE 0x00080000 /* file type */ #define F_UID 0x00100000 /* uid */ #define F_UNAME 0x00200000 /* user name */ #define F_VISIT 0x00400000 /* file visited */ #define F_SHA256 0x00800000 /* SHA-256 digest */ #define F_SHA384 0x01000000 /* SHA-384 digest */ #define F_SHA512 0x02000000 /* SHA-512 digest */ #define F_INO 0x04000000 /* inode number */ #define F_RESDEV 0x08000000 /* device ID on which the * entry resides */ /* Options */ int dironly; /* If it is set, ignore all files except * directory files, like mtree(8) -d option. */ int indent; /* If it is set, indent output data. */ int output_global_set; /* If it is set, use /set keyword to set * global values. When generating mtree * classic format, it is set by default. */ }; #define DEFAULT_KEYS (F_DEV | F_FLAGS | F_GID | F_GNAME | F_SLINK | F_MODE\ | F_NLINK | F_SIZE | F_TIME | F_TYPE | F_UID\ | F_UNAME) #define attr_counter_set_reset attr_counter_set_free static void attr_counter_free(struct attr_counter **); static int attr_counter_inc(struct attr_counter **, struct attr_counter *, struct attr_counter *, struct mtree_entry *); static struct attr_counter * attr_counter_new(struct mtree_entry *, struct attr_counter *); static int attr_counter_set_collect(struct mtree_writer *, struct mtree_entry *); static void attr_counter_set_free(struct mtree_writer *); static int get_global_set_keys(struct mtree_writer *, struct mtree_entry *); static int mtree_entry_add_child_tail(struct mtree_entry *, struct mtree_entry *); static int mtree_entry_create_virtual_dir(struct archive_write *, const char *, struct mtree_entry **); static int mtree_entry_cmp_node(const struct archive_rb_node *, const struct archive_rb_node *); static int mtree_entry_cmp_key(const struct archive_rb_node *, const void *); static int mtree_entry_exchange_same_entry(struct archive_write *, struct mtree_entry *, struct mtree_entry *); static void mtree_entry_free(struct mtree_entry *); static int mtree_entry_new(struct archive_write *, struct archive_entry *, struct mtree_entry **); static void mtree_entry_register_free(struct mtree_writer *); static void mtree_entry_register_init(struct mtree_writer *); static int mtree_entry_setup_filenames(struct archive_write *, struct mtree_entry *, struct archive_entry *); static int mtree_entry_tree_add(struct archive_write *, struct mtree_entry **); static void sum_init(struct mtree_writer *); static void sum_update(struct mtree_writer *, const void *, size_t); static void sum_final(struct mtree_writer *, struct reg_info *); static void sum_write(struct archive_string *, struct reg_info *); static int write_mtree_entry(struct archive_write *, struct mtree_entry *); static int write_dot_dot_entry(struct archive_write *, struct mtree_entry *); #define COMPUTE_CRC(var, ch) (var) = (var) << 8 ^ crctab[(var) >> 24 ^ (ch)] static const uint32_t crctab[] = { 0x0, 0x04c11db7, 0x09823b6e, 0x0d4326d9, 0x130476dc, 0x17c56b6b, 0x1a864db2, 0x1e475005, 0x2608edb8, 0x22c9f00f, 0x2f8ad6d6, 0x2b4bcb61, 0x350c9b64, 0x31cd86d3, 0x3c8ea00a, 0x384fbdbd, 0x4c11db70, 0x48d0c6c7, 0x4593e01e, 0x4152fda9, 0x5f15adac, 0x5bd4b01b, 0x569796c2, 0x52568b75, 0x6a1936c8, 0x6ed82b7f, 0x639b0da6, 0x675a1011, 0x791d4014, 0x7ddc5da3, 0x709f7b7a, 0x745e66cd, 0x9823b6e0, 0x9ce2ab57, 0x91a18d8e, 0x95609039, 0x8b27c03c, 0x8fe6dd8b, 0x82a5fb52, 0x8664e6e5, 0xbe2b5b58, 0xbaea46ef, 0xb7a96036, 0xb3687d81, 0xad2f2d84, 0xa9ee3033, 0xa4ad16ea, 0xa06c0b5d, 0xd4326d90, 0xd0f37027, 0xddb056fe, 0xd9714b49, 0xc7361b4c, 0xc3f706fb, 0xceb42022, 0xca753d95, 0xf23a8028, 0xf6fb9d9f, 0xfbb8bb46, 0xff79a6f1, 0xe13ef6f4, 0xe5ffeb43, 0xe8bccd9a, 0xec7dd02d, 0x34867077, 0x30476dc0, 0x3d044b19, 0x39c556ae, 0x278206ab, 0x23431b1c, 0x2e003dc5, 0x2ac12072, 0x128e9dcf, 0x164f8078, 0x1b0ca6a1, 0x1fcdbb16, 0x018aeb13, 0x054bf6a4, 0x0808d07d, 0x0cc9cdca, 0x7897ab07, 0x7c56b6b0, 0x71159069, 0x75d48dde, 0x6b93dddb, 0x6f52c06c, 0x6211e6b5, 0x66d0fb02, 0x5e9f46bf, 0x5a5e5b08, 0x571d7dd1, 0x53dc6066, 0x4d9b3063, 0x495a2dd4, 0x44190b0d, 0x40d816ba, 0xaca5c697, 0xa864db20, 0xa527fdf9, 0xa1e6e04e, 0xbfa1b04b, 0xbb60adfc, 0xb6238b25, 0xb2e29692, 0x8aad2b2f, 0x8e6c3698, 0x832f1041, 0x87ee0df6, 0x99a95df3, 0x9d684044, 0x902b669d, 0x94ea7b2a, 0xe0b41de7, 0xe4750050, 0xe9362689, 0xedf73b3e, 0xf3b06b3b, 0xf771768c, 0xfa325055, 0xfef34de2, 0xc6bcf05f, 0xc27dede8, 0xcf3ecb31, 0xcbffd686, 0xd5b88683, 0xd1799b34, 0xdc3abded, 0xd8fba05a, 0x690ce0ee, 0x6dcdfd59, 0x608edb80, 0x644fc637, 0x7a089632, 0x7ec98b85, 0x738aad5c, 0x774bb0eb, 0x4f040d56, 0x4bc510e1, 0x46863638, 0x42472b8f, 0x5c007b8a, 0x58c1663d, 0x558240e4, 0x51435d53, 0x251d3b9e, 0x21dc2629, 0x2c9f00f0, 0x285e1d47, 0x36194d42, 0x32d850f5, 0x3f9b762c, 0x3b5a6b9b, 0x0315d626, 0x07d4cb91, 0x0a97ed48, 0x0e56f0ff, 0x1011a0fa, 0x14d0bd4d, 0x19939b94, 0x1d528623, 0xf12f560e, 0xf5ee4bb9, 0xf8ad6d60, 0xfc6c70d7, 0xe22b20d2, 0xe6ea3d65, 0xeba91bbc, 0xef68060b, 0xd727bbb6, 0xd3e6a601, 0xdea580d8, 0xda649d6f, 0xc423cd6a, 0xc0e2d0dd, 0xcda1f604, 0xc960ebb3, 0xbd3e8d7e, 0xb9ff90c9, 0xb4bcb610, 0xb07daba7, 0xae3afba2, 0xaafbe615, 0xa7b8c0cc, 0xa379dd7b, 0x9b3660c6, 0x9ff77d71, 0x92b45ba8, 0x9675461f, 0x8832161a, 0x8cf30bad, 0x81b02d74, 0x857130c3, 0x5d8a9099, 0x594b8d2e, 0x5408abf7, 0x50c9b640, 0x4e8ee645, 0x4a4ffbf2, 0x470cdd2b, 0x43cdc09c, 0x7b827d21, 0x7f436096, 0x7200464f, 0x76c15bf8, 0x68860bfd, 0x6c47164a, 0x61043093, 0x65c52d24, 0x119b4be9, 0x155a565e, 0x18197087, 0x1cd86d30, 0x029f3d35, 0x065e2082, 0x0b1d065b, 0x0fdc1bec, 0x3793a651, 0x3352bbe6, 0x3e119d3f, 0x3ad08088, 0x2497d08d, 0x2056cd3a, 0x2d15ebe3, 0x29d4f654, 0xc5a92679, 0xc1683bce, 0xcc2b1d17, 0xc8ea00a0, 0xd6ad50a5, 0xd26c4d12, 0xdf2f6bcb, 0xdbee767c, 0xe3a1cbc1, 0xe760d676, 0xea23f0af, 0xeee2ed18, 0xf0a5bd1d, 0xf464a0aa, 0xf9278673, 0xfde69bc4, 0x89b8fd09, 0x8d79e0be, 0x803ac667, 0x84fbdbd0, 0x9abc8bd5, 0x9e7d9662, 0x933eb0bb, 0x97ffad0c, 0xafb010b1, 0xab710d06, 0xa6322bdf, 0xa2f33668, 0xbcb4666d, 0xb8757bda, 0xb5365d03, 0xb1f740b4 }; static const unsigned char safe_char[256] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ /* !"$%&'()*+,-./ EXCLUSION:0x20( ) 0x23(#) */ 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 20 - 2F */ /* 0123456789:;<>? EXCLUSION:0x3d(=) */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, /* 30 - 3F */ /* @ABCDEFGHIJKLMNO */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 40 - 4F */ /* PQRSTUVWXYZ[]^_ EXCLUSION:0x5c(\) */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, /* 50 - 5F */ /* `abcdefghijklmno */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 60 - 6F */ /* pqrstuvwxyz{|}~ */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, /* 70 - 7F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ }; static void mtree_quote(struct archive_string *s, const char *str) { const char *start; char buf[4]; unsigned char c; for (start = str; *str != '\0'; ++str) { if (safe_char[*(const unsigned char *)str]) continue; if (start != str) archive_strncat(s, start, str - start); c = (unsigned char)*str; buf[0] = '\\'; buf[1] = (c / 64) + '0'; buf[2] = (c / 8 % 8) + '0'; buf[3] = (c % 8) + '0'; archive_strncat(s, buf, 4); start = str + 1; } if (start != str) archive_strncat(s, start, str - start); } /* * Indent a line as mtree utility to be readable for people. */ static void mtree_indent(struct mtree_writer *mtree) { int i, fn, nd, pd; const char *r, *s, *x; if (mtree->classic) { if (mtree->indent) { nd = 0; pd = mtree->depth * 4; } else { nd = mtree->depth?4:0; pd = 0; } } else nd = pd = 0; fn = 1; s = r = mtree->ebuf.s; x = NULL; while (*r == ' ') r++; while ((r = strchr(r, ' ')) != NULL) { if (fn) { fn = 0; for (i = 0; i < nd + pd; i++) archive_strappend_char(&mtree->buf, ' '); archive_strncat(&mtree->buf, s, r - s); if (nd + (r -s) > INDENTNAMELEN) { archive_strncat(&mtree->buf, " \\\n", 3); for (i = 0; i < (INDENTNAMELEN + 1 + pd); i++) archive_strappend_char(&mtree->buf, ' '); } else { for (i = (int)(r -s + nd); i < (INDENTNAMELEN + 1); i++) archive_strappend_char(&mtree->buf, ' '); } s = ++r; x = NULL; continue; } if (pd + (r - s) <= MAXLINELEN - 3 - INDENTNAMELEN) x = r++; else { if (x == NULL) x = r; archive_strncat(&mtree->buf, s, x - s); archive_strncat(&mtree->buf, " \\\n", 3); for (i = 0; i < (INDENTNAMELEN + 1 + pd); i++) archive_strappend_char(&mtree->buf, ' '); s = r = ++x; x = NULL; } } if (fn) { for (i = 0; i < nd + pd; i++) archive_strappend_char(&mtree->buf, ' '); archive_strcat(&mtree->buf, s); s += strlen(s); } if (x != NULL && pd + strlen(s) > MAXLINELEN - 3 - INDENTNAMELEN) { /* Last keyword is longer. */ archive_strncat(&mtree->buf, s, x - s); archive_strncat(&mtree->buf, " \\\n", 3); for (i = 0; i < (INDENTNAMELEN + 1 + pd); i++) archive_strappend_char(&mtree->buf, ' '); s = ++x; } archive_strcat(&mtree->buf, s); archive_string_empty(&mtree->ebuf); } /* * Write /set keyword. * Set most used value of uid,gid,mode and fflags, which are * collected by attr_counter_set_collect() function. */ static void write_global(struct mtree_writer *mtree) { struct archive_string setstr; struct archive_string unsetstr; struct att_counter_set *acs; int keys, oldkeys, effkeys; archive_string_init(&setstr); archive_string_init(&unsetstr); keys = mtree->keys & SET_KEYS; oldkeys = mtree->set.keys; effkeys = keys; acs = &mtree->acs; if (mtree->set.processing) { /* * Check if the global data needs updating. */ effkeys &= ~F_TYPE; if (acs->uid_list == NULL) effkeys &= ~(F_UNAME | F_UID); else if (oldkeys & (F_UNAME | F_UID)) { if (acs->uid_list->count < 2 || mtree->set.uid == acs->uid_list->m_entry->uid) effkeys &= ~(F_UNAME | F_UID); } if (acs->gid_list == NULL) effkeys &= ~(F_GNAME | F_GID); else if (oldkeys & (F_GNAME | F_GID)) { if (acs->gid_list->count < 2 || mtree->set.gid == acs->gid_list->m_entry->gid) effkeys &= ~(F_GNAME | F_GID); } if (acs->mode_list == NULL) effkeys &= ~F_MODE; else if (oldkeys & F_MODE) { if (acs->mode_list->count < 2 || mtree->set.mode == acs->mode_list->m_entry->mode) effkeys &= ~F_MODE; } if (acs->flags_list == NULL) effkeys &= ~F_FLAGS; else if ((oldkeys & F_FLAGS) != 0) { if (acs->flags_list->count < 2 || (acs->flags_list->m_entry->fflags_set == mtree->set.fflags_set && acs->flags_list->m_entry->fflags_clear == mtree->set.fflags_clear)) effkeys &= ~F_FLAGS; } } else { if (acs->uid_list == NULL) keys &= ~(F_UNAME | F_UID); if (acs->gid_list == NULL) keys &= ~(F_GNAME | F_GID); if (acs->mode_list == NULL) keys &= ~F_MODE; if (acs->flags_list == NULL) keys &= ~F_FLAGS; } if ((keys & effkeys & F_TYPE) != 0) { if (mtree->dironly) { archive_strcat(&setstr, " type=dir"); mtree->set.type = AE_IFDIR; } else { archive_strcat(&setstr, " type=file"); mtree->set.type = AE_IFREG; } } if ((keys & effkeys & F_UNAME) != 0) { if (archive_strlen(&(acs->uid_list->m_entry->uname)) > 0) { archive_strcat(&setstr, " uname="); mtree_quote(&setstr, acs->uid_list->m_entry->uname.s); } else { keys &= ~F_UNAME; if ((oldkeys & F_UNAME) != 0) archive_strcat(&unsetstr, " uname"); } } if ((keys & effkeys & F_UID) != 0) { mtree->set.uid = acs->uid_list->m_entry->uid; archive_string_sprintf(&setstr, " uid=%jd", (intmax_t)mtree->set.uid); } if ((keys & effkeys & F_GNAME) != 0) { if (archive_strlen(&(acs->gid_list->m_entry->gname)) > 0) { archive_strcat(&setstr, " gname="); mtree_quote(&setstr, acs->gid_list->m_entry->gname.s); } else { keys &= ~F_GNAME; if ((oldkeys & F_GNAME) != 0) archive_strcat(&unsetstr, " gname"); } } if ((keys & effkeys & F_GID) != 0) { mtree->set.gid = acs->gid_list->m_entry->gid; archive_string_sprintf(&setstr, " gid=%jd", (intmax_t)mtree->set.gid); } if ((keys & effkeys & F_MODE) != 0) { mtree->set.mode = acs->mode_list->m_entry->mode; archive_string_sprintf(&setstr, " mode=%o", (unsigned int)mtree->set.mode); } if ((keys & effkeys & F_FLAGS) != 0) { if (archive_strlen( &(acs->flags_list->m_entry->fflags_text)) > 0) { archive_strcat(&setstr, " flags="); mtree_quote(&setstr, acs->flags_list->m_entry->fflags_text.s); mtree->set.fflags_set = acs->flags_list->m_entry->fflags_set; mtree->set.fflags_clear = acs->flags_list->m_entry->fflags_clear; } else { keys &= ~F_FLAGS; if ((oldkeys & F_FLAGS) != 0) archive_strcat(&unsetstr, " flags"); } } if (unsetstr.length > 0) archive_string_sprintf(&mtree->buf, "/unset%s\n", unsetstr.s); archive_string_free(&unsetstr); if (setstr.length > 0) archive_string_sprintf(&mtree->buf, "/set%s\n", setstr.s); archive_string_free(&setstr); mtree->set.keys = keys; mtree->set.processing = 1; } static struct attr_counter * attr_counter_new(struct mtree_entry *me, struct attr_counter *prev) { struct attr_counter *ac; ac = malloc(sizeof(*ac)); if (ac != NULL) { ac->prev = prev; ac->next = NULL; ac->count = 1; ac->m_entry = me; } return (ac); } static void attr_counter_free(struct attr_counter **top) { struct attr_counter *ac, *tac; if (*top == NULL) return; ac = *top; while (ac != NULL) { tac = ac->next; free(ac); ac = tac; } *top = NULL; } static int attr_counter_inc(struct attr_counter **top, struct attr_counter *ac, struct attr_counter *last, struct mtree_entry *me) { struct attr_counter *pac; if (ac != NULL) { ac->count++; if (*top == ac || ac->prev->count >= ac->count) return (0); for (pac = ac->prev; pac; pac = pac->prev) { if (pac->count >= ac->count) break; } ac->prev->next = ac->next; if (ac->next != NULL) ac->next->prev = ac->prev; if (pac != NULL) { ac->prev = pac; ac->next = pac->next; pac->next = ac; if (ac->next != NULL) ac->next->prev = ac; } else { ac->prev = NULL; ac->next = *top; *top = ac; ac->next->prev = ac; } } else if (last != NULL) { ac = attr_counter_new(me, last); if (ac == NULL) return (-1); last->next = ac; } return (0); } /* * Tabulate uid,gid,mode and fflags of a entry in order to be used for /set. */ static int attr_counter_set_collect(struct mtree_writer *mtree, struct mtree_entry *me) { struct attr_counter *ac, *last; struct att_counter_set *acs = &mtree->acs; int keys = mtree->keys; if (keys & (F_UNAME | F_UID)) { if (acs->uid_list == NULL) { acs->uid_list = attr_counter_new(me, NULL); if (acs->uid_list == NULL) return (-1); } else { last = NULL; for (ac = acs->uid_list; ac; ac = ac->next) { if (ac->m_entry->uid == me->uid) break; last = ac; } if (attr_counter_inc(&acs->uid_list, ac, last, me) < 0) return (-1); } } if (keys & (F_GNAME | F_GID)) { if (acs->gid_list == NULL) { acs->gid_list = attr_counter_new(me, NULL); if (acs->gid_list == NULL) return (-1); } else { last = NULL; for (ac = acs->gid_list; ac; ac = ac->next) { if (ac->m_entry->gid == me->gid) break; last = ac; } if (attr_counter_inc(&acs->gid_list, ac, last, me) < 0) return (-1); } } if (keys & F_MODE) { if (acs->mode_list == NULL) { acs->mode_list = attr_counter_new(me, NULL); if (acs->mode_list == NULL) return (-1); } else { last = NULL; for (ac = acs->mode_list; ac; ac = ac->next) { if (ac->m_entry->mode == me->mode) break; last = ac; } if (attr_counter_inc(&acs->mode_list, ac, last, me) < 0) return (-1); } } if (keys & F_FLAGS) { if (acs->flags_list == NULL) { acs->flags_list = attr_counter_new(me, NULL); if (acs->flags_list == NULL) return (-1); } else { last = NULL; for (ac = acs->flags_list; ac; ac = ac->next) { if (ac->m_entry->fflags_set == me->fflags_set && ac->m_entry->fflags_clear == me->fflags_clear) break; last = ac; } if (attr_counter_inc(&acs->flags_list, ac, last, me) < 0) return (-1); } } return (0); } static void attr_counter_set_free(struct mtree_writer *mtree) { struct att_counter_set *acs = &mtree->acs; attr_counter_free(&acs->uid_list); attr_counter_free(&acs->gid_list); attr_counter_free(&acs->mode_list); attr_counter_free(&acs->flags_list); } static int get_global_set_keys(struct mtree_writer *mtree, struct mtree_entry *me) { int keys; keys = mtree->keys; /* * If a keyword has been set by /set, we do not need to * output it. */ if (mtree->set.keys == 0) return (keys);/* /set is not used. */ if ((mtree->set.keys & (F_GNAME | F_GID)) != 0 && mtree->set.gid == me->gid) keys &= ~(F_GNAME | F_GID); if ((mtree->set.keys & (F_UNAME | F_UID)) != 0 && mtree->set.uid == me->uid) keys &= ~(F_UNAME | F_UID); if (mtree->set.keys & F_FLAGS) { if (mtree->set.fflags_set == me->fflags_set && mtree->set.fflags_clear == me->fflags_clear) keys &= ~F_FLAGS; } if ((mtree->set.keys & F_MODE) != 0 && mtree->set.mode == me->mode) keys &= ~F_MODE; switch (me->filetype) { case AE_IFLNK: case AE_IFSOCK: case AE_IFCHR: case AE_IFBLK: case AE_IFIFO: break; case AE_IFDIR: if ((mtree->set.keys & F_TYPE) != 0 && mtree->set.type == AE_IFDIR) keys &= ~F_TYPE; break; case AE_IFREG: default: /* Handle unknown file types as regular files. */ if ((mtree->set.keys & F_TYPE) != 0 && mtree->set.type == AE_IFREG) keys &= ~F_TYPE; break; } return (keys); } static int mtree_entry_new(struct archive_write *a, struct archive_entry *entry, struct mtree_entry **m_entry) { struct mtree_entry *me; const char *s; int r; static const struct archive_rb_tree_ops rb_ops = { mtree_entry_cmp_node, mtree_entry_cmp_key }; me = calloc(1, sizeof(*me)); if (me == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for a mtree entry"); *m_entry = NULL; return (ARCHIVE_FATAL); } r = mtree_entry_setup_filenames(a, me, entry); if (r < ARCHIVE_WARN) { mtree_entry_free(me); *m_entry = NULL; return (r); } if ((s = archive_entry_symlink(entry)) != NULL) archive_strcpy(&me->symlink, s); me->nlink = archive_entry_nlink(entry); me->filetype = archive_entry_filetype(entry); me->mode = archive_entry_mode(entry) & 07777; me->uid = archive_entry_uid(entry); me->gid = archive_entry_gid(entry); if ((s = archive_entry_uname(entry)) != NULL) archive_strcpy(&me->uname, s); if ((s = archive_entry_gname(entry)) != NULL) archive_strcpy(&me->gname, s); if ((s = archive_entry_fflags_text(entry)) != NULL) archive_strcpy(&me->fflags_text, s); archive_entry_fflags(entry, &me->fflags_set, &me->fflags_clear); me->mtime = archive_entry_mtime(entry); me->mtime_nsec = archive_entry_mtime_nsec(entry); me->rdevmajor = archive_entry_rdevmajor(entry); me->rdevminor = archive_entry_rdevminor(entry); me->devmajor = archive_entry_devmajor(entry); me->devminor = archive_entry_devminor(entry); me->ino = archive_entry_ino(entry); me->size = archive_entry_size(entry); if (me->filetype == AE_IFDIR) { me->dir_info = calloc(1, sizeof(*me->dir_info)); if (me->dir_info == NULL) { mtree_entry_free(me); archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for a mtree entry"); *m_entry = NULL; return (ARCHIVE_FATAL); } __archive_rb_tree_init(&me->dir_info->rbtree, &rb_ops); me->dir_info->children.first = NULL; me->dir_info->children.last = &(me->dir_info->children.first); me->dir_info->chnext = NULL; } else if (me->filetype == AE_IFREG) { me->reg_info = calloc(1, sizeof(*me->reg_info)); if (me->reg_info == NULL) { mtree_entry_free(me); archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for a mtree entry"); *m_entry = NULL; return (ARCHIVE_FATAL); } me->reg_info->compute_sum = 0; } *m_entry = me; return (ARCHIVE_OK); } static void mtree_entry_free(struct mtree_entry *me) { archive_string_free(&me->parentdir); archive_string_free(&me->basename); archive_string_free(&me->pathname); archive_string_free(&me->symlink); archive_string_free(&me->uname); archive_string_free(&me->gname); archive_string_free(&me->fflags_text); free(me->dir_info); free(me->reg_info); free(me); } static int archive_write_mtree_header(struct archive_write *a, struct archive_entry *entry) { struct mtree_writer *mtree= a->format_data; struct mtree_entry *mtree_entry; int r, r2; if (mtree->first) { mtree->first = 0; archive_strcat(&mtree->buf, "#mtree\n"); if ((mtree->keys & SET_KEYS) == 0) mtree->output_global_set = 0;/* Disabled. */ } mtree->entry_bytes_remaining = archive_entry_size(entry); /* While directory only mode, we do not handle non directory files. */ if (mtree->dironly && archive_entry_filetype(entry) != AE_IFDIR) return (ARCHIVE_OK); r2 = mtree_entry_new(a, entry, &mtree_entry); if (r2 < ARCHIVE_WARN) return (r2); r = mtree_entry_tree_add(a, &mtree_entry); if (r < ARCHIVE_WARN) { mtree_entry_free(mtree_entry); return (r); } mtree->mtree_entry = mtree_entry; /* If the current file is a regular file, we have to * compute the sum of its content. * Initialize a bunch of sum check context. */ if (mtree_entry->reg_info) sum_init(mtree); return (r2); } static int write_mtree_entry(struct archive_write *a, struct mtree_entry *me) { struct mtree_writer *mtree = a->format_data; struct archive_string *str; int keys, ret; if (me->dir_info) { if (mtree->classic) { /* * Output a comment line to describe the full * pathname of the entry as mtree utility does * while generating classic format. */ if (!mtree->dironly) archive_strappend_char(&mtree->buf, '\n'); if (me->parentdir.s) archive_string_sprintf(&mtree->buf, "# %s/%s\n", me->parentdir.s, me->basename.s); else archive_string_sprintf(&mtree->buf, "# %s\n", me->basename.s); } if (mtree->output_global_set) write_global(mtree); } archive_string_empty(&mtree->ebuf); str = (mtree->indent || mtree->classic)? &mtree->ebuf : &mtree->buf; if (!mtree->classic && me->parentdir.s) { /* * If generating format is not classic one(v1), output * a full pathname. */ mtree_quote(str, me->parentdir.s); archive_strappend_char(str, '/'); } mtree_quote(str, me->basename.s); keys = get_global_set_keys(mtree, me); if ((keys & F_NLINK) != 0 && me->nlink != 1 && me->filetype != AE_IFDIR) archive_string_sprintf(str, " nlink=%u", me->nlink); if ((keys & F_GNAME) != 0 && archive_strlen(&me->gname) > 0) { archive_strcat(str, " gname="); mtree_quote(str, me->gname.s); } if ((keys & F_UNAME) != 0 && archive_strlen(&me->uname) > 0) { archive_strcat(str, " uname="); mtree_quote(str, me->uname.s); } if ((keys & F_FLAGS) != 0) { if (archive_strlen(&me->fflags_text) > 0) { archive_strcat(str, " flags="); mtree_quote(str, me->fflags_text.s); } else if (mtree->set.processing && (mtree->set.keys & F_FLAGS) != 0) /* Overwrite the global parameter. */ archive_strcat(str, " flags=none"); } if ((keys & F_TIME) != 0) archive_string_sprintf(str, " time=%jd.%jd", (intmax_t)me->mtime, (intmax_t)me->mtime_nsec); if ((keys & F_MODE) != 0) archive_string_sprintf(str, " mode=%o", (unsigned int)me->mode); if ((keys & F_GID) != 0) archive_string_sprintf(str, " gid=%jd", (intmax_t)me->gid); if ((keys & F_UID) != 0) archive_string_sprintf(str, " uid=%jd", (intmax_t)me->uid); if ((keys & F_INO) != 0) archive_string_sprintf(str, " inode=%jd", (intmax_t)me->ino); if ((keys & F_RESDEV) != 0) { archive_string_sprintf(str, " resdevice=native,%ju,%ju", (uintmax_t)me->devmajor, (uintmax_t)me->devminor); } switch (me->filetype) { case AE_IFLNK: if ((keys & F_TYPE) != 0) archive_strcat(str, " type=link"); if ((keys & F_SLINK) != 0) { archive_strcat(str, " link="); mtree_quote(str, me->symlink.s); } break; case AE_IFSOCK: if ((keys & F_TYPE) != 0) archive_strcat(str, " type=socket"); break; case AE_IFCHR: if ((keys & F_TYPE) != 0) archive_strcat(str, " type=char"); if ((keys & F_DEV) != 0) { archive_string_sprintf(str, " device=native,%ju,%ju", (uintmax_t)me->rdevmajor, (uintmax_t)me->rdevminor); } break; case AE_IFBLK: if ((keys & F_TYPE) != 0) archive_strcat(str, " type=block"); if ((keys & F_DEV) != 0) { archive_string_sprintf(str, " device=native,%ju,%ju", (uintmax_t)me->rdevmajor, (uintmax_t)me->rdevminor); } break; case AE_IFDIR: if ((keys & F_TYPE) != 0) archive_strcat(str, " type=dir"); break; case AE_IFIFO: if ((keys & F_TYPE) != 0) archive_strcat(str, " type=fifo"); break; case AE_IFREG: default: /* Handle unknown file types as regular files. */ if ((keys & F_TYPE) != 0) archive_strcat(str, " type=file"); if ((keys & F_SIZE) != 0) archive_string_sprintf(str, " size=%jd", (intmax_t)me->size); break; } /* Write a bunch of sum. */ if (me->reg_info) sum_write(str, me->reg_info); archive_strappend_char(str, '\n'); if (mtree->indent || mtree->classic) mtree_indent(mtree); if (mtree->buf.length > 32768) { ret = __archive_write_output( a, mtree->buf.s, mtree->buf.length); archive_string_empty(&mtree->buf); } else ret = ARCHIVE_OK; return (ret); } static int write_dot_dot_entry(struct archive_write *a, struct mtree_entry *n) { struct mtree_writer *mtree = a->format_data; int ret; if (n->parentdir.s) { if (mtree->indent) { int i, pd = mtree->depth * 4; for (i = 0; i < pd; i++) archive_strappend_char(&mtree->buf, ' '); } archive_string_sprintf(&mtree->buf, "# %s/%s\n", n->parentdir.s, n->basename.s); } if (mtree->indent) { archive_string_empty(&mtree->ebuf); archive_strncat(&mtree->ebuf, "..\n\n", (mtree->dironly)?3:4); mtree_indent(mtree); } else archive_strncat(&mtree->buf, "..\n\n", (mtree->dironly)?3:4); if (mtree->buf.length > 32768) { ret = __archive_write_output( a, mtree->buf.s, mtree->buf.length); archive_string_empty(&mtree->buf); } else ret = ARCHIVE_OK; return (ret); } /* * Write mtree entries saved at attr_counter_set_collect() function. */ static int write_mtree_entry_tree(struct archive_write *a) { struct mtree_writer *mtree = a->format_data; struct mtree_entry *np = mtree->root; struct archive_rb_node *n; int ret; do { if (mtree->output_global_set) { /* * Collect attribute information to know which value * is frequently used among the children. */ attr_counter_set_reset(mtree); ARCHIVE_RB_TREE_FOREACH(n, &(np->dir_info->rbtree)) { struct mtree_entry *e = (struct mtree_entry *)n; if (attr_counter_set_collect(mtree, e) < 0) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } } } if (!np->dir_info->virtual || mtree->classic) { ret = write_mtree_entry(a, np); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } else { /* Whenever output_global_set is enabled * output global value(/set keywords) * even if the directory entry is not allowed * to be written because the global values * can be used for the children. */ if (mtree->output_global_set) write_global(mtree); } /* * Output the attribute of all files except directory files. */ mtree->depth++; ARCHIVE_RB_TREE_FOREACH(n, &(np->dir_info->rbtree)) { struct mtree_entry *e = (struct mtree_entry *)n; if (e->dir_info) mtree_entry_add_child_tail(np, e); else { ret = write_mtree_entry(a, e); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } } mtree->depth--; if (np->dir_info->children.first != NULL) { /* * Descend the tree. */ np = np->dir_info->children.first; if (mtree->indent) mtree->depth++; continue; } else if (mtree->classic) { /* * While printing mtree classic, if there are not * any directory files(except "." and "..") in the * directory, output two dots ".." as returning * the parent directory. */ ret = write_dot_dot_entry(a, np); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } while (np != np->parent) { if (np->dir_info->chnext == NULL) { /* * Ascend the tree; go back to the parent. */ if (mtree->indent) mtree->depth--; if (mtree->classic) { ret = write_dot_dot_entry(a, np->parent); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } np = np->parent; } else { /* * Switch to next mtree entry in the directory. */ np = np->dir_info->chnext; break; } } } while (np != np->parent); return (ARCHIVE_OK); } static int archive_write_mtree_finish_entry(struct archive_write *a) { struct mtree_writer *mtree = a->format_data; struct mtree_entry *me; if ((me = mtree->mtree_entry) == NULL) return (ARCHIVE_OK); mtree->mtree_entry = NULL; if (me->reg_info) sum_final(mtree, me->reg_info); return (ARCHIVE_OK); } static int archive_write_mtree_close(struct archive_write *a) { struct mtree_writer *mtree= a->format_data; int ret; if (mtree->root != NULL) { ret = write_mtree_entry_tree(a); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } archive_write_set_bytes_in_last_block(&a->archive, 1); return __archive_write_output(a, mtree->buf.s, mtree->buf.length); } static ssize_t archive_write_mtree_data(struct archive_write *a, const void *buff, size_t n) { struct mtree_writer *mtree= a->format_data; if (n > mtree->entry_bytes_remaining) n = (size_t)mtree->entry_bytes_remaining; mtree->entry_bytes_remaining -= n; /* We don't need to compute a regular file sum */ if (mtree->mtree_entry == NULL) return (n); if (mtree->mtree_entry->filetype == AE_IFREG) sum_update(mtree, buff, n); return (n); } static int archive_write_mtree_free(struct archive_write *a) { struct mtree_writer *mtree= a->format_data; if (mtree == NULL) return (ARCHIVE_OK); /* Make sure we dot not leave any entries. */ mtree_entry_register_free(mtree); archive_string_free(&mtree->cur_dirstr); archive_string_free(&mtree->ebuf); archive_string_free(&mtree->buf); attr_counter_set_free(mtree); free(mtree); a->format_data = NULL; return (ARCHIVE_OK); } static int archive_write_mtree_options(struct archive_write *a, const char *key, const char *value) { struct mtree_writer *mtree= a->format_data; int keybit = 0; switch (key[0]) { case 'a': if (strcmp(key, "all") == 0) keybit = ~0; break; case 'c': if (strcmp(key, "cksum") == 0) keybit = F_CKSUM; break; case 'd': if (strcmp(key, "device") == 0) keybit = F_DEV; else if (strcmp(key, "dironly") == 0) { mtree->dironly = (value != NULL)? 1: 0; return (ARCHIVE_OK); } break; case 'f': if (strcmp(key, "flags") == 0) keybit = F_FLAGS; break; case 'g': if (strcmp(key, "gid") == 0) keybit = F_GID; else if (strcmp(key, "gname") == 0) keybit = F_GNAME; break; case 'i': if (strcmp(key, "indent") == 0) { mtree->indent = (value != NULL)? 1: 0; return (ARCHIVE_OK); } else if (strcmp(key, "inode") == 0) { keybit = F_INO; } break; case 'l': if (strcmp(key, "link") == 0) keybit = F_SLINK; break; case 'm': if (strcmp(key, "md5") == 0 || strcmp(key, "md5digest") == 0) keybit = F_MD5; if (strcmp(key, "mode") == 0) keybit = F_MODE; break; case 'n': if (strcmp(key, "nlink") == 0) keybit = F_NLINK; break; case 'r': if (strcmp(key, "resdevice") == 0) { keybit = F_RESDEV; } else if (strcmp(key, "ripemd160digest") == 0 || strcmp(key, "rmd160") == 0 || strcmp(key, "rmd160digest") == 0) keybit = F_RMD160; break; case 's': if (strcmp(key, "sha1") == 0 || strcmp(key, "sha1digest") == 0) keybit = F_SHA1; if (strcmp(key, "sha256") == 0 || strcmp(key, "sha256digest") == 0) keybit = F_SHA256; if (strcmp(key, "sha384") == 0 || strcmp(key, "sha384digest") == 0) keybit = F_SHA384; if (strcmp(key, "sha512") == 0 || strcmp(key, "sha512digest") == 0) keybit = F_SHA512; if (strcmp(key, "size") == 0) keybit = F_SIZE; break; case 't': if (strcmp(key, "time") == 0) keybit = F_TIME; else if (strcmp(key, "type") == 0) keybit = F_TYPE; break; case 'u': if (strcmp(key, "uid") == 0) keybit = F_UID; else if (strcmp(key, "uname") == 0) keybit = F_UNAME; else if (strcmp(key, "use-set") == 0) { mtree->output_global_set = (value != NULL)? 1: 0; return (ARCHIVE_OK); } break; } if (keybit != 0) { if (value != NULL) mtree->keys |= keybit; else mtree->keys &= ~keybit; return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int archive_write_set_format_mtree_default(struct archive *_a, const char *fn) { struct archive_write *a = (struct archive_write *)_a; struct mtree_writer *mtree; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, fn); if (a->format_free != NULL) (a->format_free)(a); if ((mtree = calloc(1, sizeof(*mtree))) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate mtree data"); return (ARCHIVE_FATAL); } mtree->mtree_entry = NULL; mtree->first = 1; memset(&(mtree->set), 0, sizeof(mtree->set)); mtree->keys = DEFAULT_KEYS; mtree->dironly = 0; mtree->indent = 0; archive_string_init(&mtree->ebuf); archive_string_init(&mtree->buf); mtree_entry_register_init(mtree); a->format_data = mtree; a->format_free = archive_write_mtree_free; a->format_name = "mtree"; a->format_options = archive_write_mtree_options; a->format_write_header = archive_write_mtree_header; a->format_close = archive_write_mtree_close; a->format_write_data = archive_write_mtree_data; a->format_finish_entry = archive_write_mtree_finish_entry; a->archive.archive_format = ARCHIVE_FORMAT_MTREE; a->archive.archive_format_name = "mtree"; return (ARCHIVE_OK); } int archive_write_set_format_mtree(struct archive *_a) { return archive_write_set_format_mtree_default(_a, "archive_write_set_format_mtree"); } int archive_write_set_format_mtree_classic(struct archive *_a) { int r; r = archive_write_set_format_mtree_default(_a, "archive_write_set_format_mtree_classic"); if (r == ARCHIVE_OK) { struct archive_write *a = (struct archive_write *)_a; struct mtree_writer *mtree; mtree = (struct mtree_writer *)a->format_data; /* Set to output a mtree archive in classic format. */ mtree->classic = 1; /* Basically, mtree classic format uses '/set' global * value. */ mtree->output_global_set = 1; } return (r); } static void sum_init(struct mtree_writer *mtree) { mtree->compute_sum = 0; if (mtree->keys & F_CKSUM) { mtree->compute_sum |= F_CKSUM; mtree->crc = 0; mtree->crc_len = 0; } #ifdef ARCHIVE_HAS_MD5 if (mtree->keys & F_MD5) { if (archive_md5_init(&mtree->md5ctx) == ARCHIVE_OK) mtree->compute_sum |= F_MD5; else mtree->keys &= ~F_MD5;/* Not supported. */ } #endif #ifdef ARCHIVE_HAS_RMD160 if (mtree->keys & F_RMD160) { if (archive_rmd160_init(&mtree->rmd160ctx) == ARCHIVE_OK) mtree->compute_sum |= F_RMD160; else mtree->keys &= ~F_RMD160;/* Not supported. */ } #endif #ifdef ARCHIVE_HAS_SHA1 if (mtree->keys & F_SHA1) { if (archive_sha1_init(&mtree->sha1ctx) == ARCHIVE_OK) mtree->compute_sum |= F_SHA1; else mtree->keys &= ~F_SHA1;/* Not supported. */ } #endif #ifdef ARCHIVE_HAS_SHA256 if (mtree->keys & F_SHA256) { if (archive_sha256_init(&mtree->sha256ctx) == ARCHIVE_OK) mtree->compute_sum |= F_SHA256; else mtree->keys &= ~F_SHA256;/* Not supported. */ } #endif #ifdef ARCHIVE_HAS_SHA384 if (mtree->keys & F_SHA384) { if (archive_sha384_init(&mtree->sha384ctx) == ARCHIVE_OK) mtree->compute_sum |= F_SHA384; else mtree->keys &= ~F_SHA384;/* Not supported. */ } #endif #ifdef ARCHIVE_HAS_SHA512 if (mtree->keys & F_SHA512) { if (archive_sha512_init(&mtree->sha512ctx) == ARCHIVE_OK) mtree->compute_sum |= F_SHA512; else mtree->keys &= ~F_SHA512;/* Not supported. */ } #endif } static void sum_update(struct mtree_writer *mtree, const void *buff, size_t n) { if (mtree->compute_sum & F_CKSUM) { /* * Compute a POSIX 1003.2 checksum */ const unsigned char *p; size_t nn; for (nn = n, p = buff; nn--; ++p) COMPUTE_CRC(mtree->crc, *p); mtree->crc_len += n; } #ifdef ARCHIVE_HAS_MD5 if (mtree->compute_sum & F_MD5) archive_md5_update(&mtree->md5ctx, buff, n); #endif #ifdef ARCHIVE_HAS_RMD160 if (mtree->compute_sum & F_RMD160) archive_rmd160_update(&mtree->rmd160ctx, buff, n); #endif #ifdef ARCHIVE_HAS_SHA1 if (mtree->compute_sum & F_SHA1) archive_sha1_update(&mtree->sha1ctx, buff, n); #endif #ifdef ARCHIVE_HAS_SHA256 if (mtree->compute_sum & F_SHA256) archive_sha256_update(&mtree->sha256ctx, buff, n); #endif #ifdef ARCHIVE_HAS_SHA384 if (mtree->compute_sum & F_SHA384) archive_sha384_update(&mtree->sha384ctx, buff, n); #endif #ifdef ARCHIVE_HAS_SHA512 if (mtree->compute_sum & F_SHA512) archive_sha512_update(&mtree->sha512ctx, buff, n); #endif } static void sum_final(struct mtree_writer *mtree, struct reg_info *reg) { if (mtree->compute_sum & F_CKSUM) { uint64_t len; /* Include the length of the file. */ for (len = mtree->crc_len; len != 0; len >>= 8) COMPUTE_CRC(mtree->crc, len & 0xff); reg->crc = ~mtree->crc; } #ifdef ARCHIVE_HAS_MD5 if (mtree->compute_sum & F_MD5) archive_md5_final(&mtree->md5ctx, reg->buf_md5); #endif #ifdef ARCHIVE_HAS_RMD160 if (mtree->compute_sum & F_RMD160) archive_rmd160_final(&mtree->rmd160ctx, reg->buf_rmd160); #endif #ifdef ARCHIVE_HAS_SHA1 if (mtree->compute_sum & F_SHA1) archive_sha1_final(&mtree->sha1ctx, reg->buf_sha1); #endif #ifdef ARCHIVE_HAS_SHA256 if (mtree->compute_sum & F_SHA256) archive_sha256_final(&mtree->sha256ctx, reg->buf_sha256); #endif #ifdef ARCHIVE_HAS_SHA384 if (mtree->compute_sum & F_SHA384) archive_sha384_final(&mtree->sha384ctx, reg->buf_sha384); #endif #ifdef ARCHIVE_HAS_SHA512 if (mtree->compute_sum & F_SHA512) archive_sha512_final(&mtree->sha512ctx, reg->buf_sha512); #endif /* Save what types of sum are computed. */ reg->compute_sum = mtree->compute_sum; } #if defined(ARCHIVE_HAS_MD5) || defined(ARCHIVE_HAS_RMD160) || \ defined(ARCHIVE_HAS_SHA1) || defined(ARCHIVE_HAS_SHA256) || \ defined(ARCHIVE_HAS_SHA384) || defined(ARCHIVE_HAS_SHA512) static void strappend_bin(struct archive_string *s, const unsigned char *bin, int n) { static const char hex[] = "0123456789abcdef"; int i; for (i = 0; i < n; i++) { archive_strappend_char(s, hex[bin[i] >> 4]); archive_strappend_char(s, hex[bin[i] & 0x0f]); } } #endif static void sum_write(struct archive_string *str, struct reg_info *reg) { if (reg->compute_sum & F_CKSUM) { archive_string_sprintf(str, " cksum=%ju", (uintmax_t)reg->crc); } #ifdef ARCHIVE_HAS_MD5 if (reg->compute_sum & F_MD5) { archive_strcat(str, " md5digest="); strappend_bin(str, reg->buf_md5, sizeof(reg->buf_md5)); } #endif #ifdef ARCHIVE_HAS_RMD160 if (reg->compute_sum & F_RMD160) { archive_strcat(str, " rmd160digest="); strappend_bin(str, reg->buf_rmd160, sizeof(reg->buf_rmd160)); } #endif #ifdef ARCHIVE_HAS_SHA1 if (reg->compute_sum & F_SHA1) { archive_strcat(str, " sha1digest="); strappend_bin(str, reg->buf_sha1, sizeof(reg->buf_sha1)); } #endif #ifdef ARCHIVE_HAS_SHA256 if (reg->compute_sum & F_SHA256) { archive_strcat(str, " sha256digest="); strappend_bin(str, reg->buf_sha256, sizeof(reg->buf_sha256)); } #endif #ifdef ARCHIVE_HAS_SHA384 if (reg->compute_sum & F_SHA384) { archive_strcat(str, " sha384digest="); strappend_bin(str, reg->buf_sha384, sizeof(reg->buf_sha384)); } #endif #ifdef ARCHIVE_HAS_SHA512 if (reg->compute_sum & F_SHA512) { archive_strcat(str, " sha512digest="); strappend_bin(str, reg->buf_sha512, sizeof(reg->buf_sha512)); } #endif } static int mtree_entry_cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct mtree_entry *e1 = (const struct mtree_entry *)n1; const struct mtree_entry *e2 = (const struct mtree_entry *)n2; return (strcmp(e2->basename.s, e1->basename.s)); } static int mtree_entry_cmp_key(const struct archive_rb_node *n, const void *key) { const struct mtree_entry *e = (const struct mtree_entry *)n; return (strcmp((const char *)key, e->basename.s)); } #if defined(_WIN32) || defined(__CYGWIN__) static int cleanup_backslash_1(char *p) { int mb, dos; mb = dos = 0; while (*p) { if (*(unsigned char *)p > 127) mb = 1; if (*p == '\\') { /* If we have not met any multi-byte characters, * we can replace '\' with '/'. */ if (!mb) *p = '/'; dos = 1; } p++; } if (!mb || !dos) return (0); return (-1); } static void cleanup_backslash_2(wchar_t *p) { /* Convert a path-separator from '\' to '/' */ while (*p != L'\0') { if (*p == L'\\') *p = L'/'; p++; } } #endif /* * Generate a parent directory name and a base name from a pathname. */ static int mtree_entry_setup_filenames(struct archive_write *a, struct mtree_entry *file, struct archive_entry *entry) { const char *pathname; char *p, *dirname, *slash; size_t len; int ret = ARCHIVE_OK; archive_strcpy(&file->pathname, archive_entry_pathname(entry)); #if defined(_WIN32) || defined(__CYGWIN__) /* * Convert a path-separator from '\' to '/' */ if (cleanup_backslash_1(file->pathname.s) != 0) { const wchar_t *wp = archive_entry_pathname_w(entry); struct archive_wstring ws; if (wp != NULL) { int r; archive_string_init(&ws); archive_wstrcpy(&ws, wp); cleanup_backslash_2(ws.s); archive_string_empty(&(file->pathname)); r = archive_string_append_from_wcs(&(file->pathname), ws.s, ws.length); archive_wstring_free(&ws); if (r < 0 && errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } } } #else (void)a; /* UNUSED */ #endif pathname = file->pathname.s; if (strcmp(pathname, ".") == 0) { archive_strcpy(&file->basename, "."); return (ARCHIVE_OK); } archive_strcpy(&(file->parentdir), pathname); len = file->parentdir.length; p = dirname = file->parentdir.s; /* * Remove leading '/' and '../' elements */ while (*p) { if (p[0] == '/') { p++; len--; } else if (p[0] != '.') break; else if (p[1] == '.' && p[2] == '/') { p += 3; len -= 3; } else break; } if (p != dirname) { memmove(dirname, p, len+1); p = dirname; } /* * Remove "/","/." and "/.." elements from tail. */ while (len > 0) { size_t ll = len; if (len > 0 && p[len-1] == '/') { p[len-1] = '\0'; len--; } if (len > 1 && p[len-2] == '/' && p[len-1] == '.') { p[len-2] = '\0'; len -= 2; } if (len > 2 && p[len-3] == '/' && p[len-2] == '.' && p[len-1] == '.') { p[len-3] = '\0'; len -= 3; } if (ll == len) break; } while (*p) { if (p[0] == '/') { if (p[1] == '/') /* Convert '//' --> '/' */ strcpy(p, p+1); else if (p[1] == '.' && p[2] == '/') /* Convert '/./' --> '/' */ strcpy(p, p+2); else if (p[1] == '.' && p[2] == '.' && p[3] == '/') { /* Convert 'dir/dir1/../dir2/' * --> 'dir/dir2/' */ char *rp = p -1; while (rp >= dirname) { if (*rp == '/') break; --rp; } if (rp > dirname) { strcpy(rp, p+3); p = rp; } else { strcpy(dirname, p+4); p = dirname; } } else p++; } else p++; } p = dirname; len = strlen(p); /* - * Add "./" prefiex. + * Add "./" prefix. * NOTE: If the pathname does not have a path separator, we have - * to add "./" to the head of the pathename because mtree reader + * to add "./" to the head of the pathname because mtree reader * will suppose that it is v1(a.k.a classic) mtree format and * change the directory unexpectedly and so it will make a wrong * path. */ if (strcmp(p, ".") != 0 && strncmp(p, "./", 2) != 0) { struct archive_string as; archive_string_init(&as); archive_strcpy(&as, "./"); archive_strncat(&as, p, len); archive_string_empty(&file->parentdir); archive_string_concat(&file->parentdir, &as); archive_string_free(&as); p = file->parentdir.s; len = archive_strlen(&file->parentdir); } /* * Find out the position which points the last position of * path separator('/'). */ slash = NULL; for (; *p != '\0'; p++) { if (*p == '/') slash = p; } if (slash == NULL) { /* The pathname doesn't have a parent directory. */ file->parentdir.length = len; archive_string_copy(&(file->basename), &(file->parentdir)); archive_string_empty(&(file->parentdir)); *file->parentdir.s = '\0'; return (ret); } /* Make a basename from file->parentdir.s and slash */ *slash = '\0'; file->parentdir.length = slash - file->parentdir.s; archive_strcpy(&(file->basename), slash + 1); return (ret); } static int mtree_entry_create_virtual_dir(struct archive_write *a, const char *pathname, struct mtree_entry **m_entry) { struct archive_entry *entry; struct mtree_entry *file; int r; entry = archive_entry_new(); if (entry == NULL) { *m_entry = NULL; archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } archive_entry_copy_pathname(entry, pathname); archive_entry_set_mode(entry, AE_IFDIR | 0755); archive_entry_set_mtime(entry, time(NULL), 0); r = mtree_entry_new(a, entry, &file); archive_entry_free(entry); if (r < ARCHIVE_WARN) { *m_entry = NULL; archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } file->dir_info->virtual = 1; *m_entry = file; return (ARCHIVE_OK); } static void mtree_entry_register_add(struct mtree_writer *mtree, struct mtree_entry *file) { file->next = NULL; *mtree->file_list.last = file; mtree->file_list.last = &(file->next); } static void mtree_entry_register_init(struct mtree_writer *mtree) { mtree->file_list.first = NULL; mtree->file_list.last = &(mtree->file_list.first); } static void mtree_entry_register_free(struct mtree_writer *mtree) { struct mtree_entry *file, *file_next; file = mtree->file_list.first; while (file != NULL) { file_next = file->next; mtree_entry_free(file); file = file_next; } } static int mtree_entry_add_child_tail(struct mtree_entry *parent, struct mtree_entry *child) { child->dir_info->chnext = NULL; *parent->dir_info->children.last = child; parent->dir_info->children.last = &(child->dir_info->chnext); return (1); } /* * Find a entry from a parent entry with the name. */ static struct mtree_entry * mtree_entry_find_child(struct mtree_entry *parent, const char *child_name) { struct mtree_entry *np; if (parent == NULL) return (NULL); np = (struct mtree_entry *)__archive_rb_tree_find_node( &(parent->dir_info->rbtree), child_name); return (np); } static int get_path_component(char *name, size_t n, const char *fn) { char *p; size_t l; p = strchr(fn, '/'); if (p == NULL) { if ((l = strlen(fn)) == 0) return (0); } else l = p - fn; if (l > n -1) return (-1); memcpy(name, fn, l); name[l] = '\0'; return ((int)l); } /* * Add a new entry into the tree. */ static int mtree_entry_tree_add(struct archive_write *a, struct mtree_entry **filep) { #if defined(_WIN32) && !defined(__CYGWIN__) char name[_MAX_FNAME];/* Included null terminator size. */ #elif defined(NAME_MAX) && NAME_MAX >= 255 char name[NAME_MAX+1]; #else char name[256]; #endif struct mtree_writer *mtree = (struct mtree_writer *)a->format_data; struct mtree_entry *dent, *file, *np; const char *fn, *p; int l, r; file = *filep; if (file->parentdir.length == 0 && file->basename.length == 1 && file->basename.s[0] == '.') { file->parent = file; if (mtree->root != NULL) { np = mtree->root; goto same_entry; } mtree->root = file; mtree_entry_register_add(mtree, file); return (ARCHIVE_OK); } if (file->parentdir.length == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Internal programing error " "in generating canonical name for %s", file->pathname.s); return (ARCHIVE_FAILED); } fn = p = file->parentdir.s; /* * If the path of the parent directory of `file' entry is * the same as the path of `cur_dirent', add `file' entry to * `cur_dirent'. */ if (archive_strlen(&(mtree->cur_dirstr)) == archive_strlen(&(file->parentdir)) && strcmp(mtree->cur_dirstr.s, fn) == 0) { if (!__archive_rb_tree_insert_node( &(mtree->cur_dirent->dir_info->rbtree), (struct archive_rb_node *)file)) { /* There is the same name in the tree. */ np = (struct mtree_entry *)__archive_rb_tree_find_node( &(mtree->cur_dirent->dir_info->rbtree), file->basename.s); goto same_entry; } file->parent = mtree->cur_dirent; mtree_entry_register_add(mtree, file); return (ARCHIVE_OK); } dent = mtree->root; for (;;) { l = get_path_component(name, sizeof(name), fn); if (l == 0) { np = NULL; break; } if (l < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "A name buffer is too small"); return (ARCHIVE_FATAL); } if (l == 1 && name[0] == '.' && dent != NULL && dent == mtree->root) { fn += l; if (fn[0] == '/') fn++; continue; } np = mtree_entry_find_child(dent, name); if (np == NULL || fn[0] == '\0') break; /* Find next sub directory. */ if (!np->dir_info) { /* NOT Directory! */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "`%s' is not directory, we cannot insert `%s' ", np->pathname.s, file->pathname.s); return (ARCHIVE_FAILED); } fn += l; if (fn[0] == '/') fn++; dent = np; } if (np == NULL) { /* * Create virtual parent directories. */ while (fn[0] != '\0') { struct mtree_entry *vp; struct archive_string as; archive_string_init(&as); archive_strncat(&as, p, fn - p + l); if (as.s[as.length-1] == '/') { as.s[as.length-1] = '\0'; as.length--; } r = mtree_entry_create_virtual_dir(a, as.s, &vp); archive_string_free(&as); if (r < ARCHIVE_WARN) return (r); if (strcmp(vp->pathname.s, ".") == 0) { vp->parent = vp; mtree->root = vp; } else { __archive_rb_tree_insert_node( &(dent->dir_info->rbtree), (struct archive_rb_node *)vp); vp->parent = dent; } mtree_entry_register_add(mtree, vp); np = vp; fn += l; if (fn[0] == '/') fn++; l = get_path_component(name, sizeof(name), fn); if (l < 0) { archive_string_free(&as); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "A name buffer is too small"); return (ARCHIVE_FATAL); } dent = np; } /* Found out the parent directory where `file' can be * inserted. */ mtree->cur_dirent = dent; archive_string_empty(&(mtree->cur_dirstr)); archive_string_ensure(&(mtree->cur_dirstr), archive_strlen(&(dent->parentdir)) + archive_strlen(&(dent->basename)) + 2); if (archive_strlen(&(dent->parentdir)) + archive_strlen(&(dent->basename)) == 0) mtree->cur_dirstr.s[0] = 0; else { if (archive_strlen(&(dent->parentdir)) > 0) { archive_string_copy(&(mtree->cur_dirstr), &(dent->parentdir)); archive_strappend_char( &(mtree->cur_dirstr), '/'); } archive_string_concat(&(mtree->cur_dirstr), &(dent->basename)); } if (!__archive_rb_tree_insert_node( &(dent->dir_info->rbtree), (struct archive_rb_node *)file)) { np = (struct mtree_entry *)__archive_rb_tree_find_node( &(dent->dir_info->rbtree), file->basename.s); goto same_entry; } file->parent = dent; mtree_entry_register_add(mtree, file); return (ARCHIVE_OK); } same_entry: /* * We have already has the entry the filename of which is * the same. */ r = mtree_entry_exchange_same_entry(a, np, file); if (r < ARCHIVE_WARN) return (r); if (np->dir_info) np->dir_info->virtual = 0; *filep = np; mtree_entry_free(file); return (ARCHIVE_WARN); } static int mtree_entry_exchange_same_entry(struct archive_write *a, struct mtree_entry *np, struct mtree_entry *file) { if ((np->mode & AE_IFMT) != (file->mode & AE_IFMT)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Found duplicate entries `%s' and its file type is " "different", np->pathname.s); return (ARCHIVE_FAILED); } /* Update the existent mtree entry's attributes by the new one's. */ archive_string_empty(&np->symlink); archive_string_concat(&np->symlink, &file->symlink); archive_string_empty(&np->uname); archive_string_concat(&np->uname, &file->uname); archive_string_empty(&np->gname); archive_string_concat(&np->gname, &file->gname); archive_string_empty(&np->fflags_text); archive_string_concat(&np->fflags_text, &file->fflags_text); np->nlink = file->nlink; np->filetype = file->filetype; np->mode = file->mode; np->size = file->size; np->uid = file->uid; np->gid = file->gid; np->fflags_set = file->fflags_set; np->fflags_clear = file->fflags_clear; np->mtime = file->mtime; np->mtime_nsec = file->mtime_nsec; np->rdevmajor = file->rdevmajor; np->rdevminor = file->rdevminor; np->devmajor = file->devmajor; np->devminor = file->devminor; np->ino = file->ino; return (ARCHIVE_WARN); } Index: head/contrib/libarchive/libarchive/archive_write_set_format_pax.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_pax.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_write_set_format_pax.c (revision 309590) @@ -1,1907 +1,1907 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * Copyright (c) 2010-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 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 #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_write_private.h" struct sparse_block { struct sparse_block *next; int is_hole; uint64_t offset; uint64_t remaining; }; struct pax { uint64_t entry_bytes_remaining; uint64_t entry_padding; struct archive_string l_url_encoded_name; struct archive_string pax_header; struct archive_string sparse_map; size_t sparse_map_padding; struct sparse_block *sparse_list; struct sparse_block *sparse_tail; struct archive_string_conv *sconv_utf8; int opt_binary; }; static void add_pax_attr(struct archive_string *, const char *key, const char *value); static void add_pax_attr_int(struct archive_string *, const char *key, int64_t value); static void add_pax_attr_time(struct archive_string *, const char *key, int64_t sec, unsigned long nanos); static ssize_t archive_write_pax_data(struct archive_write *, const void *, size_t); static int archive_write_pax_close(struct archive_write *); static int archive_write_pax_free(struct archive_write *); static int archive_write_pax_finish_entry(struct archive_write *); static int archive_write_pax_header(struct archive_write *, struct archive_entry *); static int archive_write_pax_options(struct archive_write *, const char *, const char *); static char *base64_encode(const char *src, size_t len); static char *build_gnu_sparse_name(char *dest, const char *src); static char *build_pax_attribute_name(char *dest, const char *src); static char *build_ustar_entry_name(char *dest, const char *src, size_t src_length, const char *insert); static char *format_int(char *dest, int64_t); static int has_non_ASCII(const char *); static void sparse_list_clear(struct pax *); static int sparse_list_add(struct pax *, int64_t, int64_t); static char *url_encode(const char *in); /* * Set output format to 'restricted pax' format. * * This is the same as normal 'pax', but tries to suppress * the pax header whenever possible. This is the default for * bsdtar, for instance. */ int archive_write_set_format_pax_restricted(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int r; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_pax_restricted"); r = archive_write_set_format_pax(&a->archive); a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_RESTRICTED; a->archive.archive_format_name = "restricted POSIX pax interchange"; return (r); } /* * Set output format to 'pax' format. */ int archive_write_set_format_pax(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct pax *pax; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_pax"); if (a->format_free != NULL) (a->format_free)(a); pax = (struct pax *)malloc(sizeof(*pax)); if (pax == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate pax data"); return (ARCHIVE_FATAL); } memset(pax, 0, sizeof(*pax)); a->format_data = pax; a->format_name = "pax"; a->format_options = archive_write_pax_options; a->format_write_header = archive_write_pax_header; a->format_write_data = archive_write_pax_data; a->format_close = archive_write_pax_close; a->format_free = archive_write_pax_free; a->format_finish_entry = archive_write_pax_finish_entry; a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "POSIX pax interchange"; return (ARCHIVE_OK); } static int archive_write_pax_options(struct archive_write *a, const char *key, const char *val) { struct pax *pax = (struct pax *)a->format_data; int ret = ARCHIVE_FAILED; if (strcmp(key, "hdrcharset") == 0) { /* * The character-set we can use are defined in * IEEE Std 1003.1-2001 */ if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "pax: hdrcharset option needs a character-set name"); else if (strcmp(val, "BINARY") == 0 || strcmp(val, "binary") == 0) { /* * Specify binary mode. We will not convert * filenames, uname and gname to any charsets. */ pax->opt_binary = 1; ret = ARCHIVE_OK; } else if (strcmp(val, "UTF-8") == 0) { /* * Specify UTF-8 character-set to be used for * filenames. This is almost the test that * running platform supports the string conversion. * Especially libarchive_test needs this trick for * its test. */ pax->sconv_utf8 = archive_string_conversion_to_charset( &(a->archive), "UTF-8", 0); if (pax->sconv_utf8 == NULL) ret = ARCHIVE_FATAL; else ret = ARCHIVE_OK; } else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "pax: invalid charset name"); return (ret); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } /* * Note: This code assumes that 'nanos' has the same sign as 'sec', * which implies that sec=-1, nanos=200000000 represents -1.2 seconds * and not -0.8 seconds. This is a pretty pedantic point, as we're * unlikely to encounter many real files created before Jan 1, 1970, * much less ones with timestamps recorded to sub-second resolution. */ static void add_pax_attr_time(struct archive_string *as, const char *key, int64_t sec, unsigned long nanos) { int digit, i; char *t; /* * Note that each byte contributes fewer than 3 base-10 * digits, so this will always be big enough. */ char tmp[1 + 3*sizeof(sec) + 1 + 3*sizeof(nanos)]; tmp[sizeof(tmp) - 1] = 0; t = tmp + sizeof(tmp) - 1; /* Skip trailing zeros in the fractional part. */ for (digit = 0, i = 10; i > 0 && digit == 0; i--) { digit = nanos % 10; nanos /= 10; } /* Only format the fraction if it's non-zero. */ if (i > 0) { while (i > 0) { *--t = "0123456789"[digit]; digit = nanos % 10; nanos /= 10; i--; } *--t = '.'; } t = format_int(t, sec); add_pax_attr(as, key, t); } static char * format_int(char *t, int64_t i) { uint64_t ui; if (i < 0) ui = (i == INT64_MIN) ? (uint64_t)(INT64_MAX) + 1 : (uint64_t)(-i); else ui = i; do { *--t = "0123456789"[ui % 10]; } while (ui /= 10); if (i < 0) *--t = '-'; return (t); } static void add_pax_attr_int(struct archive_string *as, const char *key, int64_t value) { char tmp[1 + 3 * sizeof(value)]; tmp[sizeof(tmp) - 1] = 0; add_pax_attr(as, key, format_int(tmp + sizeof(tmp) - 1, value)); } /* * Add a key/value attribute to the pax header. This function handles * the length field and various other syntactic requirements. */ static void add_pax_attr(struct archive_string *as, const char *key, const char *value) { int digits, i, len, next_ten; char tmp[1 + 3 * sizeof(int)]; /* < 3 base-10 digits per byte */ /*- * PAX attributes have the following layout: * <=> */ len = 1 + (int)strlen(key) + 1 + (int)strlen(value) + 1; /* * The field includes the length of the field, so * computing the correct length is tricky. I start by * counting the number of base-10 digits in 'len' and * computing the next higher power of 10. */ next_ten = 1; digits = 0; i = len; while (i > 0) { i = i / 10; digits++; next_ten = next_ten * 10; } /* * For example, if string without the length field is 99 * chars, then adding the 2 digit length "99" will force the * total length past 100, requiring an extra digit. The next * statement adjusts for this effect. */ if (len + digits >= next_ten) digits++; /* Now, we have the right length so we can build the line. */ tmp[sizeof(tmp) - 1] = 0; /* Null-terminate the work area. */ archive_strcat(as, format_int(tmp + sizeof(tmp) - 1, len + digits)); archive_strappend_char(as, ' '); archive_strcat(as, key); archive_strappend_char(as, '='); archive_strcat(as, value); archive_strappend_char(as, '\n'); } static int archive_write_pax_header_xattrs(struct archive_write *a, struct pax *pax, struct archive_entry *entry) { struct archive_string s; int i = archive_entry_xattr_reset(entry); while (i--) { const char *name; const void *value; char *encoded_value; char *url_encoded_name = NULL, *encoded_name = NULL; size_t size; int r; archive_entry_xattr_next(entry, &name, &value, &size); url_encoded_name = url_encode(name); if (url_encoded_name != NULL) { /* Convert narrow-character to UTF-8. */ r = archive_strcpy_l(&(pax->l_url_encoded_name), url_encoded_name, pax->sconv_utf8); free(url_encoded_name); /* Done with this. */ if (r == 0) encoded_name = pax->l_url_encoded_name.s; else if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } } encoded_value = base64_encode((const char *)value, size); if (encoded_name != NULL && encoded_value != NULL) { archive_string_init(&s); archive_strcpy(&s, "LIBARCHIVE.xattr."); archive_strcat(&s, encoded_name); add_pax_attr(&(pax->pax_header), s.s, encoded_value); archive_string_free(&s); } free(encoded_value); } return (ARCHIVE_OK); } static int get_entry_hardlink(struct archive_write *a, struct archive_entry *entry, const char **name, size_t *length, struct archive_string_conv *sc) { int r; r = archive_entry_hardlink_l(entry, name, length, sc); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } return (ARCHIVE_WARN); } return (ARCHIVE_OK); } static int get_entry_pathname(struct archive_write *a, struct archive_entry *entry, const char **name, size_t *length, struct archive_string_conv *sc) { int r; r = archive_entry_pathname_l(entry, name, length, sc); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } return (ARCHIVE_WARN); } return (ARCHIVE_OK); } static int get_entry_uname(struct archive_write *a, struct archive_entry *entry, const char **name, size_t *length, struct archive_string_conv *sc) { int r; r = archive_entry_uname_l(entry, name, length, sc); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Uname"); return (ARCHIVE_FATAL); } return (ARCHIVE_WARN); } return (ARCHIVE_OK); } static int get_entry_gname(struct archive_write *a, struct archive_entry *entry, const char **name, size_t *length, struct archive_string_conv *sc) { int r; r = archive_entry_gname_l(entry, name, length, sc); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Gname"); return (ARCHIVE_FATAL); } return (ARCHIVE_WARN); } return (ARCHIVE_OK); } static int get_entry_symlink(struct archive_write *a, struct archive_entry *entry, const char **name, size_t *length, struct archive_string_conv *sc) { int r; r = archive_entry_symlink_l(entry, name, length, sc); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } return (ARCHIVE_WARN); } return (ARCHIVE_OK); } /* * TODO: Consider adding 'comment' and 'charset' fields to * archive_entry so that clients can specify them. Also, consider * adding generic key/value tags so clients can add arbitrary * key/value data. * * TODO: Break up this 700-line function!!!! Yowza! */ static int archive_write_pax_header(struct archive_write *a, struct archive_entry *entry_original) { struct archive_entry *entry_main; const char *p; const char *suffix; int need_extension, r, ret; int sparse_count; uint64_t sparse_total, real_size; struct pax *pax; const char *hardlink; const char *path = NULL, *linkpath = NULL; const char *uname = NULL, *gname = NULL; const void *mac_metadata; size_t mac_metadata_size; struct archive_string_conv *sconv; size_t hardlink_length, path_length, linkpath_length; size_t uname_length, gname_length; char paxbuff[512]; char ustarbuff[512]; char ustar_entry_name[256]; char pax_entry_name[256]; char gnu_sparse_name[256]; struct archive_string entry_name; ret = ARCHIVE_OK; need_extension = 0; pax = (struct pax *)a->format_data; /* Sanity check. */ if (archive_entry_pathname(entry_original) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't record entry in tar file without pathname"); return (ARCHIVE_FAILED); } /* * Choose a header encoding. */ if (pax->opt_binary) sconv = NULL;/* Binary mode. */ else { /* Header encoding is UTF-8. */ if (pax->sconv_utf8 == NULL) { /* Initialize the string conversion object * we must need */ pax->sconv_utf8 = archive_string_conversion_to_charset( &(a->archive), "UTF-8", 1); if (pax->sconv_utf8 == NULL) /* Couldn't allocate memory */ return (ARCHIVE_FAILED); } sconv = pax->sconv_utf8; } r = get_entry_hardlink(a, entry_original, &hardlink, &hardlink_length, sconv); if (r == ARCHIVE_FATAL) return (r); else if (r != ARCHIVE_OK) { r = get_entry_hardlink(a, entry_original, &hardlink, &hardlink_length, NULL); if (r == ARCHIVE_FATAL) return (r); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate linkname '%s' to %s", hardlink, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; sconv = NULL;/* The header charset switches to binary mode. */ } /* Make sure this is a type of entry that we can handle here */ if (hardlink == NULL) { switch (archive_entry_filetype(entry_original)) { case AE_IFBLK: case AE_IFCHR: case AE_IFIFO: case AE_IFLNK: case AE_IFREG: break; case AE_IFDIR: { /* * Ensure a trailing '/'. Modify the original * entry so the client sees the change. */ #if defined(_WIN32) && !defined(__CYGWIN__) const wchar_t *wp; wp = archive_entry_pathname_w(entry_original); if (wp != NULL && wp[wcslen(wp) -1] != L'/') { struct archive_wstring ws; archive_string_init(&ws); path_length = wcslen(wp); if (archive_wstring_ensure(&ws, path_length + 2) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate pax data"); archive_wstring_free(&ws); return(ARCHIVE_FATAL); } /* Should we keep '\' ? */ if (wp[path_length -1] == L'\\') path_length--; archive_wstrncpy(&ws, wp, path_length); archive_wstrappend_wchar(&ws, L'/'); archive_entry_copy_pathname_w( entry_original, ws.s); archive_wstring_free(&ws); p = NULL; } else #endif p = archive_entry_pathname(entry_original); /* * On Windows, this is a backup operation just in * case getting WCS failed. On POSIX, this is a * normal operation. */ if (p != NULL && p[strlen(p) - 1] != '/') { struct archive_string as; archive_string_init(&as); path_length = strlen(p); if (archive_string_ensure(&as, path_length + 2) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate pax data"); archive_string_free(&as); return(ARCHIVE_FATAL); } #if defined(_WIN32) && !defined(__CYGWIN__) /* NOTE: This might break the pathname * if the current code page is CP932 and * the pathname includes a character '\' * as a part of its multibyte pathname. */ if (p[strlen(p) -1] == '\\') path_length--; else #endif archive_strncpy(&as, p, path_length); archive_strappend_char(&as, '/'); archive_entry_copy_pathname( entry_original, as.s); archive_string_free(&as); } break; } case AE_IFSOCK: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive socket"); return (ARCHIVE_FAILED); default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive this (type=0%lo)", (unsigned long) archive_entry_filetype(entry_original)); return (ARCHIVE_FAILED); } } /* * If Mac OS metadata blob is here, recurse to write that * as a separate entry. This is really a pretty poor design: * In particular, it doubles the overhead for long filenames. * TODO: Help Apple folks design something better and figure * out how to transition from this legacy format. * * Note that this code is present on every platform; clients * on non-Mac are unlikely to ever provide this data, but * applications that copy entries from one archive to another * should not lose data just because the local filesystem * can't store it. */ mac_metadata = archive_entry_mac_metadata(entry_original, &mac_metadata_size); if (mac_metadata != NULL) { const char *oname; char *name, *bname; size_t name_length; struct archive_entry *extra = archive_entry_new2(&a->archive); oname = archive_entry_pathname(entry_original); name_length = strlen(oname); name = malloc(name_length + 3); if (name == NULL || extra == NULL) { /* XXX error message */ archive_entry_free(extra); free(name); return (ARCHIVE_FAILED); } strcpy(name, oname); /* Find last '/'; strip trailing '/' characters */ bname = strrchr(name, '/'); while (bname != NULL && bname[1] == '\0') { *bname = '\0'; bname = strrchr(name, '/'); } if (bname == NULL) { memmove(name + 2, name, name_length + 1); memmove(name, "._", 2); } else { bname += 1; memmove(bname + 2, bname, strlen(bname) + 1); memmove(bname, "._", 2); } archive_entry_copy_pathname(extra, name); free(name); archive_entry_set_size(extra, mac_metadata_size); archive_entry_set_filetype(extra, AE_IFREG); archive_entry_set_perm(extra, archive_entry_perm(entry_original)); archive_entry_set_mtime(extra, archive_entry_mtime(entry_original), archive_entry_mtime_nsec(entry_original)); archive_entry_set_gid(extra, archive_entry_gid(entry_original)); archive_entry_set_gname(extra, archive_entry_gname(entry_original)); archive_entry_set_uid(extra, archive_entry_uid(entry_original)); archive_entry_set_uname(extra, archive_entry_uname(entry_original)); /* Recurse to write the special copyfile entry. */ r = archive_write_pax_header(a, extra); archive_entry_free(extra); if (r < ARCHIVE_WARN) return (r); if (r < ret) ret = r; r = (int)archive_write_pax_data(a, mac_metadata, mac_metadata_size); if (r < ARCHIVE_WARN) return (r); if (r < ret) ret = r; r = archive_write_pax_finish_entry(a); if (r < ARCHIVE_WARN) return (r); if (r < ret) ret = r; } /* Copy entry so we can modify it as needed. */ #if defined(_WIN32) && !defined(__CYGWIN__) - /* Make sure the path separators in pahtname, hardlink and symlink + /* Make sure the path separators in pathname, hardlink and symlink * are all slash '/', not the Windows path separator '\'. */ entry_main = __la_win_entry_in_posix_pathseparator(entry_original); if (entry_main == entry_original) entry_main = archive_entry_clone(entry_original); #else entry_main = archive_entry_clone(entry_original); #endif if (entry_main == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate pax data"); return(ARCHIVE_FATAL); } archive_string_empty(&(pax->pax_header)); /* Blank our work area. */ archive_string_empty(&(pax->sparse_map)); sparse_total = 0; sparse_list_clear(pax); if (hardlink == NULL && archive_entry_filetype(entry_main) == AE_IFREG) sparse_count = archive_entry_sparse_reset(entry_main); else sparse_count = 0; if (sparse_count) { int64_t offset, length, last_offset = 0; /* Get the last entry of sparse block. */ while (archive_entry_sparse_next( entry_main, &offset, &length) == ARCHIVE_OK) last_offset = offset + length; /* If the last sparse block does not reach the end of file, * We have to add a empty sparse block as the last entry to * manage storing file data. */ if (last_offset < archive_entry_size(entry_main)) archive_entry_sparse_add_entry(entry_main, archive_entry_size(entry_main), 0); sparse_count = archive_entry_sparse_reset(entry_main); } /* * First, check the name fields and see if any of them * require binary coding. If any of them does, then all of * them do. */ r = get_entry_pathname(a, entry_main, &path, &path_length, sconv); if (r == ARCHIVE_FATAL) return (r); else if (r != ARCHIVE_OK) { r = get_entry_pathname(a, entry_main, &path, &path_length, NULL); if (r == ARCHIVE_FATAL) return (r); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate pathname '%s' to %s", path, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; sconv = NULL;/* The header charset switches to binary mode. */ } r = get_entry_uname(a, entry_main, &uname, &uname_length, sconv); if (r == ARCHIVE_FATAL) return (r); else if (r != ARCHIVE_OK) { r = get_entry_uname(a, entry_main, &uname, &uname_length, NULL); if (r == ARCHIVE_FATAL) return (r); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate uname '%s' to %s", uname, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; sconv = NULL;/* The header charset switches to binary mode. */ } r = get_entry_gname(a, entry_main, &gname, &gname_length, sconv); if (r == ARCHIVE_FATAL) return (r); else if (r != ARCHIVE_OK) { r = get_entry_gname(a, entry_main, &gname, &gname_length, NULL); if (r == ARCHIVE_FATAL) return (r); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate gname '%s' to %s", gname, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; sconv = NULL;/* The header charset switches to binary mode. */ } linkpath = hardlink; linkpath_length = hardlink_length; if (linkpath == NULL) { r = get_entry_symlink(a, entry_main, &linkpath, &linkpath_length, sconv); if (r == ARCHIVE_FATAL) return (r); else if (r != ARCHIVE_OK) { r = get_entry_symlink(a, entry_main, &linkpath, &linkpath_length, NULL); if (r == ARCHIVE_FATAL) return (r); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate linkname '%s' to %s", linkpath, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; sconv = NULL; } } /* If any string conversions failed, get all attributes * in binary-mode. */ if (sconv == NULL && !pax->opt_binary) { if (hardlink != NULL) { r = get_entry_hardlink(a, entry_main, &hardlink, &hardlink_length, NULL); if (r == ARCHIVE_FATAL) return (r); linkpath = hardlink; linkpath_length = hardlink_length; } r = get_entry_pathname(a, entry_main, &path, &path_length, NULL); if (r == ARCHIVE_FATAL) return (r); r = get_entry_uname(a, entry_main, &uname, &uname_length, NULL); if (r == ARCHIVE_FATAL) return (r); r = get_entry_gname(a, entry_main, &gname, &gname_length, NULL); if (r == ARCHIVE_FATAL) return (r); } /* Store the header encoding first, to be nice to readers. */ if (sconv == NULL) add_pax_attr(&(pax->pax_header), "hdrcharset", "BINARY"); /* * If name is too long, or has non-ASCII characters, add * 'path' to pax extended attrs. (Note that an unconvertible * name must have non-ASCII characters.) */ if (has_non_ASCII(path)) { /* We have non-ASCII characters. */ add_pax_attr(&(pax->pax_header), "path", path); archive_entry_set_pathname(entry_main, build_ustar_entry_name(ustar_entry_name, path, path_length, NULL)); need_extension = 1; } else { /* We have an all-ASCII path; we'd like to just store * it in the ustar header if it will fit. Yes, this * duplicates some of the logic in * archive_write_set_format_ustar.c */ if (path_length <= 100) { /* Fits in the old 100-char tar name field. */ } else { /* Find largest suffix that will fit. */ /* Note: strlen() > 100, so strlen() - 100 - 1 >= 0 */ suffix = strchr(path + path_length - 100 - 1, '/'); /* Don't attempt an empty prefix. */ if (suffix == path) suffix = strchr(suffix + 1, '/'); /* We can put it in the ustar header if it's * all ASCII and it's either <= 100 characters * or can be split at a '/' into a prefix <= * 155 chars and a suffix <= 100 chars. (Note * the strchr() above will return NULL exactly * when the path can't be split.) */ if (suffix == NULL /* Suffix > 100 chars. */ || suffix[1] == '\0' /* empty suffix */ || suffix - path > 155) /* Prefix > 155 chars */ { add_pax_attr(&(pax->pax_header), "path", path); archive_entry_set_pathname(entry_main, build_ustar_entry_name(ustar_entry_name, path, path_length, NULL)); need_extension = 1; } } } if (linkpath != NULL) { /* If link name is too long or has non-ASCII characters, add * 'linkpath' to pax extended attrs. */ if (linkpath_length > 100 || has_non_ASCII(linkpath)) { add_pax_attr(&(pax->pax_header), "linkpath", linkpath); if (linkpath_length > 100) { if (hardlink != NULL) archive_entry_set_hardlink(entry_main, "././@LongHardLink"); else archive_entry_set_symlink(entry_main, "././@LongSymLink"); } need_extension = 1; } } /* Save a pathname since it will be renamed if `entry_main` has * sparse blocks. */ archive_string_init(&entry_name); archive_strcpy(&entry_name, archive_entry_pathname(entry_main)); /* If file size is too large, add 'size' to pax extended attrs. */ if (archive_entry_size(entry_main) >= (((int64_t)1) << 33)) { add_pax_attr_int(&(pax->pax_header), "size", archive_entry_size(entry_main)); need_extension = 1; } /* If numeric GID is too large, add 'gid' to pax extended attrs. */ if ((unsigned int)archive_entry_gid(entry_main) >= (1 << 18)) { add_pax_attr_int(&(pax->pax_header), "gid", archive_entry_gid(entry_main)); need_extension = 1; } /* If group name is too large or has non-ASCII characters, add * 'gname' to pax extended attrs. */ if (gname != NULL) { if (gname_length > 31 || has_non_ASCII(gname)) { add_pax_attr(&(pax->pax_header), "gname", gname); need_extension = 1; } } /* If numeric UID is too large, add 'uid' to pax extended attrs. */ if ((unsigned int)archive_entry_uid(entry_main) >= (1 << 18)) { add_pax_attr_int(&(pax->pax_header), "uid", archive_entry_uid(entry_main)); need_extension = 1; } /* Add 'uname' to pax extended attrs if necessary. */ if (uname != NULL) { if (uname_length > 31 || has_non_ASCII(uname)) { add_pax_attr(&(pax->pax_header), "uname", uname); need_extension = 1; } } /* * POSIX/SUSv3 doesn't provide a standard key for large device * numbers. I use the same keys here that Joerg Schilling * used for 'star.' (Which, somewhat confusingly, are called * "devXXX" even though they code "rdev" values.) No doubt, * other implementations use other keys. Note that there's no * reason we can't write the same information into a number of * different keys. * * Of course, this is only needed for block or char device entries. */ if (archive_entry_filetype(entry_main) == AE_IFBLK || archive_entry_filetype(entry_main) == AE_IFCHR) { /* * If rdevmajor is too large, add 'SCHILY.devmajor' to * extended attributes. */ int rdevmajor, rdevminor; rdevmajor = archive_entry_rdevmajor(entry_main); rdevminor = archive_entry_rdevminor(entry_main); if (rdevmajor >= (1 << 18)) { add_pax_attr_int(&(pax->pax_header), "SCHILY.devmajor", rdevmajor); /* * Non-strict formatting below means we don't * have to truncate here. Not truncating improves * the chance that some more modern tar archivers * (such as GNU tar 1.13) can restore the full * value even if they don't understand the pax * extended attributes. See my rant below about * file size fields for additional details. */ /* archive_entry_set_rdevmajor(entry_main, rdevmajor & ((1 << 18) - 1)); */ need_extension = 1; } /* * If devminor is too large, add 'SCHILY.devminor' to * extended attributes. */ if (rdevminor >= (1 << 18)) { add_pax_attr_int(&(pax->pax_header), "SCHILY.devminor", rdevminor); /* Truncation is not necessary here, either. */ /* archive_entry_set_rdevminor(entry_main, rdevminor & ((1 << 18) - 1)); */ need_extension = 1; } } /* * Technically, the mtime field in the ustar header can * support 33 bits, but many platforms use signed 32-bit time * values. The cutoff of 0x7fffffff here is a compromise. * Yes, this check is duplicated just below; this helps to * avoid writing an mtime attribute just to handle a * high-resolution timestamp in "restricted pax" mode. */ if (!need_extension && ((archive_entry_mtime(entry_main) < 0) || (archive_entry_mtime(entry_main) >= 0x7fffffff))) need_extension = 1; /* I use a star-compatible file flag attribute. */ p = archive_entry_fflags_text(entry_main); if (!need_extension && p != NULL && *p != '\0') need_extension = 1; /* If there are non-trivial ACL entries, we need an extension. */ if (!need_extension && archive_entry_acl_count(entry_original, ARCHIVE_ENTRY_ACL_TYPE_ACCESS) > 0) need_extension = 1; /* If there are non-trivial ACL entries, we need an extension. */ if (!need_extension && archive_entry_acl_count(entry_original, ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) > 0) need_extension = 1; /* If there are extended attributes, we need an extension */ if (!need_extension && archive_entry_xattr_count(entry_original) > 0) need_extension = 1; /* If there are sparse info, we need an extension */ if (!need_extension && sparse_count > 0) need_extension = 1; /* * Libarchive used to include these in extended headers for * restricted pax format, but that confused people who * expected ustar-like time semantics. So now we only include * them in full pax format. */ if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_RESTRICTED) { if (archive_entry_ctime(entry_main) != 0 || archive_entry_ctime_nsec(entry_main) != 0) add_pax_attr_time(&(pax->pax_header), "ctime", archive_entry_ctime(entry_main), archive_entry_ctime_nsec(entry_main)); if (archive_entry_atime(entry_main) != 0 || archive_entry_atime_nsec(entry_main) != 0) add_pax_attr_time(&(pax->pax_header), "atime", archive_entry_atime(entry_main), archive_entry_atime_nsec(entry_main)); /* Store birth/creationtime only if it's earlier than mtime */ if (archive_entry_birthtime_is_set(entry_main) && archive_entry_birthtime(entry_main) < archive_entry_mtime(entry_main)) add_pax_attr_time(&(pax->pax_header), "LIBARCHIVE.creationtime", archive_entry_birthtime(entry_main), archive_entry_birthtime_nsec(entry_main)); } /* * The following items are handled differently in "pax * restricted" format. In particular, in "pax restricted" * format they won't be added unless need_extension is * already set (we're already generating an extended header, so * may as well include these). */ if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_RESTRICTED || need_extension) { if (archive_entry_mtime(entry_main) < 0 || archive_entry_mtime(entry_main) >= 0x7fffffff || archive_entry_mtime_nsec(entry_main) != 0) add_pax_attr_time(&(pax->pax_header), "mtime", archive_entry_mtime(entry_main), archive_entry_mtime_nsec(entry_main)); /* I use a star-compatible file flag attribute. */ p = archive_entry_fflags_text(entry_main); if (p != NULL && *p != '\0') add_pax_attr(&(pax->pax_header), "SCHILY.fflags", p); /* I use star-compatible ACL attributes. */ r = archive_entry_acl_text_l(entry_original, ARCHIVE_ENTRY_ACL_TYPE_ACCESS | ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID, &p, NULL, pax->sconv_utf8); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for " "ACL.access"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate ACL.access to UTF-8"); ret = ARCHIVE_WARN; } else if (p != NULL && *p != '\0') { add_pax_attr(&(pax->pax_header), "SCHILY.acl.access", p); } r = archive_entry_acl_text_l(entry_original, ARCHIVE_ENTRY_ACL_TYPE_DEFAULT | ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID, &p, NULL, pax->sconv_utf8); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for " "ACL.default"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate ACL.default to UTF-8"); ret = ARCHIVE_WARN; } else if (p != NULL && *p != '\0') { add_pax_attr(&(pax->pax_header), "SCHILY.acl.default", p); } /* We use GNU-tar-compatible sparse attributes. */ if (sparse_count > 0) { int64_t soffset, slength; add_pax_attr_int(&(pax->pax_header), "GNU.sparse.major", 1); add_pax_attr_int(&(pax->pax_header), "GNU.sparse.minor", 0); add_pax_attr(&(pax->pax_header), "GNU.sparse.name", entry_name.s); add_pax_attr_int(&(pax->pax_header), "GNU.sparse.realsize", archive_entry_size(entry_main)); /* Rename the file name which will be used for * ustar header to a special name, which GNU * PAX Format 1.0 requires */ archive_entry_set_pathname(entry_main, build_gnu_sparse_name(gnu_sparse_name, entry_name.s)); /* * - Make a sparse map, which will precede a file data. * - Get the total size of available data of sparse. */ archive_string_sprintf(&(pax->sparse_map), "%d\n", sparse_count); while (archive_entry_sparse_next(entry_main, &soffset, &slength) == ARCHIVE_OK) { archive_string_sprintf(&(pax->sparse_map), "%jd\n%jd\n", (intmax_t)soffset, (intmax_t)slength); sparse_total += slength; if (sparse_list_add(pax, soffset, slength) != ARCHIVE_OK) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); archive_entry_free(entry_main); archive_string_free(&entry_name); return (ARCHIVE_FATAL); } } } /* Store extended attributes */ if (archive_write_pax_header_xattrs(a, pax, entry_original) == ARCHIVE_FATAL) { archive_entry_free(entry_main); archive_string_free(&entry_name); return (ARCHIVE_FATAL); } } /* Only regular files have data. */ if (archive_entry_filetype(entry_main) != AE_IFREG) archive_entry_set_size(entry_main, 0); /* * Pax-restricted does not store data for hardlinks, in order * to improve compatibility with ustar. */ if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE && hardlink != NULL) archive_entry_set_size(entry_main, 0); /* * XXX Full pax interchange format does permit a hardlink * entry to have data associated with it. I'm not supporting * that here because the client expects me to tell them whether * or not this format expects data for hardlinks. If I * don't check here, then every pax archive will end up with * duplicated data for hardlinks. Someday, there may be * need to select this behavior, in which case the following * will need to be revisited. XXX */ if (hardlink != NULL) archive_entry_set_size(entry_main, 0); /* Save a real file size. */ real_size = archive_entry_size(entry_main); /* * Overwrite a file size by the total size of sparse blocks and * the size of sparse map info. That file size is the length of * the data, which we will exactly store into an archive file. */ if (archive_strlen(&(pax->sparse_map))) { size_t mapsize = archive_strlen(&(pax->sparse_map)); pax->sparse_map_padding = 0x1ff & (-(ssize_t)mapsize); archive_entry_set_size(entry_main, mapsize + pax->sparse_map_padding + sparse_total); } /* Format 'ustar' header for main entry. * * The trouble with file size: If the reader can't understand * the file size, they may not be able to locate the next * entry and the rest of the archive is toast. Pax-compliant * readers are supposed to ignore the file size in the main * header, so the question becomes how to maximize portability * for readers that don't support pax attribute extensions. * For maximum compatibility, I permit numeric extensions in * the main header so that the file size stored will always be * correct, even if it's in a format that only some * implementations understand. The technique used here is: * * a) If possible, follow the standard exactly. This handles * files up to 8 gigabytes minus 1. * * b) If that fails, try octal but omit the field terminator. * That handles files up to 64 gigabytes minus 1. * * c) Otherwise, use base-256 extensions. That handles files * up to 2^63 in this implementation, with the potential to * go up to 2^94. That should hold us for a while. ;-) * * The non-strict formatter uses similar logic for other * numeric fields, though they're less critical. */ if (__archive_write_format_header_ustar(a, ustarbuff, entry_main, -1, 0, NULL) == ARCHIVE_FATAL) return (ARCHIVE_FATAL); /* If we built any extended attributes, write that entry first. */ if (archive_strlen(&(pax->pax_header)) > 0) { struct archive_entry *pax_attr_entry; time_t s; int64_t uid, gid; int mode; pax_attr_entry = archive_entry_new2(&a->archive); p = entry_name.s; archive_entry_set_pathname(pax_attr_entry, build_pax_attribute_name(pax_entry_name, p)); archive_entry_set_size(pax_attr_entry, archive_strlen(&(pax->pax_header))); /* Copy uid/gid (but clip to ustar limits). */ uid = archive_entry_uid(entry_main); if (uid >= 1 << 18) uid = (1 << 18) - 1; archive_entry_set_uid(pax_attr_entry, uid); gid = archive_entry_gid(entry_main); if (gid >= 1 << 18) gid = (1 << 18) - 1; archive_entry_set_gid(pax_attr_entry, gid); /* Copy mode over (but not setuid/setgid bits) */ mode = archive_entry_mode(entry_main); #ifdef S_ISUID mode &= ~S_ISUID; #endif #ifdef S_ISGID mode &= ~S_ISGID; #endif #ifdef S_ISVTX mode &= ~S_ISVTX; #endif archive_entry_set_mode(pax_attr_entry, mode); /* Copy uname/gname. */ archive_entry_set_uname(pax_attr_entry, archive_entry_uname(entry_main)); archive_entry_set_gname(pax_attr_entry, archive_entry_gname(entry_main)); /* Copy mtime, but clip to ustar limits. */ s = archive_entry_mtime(entry_main); if (s < 0) { s = 0; } if (s >= 0x7fffffff) { s = 0x7fffffff; } archive_entry_set_mtime(pax_attr_entry, s, 0); /* Standard ustar doesn't support atime. */ archive_entry_set_atime(pax_attr_entry, 0, 0); /* Standard ustar doesn't support ctime. */ archive_entry_set_ctime(pax_attr_entry, 0, 0); r = __archive_write_format_header_ustar(a, paxbuff, pax_attr_entry, 'x', 1, NULL); archive_entry_free(pax_attr_entry); /* Note that the 'x' header shouldn't ever fail to format */ if (r < ARCHIVE_WARN) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "archive_write_pax_header: " "'x' header failed?! This can't happen.\n"); return (ARCHIVE_FATAL); } else if (r < ret) ret = r; r = __archive_write_output(a, paxbuff, 512); if (r != ARCHIVE_OK) { sparse_list_clear(pax); pax->entry_bytes_remaining = 0; pax->entry_padding = 0; return (ARCHIVE_FATAL); } pax->entry_bytes_remaining = archive_strlen(&(pax->pax_header)); pax->entry_padding = 0x1ff & (-(int64_t)pax->entry_bytes_remaining); r = __archive_write_output(a, pax->pax_header.s, archive_strlen(&(pax->pax_header))); if (r != ARCHIVE_OK) { /* If a write fails, we're pretty much toast. */ return (ARCHIVE_FATAL); } /* Pad out the end of the entry. */ r = __archive_write_nulls(a, (size_t)pax->entry_padding); if (r != ARCHIVE_OK) { /* If a write fails, we're pretty much toast. */ return (ARCHIVE_FATAL); } pax->entry_bytes_remaining = pax->entry_padding = 0; } /* Write the header for main entry. */ r = __archive_write_output(a, ustarbuff, 512); if (r != ARCHIVE_OK) return (r); /* * Inform the client of the on-disk size we're using, so * they can avoid unnecessarily writing a body for something * that we're just going to ignore. */ archive_entry_set_size(entry_original, real_size); if (pax->sparse_list == NULL && real_size > 0) { /* This is not a sparse file but we handle its data as * a sparse block. */ sparse_list_add(pax, 0, real_size); sparse_total = real_size; } pax->entry_padding = 0x1ff & (-(int64_t)sparse_total); archive_entry_free(entry_main); archive_string_free(&entry_name); return (ret); } /* * We need a valid name for the regular 'ustar' entry. This routine * tries to hack something more-or-less reasonable. * * The approach here tries to preserve leading dir names. We do so by * working with four sections: * 1) "prefix" directory names, * 2) "suffix" directory names, * 3) inserted dir name (optional), * 4) filename. * * These sections must satisfy the following requirements: * * Parts 1 & 2 together form an initial portion of the dir name. * * Part 3 is specified by the caller. (It should not contain a leading * or trailing '/'.) * * Part 4 forms an initial portion of the base filename. * * The filename must be <= 99 chars to fit the ustar 'name' field. * * Parts 2, 3, 4 together must be <= 99 chars to fit the ustar 'name' fld. * * Part 1 must be <= 155 chars to fit the ustar 'prefix' field. * * If the original name ends in a '/', the new name must also end in a '/' * * Trailing '/.' sequences may be stripped. * * Note: Recall that the ustar format does not store the '/' separating * parts 1 & 2, but does store the '/' separating parts 2 & 3. */ static char * build_ustar_entry_name(char *dest, const char *src, size_t src_length, const char *insert) { const char *prefix, *prefix_end; const char *suffix, *suffix_end; const char *filename, *filename_end; char *p; int need_slash = 0; /* Was there a trailing slash? */ size_t suffix_length = 99; size_t insert_length; /* Length of additional dir element to be added. */ if (insert == NULL) insert_length = 0; else /* +2 here allows for '/' before and after the insert. */ insert_length = strlen(insert) + 2; /* Step 0: Quick bailout in a common case. */ if (src_length < 100 && insert == NULL) { strncpy(dest, src, src_length); dest[src_length] = '\0'; return (dest); } /* Step 1: Locate filename and enforce the length restriction. */ filename_end = src + src_length; /* Remove trailing '/' chars and '/.' pairs. */ for (;;) { if (filename_end > src && filename_end[-1] == '/') { filename_end --; need_slash = 1; /* Remember to restore trailing '/'. */ continue; } if (filename_end > src + 1 && filename_end[-1] == '.' && filename_end[-2] == '/') { filename_end -= 2; need_slash = 1; /* "foo/." will become "foo/" */ continue; } break; } if (need_slash) suffix_length--; /* Find start of filename. */ filename = filename_end - 1; while ((filename > src) && (*filename != '/')) filename --; if ((*filename == '/') && (filename < filename_end - 1)) filename ++; /* Adjust filename_end so that filename + insert fits in 99 chars. */ suffix_length -= insert_length; if (filename_end > filename + suffix_length) filename_end = filename + suffix_length; /* Calculate max size for "suffix" section (#3 above). */ suffix_length -= filename_end - filename; /* Step 2: Locate the "prefix" section of the dirname, including * trailing '/'. */ prefix = src; prefix_end = prefix + 155; if (prefix_end > filename) prefix_end = filename; while (prefix_end > prefix && *prefix_end != '/') prefix_end--; if ((prefix_end < filename) && (*prefix_end == '/')) prefix_end++; /* Step 3: Locate the "suffix" section of the dirname, * including trailing '/'. */ suffix = prefix_end; suffix_end = suffix + suffix_length; /* Enforce limit. */ if (suffix_end > filename) suffix_end = filename; if (suffix_end < suffix) suffix_end = suffix; while (suffix_end > suffix && *suffix_end != '/') suffix_end--; if ((suffix_end < filename) && (*suffix_end == '/')) suffix_end++; /* Step 4: Build the new name. */ /* The OpenBSD strlcpy function is safer, but less portable. */ /* Rather than maintain two versions, just use the strncpy version. */ p = dest; if (prefix_end > prefix) { strncpy(p, prefix, prefix_end - prefix); p += prefix_end - prefix; } if (suffix_end > suffix) { strncpy(p, suffix, suffix_end - suffix); p += suffix_end - suffix; } if (insert != NULL) { /* Note: assume insert does not have leading or trailing '/' */ strcpy(p, insert); p += strlen(insert); *p++ = '/'; } strncpy(p, filename, filename_end - filename); p += filename_end - filename; if (need_slash) *p++ = '/'; *p = '\0'; return (dest); } /* * The ustar header for the pax extended attributes must have a * reasonable name: SUSv3 requires 'dirname'/PaxHeader.'pid'/'filename' * where 'pid' is the PID of the archiving process. Unfortunately, * that makes testing a pain since the output varies for each run, * so I'm sticking with the simpler 'dirname'/PaxHeader/'filename' * for now. (Someday, I'll make this settable. Then I can use the * SUS recommendation as default and test harnesses can override it * to get predictable results.) * * Joerg Schilling has argued that this is unnecessary because, in * practice, if the pax extended attributes get extracted as regular * files, no one is going to bother reading those attributes to * manually restore them. Based on this, 'star' uses * /tmp/PaxHeader/'basename' as the ustar header name. This is a * tempting argument, in part because it's simpler than the SUSv3 * recommendation, but I'm not entirely convinced. I'm also * uncomfortable with the fact that "/tmp" is a Unix-ism. * * The following routine leverages build_ustar_entry_name() above and * so is simpler than you might think. It just needs to provide the * additional path element and handle a few pathological cases). */ static char * build_pax_attribute_name(char *dest, const char *src) { char buff[64]; const char *p; /* Handle the null filename case. */ if (src == NULL || *src == '\0') { strcpy(dest, "PaxHeader/blank"); return (dest); } /* Prune final '/' and other unwanted final elements. */ p = src + strlen(src); for (;;) { /* Ends in "/", remove the '/' */ if (p > src && p[-1] == '/') { --p; continue; } /* Ends in "/.", remove the '.' */ if (p > src + 1 && p[-1] == '.' && p[-2] == '/') { --p; continue; } break; } /* Pathological case: After above, there was nothing left. * This includes "/." "/./." "/.//./." etc. */ if (p == src) { strcpy(dest, "/PaxHeader/rootdir"); return (dest); } /* Convert unadorned "." into a suitable filename. */ if (*src == '.' && p == src + 1) { strcpy(dest, "PaxHeader/currentdir"); return (dest); } /* * TODO: Push this string into the 'pax' structure to avoid * recomputing it every time. That will also open the door * to having clients override it. */ #if HAVE_GETPID && 0 /* Disable this for now; see above comment. */ sprintf(buff, "PaxHeader.%d", getpid()); #else /* If the platform can't fetch the pid, don't include it. */ strcpy(buff, "PaxHeader"); #endif /* General case: build a ustar-compatible name adding * "/PaxHeader/". */ build_ustar_entry_name(dest, src, p - src, buff); return (dest); } /* * GNU PAX Format 1.0 requires the special name, which pattern is: * /GNUSparseFile./ * * This function is used for only Sparse file, a file type of which * is regular file. */ static char * build_gnu_sparse_name(char *dest, const char *src) { char buff[64]; const char *p; /* Handle the null filename case. */ if (src == NULL || *src == '\0') { strcpy(dest, "GNUSparseFile/blank"); return (dest); } /* Prune final '/' and other unwanted final elements. */ p = src + strlen(src); for (;;) { /* Ends in "/", remove the '/' */ if (p > src && p[-1] == '/') { --p; continue; } /* Ends in "/.", remove the '.' */ if (p > src + 1 && p[-1] == '.' && p[-2] == '/') { --p; continue; } break; } #if HAVE_GETPID && 0 /* Disable this as pax attribute name. */ sprintf(buff, "GNUSparseFile.%d", getpid()); #else /* If the platform can't fetch the pid, don't include it. */ strcpy(buff, "GNUSparseFile"); #endif /* General case: build a ustar-compatible name adding * "/GNUSparseFile/". */ build_ustar_entry_name(dest, src, p - src, buff); return (dest); } /* Write two null blocks for the end of archive */ static int archive_write_pax_close(struct archive_write *a) { return (__archive_write_nulls(a, 512 * 2)); } static int archive_write_pax_free(struct archive_write *a) { struct pax *pax; pax = (struct pax *)a->format_data; if (pax == NULL) return (ARCHIVE_OK); archive_string_free(&pax->pax_header); archive_string_free(&pax->sparse_map); archive_string_free(&pax->l_url_encoded_name); sparse_list_clear(pax); free(pax); a->format_data = NULL; return (ARCHIVE_OK); } static int archive_write_pax_finish_entry(struct archive_write *a) { struct pax *pax; uint64_t remaining; int ret; pax = (struct pax *)a->format_data; remaining = pax->entry_bytes_remaining; if (remaining == 0) { while (pax->sparse_list) { struct sparse_block *sb; if (!pax->sparse_list->is_hole) remaining += pax->sparse_list->remaining; sb = pax->sparse_list->next; free(pax->sparse_list); pax->sparse_list = sb; } } ret = __archive_write_nulls(a, (size_t)(remaining + pax->entry_padding)); pax->entry_bytes_remaining = pax->entry_padding = 0; return (ret); } static ssize_t archive_write_pax_data(struct archive_write *a, const void *buff, size_t s) { struct pax *pax; size_t ws; size_t total; int ret; pax = (struct pax *)a->format_data; /* * According to GNU PAX format 1.0, write a sparse map * before the body. */ if (archive_strlen(&(pax->sparse_map))) { ret = __archive_write_output(a, pax->sparse_map.s, archive_strlen(&(pax->sparse_map))); if (ret != ARCHIVE_OK) return (ret); ret = __archive_write_nulls(a, pax->sparse_map_padding); if (ret != ARCHIVE_OK) return (ret); archive_string_empty(&(pax->sparse_map)); } total = 0; while (total < s) { const unsigned char *p; while (pax->sparse_list != NULL && pax->sparse_list->remaining == 0) { struct sparse_block *sb = pax->sparse_list->next; free(pax->sparse_list); pax->sparse_list = sb; } if (pax->sparse_list == NULL) return (total); p = ((const unsigned char *)buff) + total; ws = s - total; if (ws > pax->sparse_list->remaining) ws = (size_t)pax->sparse_list->remaining; if (pax->sparse_list->is_hole) { /* Current block is hole thus we do not write * the body. */ pax->sparse_list->remaining -= ws; total += ws; continue; } ret = __archive_write_output(a, p, ws); pax->sparse_list->remaining -= ws; total += ws; if (ret != ARCHIVE_OK) return (ret); } return (total); } static int has_non_ASCII(const char *_p) { const unsigned char *p = (const unsigned char *)_p; if (p == NULL) return (1); while (*p != '\0' && *p < 128) p++; return (*p != '\0'); } /* * Used by extended attribute support; encodes the name * so that there will be no '=' characters in the result. */ static char * url_encode(const char *in) { const char *s; char *d; int out_len = 0; char *out; for (s = in; *s != '\0'; s++) { if (*s < 33 || *s > 126 || *s == '%' || *s == '=') out_len += 3; else out_len++; } out = (char *)malloc(out_len + 1); if (out == NULL) return (NULL); for (s = in, d = out; *s != '\0'; s++) { /* encode any non-printable ASCII character or '%' or '=' */ if (*s < 33 || *s > 126 || *s == '%' || *s == '=') { /* URL encoding is '%' followed by two hex digits */ *d++ = '%'; *d++ = "0123456789ABCDEF"[0x0f & (*s >> 4)]; *d++ = "0123456789ABCDEF"[0x0f & *s]; } else { *d++ = *s; } } *d = '\0'; return (out); } /* * Encode a sequence of bytes into a C string using base-64 encoding. * * Returns a null-terminated C string allocated with malloc(); caller * is responsible for freeing the result. */ static char * base64_encode(const char *s, size_t len) { static const char digits[64] = { 'A','B','C','D','E','F','G','H','I','J','K','L','M','N','O', 'P','Q','R','S','T','U','V','W','X','Y','Z','a','b','c','d', 'e','f','g','h','i','j','k','l','m','n','o','p','q','r','s', 't','u','v','w','x','y','z','0','1','2','3','4','5','6','7', '8','9','+','/' }; int v; char *d, *out; /* 3 bytes becomes 4 chars, but round up and allow for trailing NUL */ out = (char *)malloc((len * 4 + 2) / 3 + 1); if (out == NULL) return (NULL); d = out; /* Convert each group of 3 bytes into 4 characters. */ while (len >= 3) { v = (((int)s[0] << 16) & 0xff0000) | (((int)s[1] << 8) & 0xff00) | (((int)s[2]) & 0x00ff); s += 3; len -= 3; *d++ = digits[(v >> 18) & 0x3f]; *d++ = digits[(v >> 12) & 0x3f]; *d++ = digits[(v >> 6) & 0x3f]; *d++ = digits[(v) & 0x3f]; } /* Handle final group of 1 byte (2 chars) or 2 bytes (3 chars). */ switch (len) { case 0: break; case 1: v = (((int)s[0] << 16) & 0xff0000); *d++ = digits[(v >> 18) & 0x3f]; *d++ = digits[(v >> 12) & 0x3f]; break; case 2: v = (((int)s[0] << 16) & 0xff0000) | (((int)s[1] << 8) & 0xff00); *d++ = digits[(v >> 18) & 0x3f]; *d++ = digits[(v >> 12) & 0x3f]; *d++ = digits[(v >> 6) & 0x3f]; break; } /* Add trailing NUL character so output is a valid C string. */ *d = '\0'; return (out); } static void sparse_list_clear(struct pax *pax) { while (pax->sparse_list != NULL) { struct sparse_block *sb = pax->sparse_list; pax->sparse_list = sb->next; free(sb); } pax->sparse_tail = NULL; } static int _sparse_list_add_block(struct pax *pax, int64_t offset, int64_t length, int is_hole) { struct sparse_block *sb; sb = (struct sparse_block *)malloc(sizeof(*sb)); if (sb == NULL) return (ARCHIVE_FATAL); sb->next = NULL; sb->is_hole = is_hole; sb->offset = offset; sb->remaining = length; if (pax->sparse_list == NULL || pax->sparse_tail == NULL) pax->sparse_list = pax->sparse_tail = sb; else { pax->sparse_tail->next = sb; pax->sparse_tail = sb; } return (ARCHIVE_OK); } static int sparse_list_add(struct pax *pax, int64_t offset, int64_t length) { int64_t last_offset; int r; if (pax->sparse_tail == NULL) last_offset = 0; else { last_offset = pax->sparse_tail->offset + pax->sparse_tail->remaining; } if (last_offset < offset) { /* Add a hole block. */ r = _sparse_list_add_block(pax, last_offset, offset - last_offset, 1); if (r != ARCHIVE_OK) return (r); } /* Add data block. */ return (_sparse_list_add_block(pax, offset, length, 0)); } Index: head/contrib/libarchive/libarchive/archive_write_set_format_ustar.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_ustar.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_write_set_format_ustar.c (revision 309590) @@ -1,764 +1,764 @@ /*- * Copyright (c) 2003-2007 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("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_write_private.h" struct ustar { uint64_t entry_bytes_remaining; uint64_t entry_padding; struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv_default; int init_default_conversion; }; /* * Define structure of POSIX 'ustar' tar header. */ #define USTAR_name_offset 0 #define USTAR_name_size 100 #define USTAR_mode_offset 100 #define USTAR_mode_size 6 #define USTAR_mode_max_size 8 #define USTAR_uid_offset 108 #define USTAR_uid_size 6 #define USTAR_uid_max_size 8 #define USTAR_gid_offset 116 #define USTAR_gid_size 6 #define USTAR_gid_max_size 8 #define USTAR_size_offset 124 #define USTAR_size_size 11 #define USTAR_size_max_size 12 #define USTAR_mtime_offset 136 #define USTAR_mtime_size 11 #define USTAR_mtime_max_size 11 #define USTAR_checksum_offset 148 #define USTAR_checksum_size 8 #define USTAR_typeflag_offset 156 #define USTAR_typeflag_size 1 #define USTAR_linkname_offset 157 #define USTAR_linkname_size 100 #define USTAR_magic_offset 257 #define USTAR_magic_size 6 #define USTAR_version_offset 263 #define USTAR_version_size 2 #define USTAR_uname_offset 265 #define USTAR_uname_size 32 #define USTAR_gname_offset 297 #define USTAR_gname_size 32 #define USTAR_rdevmajor_offset 329 #define USTAR_rdevmajor_size 6 #define USTAR_rdevmajor_max_size 8 #define USTAR_rdevminor_offset 337 #define USTAR_rdevminor_size 6 #define USTAR_rdevminor_max_size 8 #define USTAR_prefix_offset 345 #define USTAR_prefix_size 155 #define USTAR_padding_offset 500 #define USTAR_padding_size 12 /* * A filled-in copy of the header for initialization. */ static const char template_header[] = { /* name: 100 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0, /* Mode, space-null termination: 8 bytes */ '0','0','0','0','0','0', ' ','\0', /* uid, space-null termination: 8 bytes */ '0','0','0','0','0','0', ' ','\0', /* gid, space-null termination: 8 bytes */ '0','0','0','0','0','0', ' ','\0', - /* size, space termation: 12 bytes */ + /* size, space termination: 12 bytes */ '0','0','0','0','0','0','0','0','0','0','0', ' ', - /* mtime, space termation: 12 bytes */ + /* mtime, space termination: 12 bytes */ '0','0','0','0','0','0','0','0','0','0','0', ' ', /* Initial checksum value: 8 spaces */ ' ',' ',' ',' ',' ',' ',' ',' ', /* Typeflag: 1 byte */ '0', /* '0' = regular file */ /* Linkname: 100 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0, /* Magic: 6 bytes, Version: 2 bytes */ 'u','s','t','a','r','\0', '0','0', /* Uname: 32 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, /* Gname: 32 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, /* rdevmajor + space/null padding: 8 bytes */ '0','0','0','0','0','0', ' ','\0', /* rdevminor + space/null padding: 8 bytes */ '0','0','0','0','0','0', ' ','\0', /* Prefix: 155 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0, /* Padding: 12 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0 }; static ssize_t archive_write_ustar_data(struct archive_write *a, const void *buff, size_t s); static int archive_write_ustar_free(struct archive_write *); static int archive_write_ustar_close(struct archive_write *); static int archive_write_ustar_finish_entry(struct archive_write *); static int archive_write_ustar_header(struct archive_write *, struct archive_entry *entry); static int archive_write_ustar_options(struct archive_write *, const char *, const char *); static int format_256(int64_t, char *, int); static int format_number(int64_t, char *, int size, int max, int strict); static int format_octal(int64_t, char *, int); /* * Set output format to 'ustar' format. */ int archive_write_set_format_ustar(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct ustar *ustar; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_ustar"); /* If someone else was already registered, unregister them. */ if (a->format_free != NULL) (a->format_free)(a); /* Basic internal sanity test. */ if (sizeof(template_header) != 512) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Internal: template_header wrong size: %zu should be 512", sizeof(template_header)); return (ARCHIVE_FATAL); } ustar = (struct ustar *)malloc(sizeof(*ustar)); if (ustar == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate ustar data"); return (ARCHIVE_FATAL); } memset(ustar, 0, sizeof(*ustar)); a->format_data = ustar; a->format_name = "ustar"; a->format_options = archive_write_ustar_options; a->format_write_header = archive_write_ustar_header; a->format_write_data = archive_write_ustar_data; a->format_close = archive_write_ustar_close; a->format_free = archive_write_ustar_free; a->format_finish_entry = archive_write_ustar_finish_entry; a->archive.archive_format = ARCHIVE_FORMAT_TAR_USTAR; a->archive.archive_format_name = "POSIX ustar"; return (ARCHIVE_OK); } static int archive_write_ustar_options(struct archive_write *a, const char *key, const char *val) { struct ustar *ustar = (struct ustar *)a->format_data; int ret = ARCHIVE_FAILED; if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: hdrcharset option needs a character-set name", a->format_name); else { ustar->opt_sconv = archive_string_conversion_to_charset( &a->archive, val, 0); if (ustar->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int archive_write_ustar_header(struct archive_write *a, struct archive_entry *entry) { char buff[512]; int ret, ret2; struct ustar *ustar; struct archive_entry *entry_main; struct archive_string_conv *sconv; ustar = (struct ustar *)a->format_data; /* Setup default string conversion. */ if (ustar->opt_sconv == NULL) { if (!ustar->init_default_conversion) { ustar->sconv_default = archive_string_default_conversion_for_write(&(a->archive)); ustar->init_default_conversion = 1; } sconv = ustar->sconv_default; } else sconv = ustar->opt_sconv; /* Sanity check. */ if (archive_entry_pathname(entry) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't record entry in tar file without pathname"); return (ARCHIVE_FAILED); } /* Only regular files (not hardlinks) have data. */ if (archive_entry_hardlink(entry) != NULL || archive_entry_symlink(entry) != NULL || !(archive_entry_filetype(entry) == AE_IFREG)) archive_entry_set_size(entry, 0); if (AE_IFDIR == archive_entry_filetype(entry)) { const char *p; size_t path_length; /* * Ensure a trailing '/'. Modify the entry so * the client sees the change. */ #if defined(_WIN32) && !defined(__CYGWIN__) const wchar_t *wp; wp = archive_entry_pathname_w(entry); if (wp != NULL && wp[wcslen(wp) -1] != L'/') { struct archive_wstring ws; archive_string_init(&ws); path_length = wcslen(wp); if (archive_wstring_ensure(&ws, path_length + 2) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate ustar data"); archive_wstring_free(&ws); return(ARCHIVE_FATAL); } /* Should we keep '\' ? */ if (wp[path_length -1] == L'\\') path_length--; archive_wstrncpy(&ws, wp, path_length); archive_wstrappend_wchar(&ws, L'/'); archive_entry_copy_pathname_w(entry, ws.s); archive_wstring_free(&ws); p = NULL; } else #endif p = archive_entry_pathname(entry); /* * On Windows, this is a backup operation just in * case getting WCS failed. On POSIX, this is a * normal operation. */ if (p != NULL && p[0] != '\0' && p[strlen(p) - 1] != '/') { struct archive_string as; archive_string_init(&as); path_length = strlen(p); if (archive_string_ensure(&as, path_length + 2) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate ustar data"); archive_string_free(&as); return(ARCHIVE_FATAL); } #if defined(_WIN32) && !defined(__CYGWIN__) /* NOTE: This might break the pathname * if the current code page is CP932 and * the pathname includes a character '\' * as a part of its multibyte pathname. */ if (p[strlen(p) -1] == '\\') path_length--; else #endif archive_strncpy(&as, p, path_length); archive_strappend_char(&as, '/'); archive_entry_copy_pathname(entry, as.s); archive_string_free(&as); } } #if defined(_WIN32) && !defined(__CYGWIN__) - /* Make sure the path separators in pahtname, hardlink and symlink + /* Make sure the path separators in pathname, hardlink and symlink * are all slash '/', not the Windows path separator '\'. */ entry_main = __la_win_entry_in_posix_pathseparator(entry); if (entry_main == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate ustar data"); return(ARCHIVE_FATAL); } if (entry != entry_main) entry = entry_main; else entry_main = NULL; #else entry_main = NULL; #endif ret = __archive_write_format_header_ustar(a, buff, entry, -1, 1, sconv); if (ret < ARCHIVE_WARN) { if (entry_main) archive_entry_free(entry_main); return (ret); } ret2 = __archive_write_output(a, buff, 512); if (ret2 < ARCHIVE_WARN) { if (entry_main) archive_entry_free(entry_main); return (ret2); } if (ret2 < ret) ret = ret2; ustar->entry_bytes_remaining = archive_entry_size(entry); ustar->entry_padding = 0x1ff & (-(int64_t)ustar->entry_bytes_remaining); if (entry_main) archive_entry_free(entry_main); return (ret); } /* * Format a basic 512-byte "ustar" header. * * Returns -1 if format failed (due to field overflow). * Note that this always formats as much of the header as possible. * If "strict" is set to zero, it will extend numeric fields as * necessary (overwriting terminators or using base-256 extensions). * * This is exported so that other 'tar' formats can use it. */ int __archive_write_format_header_ustar(struct archive_write *a, char h[512], struct archive_entry *entry, int tartype, int strict, struct archive_string_conv *sconv) { unsigned int checksum; int i, r, ret; size_t copy_length; const char *p, *pp; int mytartype; ret = 0; mytartype = -1; /* * The "template header" already includes the "ustar" * signature, various end-of-field markers and other required * elements. */ memcpy(h, &template_header, 512); /* * Because the block is already null-filled, and strings * are allowed to exactly fill their destination (without null), * I use memcpy(dest, src, strlen()) here a lot to copy strings. */ r = archive_entry_pathname_l(entry, &pp, ©_length, sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate pathname '%s' to %s", pp, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; } if (copy_length <= USTAR_name_size) memcpy(h + USTAR_name_offset, pp, copy_length); else { /* Store in two pieces, splitting at a '/'. */ p = strchr(pp + copy_length - USTAR_name_size - 1, '/'); /* * Look for the next '/' if we chose the first character * as the separator. (ustar format doesn't permit * an empty prefix.) */ if (p == pp) p = strchr(p + 1, '/'); /* Fail if the name won't fit. */ if (!p) { /* No separator. */ archive_set_error(&a->archive, ENAMETOOLONG, "Pathname too long"); ret = ARCHIVE_FAILED; } else if (p[1] == '\0') { /* * The only feasible separator is a final '/'; * this would result in a non-empty prefix and * an empty name, which POSIX doesn't * explicitly forbid, but it just feels wrong. */ archive_set_error(&a->archive, ENAMETOOLONG, "Pathname too long"); ret = ARCHIVE_FAILED; } else if (p > pp + USTAR_prefix_size) { /* Prefix is too long. */ archive_set_error(&a->archive, ENAMETOOLONG, "Pathname too long"); ret = ARCHIVE_FAILED; } else { /* Copy prefix and remainder to appropriate places */ memcpy(h + USTAR_prefix_offset, pp, p - pp); memcpy(h + USTAR_name_offset, p + 1, pp + copy_length - p - 1); } } r = archive_entry_hardlink_l(entry, &p, ©_length, sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate linkname '%s' to %s", p, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; } if (copy_length > 0) mytartype = '1'; else { r = archive_entry_symlink_l(entry, &p, ©_length, sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate linkname '%s' to %s", p, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; } } if (copy_length > 0) { if (copy_length > USTAR_linkname_size) { archive_set_error(&a->archive, ENAMETOOLONG, "Link contents too long"); ret = ARCHIVE_FAILED; copy_length = USTAR_linkname_size; } memcpy(h + USTAR_linkname_offset, p, copy_length); } r = archive_entry_uname_l(entry, &p, ©_length, sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Uname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate uname '%s' to %s", p, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; } if (copy_length > 0) { if (copy_length > USTAR_uname_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Username too long"); ret = ARCHIVE_FAILED; copy_length = USTAR_uname_size; } memcpy(h + USTAR_uname_offset, p, copy_length); } r = archive_entry_gname_l(entry, &p, ©_length, sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Gname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate gname '%s' to %s", p, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; } if (copy_length > 0) { if (strlen(p) > USTAR_gname_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Group name too long"); ret = ARCHIVE_FAILED; copy_length = USTAR_gname_size; } memcpy(h + USTAR_gname_offset, p, copy_length); } if (format_number(archive_entry_mode(entry) & 07777, h + USTAR_mode_offset, USTAR_mode_size, USTAR_mode_max_size, strict)) { archive_set_error(&a->archive, ERANGE, "Numeric mode too large"); ret = ARCHIVE_FAILED; } if (format_number(archive_entry_uid(entry), h + USTAR_uid_offset, USTAR_uid_size, USTAR_uid_max_size, strict)) { archive_set_error(&a->archive, ERANGE, "Numeric user ID too large"); ret = ARCHIVE_FAILED; } if (format_number(archive_entry_gid(entry), h + USTAR_gid_offset, USTAR_gid_size, USTAR_gid_max_size, strict)) { archive_set_error(&a->archive, ERANGE, "Numeric group ID too large"); ret = ARCHIVE_FAILED; } if (format_number(archive_entry_size(entry), h + USTAR_size_offset, USTAR_size_size, USTAR_size_max_size, strict)) { archive_set_error(&a->archive, ERANGE, "File size out of range"); ret = ARCHIVE_FAILED; } if (format_number(archive_entry_mtime(entry), h + USTAR_mtime_offset, USTAR_mtime_size, USTAR_mtime_max_size, strict)) { archive_set_error(&a->archive, ERANGE, "File modification time too large"); ret = ARCHIVE_FAILED; } if (archive_entry_filetype(entry) == AE_IFBLK || archive_entry_filetype(entry) == AE_IFCHR) { if (format_number(archive_entry_rdevmajor(entry), h + USTAR_rdevmajor_offset, USTAR_rdevmajor_size, USTAR_rdevmajor_max_size, strict)) { archive_set_error(&a->archive, ERANGE, "Major device number too large"); ret = ARCHIVE_FAILED; } if (format_number(archive_entry_rdevminor(entry), h + USTAR_rdevminor_offset, USTAR_rdevminor_size, USTAR_rdevminor_max_size, strict)) { archive_set_error(&a->archive, ERANGE, "Minor device number too large"); ret = ARCHIVE_FAILED; } } if (tartype >= 0) { h[USTAR_typeflag_offset] = tartype; } else if (mytartype >= 0) { h[USTAR_typeflag_offset] = mytartype; } else { switch (archive_entry_filetype(entry)) { case AE_IFREG: h[USTAR_typeflag_offset] = '0' ; break; case AE_IFLNK: h[USTAR_typeflag_offset] = '2' ; break; case AE_IFCHR: h[USTAR_typeflag_offset] = '3' ; break; case AE_IFBLK: h[USTAR_typeflag_offset] = '4' ; break; case AE_IFDIR: h[USTAR_typeflag_offset] = '5' ; break; case AE_IFIFO: h[USTAR_typeflag_offset] = '6' ; break; case AE_IFSOCK: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive socket"); return (ARCHIVE_FAILED); default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive this (mode=0%lo)", (unsigned long)archive_entry_mode(entry)); ret = ARCHIVE_FAILED; } } checksum = 0; for (i = 0; i < 512; i++) checksum += 255 & (unsigned int)h[i]; h[USTAR_checksum_offset + 6] = '\0'; /* Can't be pre-set in the template. */ /* h[USTAR_checksum_offset + 7] = ' '; */ /* This is pre-set in the template. */ format_octal(checksum, h + USTAR_checksum_offset, 6); return (ret); } /* * Format a number into a field, with some intelligence. */ static int format_number(int64_t v, char *p, int s, int maxsize, int strict) { int64_t limit; limit = ((int64_t)1 << (s*3)); /* "Strict" only permits octal values with proper termination. */ if (strict) return (format_octal(v, p, s)); /* * In non-strict mode, we allow the number to overwrite one or * more bytes of the field termination. Even old tar * implementations should be able to handle this with no * problem. */ if (v >= 0) { while (s <= maxsize) { if (v < limit) return (format_octal(v, p, s)); s++; limit <<= 3; } } /* Base-256 can handle any number, positive or negative. */ return (format_256(v, p, maxsize)); } /* * Format a number into the specified field using base-256. */ static int format_256(int64_t v, char *p, int s) { p += s; while (s-- > 0) { *--p = (char)(v & 0xff); v >>= 8; } *p |= 0x80; /* Set the base-256 marker bit. */ return (0); } /* * Format a number into the specified field. */ static int format_octal(int64_t v, char *p, int s) { int len; len = s; /* Octal values can't be negative, so use 0. */ if (v < 0) { while (len-- > 0) *p++ = '0'; return (-1); } p += s; /* Start at the end and work backwards. */ while (s-- > 0) { *--p = (char)('0' + (v & 7)); v >>= 3; } if (v == 0) return (0); /* If it overflowed, fill field with max value. */ while (len-- > 0) *p++ = '7'; return (-1); } static int archive_write_ustar_close(struct archive_write *a) { return (__archive_write_nulls(a, 512*2)); } static int archive_write_ustar_free(struct archive_write *a) { struct ustar *ustar; ustar = (struct ustar *)a->format_data; free(ustar); a->format_data = NULL; return (ARCHIVE_OK); } static int archive_write_ustar_finish_entry(struct archive_write *a) { struct ustar *ustar; int ret; ustar = (struct ustar *)a->format_data; ret = __archive_write_nulls(a, (size_t)(ustar->entry_bytes_remaining + ustar->entry_padding)); ustar->entry_bytes_remaining = ustar->entry_padding = 0; return (ret); } static ssize_t archive_write_ustar_data(struct archive_write *a, const void *buff, size_t s) { struct ustar *ustar; int ret; ustar = (struct ustar *)a->format_data; if (s > ustar->entry_bytes_remaining) s = (size_t)ustar->entry_bytes_remaining; ret = __archive_write_output(a, buff, s); ustar->entry_bytes_remaining -= s; if (ret != ARCHIVE_OK) return (ret); return (s); } Index: head/contrib/libarchive/libarchive/archive_write_set_format_v7tar.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_v7tar.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_write_set_format_v7tar.c (revision 309590) @@ -1,661 +1,661 @@ /*- * Copyright (c) 2003-2007 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("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_write_private.h" struct v7tar { uint64_t entry_bytes_remaining; uint64_t entry_padding; struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv_default; int init_default_conversion; }; /* * Define structure of POSIX 'v7tar' tar header. */ #define V7TAR_name_offset 0 #define V7TAR_name_size 100 #define V7TAR_mode_offset 100 #define V7TAR_mode_size 6 #define V7TAR_mode_max_size 8 #define V7TAR_uid_offset 108 #define V7TAR_uid_size 6 #define V7TAR_uid_max_size 8 #define V7TAR_gid_offset 116 #define V7TAR_gid_size 6 #define V7TAR_gid_max_size 8 #define V7TAR_size_offset 124 #define V7TAR_size_size 11 #define V7TAR_size_max_size 12 #define V7TAR_mtime_offset 136 #define V7TAR_mtime_size 11 #define V7TAR_mtime_max_size 12 #define V7TAR_checksum_offset 148 #define V7TAR_checksum_size 8 #define V7TAR_typeflag_offset 156 #define V7TAR_typeflag_size 1 #define V7TAR_linkname_offset 157 #define V7TAR_linkname_size 100 #define V7TAR_padding_offset 257 #define V7TAR_padding_size 255 /* * A filled-in copy of the header for initialization. */ static const char template_header[] = { /* name: 100 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0, /* Mode, space-null termination: 8 bytes */ '0','0','0','0','0','0', ' ','\0', /* uid, space-null termination: 8 bytes */ '0','0','0','0','0','0', ' ','\0', /* gid, space-null termination: 8 bytes */ '0','0','0','0','0','0', ' ','\0', - /* size, space termation: 12 bytes */ + /* size, space termination: 12 bytes */ '0','0','0','0','0','0','0','0','0','0','0', ' ', - /* mtime, space termation: 12 bytes */ + /* mtime, space termination: 12 bytes */ '0','0','0','0','0','0','0','0','0','0','0', ' ', /* Initial checksum value: 8 spaces */ ' ',' ',' ',' ',' ',' ',' ',' ', /* Typeflag: 1 byte */ 0, /* Linkname: 100 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0, /* Padding: 255 bytes */ 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0 }; static ssize_t archive_write_v7tar_data(struct archive_write *a, const void *buff, size_t s); static int archive_write_v7tar_free(struct archive_write *); static int archive_write_v7tar_close(struct archive_write *); static int archive_write_v7tar_finish_entry(struct archive_write *); static int archive_write_v7tar_header(struct archive_write *, struct archive_entry *entry); static int archive_write_v7tar_options(struct archive_write *, const char *, const char *); static int format_256(int64_t, char *, int); static int format_number(int64_t, char *, int size, int max, int strict); static int format_octal(int64_t, char *, int); static int format_header_v7tar(struct archive_write *, char h[512], struct archive_entry *, int, struct archive_string_conv *); /* * Set output format to 'v7tar' format. */ int archive_write_set_format_v7tar(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct v7tar *v7tar; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_v7tar"); /* If someone else was already registered, unregister them. */ if (a->format_free != NULL) (a->format_free)(a); /* Basic internal sanity test. */ if (sizeof(template_header) != 512) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Internal: template_header wrong size: %zu should be 512", sizeof(template_header)); return (ARCHIVE_FATAL); } v7tar = (struct v7tar *)malloc(sizeof(*v7tar)); if (v7tar == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate v7tar data"); return (ARCHIVE_FATAL); } memset(v7tar, 0, sizeof(*v7tar)); a->format_data = v7tar; a->format_name = "tar (non-POSIX)"; a->format_options = archive_write_v7tar_options; a->format_write_header = archive_write_v7tar_header; a->format_write_data = archive_write_v7tar_data; a->format_close = archive_write_v7tar_close; a->format_free = archive_write_v7tar_free; a->format_finish_entry = archive_write_v7tar_finish_entry; a->archive.archive_format = ARCHIVE_FORMAT_TAR; a->archive.archive_format_name = "tar (non-POSIX)"; return (ARCHIVE_OK); } static int archive_write_v7tar_options(struct archive_write *a, const char *key, const char *val) { struct v7tar *v7tar = (struct v7tar *)a->format_data; int ret = ARCHIVE_FAILED; if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: hdrcharset option needs a character-set name", a->format_name); else { v7tar->opt_sconv = archive_string_conversion_to_charset( &a->archive, val, 0); if (v7tar->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int archive_write_v7tar_header(struct archive_write *a, struct archive_entry *entry) { char buff[512]; int ret, ret2; struct v7tar *v7tar; struct archive_entry *entry_main; struct archive_string_conv *sconv; v7tar = (struct v7tar *)a->format_data; /* Setup default string conversion. */ if (v7tar->opt_sconv == NULL) { if (!v7tar->init_default_conversion) { v7tar->sconv_default = archive_string_default_conversion_for_write( &(a->archive)); v7tar->init_default_conversion = 1; } sconv = v7tar->sconv_default; } else sconv = v7tar->opt_sconv; /* Sanity check. */ if (archive_entry_pathname(entry) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't record entry in tar file without pathname"); return (ARCHIVE_FAILED); } /* Only regular files (not hardlinks) have data. */ if (archive_entry_hardlink(entry) != NULL || archive_entry_symlink(entry) != NULL || !(archive_entry_filetype(entry) == AE_IFREG)) archive_entry_set_size(entry, 0); if (AE_IFDIR == archive_entry_filetype(entry)) { const char *p; size_t path_length; /* * Ensure a trailing '/'. Modify the entry so * the client sees the change. */ #if defined(_WIN32) && !defined(__CYGWIN__) const wchar_t *wp; wp = archive_entry_pathname_w(entry); if (wp != NULL && wp[wcslen(wp) -1] != L'/') { struct archive_wstring ws; archive_string_init(&ws); path_length = wcslen(wp); if (archive_wstring_ensure(&ws, path_length + 2) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate v7tar data"); archive_wstring_free(&ws); return(ARCHIVE_FATAL); } /* Should we keep '\' ? */ if (wp[path_length -1] == L'\\') path_length--; archive_wstrncpy(&ws, wp, path_length); archive_wstrappend_wchar(&ws, L'/'); archive_entry_copy_pathname_w(entry, ws.s); archive_wstring_free(&ws); p = NULL; } else #endif p = archive_entry_pathname(entry); /* * On Windows, this is a backup operation just in * case getting WCS failed. On POSIX, this is a * normal operation. */ if (p != NULL && p[strlen(p) - 1] != '/') { struct archive_string as; archive_string_init(&as); path_length = strlen(p); if (archive_string_ensure(&as, path_length + 2) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate v7tar data"); archive_string_free(&as); return(ARCHIVE_FATAL); } #if defined(_WIN32) && !defined(__CYGWIN__) /* NOTE: This might break the pathname * if the current code page is CP932 and * the pathname includes a character '\' * as a part of its multibyte pathname. */ if (p[strlen(p) -1] == '\\') path_length--; else #endif archive_strncpy(&as, p, path_length); archive_strappend_char(&as, '/'); archive_entry_copy_pathname(entry, as.s); archive_string_free(&as); } } #if defined(_WIN32) && !defined(__CYGWIN__) - /* Make sure the path separators in pahtname, hardlink and symlink + /* Make sure the path separators in pathname, hardlink and symlink * are all slash '/', not the Windows path separator '\'. */ entry_main = __la_win_entry_in_posix_pathseparator(entry); if (entry_main == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate v7tar data"); return(ARCHIVE_FATAL); } if (entry != entry_main) entry = entry_main; else entry_main = NULL; #else entry_main = NULL; #endif ret = format_header_v7tar(a, buff, entry, 1, sconv); if (ret < ARCHIVE_WARN) { if (entry_main) archive_entry_free(entry_main); return (ret); } ret2 = __archive_write_output(a, buff, 512); if (ret2 < ARCHIVE_WARN) { if (entry_main) archive_entry_free(entry_main); return (ret2); } if (ret2 < ret) ret = ret2; v7tar->entry_bytes_remaining = archive_entry_size(entry); v7tar->entry_padding = 0x1ff & (-(int64_t)v7tar->entry_bytes_remaining); if (entry_main) archive_entry_free(entry_main); return (ret); } /* * Format a basic 512-byte "v7tar" header. * * Returns -1 if format failed (due to field overflow). * Note that this always formats as much of the header as possible. * If "strict" is set to zero, it will extend numeric fields as * necessary (overwriting terminators or using base-256 extensions). * */ static int format_header_v7tar(struct archive_write *a, char h[512], struct archive_entry *entry, int strict, struct archive_string_conv *sconv) { unsigned int checksum; int i, r, ret; size_t copy_length; const char *p, *pp; int mytartype; ret = 0; mytartype = -1; /* * The "template header" already includes the "v7tar" * signature, various end-of-field markers and other required * elements. */ memcpy(h, &template_header, 512); /* * Because the block is already null-filled, and strings * are allowed to exactly fill their destination (without null), * I use memcpy(dest, src, strlen()) here a lot to copy strings. */ r = archive_entry_pathname_l(entry, &pp, ©_length, sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate pathname '%s' to %s", pp, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; } if (strict && copy_length < V7TAR_name_size) memcpy(h + V7TAR_name_offset, pp, copy_length); else if (!strict && copy_length <= V7TAR_name_size) memcpy(h + V7TAR_name_offset, pp, copy_length); else { /* Prefix is too long. */ archive_set_error(&a->archive, ENAMETOOLONG, "Pathname too long"); ret = ARCHIVE_FAILED; } r = archive_entry_hardlink_l(entry, &p, ©_length, sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate linkname '%s' to %s", p, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; } if (copy_length > 0) mytartype = '1'; else { r = archive_entry_symlink_l(entry, &p, ©_length, sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate linkname '%s' to %s", p, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; } } if (copy_length > 0) { if (copy_length >= V7TAR_linkname_size) { archive_set_error(&a->archive, ENAMETOOLONG, "Link contents too long"); ret = ARCHIVE_FAILED; copy_length = V7TAR_linkname_size; } memcpy(h + V7TAR_linkname_offset, p, copy_length); } if (format_number(archive_entry_mode(entry) & 07777, h + V7TAR_mode_offset, V7TAR_mode_size, V7TAR_mode_max_size, strict)) { archive_set_error(&a->archive, ERANGE, "Numeric mode too large"); ret = ARCHIVE_FAILED; } if (format_number(archive_entry_uid(entry), h + V7TAR_uid_offset, V7TAR_uid_size, V7TAR_uid_max_size, strict)) { archive_set_error(&a->archive, ERANGE, "Numeric user ID too large"); ret = ARCHIVE_FAILED; } if (format_number(archive_entry_gid(entry), h + V7TAR_gid_offset, V7TAR_gid_size, V7TAR_gid_max_size, strict)) { archive_set_error(&a->archive, ERANGE, "Numeric group ID too large"); ret = ARCHIVE_FAILED; } if (format_number(archive_entry_size(entry), h + V7TAR_size_offset, V7TAR_size_size, V7TAR_size_max_size, strict)) { archive_set_error(&a->archive, ERANGE, "File size out of range"); ret = ARCHIVE_FAILED; } if (format_number(archive_entry_mtime(entry), h + V7TAR_mtime_offset, V7TAR_mtime_size, V7TAR_mtime_max_size, strict)) { archive_set_error(&a->archive, ERANGE, "File modification time too large"); ret = ARCHIVE_FAILED; } if (mytartype >= 0) { h[V7TAR_typeflag_offset] = mytartype; } else { switch (archive_entry_filetype(entry)) { case AE_IFREG: case AE_IFDIR: break; case AE_IFLNK: h[V7TAR_typeflag_offset] = '2'; break; case AE_IFCHR: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive character device"); return (ARCHIVE_FAILED); case AE_IFBLK: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive block device"); return (ARCHIVE_FAILED); case AE_IFIFO: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive fifo"); return (ARCHIVE_FAILED); case AE_IFSOCK: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive socket"); return (ARCHIVE_FAILED); default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive this (mode=0%lo)", (unsigned long)archive_entry_mode(entry)); ret = ARCHIVE_FAILED; } } checksum = 0; for (i = 0; i < 512; i++) checksum += 255 & (unsigned int)h[i]; format_octal(checksum, h + V7TAR_checksum_offset, 6); /* Can't be pre-set in the template. */ h[V7TAR_checksum_offset + 6] = '\0'; return (ret); } /* * Format a number into a field, with some intelligence. */ static int format_number(int64_t v, char *p, int s, int maxsize, int strict) { int64_t limit; limit = ((int64_t)1 << (s*3)); /* "Strict" only permits octal values with proper termination. */ if (strict) return (format_octal(v, p, s)); /* * In non-strict mode, we allow the number to overwrite one or * more bytes of the field termination. Even old tar * implementations should be able to handle this with no * problem. */ if (v >= 0) { while (s <= maxsize) { if (v < limit) return (format_octal(v, p, s)); s++; limit <<= 3; } } /* Base-256 can handle any number, positive or negative. */ return (format_256(v, p, maxsize)); } /* * Format a number into the specified field using base-256. */ static int format_256(int64_t v, char *p, int s) { p += s; while (s-- > 0) { *--p = (char)(v & 0xff); v >>= 8; } *p |= 0x80; /* Set the base-256 marker bit. */ return (0); } /* * Format a number into the specified field. */ static int format_octal(int64_t v, char *p, int s) { int len; len = s; /* Octal values can't be negative, so use 0. */ if (v < 0) { while (len-- > 0) *p++ = '0'; return (-1); } p += s; /* Start at the end and work backwards. */ while (s-- > 0) { *--p = (char)('0' + (v & 7)); v >>= 3; } if (v == 0) return (0); /* If it overflowed, fill field with max value. */ while (len-- > 0) *p++ = '7'; return (-1); } static int archive_write_v7tar_close(struct archive_write *a) { return (__archive_write_nulls(a, 512*2)); } static int archive_write_v7tar_free(struct archive_write *a) { struct v7tar *v7tar; v7tar = (struct v7tar *)a->format_data; free(v7tar); a->format_data = NULL; return (ARCHIVE_OK); } static int archive_write_v7tar_finish_entry(struct archive_write *a) { struct v7tar *v7tar; int ret; v7tar = (struct v7tar *)a->format_data; ret = __archive_write_nulls(a, (size_t)(v7tar->entry_bytes_remaining + v7tar->entry_padding)); v7tar->entry_bytes_remaining = v7tar->entry_padding = 0; return (ret); } static ssize_t archive_write_v7tar_data(struct archive_write *a, const void *buff, size_t s) { struct v7tar *v7tar; int ret; v7tar = (struct v7tar *)a->format_data; if (s > v7tar->entry_bytes_remaining) s = (size_t)v7tar->entry_bytes_remaining; ret = __archive_write_output(a, buff, s); v7tar->entry_bytes_remaining -= s; if (ret != ARCHIVE_OK) return (ret); return (s); } Index: head/contrib/libarchive/libarchive/archive_write_set_format_zip.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_zip.c (revision 309589) +++ head/contrib/libarchive/libarchive/archive_write_set_format_zip.c (revision 309590) @@ -1,1678 +1,1678 @@ /*- * Copyright (c) 2008 Anselm Strauss * Copyright (c) 2009 Joerg Sonnenberger * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /* * Development supported by Google Summer of Code 2008. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LANGINFO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #include "archive.h" #include "archive_cryptor_private.h" #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_hmac_private.h" #include "archive_private.h" #include "archive_random_private.h" #include "archive_write_private.h" #ifndef HAVE_ZLIB_H #include "archive_crc32.h" #endif #define ZIP_ENTRY_FLAG_ENCRYPTED (1<<0) #define ZIP_ENTRY_FLAG_LENGTH_AT_END (1<<3) #define ZIP_ENTRY_FLAG_UTF8_NAME (1 << 11) #define ZIP_4GB_MAX ARCHIVE_LITERAL_LL(0xffffffff) #define ZIP_4GB_MAX_UNCOMPRESSED ARCHIVE_LITERAL_LL(0xff000000) enum compression { COMPRESSION_UNSPECIFIED = -1, COMPRESSION_STORE = 0, COMPRESSION_DEFLATE = 8 }; #ifdef HAVE_ZLIB_H #define COMPRESSION_DEFAULT COMPRESSION_DEFLATE #else #define COMPRESSION_DEFAULT COMPRESSION_STORE #endif enum encryption { ENCRYPTION_NONE = 0, ENCRYPTION_TRADITIONAL, /* Traditional PKWARE encryption. */ ENCRYPTION_WINZIP_AES128, /* WinZIP AES-128 encryption. */ ENCRYPTION_WINZIP_AES256, /* WinZIP AES-256 encryption. */ }; #define TRAD_HEADER_SIZE 12 /* * See "WinZip - AES Encryption Information" * http://www.winzip.com/aes_info.htm */ /* Value used in compression method. */ #define WINZIP_AES_ENCRYPTION 99 /* A WinZip AES header size which is stored at the beginning of * file contents. */ #define WINZIP_AES128_HEADER_SIZE (8 + 2) #define WINZIP_AES256_HEADER_SIZE (16 + 2) /* AES vendor version. */ #define AES_VENDOR_AE_1 0x0001 #define AES_VENDOR_AE_2 0x0002 /* Authentication code size. */ #define AUTH_CODE_SIZE 10 /**/ #define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) struct cd_segment { struct cd_segment *next; size_t buff_size; unsigned char *buff; unsigned char *p; }; struct trad_enc_ctx { uint32_t keys[3]; }; struct zip { int64_t entry_offset; int64_t entry_compressed_size; int64_t entry_uncompressed_size; int64_t entry_compressed_written; int64_t entry_uncompressed_written; int64_t entry_uncompressed_limit; struct archive_entry *entry; uint32_t entry_crc32; enum compression entry_compression; enum encryption entry_encryption; int entry_flags; int entry_uses_zip64; int experiments; struct trad_enc_ctx tctx; char tctx_valid; unsigned char trad_chkdat; unsigned aes_vendor; archive_crypto_ctx cctx; char cctx_valid; archive_hmac_sha1_ctx hctx; char hctx_valid; unsigned char *file_header; size_t file_header_extra_offset; unsigned long (*crc32func)(unsigned long crc, const void *buff, size_t len); struct cd_segment *central_directory; struct cd_segment *central_directory_last; size_t central_directory_bytes; size_t central_directory_entries; int64_t written_bytes; /* Overall position in file. */ struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv_default; enum compression requested_compression; int deflate_compression_level; int init_default_conversion; enum encryption encryption_type; #define ZIP_FLAG_AVOID_ZIP64 1 #define ZIP_FLAG_FORCE_ZIP64 2 #define ZIP_FLAG_EXPERIMENT_xl 4 int flags; #ifdef HAVE_ZLIB_H z_stream stream; #endif size_t len_buf; unsigned char *buf; }; /* Don't call this min or MIN, since those are already defined on lots of platforms (but not all). */ #define zipmin(a, b) ((a) > (b) ? (b) : (a)) static ssize_t archive_write_zip_data(struct archive_write *, const void *buff, size_t s); static int archive_write_zip_close(struct archive_write *); static int archive_write_zip_free(struct archive_write *); static int archive_write_zip_finish_entry(struct archive_write *); static int archive_write_zip_header(struct archive_write *, struct archive_entry *); static int archive_write_zip_options(struct archive_write *, const char *, const char *); static unsigned int dos_time(const time_t); static size_t path_length(struct archive_entry *); static int write_path(struct archive_entry *, struct archive_write *); static void copy_path(struct archive_entry *, unsigned char *); static struct archive_string_conv *get_sconv(struct archive_write *, struct zip *); static int trad_enc_init(struct trad_enc_ctx *, const char *, size_t); static unsigned trad_enc_encrypt_update(struct trad_enc_ctx *, const uint8_t *, size_t, uint8_t *, size_t); static int init_traditional_pkware_encryption(struct archive_write *); static int is_traditional_pkware_encryption_supported(void); static int init_winzip_aes_encryption(struct archive_write *); static int is_winzip_aes_encryption_supported(int encryption); static unsigned char * cd_alloc(struct zip *zip, size_t length) { unsigned char *p; if (zip->central_directory == NULL || (zip->central_directory_last->p + length > zip->central_directory_last->buff + zip->central_directory_last->buff_size)) { struct cd_segment *segment = calloc(1, sizeof(*segment)); if (segment == NULL) return NULL; segment->buff_size = 64 * 1024; segment->buff = malloc(segment->buff_size); if (segment->buff == NULL) { free(segment); return NULL; } segment->p = segment->buff; if (zip->central_directory == NULL) { zip->central_directory = zip->central_directory_last = segment; } else { zip->central_directory_last->next = segment; zip->central_directory_last = segment; } } p = zip->central_directory_last->p; zip->central_directory_last->p += length; zip->central_directory_bytes += length; return (p); } static unsigned long real_crc32(unsigned long crc, const void *buff, size_t len) { return crc32(crc, buff, (unsigned int)len); } static unsigned long fake_crc32(unsigned long crc, const void *buff, size_t len) { (void)crc; /* UNUSED */ (void)buff; /* UNUSED */ (void)len; /* UNUSED */ return 0; } static int archive_write_zip_options(struct archive_write *a, const char *key, const char *val) { struct zip *zip = a->format_data; int ret = ARCHIVE_FAILED; if (strcmp(key, "compression") == 0) { /* * Set compression to use on all future entries. * This only affects regular files. */ if (val == NULL || val[0] == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: compression option needs a compression name", a->format_name); } else if (strcmp(val, "deflate") == 0) { #ifdef HAVE_ZLIB_H zip->requested_compression = COMPRESSION_DEFLATE; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "deflate compression not supported"); #endif } else if (strcmp(val, "store") == 0) { zip->requested_compression = COMPRESSION_STORE; ret = ARCHIVE_OK; } return (ret); } else if (strcmp(key, "compression-level") == 0) { if (val == NULL || !(val[0] >= '0' && val[0] <= '9') || val[1] != '\0') { return ARCHIVE_WARN; } if (val[0] == '0') { zip->requested_compression = COMPRESSION_STORE; return ARCHIVE_OK; } else { #ifdef HAVE_ZLIB_H zip->requested_compression = COMPRESSION_DEFLATE; zip->deflate_compression_level = val[0] - '0'; return ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "deflate compression not supported"); #endif } } else if (strcmp(key, "encryption") == 0) { if (val == NULL) { zip->encryption_type = ENCRYPTION_NONE; ret = ARCHIVE_OK; } else if (val[0] == '1' || strcmp(val, "traditional") == 0 || strcmp(val, "zipcrypt") == 0 || strcmp(val, "ZipCrypt") == 0) { if (is_traditional_pkware_encryption_supported()) { zip->encryption_type = ENCRYPTION_TRADITIONAL; ret = ARCHIVE_OK; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "encryption not supported"); } } else if (strcmp(val, "aes128") == 0) { if (is_winzip_aes_encryption_supported( ENCRYPTION_WINZIP_AES128)) { zip->encryption_type = ENCRYPTION_WINZIP_AES128; ret = ARCHIVE_OK; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "encryption not supported"); } } else if (strcmp(val, "aes256") == 0) { if (is_winzip_aes_encryption_supported( ENCRYPTION_WINZIP_AES256)) { zip->encryption_type = ENCRYPTION_WINZIP_AES256; ret = ARCHIVE_OK; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "encryption not supported"); } } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: unknown encryption '%s'", a->format_name, val); } return (ret); } else if (strcmp(key, "experimental") == 0) { if (val == NULL || val[0] == 0) { zip->flags &= ~ ZIP_FLAG_EXPERIMENT_xl; } else { zip->flags |= ZIP_FLAG_EXPERIMENT_xl; } return (ARCHIVE_OK); } else if (strcmp(key, "fakecrc32") == 0) { /* * FOR TESTING ONLY: disable CRC calculation to speed up * certain complex tests. */ if (val == NULL || val[0] == 0) { zip->crc32func = real_crc32; } else { zip->crc32func = fake_crc32; } return (ARCHIVE_OK); } else if (strcmp(key, "hdrcharset") == 0) { /* * Set the character set used in translating filenames. */ if (val == NULL || val[0] == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: hdrcharset option needs a character-set name", a->format_name); } else { zip->opt_sconv = archive_string_conversion_to_charset( &a->archive, val, 0); if (zip->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } else if (strcmp(key, "zip64") == 0) { /* * Bias decisions about Zip64: force them to be * generated in certain cases where they are not * forbidden or avoid them in certain cases where they * are not strictly required. */ if (val != NULL && *val != '\0') { zip->flags |= ZIP_FLAG_FORCE_ZIP64; zip->flags &= ~ZIP_FLAG_AVOID_ZIP64; } else { zip->flags &= ~ZIP_FLAG_FORCE_ZIP64; zip->flags |= ZIP_FLAG_AVOID_ZIP64; } return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } int archive_write_zip_set_compression_deflate(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int ret = ARCHIVE_FAILED; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_zip_set_compression_deflate"); if (a->archive.archive_format != ARCHIVE_FORMAT_ZIP) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can only use archive_write_zip_set_compression_deflate" " with zip format"); ret = ARCHIVE_FATAL; } else { #ifdef HAVE_ZLIB_H struct zip *zip = a->format_data; zip->requested_compression = COMPRESSION_DEFLATE; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "deflate compression not supported"); ret = ARCHIVE_FAILED; #endif } return (ret); } int archive_write_zip_set_compression_store(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct zip *zip = a->format_data; int ret = ARCHIVE_FAILED; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_zip_set_compression_deflate"); if (a->archive.archive_format != ARCHIVE_FORMAT_ZIP) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can only use archive_write_zip_set_compression_store" " with zip format"); ret = ARCHIVE_FATAL; } else { zip->requested_compression = COMPRESSION_STORE; ret = ARCHIVE_OK; } return (ret); } int archive_write_set_format_zip(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct zip *zip; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_zip"); /* If another format was already registered, unregister it. */ if (a->format_free != NULL) (a->format_free)(a); zip = (struct zip *) calloc(1, sizeof(*zip)); if (zip == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip data"); return (ARCHIVE_FATAL); } /* "Unspecified" lets us choose the appropriate compression. */ zip->requested_compression = COMPRESSION_UNSPECIFIED; #ifdef HAVE_ZLIB_H zip->deflate_compression_level = Z_DEFAULT_COMPRESSION; #endif zip->crc32func = real_crc32; /* A buffer used for both compression and encryption. */ zip->len_buf = 65536; zip->buf = malloc(zip->len_buf); if (zip->buf == NULL) { free(zip); archive_set_error(&a->archive, ENOMEM, "Can't allocate compression buffer"); return (ARCHIVE_FATAL); } a->format_data = zip; a->format_name = "zip"; a->format_options = archive_write_zip_options; a->format_write_header = archive_write_zip_header; a->format_write_data = archive_write_zip_data; a->format_finish_entry = archive_write_zip_finish_entry; a->format_close = archive_write_zip_close; a->format_free = archive_write_zip_free; a->archive.archive_format = ARCHIVE_FORMAT_ZIP; a->archive.archive_format_name = "ZIP"; return (ARCHIVE_OK); } static int is_all_ascii(const char *p) { const unsigned char *pp = (const unsigned char *)p; while (*pp) { if (*pp++ > 127) return (0); } return (1); } static int archive_write_zip_header(struct archive_write *a, struct archive_entry *entry) { unsigned char local_header[32]; unsigned char local_extra[144]; struct zip *zip = a->format_data; unsigned char *e; unsigned char *cd_extra; size_t filename_length; const char *slink = NULL; size_t slink_size = 0; struct archive_string_conv *sconv = get_sconv(a, zip); int ret, ret2 = ARCHIVE_OK; mode_t type; int version_needed = 10; /* Ignore types of entries that we don't support. */ type = archive_entry_filetype(entry); if (type != AE_IFREG && type != AE_IFDIR && type != AE_IFLNK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Filetype not supported"); return ARCHIVE_FAILED; }; /* If we're not using Zip64, reject large files. */ if (zip->flags & ZIP_FLAG_AVOID_ZIP64) { /* Reject entries over 4GB. */ if (archive_entry_size_is_set(entry) && (archive_entry_size(entry) > ZIP_4GB_MAX)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Files > 4GB require Zip64 extensions"); return ARCHIVE_FAILED; } /* Reject entries if archive is > 4GB. */ if (zip->written_bytes > ZIP_4GB_MAX) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Archives > 4GB require Zip64 extensions"); return ARCHIVE_FAILED; } } /* Only regular files can have size > 0. */ if (type != AE_IFREG) archive_entry_set_size(entry, 0); /* Reset information from last entry. */ zip->entry_offset = zip->written_bytes; zip->entry_uncompressed_limit = INT64_MAX; zip->entry_compressed_size = 0; zip->entry_uncompressed_size = 0; zip->entry_compressed_written = 0; zip->entry_uncompressed_written = 0; zip->entry_flags = 0; zip->entry_uses_zip64 = 0; zip->entry_crc32 = zip->crc32func(0, NULL, 0); zip->entry_encryption = 0; if (zip->entry != NULL) { archive_entry_free(zip->entry); zip->entry = NULL; } if (zip->cctx_valid) archive_encrypto_aes_ctr_release(&zip->cctx); if (zip->hctx_valid) archive_hmac_sha1_cleanup(&zip->hctx); zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; if (type == AE_IFREG &&(!archive_entry_size_is_set(entry) || archive_entry_size(entry) > 0)) { switch (zip->encryption_type) { case ENCRYPTION_TRADITIONAL: case ENCRYPTION_WINZIP_AES128: case ENCRYPTION_WINZIP_AES256: zip->entry_flags |= ZIP_ENTRY_FLAG_ENCRYPTED; zip->entry_encryption = zip->encryption_type; break; default: break; } } #if defined(_WIN32) && !defined(__CYGWIN__) - /* Make sure the path separators in pahtname, hardlink and symlink + /* Make sure the path separators in pathname, hardlink and symlink * are all slash '/', not the Windows path separator '\'. */ zip->entry = __la_win_entry_in_posix_pathseparator(entry); if (zip->entry == entry) zip->entry = archive_entry_clone(entry); #else zip->entry = archive_entry_clone(entry); #endif if (zip->entry == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip header data"); return (ARCHIVE_FATAL); } if (sconv != NULL) { const char *p; size_t len; if (archive_entry_pathname_l(entry, &p, &len, sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate Pathname '%s' to %s", archive_entry_pathname(entry), archive_string_conversion_charset_name(sconv)); ret2 = ARCHIVE_WARN; } if (len > 0) archive_entry_set_pathname(zip->entry, p); /* * There is no standard for symlink handling; we convert * it using the same character-set translation that we use * for filename. */ if (type == AE_IFLNK) { if (archive_entry_symlink_l(entry, &p, &len, sconv)) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory " " for Symlink"); return (ARCHIVE_FATAL); } /* No error if we can't convert. */ } else if (len > 0) archive_entry_set_symlink(zip->entry, p); } } /* If filename isn't ASCII and we can use UTF-8, set the UTF-8 flag. */ if (!is_all_ascii(archive_entry_pathname(zip->entry))) { if (zip->opt_sconv != NULL) { if (strcmp(archive_string_conversion_charset_name( zip->opt_sconv), "UTF-8") == 0) zip->entry_flags |= ZIP_ENTRY_FLAG_UTF8_NAME; #if HAVE_NL_LANGINFO } else if (strcmp(nl_langinfo(CODESET), "UTF-8") == 0) { zip->entry_flags |= ZIP_ENTRY_FLAG_UTF8_NAME; #endif } } filename_length = path_length(zip->entry); /* Determine appropriate compression and size for this entry. */ if (type == AE_IFLNK) { slink = archive_entry_symlink(zip->entry); if (slink != NULL) slink_size = strlen(slink); else slink_size = 0; zip->entry_uncompressed_limit = slink_size; zip->entry_compressed_size = slink_size; zip->entry_uncompressed_size = slink_size; zip->entry_crc32 = zip->crc32func(zip->entry_crc32, (const unsigned char *)slink, slink_size); zip->entry_compression = COMPRESSION_STORE; version_needed = 20; } else if (type != AE_IFREG) { zip->entry_compression = COMPRESSION_STORE; zip->entry_uncompressed_limit = 0; version_needed = 20; } else if (archive_entry_size_is_set(zip->entry)) { int64_t size = archive_entry_size(zip->entry); int64_t additional_size = 0; zip->entry_uncompressed_limit = size; zip->entry_compression = zip->requested_compression; if (zip->entry_compression == COMPRESSION_UNSPECIFIED) { zip->entry_compression = COMPRESSION_DEFAULT; } if (zip->entry_compression == COMPRESSION_STORE) { zip->entry_compressed_size = size; zip->entry_uncompressed_size = size; version_needed = 10; } else { zip->entry_uncompressed_size = size; version_needed = 20; } if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { switch (zip->entry_encryption) { case ENCRYPTION_TRADITIONAL: additional_size = TRAD_HEADER_SIZE; version_needed = 20; break; case ENCRYPTION_WINZIP_AES128: additional_size = WINZIP_AES128_HEADER_SIZE + AUTH_CODE_SIZE; version_needed = 20; break; case ENCRYPTION_WINZIP_AES256: additional_size = WINZIP_AES256_HEADER_SIZE + AUTH_CODE_SIZE; version_needed = 20; break; default: break; } if (zip->entry_compression == COMPRESSION_STORE) zip->entry_compressed_size += additional_size; } /* * Set Zip64 extension in any of the following cases * (this was suggested by discussion on info-zip-dev * mailing list): * = Zip64 is being forced by user * = File is over 4GiB uncompressed * (including encryption header, if any) * = File is close to 4GiB and is being compressed * (compression might make file larger) */ if ((zip->flags & ZIP_FLAG_FORCE_ZIP64) || (zip->entry_uncompressed_size + additional_size > ZIP_4GB_MAX) || (zip->entry_uncompressed_size > ZIP_4GB_MAX_UNCOMPRESSED && zip->entry_compression != COMPRESSION_STORE)) { zip->entry_uses_zip64 = 1; version_needed = 45; } /* We may know the size, but never the CRC. */ zip->entry_flags |= ZIP_ENTRY_FLAG_LENGTH_AT_END; } else { /* We don't know the size. In this case, we prefer * deflate (it has a clear end-of-data marker which * makes length-at-end more reliable) and will * enable Zip64 extensions unless we're told not to. */ zip->entry_compression = COMPRESSION_DEFAULT; zip->entry_flags |= ZIP_ENTRY_FLAG_LENGTH_AT_END; if ((zip->flags & ZIP_FLAG_AVOID_ZIP64) == 0) { zip->entry_uses_zip64 = 1; version_needed = 45; } else if (zip->entry_compression == COMPRESSION_STORE) { version_needed = 10; } else { version_needed = 20; } if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { switch (zip->entry_encryption) { case ENCRYPTION_TRADITIONAL: case ENCRYPTION_WINZIP_AES128: case ENCRYPTION_WINZIP_AES256: if (version_needed < 20) version_needed = 20; break; default: break; } } } /* Format the local header. */ memset(local_header, 0, sizeof(local_header)); memcpy(local_header, "PK\003\004", 4); archive_le16enc(local_header + 4, version_needed); archive_le16enc(local_header + 6, zip->entry_flags); if (zip->entry_encryption == ENCRYPTION_WINZIP_AES128 || zip->entry_encryption == ENCRYPTION_WINZIP_AES256) archive_le16enc(local_header + 8, WINZIP_AES_ENCRYPTION); else archive_le16enc(local_header + 8, zip->entry_compression); archive_le32enc(local_header + 10, dos_time(archive_entry_mtime(zip->entry))); archive_le32enc(local_header + 14, zip->entry_crc32); if (zip->entry_uses_zip64) { /* Zip64 data in the local header "must" include both * compressed and uncompressed sizes AND those fields * are included only if these are 0xffffffff; * THEREFORE these must be set this way, even if we * know one of them is smaller. */ archive_le32enc(local_header + 18, ZIP_4GB_MAX); archive_le32enc(local_header + 22, ZIP_4GB_MAX); } else { archive_le32enc(local_header + 18, (uint32_t)zip->entry_compressed_size); archive_le32enc(local_header + 22, (uint32_t)zip->entry_uncompressed_size); } archive_le16enc(local_header + 26, (uint16_t)filename_length); if (zip->entry_encryption == ENCRYPTION_TRADITIONAL) { if (zip->entry_flags & ZIP_ENTRY_FLAG_LENGTH_AT_END) zip->trad_chkdat = local_header[11]; else zip->trad_chkdat = local_header[17]; } /* Format as much of central directory file header as we can: */ zip->file_header = cd_alloc(zip, 46); /* If (zip->file_header == NULL) XXXX */ ++zip->central_directory_entries; memset(zip->file_header, 0, 46); memcpy(zip->file_header, "PK\001\002", 4); /* "Made by PKZip 2.0 on Unix." */ archive_le16enc(zip->file_header + 4, 3 * 256 + version_needed); archive_le16enc(zip->file_header + 6, version_needed); archive_le16enc(zip->file_header + 8, zip->entry_flags); if (zip->entry_encryption == ENCRYPTION_WINZIP_AES128 || zip->entry_encryption == ENCRYPTION_WINZIP_AES256) archive_le16enc(zip->file_header + 10, WINZIP_AES_ENCRYPTION); else archive_le16enc(zip->file_header + 10, zip->entry_compression); archive_le32enc(zip->file_header + 12, dos_time(archive_entry_mtime(zip->entry))); archive_le16enc(zip->file_header + 28, (uint16_t)filename_length); /* Following Info-Zip, store mode in the "external attributes" field. */ archive_le32enc(zip->file_header + 38, ((uint32_t)archive_entry_mode(zip->entry)) << 16); e = cd_alloc(zip, filename_length); /* If (e == NULL) XXXX */ copy_path(zip->entry, e); /* Format extra data. */ memset(local_extra, 0, sizeof(local_extra)); e = local_extra; /* First, extra blocks that are the same between * the local file header and the central directory. * We format them once and then duplicate them. */ /* UT timestamp, length depends on what timestamps are set. */ memcpy(e, "UT", 2); archive_le16enc(e + 2, 1 + (archive_entry_mtime_is_set(entry) ? 4 : 0) + (archive_entry_atime_is_set(entry) ? 4 : 0) + (archive_entry_ctime_is_set(entry) ? 4 : 0)); e += 4; *e++ = (archive_entry_mtime_is_set(entry) ? 1 : 0) | (archive_entry_atime_is_set(entry) ? 2 : 0) | (archive_entry_ctime_is_set(entry) ? 4 : 0); if (archive_entry_mtime_is_set(entry)) { archive_le32enc(e, (uint32_t)archive_entry_mtime(entry)); e += 4; } if (archive_entry_atime_is_set(entry)) { archive_le32enc(e, (uint32_t)archive_entry_atime(entry)); e += 4; } if (archive_entry_ctime_is_set(entry)) { archive_le32enc(e, (uint32_t)archive_entry_ctime(entry)); e += 4; } /* ux Unix extra data, length 11, version 1 */ /* TODO: If uid < 64k, use 2 bytes, ditto for gid. */ memcpy(e, "ux\013\000\001", 5); e += 5; *e++ = 4; /* Length of following UID */ archive_le32enc(e, (uint32_t)archive_entry_uid(entry)); e += 4; *e++ = 4; /* Length of following GID */ archive_le32enc(e, (uint32_t)archive_entry_gid(entry)); e += 4; /* AES extra data field: WinZIP AES information, ID=0x9901 */ if ((zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) && (zip->entry_encryption == ENCRYPTION_WINZIP_AES128 || zip->entry_encryption == ENCRYPTION_WINZIP_AES256)) { memcpy(e, "\001\231\007\000\001\000AE", 8); /* AES vendoer version AE-2 does not store a CRC. * WinZip 11 uses AE-1, which does store the CRC, * but it does not store the CRC when the file size * is less than 20 bytes. So we simulate what * WinZip 11 does. * NOTE: WinZip 9.0 and 10.0 uses AE-2 by default. */ if (archive_entry_size_is_set(zip->entry) && archive_entry_size(zip->entry) < 20) { archive_le16enc(e+4, AES_VENDOR_AE_2); zip->aes_vendor = AES_VENDOR_AE_2;/* no CRC. */ } else zip->aes_vendor = AES_VENDOR_AE_1; e += 8; /* AES encryption strength. */ *e++ = (zip->entry_encryption == ENCRYPTION_WINZIP_AES128)?1:3; /* Actual compression method. */ archive_le16enc(e, zip->entry_compression); e += 2; } /* Copy UT ,ux, and AES-extra into central directory as well. */ zip->file_header_extra_offset = zip->central_directory_bytes; cd_extra = cd_alloc(zip, e - local_extra); memcpy(cd_extra, local_extra, e - local_extra); /* * Following extra blocks vary between local header and * central directory. These are the local header versions. * Central directory versions get formatted in * archive_write_zip_finish_entry() below. */ /* "[Zip64 entry] in the local header MUST include BOTH * original [uncompressed] and compressed size fields." */ if (zip->entry_uses_zip64) { unsigned char *zip64_start = e; memcpy(e, "\001\000\020\000", 4); e += 4; archive_le64enc(e, zip->entry_uncompressed_size); e += 8; archive_le64enc(e, zip->entry_compressed_size); e += 8; archive_le16enc(zip64_start + 2, (uint16_t)(e - (zip64_start + 4))); } if (zip->flags & ZIP_FLAG_EXPERIMENT_xl) { /* Experimental 'xl' extension to improve streaming. */ unsigned char *external_info = e; int included = 7; memcpy(e, "xl\000\000", 4); // 0x6c65 + 2-byte length e += 4; e[0] = included; /* bitmap of included fields */ e += 1; if (included & 1) { archive_le16enc(e, /* "Version created by" */ 3 * 256 + version_needed); e += 2; } if (included & 2) { archive_le16enc(e, 0); /* internal file attributes */ e += 2; } if (included & 4) { archive_le32enc(e, /* external file attributes */ ((uint32_t)archive_entry_mode(zip->entry)) << 16); e += 4; } if (included & 8) { // Libarchive does not currently support file comments. } archive_le16enc(external_info + 2, (uint16_t)(e - (external_info + 4))); } /* Update local header with size of extra data and write it all out: */ archive_le16enc(local_header + 28, (uint16_t)(e - local_extra)); ret = __archive_write_output(a, local_header, 30); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += 30; ret = write_path(zip->entry, a); if (ret <= ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += ret; ret = __archive_write_output(a, local_extra, e - local_extra); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += e - local_extra; /* For symlinks, write the body now. */ if (slink != NULL) { ret = __archive_write_output(a, slink, slink_size); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->entry_compressed_written += slink_size; zip->entry_uncompressed_written += slink_size; zip->written_bytes += slink_size; } #ifdef HAVE_ZLIB_H if (zip->entry_compression == COMPRESSION_DEFLATE) { zip->stream.zalloc = Z_NULL; zip->stream.zfree = Z_NULL; zip->stream.opaque = Z_NULL; zip->stream.next_out = zip->buf; zip->stream.avail_out = (uInt)zip->len_buf; if (deflateInit2(&zip->stream, zip->deflate_compression_level, Z_DEFLATED, -15, 8, Z_DEFAULT_STRATEGY) != Z_OK) { archive_set_error(&a->archive, ENOMEM, "Can't init deflate compressor"); return (ARCHIVE_FATAL); } } #endif return (ret2); } static ssize_t archive_write_zip_data(struct archive_write *a, const void *buff, size_t s) { int ret; struct zip *zip = a->format_data; if ((int64_t)s > zip->entry_uncompressed_limit) s = (size_t)zip->entry_uncompressed_limit; zip->entry_uncompressed_written += s; if (s == 0) return 0; if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { switch (zip->entry_encryption) { case ENCRYPTION_TRADITIONAL: /* Initialize traditoinal PKWARE encryption context. */ if (!zip->tctx_valid) { ret = init_traditional_pkware_encryption(a); if (ret != ARCHIVE_OK) return (ret); zip->tctx_valid = 1; } break; case ENCRYPTION_WINZIP_AES128: case ENCRYPTION_WINZIP_AES256: if (!zip->cctx_valid) { ret = init_winzip_aes_encryption(a); if (ret != ARCHIVE_OK) return (ret); zip->cctx_valid = zip->hctx_valid = 1; } break; default: break; } } switch (zip->entry_compression) { case COMPRESSION_STORE: if (zip->tctx_valid || zip->cctx_valid) { const uint8_t *rb = (const uint8_t *)buff; const uint8_t * const re = rb + s; while (rb < re) { size_t l; if (zip->tctx_valid) { l = trad_enc_encrypt_update(&zip->tctx, rb, re - rb, zip->buf, zip->len_buf); } else { l = zip->len_buf; ret = archive_encrypto_aes_ctr_update( &zip->cctx, rb, re - rb, zip->buf, &l); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, l); } ret = __archive_write_output(a, zip->buf, l); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += l; zip->written_bytes += l; rb += l; } } else { ret = __archive_write_output(a, buff, s); if (ret != ARCHIVE_OK) return (ret); zip->written_bytes += s; zip->entry_compressed_written += s; } break; #if HAVE_ZLIB_H case COMPRESSION_DEFLATE: zip->stream.next_in = (unsigned char*)(uintptr_t)buff; zip->stream.avail_in = (uInt)s; do { ret = deflate(&zip->stream, Z_NO_FLUSH); if (ret == Z_STREAM_ERROR) return (ARCHIVE_FATAL); if (zip->stream.avail_out == 0) { if (zip->tctx_valid) { trad_enc_encrypt_update(&zip->tctx, zip->buf, zip->len_buf, zip->buf, zip->len_buf); } else if (zip->cctx_valid) { size_t outl = zip->len_buf; ret = archive_encrypto_aes_ctr_update( &zip->cctx, zip->buf, zip->len_buf, zip->buf, &outl); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, zip->len_buf); } ret = __archive_write_output(a, zip->buf, zip->len_buf); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += zip->len_buf; zip->written_bytes += zip->len_buf; zip->stream.next_out = zip->buf; zip->stream.avail_out = (uInt)zip->len_buf; } } while (zip->stream.avail_in != 0); break; #endif default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid ZIP compression type"); return ARCHIVE_FATAL; } zip->entry_uncompressed_limit -= s; if (!zip->cctx_valid || zip->aes_vendor != AES_VENDOR_AE_2) zip->entry_crc32 = zip->crc32func(zip->entry_crc32, buff, (unsigned)s); return (s); } static int archive_write_zip_finish_entry(struct archive_write *a) { struct zip *zip = a->format_data; int ret; #if HAVE_ZLIB_H if (zip->entry_compression == COMPRESSION_DEFLATE) { for (;;) { size_t remainder; ret = deflate(&zip->stream, Z_FINISH); if (ret == Z_STREAM_ERROR) return (ARCHIVE_FATAL); remainder = zip->len_buf - zip->stream.avail_out; if (zip->tctx_valid) { trad_enc_encrypt_update(&zip->tctx, zip->buf, remainder, zip->buf, remainder); } else if (zip->cctx_valid) { size_t outl = remainder; ret = archive_encrypto_aes_ctr_update( &zip->cctx, zip->buf, remainder, zip->buf, &outl); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, remainder); } ret = __archive_write_output(a, zip->buf, remainder); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += remainder; zip->written_bytes += remainder; zip->stream.next_out = zip->buf; if (zip->stream.avail_out != 0) break; zip->stream.avail_out = (uInt)zip->len_buf; } deflateEnd(&zip->stream); } #endif if (zip->hctx_valid) { uint8_t hmac[20]; size_t hmac_len = 20; archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); ret = __archive_write_output(a, hmac, AUTH_CODE_SIZE); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += AUTH_CODE_SIZE; zip->written_bytes += AUTH_CODE_SIZE; } /* Write trailing data descriptor. */ if ((zip->entry_flags & ZIP_ENTRY_FLAG_LENGTH_AT_END) != 0) { char d[24]; memcpy(d, "PK\007\010", 4); if (zip->cctx_valid && zip->aes_vendor == AES_VENDOR_AE_2) archive_le32enc(d + 4, 0);/* no CRC.*/ else archive_le32enc(d + 4, zip->entry_crc32); if (zip->entry_uses_zip64) { archive_le64enc(d + 8, (uint64_t)zip->entry_compressed_written); archive_le64enc(d + 16, (uint64_t)zip->entry_uncompressed_written); ret = __archive_write_output(a, d, 24); zip->written_bytes += 24; } else { archive_le32enc(d + 8, (uint32_t)zip->entry_compressed_written); archive_le32enc(d + 12, (uint32_t)zip->entry_uncompressed_written); ret = __archive_write_output(a, d, 16); zip->written_bytes += 16; } if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } /* Append Zip64 extra data to central directory information. */ if (zip->entry_compressed_written > ZIP_4GB_MAX || zip->entry_uncompressed_written > ZIP_4GB_MAX || zip->entry_offset > ZIP_4GB_MAX) { unsigned char zip64[32]; unsigned char *z = zip64, *zd; memcpy(z, "\001\000\000\000", 4); z += 4; if (zip->entry_uncompressed_written >= ZIP_4GB_MAX) { archive_le64enc(z, zip->entry_uncompressed_written); z += 8; } if (zip->entry_compressed_written >= ZIP_4GB_MAX) { archive_le64enc(z, zip->entry_compressed_written); z += 8; } if (zip->entry_offset >= ZIP_4GB_MAX) { archive_le64enc(z, zip->entry_offset); z += 8; } archive_le16enc(zip64 + 2, (uint16_t)(z - (zip64 + 4))); zd = cd_alloc(zip, z - zip64); if (zd == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip data"); return (ARCHIVE_FATAL); } memcpy(zd, zip64, z - zip64); /* Zip64 means version needs to be set to at least 4.5 */ if (archive_le16dec(zip->file_header + 6) < 45) archive_le16enc(zip->file_header + 6, 45); } /* Fix up central directory file header. */ if (zip->cctx_valid && zip->aes_vendor == AES_VENDOR_AE_2) archive_le32enc(zip->file_header + 16, 0);/* no CRC.*/ else archive_le32enc(zip->file_header + 16, zip->entry_crc32); archive_le32enc(zip->file_header + 20, (uint32_t)zipmin(zip->entry_compressed_written, ZIP_4GB_MAX)); archive_le32enc(zip->file_header + 24, (uint32_t)zipmin(zip->entry_uncompressed_written, ZIP_4GB_MAX)); archive_le16enc(zip->file_header + 30, (uint16_t)(zip->central_directory_bytes - zip->file_header_extra_offset)); archive_le32enc(zip->file_header + 42, (uint32_t)zipmin(zip->entry_offset, ZIP_4GB_MAX)); return (ARCHIVE_OK); } static int archive_write_zip_close(struct archive_write *a) { uint8_t buff[64]; int64_t offset_start, offset_end; struct zip *zip = a->format_data; struct cd_segment *segment; int ret; offset_start = zip->written_bytes; segment = zip->central_directory; while (segment != NULL) { ret = __archive_write_output(a, segment->buff, segment->p - segment->buff); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += segment->p - segment->buff; segment = segment->next; } offset_end = zip->written_bytes; /* If central dir info is too large, write Zip64 end-of-cd */ if (offset_end - offset_start > ZIP_4GB_MAX || offset_start > ZIP_4GB_MAX || zip->central_directory_entries > 0xffffUL || (zip->flags & ZIP_FLAG_FORCE_ZIP64)) { /* Zip64 end-of-cd record */ memset(buff, 0, 56); memcpy(buff, "PK\006\006", 4); archive_le64enc(buff + 4, 44); archive_le16enc(buff + 12, 45); archive_le16enc(buff + 14, 45); /* This is disk 0 of 0. */ archive_le64enc(buff + 24, zip->central_directory_entries); archive_le64enc(buff + 32, zip->central_directory_entries); archive_le64enc(buff + 40, offset_end - offset_start); archive_le64enc(buff + 48, offset_start); ret = __archive_write_output(a, buff, 56); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += 56; /* Zip64 end-of-cd locator record. */ memset(buff, 0, 20); memcpy(buff, "PK\006\007", 4); archive_le32enc(buff + 4, 0); archive_le64enc(buff + 8, offset_end); archive_le32enc(buff + 16, 1); ret = __archive_write_output(a, buff, 20); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += 20; } /* Format and write end of central directory. */ memset(buff, 0, sizeof(buff)); memcpy(buff, "PK\005\006", 4); archive_le16enc(buff + 8, (uint16_t)zipmin(0xffffU, zip->central_directory_entries)); archive_le16enc(buff + 10, (uint16_t)zipmin(0xffffU, zip->central_directory_entries)); archive_le32enc(buff + 12, (uint32_t)zipmin(ZIP_4GB_MAX, (offset_end - offset_start))); archive_le32enc(buff + 16, (uint32_t)zipmin(ZIP_4GB_MAX, offset_start)); ret = __archive_write_output(a, buff, 22); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += 22; return (ARCHIVE_OK); } static int archive_write_zip_free(struct archive_write *a) { struct zip *zip; struct cd_segment *segment; zip = a->format_data; while (zip->central_directory != NULL) { segment = zip->central_directory; zip->central_directory = segment->next; free(segment->buff); free(segment); } free(zip->buf); archive_entry_free(zip->entry); if (zip->cctx_valid) archive_encrypto_aes_ctr_release(&zip->cctx); if (zip->hctx_valid) archive_hmac_sha1_cleanup(&zip->hctx); /* TODO: Free opt_sconv, sconv_default */ free(zip); a->format_data = NULL; return (ARCHIVE_OK); } /* Convert into MSDOS-style date/time. */ static unsigned int dos_time(const time_t unix_time) { struct tm *t; unsigned int dt; /* This will not preserve time when creating/extracting the archive * on two systems with different time zones. */ t = localtime(&unix_time); /* MSDOS-style date/time is only between 1980-01-01 and 2107-12-31 */ if (t->tm_year < 1980 - 1900) /* Set minimum date/time '1980-01-01 00:00:00'. */ dt = 0x00210000U; else if (t->tm_year > 2107 - 1900) /* Set maximum date/time '2107-12-31 23:59:58'. */ dt = 0xff9fbf7dU; else { dt = 0; dt += ((t->tm_year - 80) & 0x7f) << 9; dt += ((t->tm_mon + 1) & 0x0f) << 5; dt += (t->tm_mday & 0x1f); dt <<= 16; dt += (t->tm_hour & 0x1f) << 11; dt += (t->tm_min & 0x3f) << 5; dt += (t->tm_sec & 0x3e) >> 1; /* Only counting every 2 seconds. */ } return dt; } static size_t path_length(struct archive_entry *entry) { mode_t type; const char *path; type = archive_entry_filetype(entry); path = archive_entry_pathname(entry); if (path == NULL) return (0); if (type == AE_IFDIR && (path[0] == '\0' || path[strlen(path) - 1] != '/')) { return strlen(path) + 1; } else { return strlen(path); } } static int write_path(struct archive_entry *entry, struct archive_write *archive) { int ret; const char *path; mode_t type; size_t written_bytes; path = archive_entry_pathname(entry); type = archive_entry_filetype(entry); written_bytes = 0; ret = __archive_write_output(archive, path, strlen(path)); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); written_bytes += strlen(path); /* Folders are recognized by a trailing slash. */ if ((type == AE_IFDIR) & (path[strlen(path) - 1] != '/')) { ret = __archive_write_output(archive, "/", 1); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); written_bytes += 1; } return ((int)written_bytes); } static void copy_path(struct archive_entry *entry, unsigned char *p) { const char *path; size_t pathlen; mode_t type; path = archive_entry_pathname(entry); pathlen = strlen(path); type = archive_entry_filetype(entry); memcpy(p, path, pathlen); /* Folders are recognized by a trailing slash. */ if ((type == AE_IFDIR) & (path[pathlen - 1] != '/')) { p[pathlen] = '/'; p[pathlen + 1] = '\0'; } } static struct archive_string_conv * get_sconv(struct archive_write *a, struct zip *zip) { if (zip->opt_sconv != NULL) return (zip->opt_sconv); if (!zip->init_default_conversion) { zip->sconv_default = archive_string_default_conversion_for_write(&(a->archive)); zip->init_default_conversion = 1; } return (zip->sconv_default); } /* Traditional PKWARE Decryption functions. */ static void trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) { uint8_t t; #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) ctx->keys[0] = CRC32(ctx->keys[0], c); ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; t = (ctx->keys[1] >> 24) & 0xff; ctx->keys[2] = CRC32(ctx->keys[2], t); #undef CRC32 } static uint8_t trad_enc_decypt_byte(struct trad_enc_ctx *ctx) { unsigned temp = ctx->keys[2] | 2; return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; } static unsigned trad_enc_encrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, size_t in_len, uint8_t *out, size_t out_len) { unsigned i, max; max = (unsigned)((in_len < out_len)? in_len: out_len); for (i = 0; i < max; i++) { uint8_t t = in[i]; out[i] = t ^ trad_enc_decypt_byte(ctx); trad_enc_update_keys(ctx, t); } return i; } static int trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len) { ctx->keys[0] = 305419896L; ctx->keys[1] = 591751049L; ctx->keys[2] = 878082192L; for (;pw_len; --pw_len) trad_enc_update_keys(ctx, *pw++); return 0; } static int is_traditional_pkware_encryption_supported(void) { uint8_t key[TRAD_HEADER_SIZE]; if (archive_random(key, sizeof(key)-1) != ARCHIVE_OK) return (0); return (1); } static int init_traditional_pkware_encryption(struct archive_write *a) { struct zip *zip = a->format_data; const char *passphrase; uint8_t key[TRAD_HEADER_SIZE]; uint8_t key_encrypted[TRAD_HEADER_SIZE]; int ret; passphrase = __archive_write_get_passphrase(a); if (passphrase == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Encryption needs passphrase"); return ARCHIVE_FAILED; } if (archive_random(key, sizeof(key)-1) != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't generate random number for encryption"); return ARCHIVE_FATAL; } trad_enc_init(&zip->tctx, passphrase, strlen(passphrase)); /* Set the last key code which will be used as a check code * for verifying passphrase in decryption. */ key[TRAD_HEADER_SIZE-1] = zip->trad_chkdat; trad_enc_encrypt_update(&zip->tctx, key, TRAD_HEADER_SIZE, key_encrypted, TRAD_HEADER_SIZE); /* Write encrypted keys in the top of the file content. */ ret = __archive_write_output(a, key_encrypted, TRAD_HEADER_SIZE); if (ret != ARCHIVE_OK) return (ret); zip->written_bytes += TRAD_HEADER_SIZE; zip->entry_compressed_written += TRAD_HEADER_SIZE; return (ret); } static int init_winzip_aes_encryption(struct archive_write *a) { struct zip *zip = a->format_data; const char *passphrase; size_t key_len, salt_len; uint8_t salt[16 + 2]; uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; int ret; passphrase = __archive_write_get_passphrase(a); if (passphrase == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Encryption needs passphrase"); return (ARCHIVE_FAILED); } if (zip->entry_encryption == ENCRYPTION_WINZIP_AES128) { salt_len = 8; key_len = 16; } else { /* AES 256 */ salt_len = 16; key_len = 32; } if (archive_random(salt, salt_len) != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't generate random number for encryption"); return (ARCHIVE_FATAL); } archive_pbkdf2_sha1(passphrase, strlen(passphrase), salt, salt_len, 1000, derived_key, key_len * 2 + 2); ret = archive_encrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); if (ret != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Decryption is unsupported due to lack of crypto library"); return (ARCHIVE_FAILED); } ret = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); if (ret != 0) { archive_encrypto_aes_ctr_release(&zip->cctx); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to initialize HMAC-SHA1"); return (ARCHIVE_FAILED); } /* Set a passowrd verification value after the 'salt'. */ salt[salt_len] = derived_key[key_len * 2]; salt[salt_len + 1] = derived_key[key_len * 2 + 1]; /* Write encrypted keys in the top of the file content. */ ret = __archive_write_output(a, salt, salt_len + 2); if (ret != ARCHIVE_OK) return (ret); zip->written_bytes += salt_len + 2; zip->entry_compressed_written += salt_len + 2; return (ARCHIVE_OK); } static int is_winzip_aes_encryption_supported(int encryption) { size_t key_len, salt_len; uint8_t salt[16 + 2]; uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; archive_crypto_ctx cctx; archive_hmac_sha1_ctx hctx; int ret; if (encryption == ENCRYPTION_WINZIP_AES128) { salt_len = 8; key_len = 16; } else { /* AES 256 */ salt_len = 16; key_len = 32; } if (archive_random(salt, salt_len) != ARCHIVE_OK) return (0); ret = archive_pbkdf2_sha1("p", 1, salt, salt_len, 1000, derived_key, key_len * 2 + 2); if (ret != 0) return (0); ret = archive_encrypto_aes_ctr_init(&cctx, derived_key, key_len); if (ret != 0) return (0); ret = archive_hmac_sha1_init(&hctx, derived_key + key_len, key_len); archive_encrypto_aes_ctr_release(&cctx); if (ret != 0) return (0); archive_hmac_sha1_cleanup(&hctx); return (1); } Index: head/contrib/libarchive/libarchive/test/main.c =================================================================== --- head/contrib/libarchive/libarchive/test/main.c (revision 309589) +++ head/contrib/libarchive/libarchive/test/main.c (revision 309590) @@ -1,3071 +1,3071 @@ /* * Copyright (c) 2003-2009 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" #include "test_utils.h" #ifdef HAVE_SYS_IOCTL_H #include #endif #ifdef HAVE_SYS_TIME_H #include #endif #include #ifdef HAVE_ICONV_H #include #endif /* * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. * As the include guards don't agree, the order of include is important. */ #ifdef HAVE_LINUX_EXT2_FS_H #include /* for Linux file flags */ #endif #if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) #include /* Linux file flags, broken on Cygwin */ #endif #include #include #ifdef HAVE_SIGNAL_H #include #endif #include #include /* * This same file is used pretty much verbatim for all test harnesses. * * The next few lines are the only differences. * TODO: Move this into a separate configuration header, have all test * suites share one copy of this file. */ __FBSDID("$FreeBSD$"); #define KNOWNREF "test_compat_gtar_1.tar.uu" #define ENVBASE "LIBARCHIVE" /* Prefix for environment variables. */ #undef PROGRAM /* Testing a library, not a program. */ #define LIBRARY "libarchive" #define EXTRA_DUMP(x) archive_error_string((struct archive *)(x)) #define EXTRA_ERRNO(x) archive_errno((struct archive *)(x)) #define EXTRA_VERSION archive_version_details() /* * * Windows support routines * * Note: Configuration is a tricky issue. Using HAVE_* feature macros * in the test harness is dangerous because they cover up * configuration errors. The classic example of this is omitting a * configure check. If libarchive and libarchive_test both look for * the same feature macro, such errors are hard to detect. Platform * macros (e.g., _WIN32 or __GNUC__) are a little better, but can * easily lead to very messy code. It's best to limit yourself * to only the most generic programming techniques in the test harness * and thus avoid conditionals altogether. Where that's not possible, * try to minimize conditionals by grouping platform-specific tests in * one place (e.g., test_acl_freebsd) or by adding new assert() * functions (e.g., assertMakeHardlink()) to cover up platform * differences. Platform-specific coding in libarchive_test is often * a symptom that some capability is missing from libarchive itself. */ #if defined(_WIN32) && !defined(__CYGWIN__) #include #include #include #ifndef F_OK #define F_OK (0) #endif #ifndef S_ISDIR #define S_ISDIR(m) ((m) & _S_IFDIR) #endif #ifndef S_ISREG #define S_ISREG(m) ((m) & _S_IFREG) #endif #if !defined(__BORLANDC__) #define access _access #undef chdir #define chdir _chdir #endif #ifndef fileno #define fileno _fileno #endif /*#define fstat _fstat64*/ #if !defined(__BORLANDC__) #define getcwd _getcwd #endif #define lstat stat /*#define lstat _stat64*/ /*#define stat _stat64*/ #define rmdir _rmdir #if !defined(__BORLANDC__) #define strdup _strdup #define umask _umask #endif #define int64_t __int64 #endif #if defined(HAVE__CrtSetReportMode) # include #endif mode_t umasked(mode_t expected_mode) { mode_t mode = umask(0); umask(mode); return expected_mode & ~mode; } /* Path to working directory for current test */ const char *testworkdir; #ifdef PROGRAM /* Pathname of exe to be tested. */ const char *testprogfile; /* Name of exe to use in printf-formatted command strings. */ /* On Windows, this includes leading/trailing quotes. */ const char *testprog; #endif #if defined(_WIN32) && !defined(__CYGWIN__) static void *GetFunctionKernel32(const char *); static int my_CreateSymbolicLinkA(const char *, const char *, int); static int my_CreateHardLinkA(const char *, const char *); static int my_GetFileInformationByName(const char *, BY_HANDLE_FILE_INFORMATION *); static void * GetFunctionKernel32(const char *name) { static HINSTANCE lib; static int set; if (!set) { set = 1; lib = LoadLibrary("kernel32.dll"); } if (lib == NULL) { fprintf(stderr, "Can't load kernel32.dll?!\n"); exit(1); } return (void *)GetProcAddress(lib, name); } static int my_CreateSymbolicLinkA(const char *linkname, const char *target, int flags) { static BOOLEAN (WINAPI *f)(LPCSTR, LPCSTR, DWORD); static int set; if (!set) { set = 1; f = GetFunctionKernel32("CreateSymbolicLinkA"); } return f == NULL ? 0 : (*f)(linkname, target, flags); } static int my_CreateHardLinkA(const char *linkname, const char *target) { static BOOLEAN (WINAPI *f)(LPCSTR, LPCSTR, LPSECURITY_ATTRIBUTES); static int set; if (!set) { set = 1; f = GetFunctionKernel32("CreateHardLinkA"); } return f == NULL ? 0 : (*f)(linkname, target, NULL); } static int my_GetFileInformationByName(const char *path, BY_HANDLE_FILE_INFORMATION *bhfi) { HANDLE h; int r; memset(bhfi, 0, sizeof(*bhfi)); h = CreateFile(path, FILE_READ_ATTRIBUTES, 0, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); if (h == INVALID_HANDLE_VALUE) return (0); r = GetFileInformationByHandle(h, bhfi); CloseHandle(h); return (r); } #endif #if defined(HAVE__CrtSetReportMode) && !defined(__WATCOMC__) static void invalid_parameter_handler(const wchar_t * expression, const wchar_t * function, const wchar_t * file, unsigned int line, uintptr_t pReserved) { /* nop */ } #endif /* * * OPTIONS FLAGS * */ /* Enable core dump on failure. */ static int dump_on_failure = 0; /* Default is to remove temp dirs and log data for successful tests. */ static int keep_temp_files = 0; /* Default is to run the specified tests once and report errors. */ static int until_failure = 0; /* Default is to just report pass/fail for each test. */ static int verbosity = 0; #define VERBOSITY_SUMMARY_ONLY -1 /* -q */ #define VERBOSITY_PASSFAIL 0 /* Default */ #define VERBOSITY_LIGHT_REPORT 1 /* -v */ #define VERBOSITY_FULL 2 /* -vv */ /* A few places generate even more output for verbosity > VERBOSITY_FULL, * mostly for debugging the test harness itself. */ /* Cumulative count of assertion failures. */ static int failures = 0; /* Cumulative count of reported skips. */ static int skips = 0; /* Cumulative count of assertions checked. */ static int assertions = 0; /* Directory where uuencoded reference files can be found. */ static const char *refdir; /* * Report log information selectively to console and/or disk log. */ static int log_console = 0; static FILE *logfile; static void vlogprintf(const char *fmt, va_list ap) { #ifdef va_copy va_list lfap; va_copy(lfap, ap); #endif if (log_console) vfprintf(stdout, fmt, ap); if (logfile != NULL) #ifdef va_copy vfprintf(logfile, fmt, lfap); va_end(lfap); #else vfprintf(logfile, fmt, ap); #endif } static void logprintf(const char *fmt, ...) { va_list ap; va_start(ap, fmt); vlogprintf(fmt, ap); va_end(ap); } /* Set up a message to display only if next assertion fails. */ static char msgbuff[4096]; static const char *msg, *nextmsg; void failure(const char *fmt, ...) { va_list ap; if (fmt == NULL) { nextmsg = NULL; } else { va_start(ap, fmt); vsprintf(msgbuff, fmt, ap); va_end(ap); nextmsg = msgbuff; } } /* * Copy arguments into file-local variables. * This was added to permit vararg assert() functions without needing * variadic wrapper macros. Turns out that the vararg capability is almost * never used, so almost all of the vararg assertions can be simplified * by removing the vararg capability and reworking the wrapper macro to * pass __FILE__, __LINE__ directly into the function instead of using * this hook. I suspect this machinery is used so rarely that we * would be better off just removing it entirely. That would simplify * the code here noticeably. */ static const char *skipping_filename; static int skipping_line; void skipping_setup(const char *filename, int line) { skipping_filename = filename; skipping_line = line; } /* Called at the beginning of each assert() function. */ static void assertion_count(const char *file, int line) { (void)file; /* UNUSED */ (void)line; /* UNUSED */ ++assertions; /* Proper handling of "failure()" message. */ msg = nextmsg; nextmsg = NULL; /* Uncomment to print file:line after every assertion. * Verbose, but occasionally useful in tracking down crashes. */ /* printf("Checked %s:%d\n", file, line); */ } /* * For each test source file, we remember how many times each * assertion was reported. Cleared before each new test, * used by test_summarize(). */ static struct line { int count; int skip; } failed_lines[10000]; const char *failed_filename; /* Count this failure, setup up log destination and handle initial report. */ static void failure_start(const char *filename, int line, const char *fmt, ...) { va_list ap; /* Record another failure for this line. */ ++failures; failed_filename = filename; failed_lines[line].count++; /* Determine whether to log header to console. */ switch (verbosity) { case VERBOSITY_LIGHT_REPORT: log_console = (failed_lines[line].count < 2); break; default: log_console = (verbosity >= VERBOSITY_FULL); } /* Log file:line header for this failure */ va_start(ap, fmt); #if _MSC_VER logprintf("%s(%d): ", filename, line); #else logprintf("%s:%d: ", filename, line); #endif vlogprintf(fmt, ap); va_end(ap); logprintf("\n"); if (msg != NULL && msg[0] != '\0') { logprintf(" Description: %s\n", msg); msg = NULL; } /* Determine whether to log details to console. */ if (verbosity == VERBOSITY_LIGHT_REPORT) log_console = 0; } /* Complete reporting of failed tests. */ /* * The 'extra' hook here is used by libarchive to include libarchive * error messages with assertion failures. It could also be used * to add strerror() output, for example. Just define the EXTRA_DUMP() * macro appropriately. */ static void failure_finish(void *extra) { (void)extra; /* UNUSED (maybe) */ #ifdef EXTRA_DUMP if (extra != NULL) { logprintf(" errno: %d\n", EXTRA_ERRNO(extra)); logprintf(" detail: %s\n", EXTRA_DUMP(extra)); } #endif if (dump_on_failure) { fprintf(stderr, " *** forcing core dump so failure can be debugged ***\n"); abort(); } } /* Inform user that we're skipping some checks. */ void test_skipping(const char *fmt, ...) { char buff[1024]; va_list ap; va_start(ap, fmt); vsprintf(buff, fmt, ap); va_end(ap); /* Use failure() message if set. */ msg = nextmsg; nextmsg = NULL; /* failure_start() isn't quite right, but is awfully convenient. */ failure_start(skipping_filename, skipping_line, "SKIPPING: %s", buff); --failures; /* Undo failures++ in failure_start() */ /* Don't failure_finish() here. */ /* Mark as skip, so doesn't count as failed test. */ failed_lines[skipping_line].skip = 1; ++skips; } /* * * ASSERTIONS * */ /* Generic assert() just displays the failed condition. */ int assertion_assert(const char *file, int line, int value, const char *condition, void *extra) { assertion_count(file, line); if (!value) { failure_start(file, line, "Assertion failed: %s", condition); failure_finish(extra); } return (value); } /* chdir() and report any errors */ int assertion_chdir(const char *file, int line, const char *pathname) { assertion_count(file, line); if (chdir(pathname) == 0) return (1); failure_start(file, line, "chdir(\"%s\")", pathname); failure_finish(NULL); return (0); } /* Verify two integers are equal. */ int assertion_equal_int(const char *file, int line, long long v1, const char *e1, long long v2, const char *e2, void *extra) { assertion_count(file, line); if (v1 == v2) return (1); failure_start(file, line, "%s != %s", e1, e2); logprintf(" %s=%lld (0x%llx, 0%llo)\n", e1, v1, v1, v1); logprintf(" %s=%lld (0x%llx, 0%llo)\n", e2, v2, v2, v2); failure_finish(extra); return (0); } /* * Utility to convert a single UTF-8 sequence. */ static int _utf8_to_unicode(uint32_t *pwc, const char *s, size_t n) { static const char utf8_count[256] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 00 - 0F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 10 - 1F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 20 - 2F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 30 - 3F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 40 - 4F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 50 - 5F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 60 - 6F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 70 - 7F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 80 - 8F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 90 - 9F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* A0 - AF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* B0 - BF */ 0, 0, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* C0 - CF */ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* D0 - DF */ 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,/* E0 - EF */ 4, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 /* F0 - FF */ }; int ch; int cnt; uint32_t wc; *pwc = 0; /* Sanity check. */ if (n == 0) return (0); /* * Decode 1-4 bytes depending on the value of the first byte. */ ch = (unsigned char)*s; if (ch == 0) return (0); /* Standard: return 0 for end-of-string. */ cnt = utf8_count[ch]; - /* Invalide sequence or there are not plenty bytes. */ + /* Invalid sequence or there are not plenty bytes. */ if (n < (size_t)cnt) return (-1); /* Make a Unicode code point from a single UTF-8 sequence. */ switch (cnt) { case 1: /* 1 byte sequence. */ *pwc = ch & 0x7f; return (cnt); case 2: /* 2 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) return (-1); *pwc = ((ch & 0x1f) << 6) | (s[1] & 0x3f); return (cnt); case 3: /* 3 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) return (-1); if ((s[2] & 0xc0) != 0x80) return (-1); wc = ((ch & 0x0f) << 12) | ((s[1] & 0x3f) << 6) | (s[2] & 0x3f); if (wc < 0x800) return (-1);/* Overlong sequence. */ break; case 4: /* 4 bytes sequence. */ if (n < 4) return (-1); if ((s[1] & 0xc0) != 0x80) return (-1); if ((s[2] & 0xc0) != 0x80) return (-1); if ((s[3] & 0xc0) != 0x80) return (-1); wc = ((ch & 0x07) << 18) | ((s[1] & 0x3f) << 12) | ((s[2] & 0x3f) << 6) | (s[3] & 0x3f); if (wc < 0x10000) return (-1);/* Overlong sequence. */ break; default: return (-1); } - /* The code point larger than 0x10FFFF is not leagal + /* The code point larger than 0x10FFFF is not legal * Unicode values. */ if (wc > 0x10FFFF) return (-1); /* Correctly gets a Unicode, returns used bytes. */ *pwc = wc; return (cnt); } static void strdump(const char *e, const char *p, int ewidth, int utf8) { const char *q = p; logprintf(" %*s = ", ewidth, e); if (p == NULL) { logprintf("NULL\n"); return; } logprintf("\""); while (*p != '\0') { unsigned int c = 0xff & *p++; switch (c) { case '\a': logprintf("\\a"); break; case '\b': logprintf("\\b"); break; case '\n': logprintf("\\n"); break; case '\r': logprintf("\\r"); break; default: if (c >= 32 && c < 127) logprintf("%c", c); else logprintf("\\x%02X", c); } } logprintf("\""); logprintf(" (length %d)", q == NULL ? -1 : (int)strlen(q)); /* * If the current string is UTF-8, dump its code points. */ if (utf8) { size_t len; uint32_t uc; int n; int cnt = 0; p = q; len = strlen(p); logprintf(" ["); while ((n = _utf8_to_unicode(&uc, p, len)) > 0) { if (p != q) logprintf(" "); logprintf("%04X", uc); p += n; len -= n; cnt++; } logprintf("]"); logprintf(" (count %d", cnt); if (n < 0) { logprintf(",unknown %d bytes", len); } logprintf(")"); } logprintf("\n"); } /* Verify two strings are equal, dump them if not. */ int assertion_equal_string(const char *file, int line, const char *v1, const char *e1, const char *v2, const char *e2, void *extra, int utf8) { int l1, l2; assertion_count(file, line); if (v1 == v2 || (v1 != NULL && v2 != NULL && strcmp(v1, v2) == 0)) return (1); failure_start(file, line, "%s != %s", e1, e2); l1 = (int)strlen(e1); l2 = (int)strlen(e2); if (l1 < l2) l1 = l2; strdump(e1, v1, l1, utf8); strdump(e2, v2, l1, utf8); failure_finish(extra); return (0); } static void wcsdump(const char *e, const wchar_t *w) { logprintf(" %s = ", e); if (w == NULL) { logprintf("(null)"); return; } logprintf("\""); while (*w != L'\0') { unsigned int c = *w++; if (c >= 32 && c < 127) logprintf("%c", c); else if (c < 256) logprintf("\\x%02X", c); else if (c < 0x10000) logprintf("\\u%04X", c); else logprintf("\\U%08X", c); } logprintf("\"\n"); } #ifndef HAVE_WCSCMP static int wcscmp(const wchar_t *s1, const wchar_t *s2) { while (*s1 == *s2++) { if (*s1++ == L'\0') return 0; } if (*s1 > *--s2) return 1; else return -1; } #endif /* Verify that two wide strings are equal, dump them if not. */ int assertion_equal_wstring(const char *file, int line, const wchar_t *v1, const char *e1, const wchar_t *v2, const char *e2, void *extra) { assertion_count(file, line); if (v1 == v2) return (1); if (v1 != NULL && v2 != NULL && wcscmp(v1, v2) == 0) return (1); failure_start(file, line, "%s != %s", e1, e2); wcsdump(e1, v1); wcsdump(e2, v2); failure_finish(extra); return (0); } /* * Pretty standard hexdump routine. As a bonus, if ref != NULL, then * any bytes in p that differ from ref will be highlighted with '_' * before and after the hex value. */ static void hexdump(const char *p, const char *ref, size_t l, size_t offset) { size_t i, j; char sep; if (p == NULL) { logprintf("(null)\n"); return; } for(i=0; i < l; i+=16) { logprintf("%04x", (unsigned)(i + offset)); sep = ' '; for (j = 0; j < 16 && i + j < l; j++) { if (ref != NULL && p[i + j] != ref[i + j]) sep = '_'; logprintf("%c%02x", sep, 0xff & (int)p[i+j]); if (ref != NULL && p[i + j] == ref[i + j]) sep = ' '; } for (; j < 16; j++) { logprintf("%c ", sep); sep = ' '; } logprintf("%c", sep); for (j=0; j < 16 && i + j < l; j++) { int c = p[i + j]; if (c >= ' ' && c <= 126) logprintf("%c", c); else logprintf("."); } logprintf("\n"); } } /* Verify that two blocks of memory are the same, display the first * block of differences if they're not. */ int assertion_equal_mem(const char *file, int line, const void *_v1, const char *e1, const void *_v2, const char *e2, size_t l, const char *ld, void *extra) { const char *v1 = (const char *)_v1; const char *v2 = (const char *)_v2; size_t offset; assertion_count(file, line); if (v1 == v2 || (v1 != NULL && v2 != NULL && memcmp(v1, v2, l) == 0)) return (1); if (v1 == NULL || v2 == NULL) return (0); failure_start(file, line, "%s != %s", e1, e2); logprintf(" size %s = %d\n", ld, (int)l); /* Dump 48 bytes (3 lines) so that the first difference is * in the second line. */ offset = 0; while (l > 64 && memcmp(v1, v2, 32) == 0) { /* Two lines agree, so step forward one line. */ v1 += 16; v2 += 16; l -= 16; offset += 16; } logprintf(" Dump of %s\n", e1); hexdump(v1, v2, l < 128 ? l : 128, offset); logprintf(" Dump of %s\n", e2); hexdump(v2, v1, l < 128 ? l : 128, offset); logprintf("\n"); failure_finish(extra); return (0); } /* Verify that a block of memory is filled with the specified byte. */ int assertion_memory_filled_with(const char *file, int line, const void *_v1, const char *vd, size_t l, const char *ld, char b, const char *bd, void *extra) { const char *v1 = (const char *)_v1; size_t c = 0; size_t i; (void)ld; /* UNUSED */ assertion_count(file, line); for (i = 0; i < l; ++i) { if (v1[i] == b) { ++c; } } if (c == l) return (1); failure_start(file, line, "%s (size %d) not filled with %s", vd, (int)l, bd); logprintf(" Only %d bytes were correct\n", (int)c); failure_finish(extra); return (0); } /* Verify that the named file exists and is empty. */ int assertion_empty_file(const char *filename, int line, const char *f1) { char buff[1024]; struct stat st; ssize_t s; FILE *f; assertion_count(filename, line); if (stat(f1, &st) != 0) { failure_start(filename, line, "Stat failed: %s", f1); failure_finish(NULL); return (0); } if (st.st_size == 0) return (1); failure_start(filename, line, "File should be empty: %s", f1); logprintf(" File size: %d\n", (int)st.st_size); logprintf(" Contents:\n"); f = fopen(f1, "rb"); if (f == NULL) { logprintf(" Unable to open %s\n", f1); } else { s = ((off_t)sizeof(buff) < st.st_size) ? (ssize_t)sizeof(buff) : (ssize_t)st.st_size; s = fread(buff, 1, s, f); hexdump(buff, NULL, s, 0); fclose(f); } failure_finish(NULL); return (0); } /* Verify that the named file exists and is not empty. */ int assertion_non_empty_file(const char *filename, int line, const char *f1) { struct stat st; assertion_count(filename, line); if (stat(f1, &st) != 0) { failure_start(filename, line, "Stat failed: %s", f1); failure_finish(NULL); return (0); } if (st.st_size == 0) { failure_start(filename, line, "File empty: %s", f1); failure_finish(NULL); return (0); } return (1); } /* Verify that two files have the same contents. */ /* TODO: hexdump the first bytes that actually differ. */ int assertion_equal_file(const char *filename, int line, const char *fn1, const char *fn2) { char buff1[1024]; char buff2[1024]; FILE *f1, *f2; int n1, n2; assertion_count(filename, line); f1 = fopen(fn1, "rb"); f2 = fopen(fn2, "rb"); if (f1 == NULL || f2 == NULL) { if (f1) fclose(f1); if (f2) fclose(f2); return (0); } for (;;) { n1 = (int)fread(buff1, 1, sizeof(buff1), f1); n2 = (int)fread(buff2, 1, sizeof(buff2), f2); if (n1 != n2) break; if (n1 == 0 && n2 == 0) { fclose(f1); fclose(f2); return (1); } if (memcmp(buff1, buff2, n1) != 0) break; } fclose(f1); fclose(f2); failure_start(filename, line, "Files not identical"); logprintf(" file1=\"%s\"\n", fn1); logprintf(" file2=\"%s\"\n", fn2); failure_finish(NULL); return (0); } /* Verify that the named file does exist. */ int assertion_file_exists(const char *filename, int line, const char *f) { assertion_count(filename, line); #if defined(_WIN32) && !defined(__CYGWIN__) if (!_access(f, 0)) return (1); #else if (!access(f, F_OK)) return (1); #endif failure_start(filename, line, "File should exist: %s", f); failure_finish(NULL); return (0); } /* Verify that the named file doesn't exist. */ int assertion_file_not_exists(const char *filename, int line, const char *f) { assertion_count(filename, line); #if defined(_WIN32) && !defined(__CYGWIN__) if (_access(f, 0)) return (1); #else if (access(f, F_OK)) return (1); #endif failure_start(filename, line, "File should not exist: %s", f); failure_finish(NULL); return (0); } /* Compare the contents of a file to a block of memory. */ int assertion_file_contents(const char *filename, int line, const void *buff, int s, const char *fn) { char *contents; FILE *f; int n; assertion_count(filename, line); f = fopen(fn, "rb"); if (f == NULL) { failure_start(filename, line, "File should exist: %s", fn); failure_finish(NULL); return (0); } contents = malloc(s * 2); n = (int)fread(contents, 1, s * 2, f); fclose(f); if (n == s && memcmp(buff, contents, s) == 0) { free(contents); return (1); } failure_start(filename, line, "File contents don't match"); logprintf(" file=\"%s\"\n", fn); if (n > 0) hexdump(contents, buff, n > 512 ? 512 : n, 0); else { logprintf(" File empty, contents should be:\n"); hexdump(buff, NULL, s > 512 ? 512 : s, 0); } failure_finish(NULL); free(contents); return (0); } /* Check the contents of a text file, being tolerant of line endings. */ int assertion_text_file_contents(const char *filename, int line, const char *buff, const char *fn) { char *contents; const char *btxt, *ftxt; FILE *f; int n, s; assertion_count(filename, line); f = fopen(fn, "r"); if (f == NULL) { failure_start(filename, line, "File doesn't exist: %s", fn); failure_finish(NULL); return (0); } s = (int)strlen(buff); contents = malloc(s * 2 + 128); n = (int)fread(contents, 1, s * 2 + 128 - 1, f); if (n >= 0) contents[n] = '\0'; fclose(f); /* Compare texts. */ btxt = buff; ftxt = (const char *)contents; while (*btxt != '\0' && *ftxt != '\0') { if (*btxt == *ftxt) { ++btxt; ++ftxt; continue; } if (btxt[0] == '\n' && ftxt[0] == '\r' && ftxt[1] == '\n') { /* Pass over different new line characters. */ ++btxt; ftxt += 2; continue; } break; } if (*btxt == '\0' && *ftxt == '\0') { free(contents); return (1); } failure_start(filename, line, "Contents don't match"); logprintf(" file=\"%s\"\n", fn); if (n > 0) { hexdump(contents, buff, n, 0); logprintf(" expected\n", fn); hexdump(buff, contents, s, 0); } else { logprintf(" File empty, contents should be:\n"); hexdump(buff, NULL, s, 0); } failure_finish(NULL); free(contents); return (0); } /* Verify that a text file contains the specified lines, regardless of order */ /* This could be more efficient if we sorted both sets of lines, etc, but * since this is used only for testing and only ever deals with a dozen or so * lines at a time, this relatively crude approach is just fine. */ int assertion_file_contains_lines_any_order(const char *file, int line, const char *pathname, const char *lines[]) { char *buff; size_t buff_size; size_t expected_count, actual_count, i, j; char **expected = NULL; char *p, **actual = NULL; char c; int expected_failure = 0, actual_failure = 0; assertion_count(file, line); buff = slurpfile(&buff_size, "%s", pathname); if (buff == NULL) { failure_start(pathname, line, "Can't read file: %s", pathname); failure_finish(NULL); return (0); } /* Make a copy of the provided lines and count up the expected * file size. */ for (i = 0; lines[i] != NULL; ++i) { } expected_count = i; if (expected_count) { expected = malloc(sizeof(char *) * expected_count); if (expected == NULL) { failure_start(pathname, line, "Can't allocate memory"); failure_finish(NULL); free(expected); return (0); } for (i = 0; lines[i] != NULL; ++i) { expected[i] = strdup(lines[i]); } } /* Break the file into lines */ actual_count = 0; for (c = '\0', p = buff; p < buff + buff_size; ++p) { if (*p == '\x0d' || *p == '\x0a') *p = '\0'; if (c == '\0' && *p != '\0') ++actual_count; c = *p; } if (actual_count) { actual = calloc(sizeof(char *), actual_count); if (actual == NULL) { failure_start(pathname, line, "Can't allocate memory"); failure_finish(NULL); free(expected); return (0); } for (j = 0, p = buff; p < buff + buff_size; p += 1 + strlen(p)) { if (*p != '\0') { actual[j] = p; ++j; } } } /* Erase matching lines from both lists */ for (i = 0; i < expected_count; ++i) { if (expected[i] == NULL) continue; for (j = 0; j < actual_count; ++j) { if (actual[j] == NULL) continue; if (strcmp(expected[i], actual[j]) == 0) { free(expected[i]); expected[i] = NULL; actual[j] = NULL; break; } } } /* If there's anything left, it's a failure */ for (i = 0; i < expected_count; ++i) { if (expected[i] != NULL) ++expected_failure; } for (j = 0; j < actual_count; ++j) { if (actual[j] != NULL) ++actual_failure; } if (expected_failure == 0 && actual_failure == 0) { free(buff); free(expected); free(actual); return (1); } failure_start(file, line, "File doesn't match: %s", pathname); for (i = 0; i < expected_count; ++i) { if (expected[i] != NULL) { logprintf(" Expected but not present: %s\n", expected[i]); free(expected[i]); } } for (j = 0; j < actual_count; ++j) { if (actual[j] != NULL) logprintf(" Present but not expected: %s\n", actual[j]); } failure_finish(NULL); free(buff); free(expected); free(actual); return (0); } /* Verify that a text file does not contains the specified strings */ int assertion_file_contains_no_invalid_strings(const char *file, int line, const char *pathname, const char *strings[]) { char *buff; int i; buff = slurpfile(NULL, "%s", pathname); if (buff == NULL) { failure_start(file, line, "Can't read file: %s", pathname); failure_finish(NULL); return (0); } for (i = 0; strings[i] != NULL; ++i) { if (strstr(buff, strings[i]) != NULL) { failure_start(file, line, "Invalid string in %s: %s", pathname, strings[i]); failure_finish(NULL); free(buff); return(0); } } free(buff); return (0); } /* Test that two paths point to the same file. */ /* As a side-effect, asserts that both files exist. */ static int is_hardlink(const char *file, int line, const char *path1, const char *path2) { #if defined(_WIN32) && !defined(__CYGWIN__) BY_HANDLE_FILE_INFORMATION bhfi1, bhfi2; int r; assertion_count(file, line); r = my_GetFileInformationByName(path1, &bhfi1); if (r == 0) { failure_start(file, line, "File %s can't be inspected?", path1); failure_finish(NULL); return (0); } r = my_GetFileInformationByName(path2, &bhfi2); if (r == 0) { failure_start(file, line, "File %s can't be inspected?", path2); failure_finish(NULL); return (0); } return (bhfi1.dwVolumeSerialNumber == bhfi2.dwVolumeSerialNumber && bhfi1.nFileIndexHigh == bhfi2.nFileIndexHigh && bhfi1.nFileIndexLow == bhfi2.nFileIndexLow); #else struct stat st1, st2; int r; assertion_count(file, line); r = lstat(path1, &st1); if (r != 0) { failure_start(file, line, "File should exist: %s", path1); failure_finish(NULL); return (0); } r = lstat(path2, &st2); if (r != 0) { failure_start(file, line, "File should exist: %s", path2); failure_finish(NULL); return (0); } return (st1.st_ino == st2.st_ino && st1.st_dev == st2.st_dev); #endif } int assertion_is_hardlink(const char *file, int line, const char *path1, const char *path2) { if (is_hardlink(file, line, path1, path2)) return (1); failure_start(file, line, "Files %s and %s are not hardlinked", path1, path2); failure_finish(NULL); return (0); } int assertion_is_not_hardlink(const char *file, int line, const char *path1, const char *path2) { if (!is_hardlink(file, line, path1, path2)) return (1); failure_start(file, line, "Files %s and %s should not be hardlinked", path1, path2); failure_finish(NULL); return (0); } /* Verify a/b/mtime of 'pathname'. */ /* If 'recent', verify that it's within last 10 seconds. */ static int assertion_file_time(const char *file, int line, const char *pathname, long t, long nsec, char type, int recent) { long long filet, filet_nsec; int r; #if defined(_WIN32) && !defined(__CYGWIN__) #define EPOC_TIME (116444736000000000ULL) FILETIME fxtime, fbirthtime, fatime, fmtime; ULARGE_INTEGER wintm; HANDLE h; fxtime.dwLowDateTime = 0; fxtime.dwHighDateTime = 0; assertion_count(file, line); /* Note: FILE_FLAG_BACKUP_SEMANTICS applies to open * a directory file. If not, CreateFile() will fail when * the pathname is a directory. */ h = CreateFile(pathname, FILE_READ_ATTRIBUTES, 0, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); if (h == INVALID_HANDLE_VALUE) { failure_start(file, line, "Can't access %s\n", pathname); failure_finish(NULL); return (0); } r = GetFileTime(h, &fbirthtime, &fatime, &fmtime); switch (type) { case 'a': fxtime = fatime; break; case 'b': fxtime = fbirthtime; break; case 'm': fxtime = fmtime; break; } CloseHandle(h); if (r == 0) { failure_start(file, line, "Can't GetFileTime %s\n", pathname); failure_finish(NULL); return (0); } wintm.LowPart = fxtime.dwLowDateTime; wintm.HighPart = fxtime.dwHighDateTime; filet = (wintm.QuadPart - EPOC_TIME) / 10000000; filet_nsec = ((wintm.QuadPart - EPOC_TIME) % 10000000) * 100; nsec = (nsec / 100) * 100; /* Round the request */ #else struct stat st; assertion_count(file, line); r = lstat(pathname, &st); if (r != 0) { failure_start(file, line, "Can't stat %s\n", pathname); failure_finish(NULL); return (0); } switch (type) { case 'a': filet = st.st_atime; break; case 'm': filet = st.st_mtime; break; case 'b': filet = 0; break; default: fprintf(stderr, "INTERNAL: Bad type %c for file time", type); exit(1); } #if defined(__FreeBSD__) switch (type) { case 'a': filet_nsec = st.st_atimespec.tv_nsec; break; case 'b': filet = st.st_birthtime; /* FreeBSD filesystems that don't support birthtime * (e.g., UFS1) always return -1 here. */ if (filet == -1) { return (1); } filet_nsec = st.st_birthtimespec.tv_nsec; break; case 'm': filet_nsec = st.st_mtimespec.tv_nsec; break; default: fprintf(stderr, "INTERNAL: Bad type %c for file time", type); exit(1); } /* FreeBSD generally only stores to microsecond res, so round. */ filet_nsec = (filet_nsec / 1000) * 1000; nsec = (nsec / 1000) * 1000; #else filet_nsec = nsec = 0; /* Generic POSIX only has whole seconds. */ if (type == 'b') return (1); /* Generic POSIX doesn't have birthtime */ #if defined(__HAIKU__) if (type == 'a') return (1); /* Haiku doesn't have atime. */ #endif #endif #endif if (recent) { /* Check that requested time is up-to-date. */ time_t now = time(NULL); if (filet < now - 10 || filet > now + 1) { failure_start(file, line, "File %s has %ctime %lld, %lld seconds ago\n", pathname, type, filet, now - filet); failure_finish(NULL); return (0); } } else if (filet != t || filet_nsec != nsec) { failure_start(file, line, "File %s has %ctime %lld.%09lld, expected %lld.%09lld", pathname, type, filet, filet_nsec, t, nsec); failure_finish(NULL); return (0); } return (1); } /* Verify atime of 'pathname'. */ int assertion_file_atime(const char *file, int line, const char *pathname, long t, long nsec) { return assertion_file_time(file, line, pathname, t, nsec, 'a', 0); } /* Verify atime of 'pathname' is up-to-date. */ int assertion_file_atime_recent(const char *file, int line, const char *pathname) { return assertion_file_time(file, line, pathname, 0, 0, 'a', 1); } /* Verify birthtime of 'pathname'. */ int assertion_file_birthtime(const char *file, int line, const char *pathname, long t, long nsec) { return assertion_file_time(file, line, pathname, t, nsec, 'b', 0); } /* Verify birthtime of 'pathname' is up-to-date. */ int assertion_file_birthtime_recent(const char *file, int line, const char *pathname) { return assertion_file_time(file, line, pathname, 0, 0, 'b', 1); } /* Verify mode of 'pathname'. */ int assertion_file_mode(const char *file, int line, const char *pathname, int expected_mode) { int mode; int r; assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) failure_start(file, line, "assertFileMode not yet implemented for Windows"); (void)mode; /* UNUSED */ (void)r; /* UNUSED */ #else { struct stat st; r = lstat(pathname, &st); mode = (int)(st.st_mode & 0777); } if (r == 0 && mode == expected_mode) return (1); failure_start(file, line, "File %s has mode %o, expected %o", pathname, mode, expected_mode); #endif failure_finish(NULL); return (0); } /* Verify mtime of 'pathname'. */ int assertion_file_mtime(const char *file, int line, const char *pathname, long t, long nsec) { return assertion_file_time(file, line, pathname, t, nsec, 'm', 0); } /* Verify mtime of 'pathname' is up-to-date. */ int assertion_file_mtime_recent(const char *file, int line, const char *pathname) { return assertion_file_time(file, line, pathname, 0, 0, 'm', 1); } /* Verify number of links to 'pathname'. */ int assertion_file_nlinks(const char *file, int line, const char *pathname, int nlinks) { #if defined(_WIN32) && !defined(__CYGWIN__) BY_HANDLE_FILE_INFORMATION bhfi; int r; assertion_count(file, line); r = my_GetFileInformationByName(pathname, &bhfi); if (r != 0 && bhfi.nNumberOfLinks == (DWORD)nlinks) return (1); failure_start(file, line, "File %s has %d links, expected %d", pathname, bhfi.nNumberOfLinks, nlinks); failure_finish(NULL); return (0); #else struct stat st; int r; assertion_count(file, line); r = lstat(pathname, &st); if (r == 0 && (int)st.st_nlink == nlinks) return (1); failure_start(file, line, "File %s has %d links, expected %d", pathname, st.st_nlink, nlinks); failure_finish(NULL); return (0); #endif } /* Verify size of 'pathname'. */ int assertion_file_size(const char *file, int line, const char *pathname, long size) { int64_t filesize; int r; assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) { BY_HANDLE_FILE_INFORMATION bhfi; r = !my_GetFileInformationByName(pathname, &bhfi); filesize = ((int64_t)bhfi.nFileSizeHigh << 32) + bhfi.nFileSizeLow; } #else { struct stat st; r = lstat(pathname, &st); filesize = st.st_size; } #endif if (r == 0 && filesize == size) return (1); failure_start(file, line, "File %s has size %ld, expected %ld", pathname, (long)filesize, (long)size); failure_finish(NULL); return (0); } /* Assert that 'pathname' is a dir. If mode >= 0, verify that too. */ int assertion_is_dir(const char *file, int line, const char *pathname, int mode) { struct stat st; int r; #if defined(_WIN32) && !defined(__CYGWIN__) (void)mode; /* UNUSED */ #endif assertion_count(file, line); r = lstat(pathname, &st); if (r != 0) { failure_start(file, line, "Dir should exist: %s", pathname); failure_finish(NULL); return (0); } if (!S_ISDIR(st.st_mode)) { failure_start(file, line, "%s is not a dir", pathname); failure_finish(NULL); return (0); } #if !defined(_WIN32) || defined(__CYGWIN__) /* Windows doesn't handle permissions the same way as POSIX, * so just ignore the mode tests. */ /* TODO: Can we do better here? */ if (mode >= 0 && (mode_t)mode != (st.st_mode & 07777)) { failure_start(file, line, "Dir %s has wrong mode", pathname); logprintf(" Expected: 0%3o\n", mode); logprintf(" Found: 0%3o\n", st.st_mode & 07777); failure_finish(NULL); return (0); } #endif return (1); } /* Verify that 'pathname' is a regular file. If 'mode' is >= 0, * verify that too. */ int assertion_is_reg(const char *file, int line, const char *pathname, int mode) { struct stat st; int r; #if defined(_WIN32) && !defined(__CYGWIN__) (void)mode; /* UNUSED */ #endif assertion_count(file, line); r = lstat(pathname, &st); if (r != 0 || !S_ISREG(st.st_mode)) { failure_start(file, line, "File should exist: %s", pathname); failure_finish(NULL); return (0); } #if !defined(_WIN32) || defined(__CYGWIN__) /* Windows doesn't handle permissions the same way as POSIX, * so just ignore the mode tests. */ /* TODO: Can we do better here? */ if (mode >= 0 && (mode_t)mode != (st.st_mode & 07777)) { failure_start(file, line, "File %s has wrong mode", pathname); logprintf(" Expected: 0%3o\n", mode); logprintf(" Found: 0%3o\n", st.st_mode & 07777); failure_finish(NULL); return (0); } #endif return (1); } /* Check whether 'pathname' is a symbolic link. If 'contents' is * non-NULL, verify that the symlink has those contents. */ static int is_symlink(const char *file, int line, const char *pathname, const char *contents) { #if defined(_WIN32) && !defined(__CYGWIN__) (void)pathname; /* UNUSED */ (void)contents; /* UNUSED */ assertion_count(file, line); /* Windows sort-of has real symlinks, but they're only usable * by privileged users and are crippled even then, so there's * really not much point in bothering with this. */ return (0); #else char buff[300]; struct stat st; ssize_t linklen; int r; assertion_count(file, line); r = lstat(pathname, &st); if (r != 0) { failure_start(file, line, "Symlink should exist: %s", pathname); failure_finish(NULL); return (0); } if (!S_ISLNK(st.st_mode)) return (0); if (contents == NULL) return (1); linklen = readlink(pathname, buff, sizeof(buff)); if (linklen < 0) { failure_start(file, line, "Can't read symlink %s", pathname); failure_finish(NULL); return (0); } buff[linklen] = '\0'; if (strcmp(buff, contents) != 0) return (0); return (1); #endif } /* Assert that path is a symlink that (optionally) contains contents. */ int assertion_is_symlink(const char *file, int line, const char *path, const char *contents) { if (is_symlink(file, line, path, contents)) return (1); if (contents) failure_start(file, line, "File %s is not a symlink to %s", path, contents); else failure_start(file, line, "File %s is not a symlink", path); failure_finish(NULL); return (0); } /* Create a directory and report any errors. */ int assertion_make_dir(const char *file, int line, const char *dirname, int mode) { assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) (void)mode; /* UNUSED */ if (0 == _mkdir(dirname)) return (1); #else if (0 == mkdir(dirname, mode)) { if (0 == chmod(dirname, mode)) { assertion_file_mode(file, line, dirname, mode); return (1); } } #endif failure_start(file, line, "Could not create directory %s", dirname); failure_finish(NULL); return(0); } /* Create a file with the specified contents and report any failures. */ int assertion_make_file(const char *file, int line, const char *path, int mode, int csize, const void *contents) { #if defined(_WIN32) && !defined(__CYGWIN__) /* TODO: Rework this to set file mode as well. */ FILE *f; (void)mode; /* UNUSED */ assertion_count(file, line); f = fopen(path, "wb"); if (f == NULL) { failure_start(file, line, "Could not create file %s", path); failure_finish(NULL); return (0); } if (contents != NULL) { size_t wsize; if (csize < 0) wsize = strlen(contents); else wsize = (size_t)csize; if (wsize != fwrite(contents, 1, wsize, f)) { fclose(f); failure_start(file, line, "Could not write file %s", path); failure_finish(NULL); return (0); } } fclose(f); return (1); #else int fd; assertion_count(file, line); fd = open(path, O_CREAT | O_WRONLY, mode >= 0 ? mode : 0644); if (fd < 0) { failure_start(file, line, "Could not create %s", path); failure_finish(NULL); return (0); } if (0 != chmod(path, mode)) { failure_start(file, line, "Could not chmod %s", path); failure_finish(NULL); close(fd); return (0); } if (contents != NULL) { ssize_t wsize; if (csize < 0) wsize = (ssize_t)strlen(contents); else wsize = (ssize_t)csize; if (wsize != write(fd, contents, wsize)) { close(fd); failure_start(file, line, "Could not write to %s", path); failure_finish(NULL); close(fd); return (0); } } close(fd); assertion_file_mode(file, line, path, mode); return (1); #endif } /* Create a hardlink and report any failures. */ int assertion_make_hardlink(const char *file, int line, const char *newpath, const char *linkto) { int succeeded; assertion_count(file, line); #if defined(_WIN32) && !defined(__CYGWIN__) succeeded = my_CreateHardLinkA(newpath, linkto); #elif HAVE_LINK succeeded = !link(linkto, newpath); #else succeeded = 0; #endif if (succeeded) return (1); failure_start(file, line, "Could not create hardlink"); logprintf(" New link: %s\n", newpath); logprintf(" Old name: %s\n", linkto); failure_finish(NULL); return(0); } /* Create a symlink and report any failures. */ int assertion_make_symlink(const char *file, int line, const char *newpath, const char *linkto) { #if defined(_WIN32) && !defined(__CYGWIN__) int targetIsDir = 0; /* TODO: Fix this */ assertion_count(file, line); if (my_CreateSymbolicLinkA(newpath, linkto, targetIsDir)) return (1); #elif HAVE_SYMLINK assertion_count(file, line); if (0 == symlink(linkto, newpath)) return (1); #endif failure_start(file, line, "Could not create symlink"); logprintf(" New link: %s\n", newpath); logprintf(" Old name: %s\n", linkto); failure_finish(NULL); return(0); } /* Set umask, report failures. */ int assertion_umask(const char *file, int line, int mask) { assertion_count(file, line); (void)file; /* UNUSED */ (void)line; /* UNUSED */ umask(mask); return (1); } /* Set times, report failures. */ int assertion_utimes(const char *file, int line, const char *pathname, long at, long at_nsec, long mt, long mt_nsec) { int r; #if defined(_WIN32) && !defined(__CYGWIN__) #define WINTIME(sec, nsec) ((Int32x32To64(sec, 10000000) + EPOC_TIME)\ + (((nsec)/1000)*10)) HANDLE h; ULARGE_INTEGER wintm; FILETIME fatime, fmtime; FILETIME *pat, *pmt; assertion_count(file, line); h = CreateFileA(pathname,GENERIC_READ | GENERIC_WRITE, FILE_SHARE_READ | FILE_SHARE_WRITE, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); if (h == INVALID_HANDLE_VALUE) { failure_start(file, line, "Can't access %s\n", pathname); failure_finish(NULL); return (0); } if (at > 0 || at_nsec > 0) { wintm.QuadPart = WINTIME(at, at_nsec); fatime.dwLowDateTime = wintm.LowPart; fatime.dwHighDateTime = wintm.HighPart; pat = &fatime; } else pat = NULL; if (mt > 0 || mt_nsec > 0) { wintm.QuadPart = WINTIME(mt, mt_nsec); fmtime.dwLowDateTime = wintm.LowPart; fmtime.dwHighDateTime = wintm.HighPart; pmt = &fmtime; } else pmt = NULL; if (pat != NULL || pmt != NULL) r = SetFileTime(h, NULL, pat, pmt); else r = 1; CloseHandle(h); if (r == 0) { failure_start(file, line, "Can't SetFileTime %s\n", pathname); failure_finish(NULL); return (0); } return (1); #else /* defined(_WIN32) && !defined(__CYGWIN__) */ struct stat st; struct timeval times[2]; #if !defined(__FreeBSD__) mt_nsec = at_nsec = 0; /* Generic POSIX only has whole seconds. */ #endif if (mt == 0 && mt_nsec == 0 && at == 0 && at_nsec == 0) return (1); r = lstat(pathname, &st); if (r < 0) { failure_start(file, line, "Can't stat %s\n", pathname); failure_finish(NULL); return (0); } if (mt == 0 && mt_nsec == 0) { mt = st.st_mtime; #if defined(__FreeBSD__) mt_nsec = st.st_mtimespec.tv_nsec; /* FreeBSD generally only stores to microsecond res, so round. */ mt_nsec = (mt_nsec / 1000) * 1000; #endif } if (at == 0 && at_nsec == 0) { at = st.st_atime; #if defined(__FreeBSD__) at_nsec = st.st_atimespec.tv_nsec; /* FreeBSD generally only stores to microsecond res, so round. */ at_nsec = (at_nsec / 1000) * 1000; #endif } times[1].tv_sec = mt; times[1].tv_usec = mt_nsec / 1000; times[0].tv_sec = at; times[0].tv_usec = at_nsec / 1000; #ifdef HAVE_LUTIMES r = lutimes(pathname, times); #else r = utimes(pathname, times); #endif if (r < 0) { failure_start(file, line, "Can't utimes %s\n", pathname); failure_finish(NULL); return (0); } return (1); #endif /* defined(_WIN32) && !defined(__CYGWIN__) */ } /* Set nodump, report failures. */ int assertion_nodump(const char *file, int line, const char *pathname) { #if defined(HAVE_STRUCT_STAT_ST_FLAGS) && defined(UF_NODUMP) int r; assertion_count(file, line); r = chflags(pathname, UF_NODUMP); if (r < 0) { failure_start(file, line, "Can't set nodump %s\n", pathname); failure_finish(NULL); return (0); } #elif defined(EXT2_IOC_GETFLAGS) && defined(HAVE_WORKING_EXT2_IOC_GETFLAGS)\ && defined(EXT2_NODUMP_FL) int fd, r, flags; assertion_count(file, line); fd = open(pathname, O_RDONLY | O_NONBLOCK); if (fd < 0) { failure_start(file, line, "Can't open %s\n", pathname); failure_finish(NULL); return (0); } r = ioctl(fd, EXT2_IOC_GETFLAGS, &flags); if (r < 0) { failure_start(file, line, "Can't get flags %s\n", pathname); failure_finish(NULL); return (0); } flags |= EXT2_NODUMP_FL; r = ioctl(fd, EXT2_IOC_SETFLAGS, &flags); if (r < 0) { failure_start(file, line, "Can't set nodump %s\n", pathname); failure_finish(NULL); return (0); } close(fd); #else (void)pathname; /* UNUSED */ assertion_count(file, line); #endif return (1); } /* * * UTILITIES for use by tests. * */ /* * Check whether platform supports symlinks. This is intended * for tests to use in deciding whether to bother testing symlink * support; if the platform doesn't support symlinks, there's no point * in checking whether the program being tested can create them. * * Note that the first time this test is called, we actually go out to * disk to create and verify a symlink. This is necessary because * symlink support is actually a property of a particular filesystem * and can thus vary between directories on a single system. After * the first call, this returns the cached result from memory, so it's * safe to call it as often as you wish. */ int canSymlink(void) { /* Remember the test result */ static int value = 0, tested = 0; if (tested) return (value); ++tested; assertion_make_file(__FILE__, __LINE__, "canSymlink.0", 0644, 1, "a"); /* Note: Cygwin has its own symlink() emulation that does not * use the Win32 CreateSymbolicLink() function. */ #if defined(_WIN32) && !defined(__CYGWIN__) value = my_CreateSymbolicLinkA("canSymlink.1", "canSymlink.0", 0) && is_symlink(__FILE__, __LINE__, "canSymlink.1", "canSymlink.0"); #elif HAVE_SYMLINK value = (0 == symlink("canSymlink.0", "canSymlink.1")) && is_symlink(__FILE__, __LINE__, "canSymlink.1","canSymlink.0"); #endif return (value); } /* Platform-dependent options for hiding the output of a subcommand. */ #if defined(_WIN32) && !defined(__CYGWIN__) static const char *redirectArgs = ">NUL 2>NUL"; /* Win32 cmd.exe */ #else static const char *redirectArgs = ">/dev/null 2>/dev/null"; /* POSIX 'sh' */ #endif /* * Can this platform run the bzip2 program? */ int canBzip2(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("bzip2 -d -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the grzip program? */ int canGrzip(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("grzip -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the gzip program? */ int canGzip(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("gzip -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lrzip program? */ int canRunCommand(const char *cmd) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("%s %s", cmd, redirectArgs) == 0) value = 1; } return (value); } int canLrzip(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lrzip -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lz4 program? */ int canLz4(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lz4 -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lzip program? */ int canLzip(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lzip -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lzma program? */ int canLzma(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lzma -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the lzop program? */ int canLzop(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("lzop -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this platform run the xz program? */ int canXz(void) { static int tested = 0, value = 0; if (!tested) { tested = 1; if (systemf("xz -V %s", redirectArgs) == 0) value = 1; } return (value); } /* * Can this filesystem handle nodump flags. */ #if defined(HAVE_STRUCT_STAT_ST_FLAGS) && defined(UF_NODUMP) int canNodump(void) { const char *path = "cannodumptest"; struct stat sb; assertion_make_file(__FILE__, __LINE__, path, 0644, 0, NULL); if (chflags(path, UF_NODUMP) < 0) return (0); if (stat(path, &sb) < 0) return (0); if (sb.st_flags & UF_NODUMP) return (1); return (0); } #elif defined(EXT2_IOC_GETFLAGS) && defined(HAVE_WORKING_EXT2_IOC_GETFLAGS)\ && defined(EXT2_NODUMP_FL) int canNodump(void) { const char *path = "cannodumptest"; int fd, r, flags; assertion_make_file(__FILE__, __LINE__, path, 0644, 0, NULL); fd = open(path, O_RDONLY | O_NONBLOCK); if (fd < 0) return (0); r = ioctl(fd, EXT2_IOC_GETFLAGS, &flags); if (r < 0) return (0); flags |= EXT2_NODUMP_FL; r = ioctl(fd, EXT2_IOC_SETFLAGS, &flags); if (r < 0) return (0); close(fd); fd = open(path, O_RDONLY | O_NONBLOCK); if (fd < 0) return (0); r = ioctl(fd, EXT2_IOC_GETFLAGS, &flags); if (r < 0) return (0); close(fd); if (flags & EXT2_NODUMP_FL) return (1); return (0); } #else int canNodump() { return (0); } #endif /* * Sleep as needed; useful for verifying disk timestamp changes by * ensuring that the wall-clock time has actually changed before we * go back to re-read something from disk. */ void sleepUntilAfter(time_t t) { while (t >= time(NULL)) #if defined(_WIN32) && !defined(__CYGWIN__) Sleep(500); #else sleep(1); #endif } /* * Call standard system() call, but build up the command line using * sprintf() conventions. */ int systemf(const char *fmt, ...) { char buff[8192]; va_list ap; int r; va_start(ap, fmt); vsprintf(buff, fmt, ap); if (verbosity > VERBOSITY_FULL) logprintf("Cmd: %s\n", buff); r = system(buff); va_end(ap); return (r); } /* * Slurp a file into memory for ease of comparison and testing. * Returns size of file in 'sizep' if non-NULL, null-terminates * data in memory for ease of use. */ char * slurpfile(size_t * sizep, const char *fmt, ...) { char filename[8192]; struct stat st; va_list ap; char *p; ssize_t bytes_read; FILE *f; int r; va_start(ap, fmt); vsprintf(filename, fmt, ap); va_end(ap); f = fopen(filename, "rb"); if (f == NULL) { /* Note: No error; non-existent file is okay here. */ return (NULL); } r = fstat(fileno(f), &st); if (r != 0) { logprintf("Can't stat file %s\n", filename); fclose(f); return (NULL); } p = malloc((size_t)st.st_size + 1); if (p == NULL) { logprintf("Can't allocate %ld bytes of memory to read file %s\n", (long int)st.st_size, filename); fclose(f); return (NULL); } bytes_read = fread(p, 1, (size_t)st.st_size, f); if (bytes_read < st.st_size) { logprintf("Can't read file %s\n", filename); fclose(f); free(p); return (NULL); } p[st.st_size] = '\0'; if (sizep != NULL) *sizep = (size_t)st.st_size; fclose(f); return (p); } /* * Slurp a file into memory for ease of comparison and testing. * Returns size of file in 'sizep' if non-NULL, null-terminates * data in memory for ease of use. */ void dumpfile(const char *filename, void *data, size_t len) { ssize_t bytes_written; FILE *f; f = fopen(filename, "wb"); if (f == NULL) { logprintf("Can't open file %s for writing\n", filename); return; } bytes_written = fwrite(data, 1, len, f); if (bytes_written < (ssize_t)len) logprintf("Can't write file %s\n", filename); fclose(f); } /* Read a uuencoded file from the reference directory, decode, and * write the result into the current directory. */ #define VALID_UUDECODE(c) (c >= 32 && c <= 96) #define UUDECODE(c) (((c) - 0x20) & 0x3f) void extract_reference_file(const char *name) { char buff[1024]; FILE *in, *out; sprintf(buff, "%s/%s.uu", refdir, name); in = fopen(buff, "r"); failure("Couldn't open reference file %s", buff); assert(in != NULL); if (in == NULL) return; /* Read up to and including the 'begin' line. */ for (;;) { if (fgets(buff, sizeof(buff), in) == NULL) { /* TODO: This is a failure. */ return; } if (memcmp(buff, "begin ", 6) == 0) break; } /* Now, decode the rest and write it. */ out = fopen(name, "wb"); while (fgets(buff, sizeof(buff), in) != NULL) { char *p = buff; int bytes; if (memcmp(buff, "end", 3) == 0) break; bytes = UUDECODE(*p++); while (bytes > 0) { int n = 0; /* Write out 1-3 bytes from that. */ if (bytes > 0) { assert(VALID_UUDECODE(p[0])); assert(VALID_UUDECODE(p[1])); n = UUDECODE(*p++) << 18; n |= UUDECODE(*p++) << 12; fputc(n >> 16, out); --bytes; } if (bytes > 0) { assert(VALID_UUDECODE(p[0])); n |= UUDECODE(*p++) << 6; fputc((n >> 8) & 0xFF, out); --bytes; } if (bytes > 0) { assert(VALID_UUDECODE(p[0])); n |= UUDECODE(*p++); fputc(n & 0xFF, out); --bytes; } } } fclose(out); fclose(in); } void copy_reference_file(const char *name) { char buff[1024]; FILE *in, *out; size_t rbytes; sprintf(buff, "%s/%s", refdir, name); in = fopen(buff, "rb"); failure("Couldn't open reference file %s", buff); assert(in != NULL); if (in == NULL) return; /* Now, decode the rest and write it. */ /* Not a lot of error checking here; the input better be right. */ out = fopen(name, "wb"); while ((rbytes = fread(buff, 1, sizeof(buff), in)) > 0) { if (fwrite(buff, 1, rbytes, out) != rbytes) { logprintf("Error: fwrite\n"); break; } } fclose(out); fclose(in); } int is_LargeInode(const char *file) { #if defined(_WIN32) && !defined(__CYGWIN__) BY_HANDLE_FILE_INFORMATION bhfi; int r; r = my_GetFileInformationByName(file, &bhfi); if (r != 0) return (0); return (bhfi.nFileIndexHigh & 0x0000FFFFUL); #else struct stat st; int64_t ino; if (stat(file, &st) < 0) return (0); ino = (int64_t)st.st_ino; return (ino > 0xffffffff); #endif } void extract_reference_files(const char **names) { while (names && *names) extract_reference_file(*names++); } /* * * TEST management * */ /* * "list.h" is simply created by "grep DEFINE_TEST test_*.c"; it has * a line like * DEFINE_TEST(test_function) * for each test. */ /* Use "list.h" to declare all of the test functions. */ #undef DEFINE_TEST #define DEFINE_TEST(name) void name(void); #include "list.h" /* Use "list.h" to create a list of all tests (functions and names). */ #undef DEFINE_TEST #define DEFINE_TEST(n) { n, #n, 0 }, struct test_list_t tests[] = { #include "list.h" }; /* * Summarize repeated failures in the just-completed test. */ static void test_summarize(int failed, int skips_num) { unsigned int i; switch (verbosity) { case VERBOSITY_SUMMARY_ONLY: printf(failed ? "E" : "."); fflush(stdout); break; case VERBOSITY_PASSFAIL: printf(failed ? "FAIL\n" : skips_num ? "ok (S)\n" : "ok\n"); break; } log_console = (verbosity == VERBOSITY_LIGHT_REPORT); for (i = 0; i < sizeof(failed_lines)/sizeof(failed_lines[0]); i++) { if (failed_lines[i].count > 1 && !failed_lines[i].skip) logprintf("%s:%d: Summary: Failed %d times\n", failed_filename, i, failed_lines[i].count); } /* Clear the failure history for the next file. */ failed_filename = NULL; memset(failed_lines, 0, sizeof(failed_lines)); } /* * Actually run a single test, with appropriate setup and cleanup. */ static int test_run(int i, const char *tmpdir) { char workdir[1024]; char logfilename[64]; int failures_before = failures; int skips_before = skips; int oldumask; switch (verbosity) { case VERBOSITY_SUMMARY_ONLY: /* No per-test reports at all */ break; case VERBOSITY_PASSFAIL: /* rest of line will include ok/FAIL marker */ printf("%3d: %-64s", i, tests[i].name); fflush(stdout); break; default: /* Title of test, details will follow */ printf("%3d: %s\n", i, tests[i].name); } /* Chdir to the top-level work directory. */ if (!assertChdir(tmpdir)) { fprintf(stderr, "ERROR: Can't chdir to top work dir %s\n", tmpdir); exit(1); } /* Create a log file for this test. */ sprintf(logfilename, "%s.log", tests[i].name); logfile = fopen(logfilename, "w"); fprintf(logfile, "%s\n\n", tests[i].name); /* Chdir() to a work dir for this specific test. */ snprintf(workdir, sizeof(workdir), "%s/%s", tmpdir, tests[i].name); testworkdir = workdir; if (!assertMakeDir(testworkdir, 0755) || !assertChdir(testworkdir)) { fprintf(stderr, "ERROR: Can't chdir to work dir %s\n", testworkdir); exit(1); } /* Explicitly reset the locale before each test. */ setlocale(LC_ALL, "C"); /* Record the umask before we run the test. */ umask(oldumask = umask(0)); /* * Run the actual test. */ (*tests[i].func)(); /* * Clean up and report afterwards. */ testworkdir = NULL; /* Restore umask */ umask(oldumask); /* Reset locale. */ setlocale(LC_ALL, "C"); /* Reset directory. */ if (!assertChdir(tmpdir)) { fprintf(stderr, "ERROR: Couldn't chdir to temp dir %s\n", tmpdir); exit(1); } /* Report per-test summaries. */ tests[i].failures = failures - failures_before; test_summarize(tests[i].failures, skips - skips_before); /* Close the per-test log file. */ fclose(logfile); logfile = NULL; /* If there were no failures, we can remove the work dir and logfile. */ if (tests[i].failures == 0) { if (!keep_temp_files && assertChdir(tmpdir)) { #if defined(_WIN32) && !defined(__CYGWIN__) /* Make sure not to leave empty directories. * Sometimes a processing of closing files used by tests * is not done, then rmdir will be failed and it will * leave a empty test directory. So we should wait a few * seconds and retry rmdir. */ int r, t; for (t = 0; t < 10; t++) { if (t > 0) Sleep(1000); r = systemf("rmdir /S /Q %s", tests[i].name); if (r == 0) break; } systemf("del %s", logfilename); #else systemf("rm -rf %s", tests[i].name); systemf("rm %s", logfilename); #endif } } /* Return appropriate status. */ return (tests[i].failures); } /* * * * MAIN and support routines. * * */ static void usage(const char *program) { static const int limit = sizeof(tests) / sizeof(tests[0]); int i; printf("Usage: %s [options] ...\n", program); printf("Default is to run all tests.\n"); printf("Otherwise, specify the numbers of the tests you wish to run.\n"); printf("Options:\n"); printf(" -d Dump core after any failure, for debugging.\n"); printf(" -k Keep all temp files.\n"); printf(" Default: temp files for successful tests deleted.\n"); #ifdef PROGRAM printf(" -p Path to executable to be tested.\n"); printf(" Default: path taken from " ENVBASE " environment variable.\n"); #endif printf(" -q Quiet.\n"); printf(" -r Path to dir containing reference files.\n"); printf(" Default: Current directory.\n"); printf(" -u Keep running specifies tests until one fails.\n"); printf(" -v Verbose.\n"); printf("Available tests:\n"); for (i = 0; i < limit; i++) printf(" %d: %s\n", i, tests[i].name); exit(1); } static char * get_refdir(const char *d) { size_t tried_size, buff_size; char *buff, *tried, *pwd = NULL, *p = NULL; #ifdef PATH_MAX buff_size = PATH_MAX; #else buff_size = 8192; #endif buff = calloc(buff_size, 1); if (buff == NULL) { fprintf(stderr, "Unable to allocate memory\n"); exit(1); } /* Allocate a buffer to hold the various directories we checked. */ tried_size = buff_size * 2; tried = calloc(tried_size, 1); if (tried == NULL) { fprintf(stderr, "Unable to allocate memory\n"); exit(1); } /* If a dir was specified, try that */ if (d != NULL) { pwd = NULL; snprintf(buff, buff_size, "%s", d); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); goto failure; } /* Get the current dir. */ #ifdef PATH_MAX pwd = getcwd(NULL, PATH_MAX);/* Solaris getcwd needs the size. */ #else pwd = getcwd(NULL, 0); #endif while (pwd[strlen(pwd) - 1] == '\n') pwd[strlen(pwd) - 1] = '\0'; /* Look for a known file. */ snprintf(buff, buff_size, "%s", pwd); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); snprintf(buff, buff_size, "%s/test", pwd); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); #if defined(LIBRARY) snprintf(buff, buff_size, "%s/%s/test", pwd, LIBRARY); #else snprintf(buff, buff_size, "%s/%s/test", pwd, PROGRAM); #endif p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); #if defined(PROGRAM_ALIAS) snprintf(buff, buff_size, "%s/%s/test", pwd, PROGRAM_ALIAS); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); #endif if (memcmp(pwd, "/usr/obj", 8) == 0) { snprintf(buff, buff_size, "%s", pwd + 8); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); snprintf(buff, buff_size, "%s/test", pwd + 8); p = slurpfile(NULL, "%s/%s", buff, KNOWNREF); if (p != NULL) goto success; strncat(tried, buff, tried_size - strlen(tried) - 1); strncat(tried, "\n", tried_size - strlen(tried) - 1); } failure: printf("Unable to locate known reference file %s\n", KNOWNREF); printf(" Checked following directories:\n%s\n", tried); printf("Use -r option to specify full path to reference directory\n"); #if defined(_WIN32) && !defined(__CYGWIN__) && defined(_DEBUG) DebugBreak(); #endif exit(1); success: free(p); free(pwd); free(tried); /* Copy result into a fresh buffer to reduce memory usage. */ p = strdup(buff); free(buff); return p; } int main(int argc, char **argv) { static const int limit = sizeof(tests) / sizeof(tests[0]); int test_set[sizeof(tests) / sizeof(tests[0])]; int i = 0, j = 0, tests_run = 0, tests_failed = 0, option; time_t now; char *refdir_alloc = NULL; const char *progname; char **saved_argv; const char *tmp, *option_arg, *p; char tmpdir[256], *pwd, *testprogdir, *tmp2 = NULL, *vlevel = NULL; char tmpdir_timestamp[256]; (void)argc; /* UNUSED */ /* Get the current dir. */ #ifdef PATH_MAX pwd = getcwd(NULL, PATH_MAX);/* Solaris getcwd needs the size. */ #else pwd = getcwd(NULL, 0); #endif while (pwd[strlen(pwd) - 1] == '\n') pwd[strlen(pwd) - 1] = '\0'; #if defined(HAVE__CrtSetReportMode) && !defined(__WATCOMC__) /* To stop to run the default invalid parameter handler. */ _set_invalid_parameter_handler(invalid_parameter_handler); /* Disable annoying assertion message box. */ _CrtSetReportMode(_CRT_ASSERT, 0); #endif /* * Name of this program, used to build root of our temp directory * tree. */ progname = p = argv[0]; if ((testprogdir = (char *)malloc(strlen(progname) + 1)) == NULL) { fprintf(stderr, "ERROR: Out of memory."); exit(1); } strcpy(testprogdir, progname); while (*p != '\0') { /* Support \ or / dir separators for Windows compat. */ if (*p == '/' || *p == '\\') { progname = p + 1; i = j; } ++p; j++; } testprogdir[i] = '\0'; #if defined(_WIN32) && !defined(__CYGWIN__) if (testprogdir[0] != '/' && testprogdir[0] != '\\' && !(((testprogdir[0] >= 'a' && testprogdir[0] <= 'z') || (testprogdir[0] >= 'A' && testprogdir[0] <= 'Z')) && testprogdir[1] == ':' && (testprogdir[2] == '/' || testprogdir[2] == '\\'))) #else if (testprogdir[0] != '/') #endif { /* Fixup path for relative directories. */ if ((testprogdir = (char *)realloc(testprogdir, strlen(pwd) + 1 + strlen(testprogdir) + 1)) == NULL) { fprintf(stderr, "ERROR: Out of memory."); exit(1); } memmove(testprogdir + strlen(pwd) + 1, testprogdir, strlen(testprogdir) + 1); memcpy(testprogdir, pwd, strlen(pwd)); testprogdir[strlen(pwd)] = '/'; } #ifdef PROGRAM /* Get the target program from environment, if available. */ testprogfile = getenv(ENVBASE); #endif if (getenv("TMPDIR") != NULL) tmp = getenv("TMPDIR"); else if (getenv("TMP") != NULL) tmp = getenv("TMP"); else if (getenv("TEMP") != NULL) tmp = getenv("TEMP"); else if (getenv("TEMPDIR") != NULL) tmp = getenv("TEMPDIR"); else tmp = "/tmp"; /* Allow -d to be controlled through the environment. */ if (getenv(ENVBASE "_DEBUG") != NULL) dump_on_failure = 1; /* Allow -v to be controlled through the environment. */ if (getenv("_VERBOSITY_LEVEL") != NULL) { vlevel = getenv("_VERBOSITY_LEVEL"); verbosity = atoi(vlevel); if (verbosity < VERBOSITY_SUMMARY_ONLY || verbosity > VERBOSITY_FULL) { /* Unsupported verbosity levels are silently ignored */ vlevel = NULL; verbosity = VERBOSITY_PASSFAIL; } } /* Get the directory holding test files from environment. */ refdir = getenv(ENVBASE "_TEST_FILES"); /* * Parse options, without using getopt(), which isn't available * on all platforms. */ ++argv; /* Skip program name */ while (*argv != NULL) { if (**argv != '-') break; p = *argv++; ++p; /* Skip '-' */ while (*p != '\0') { option = *p++; option_arg = NULL; /* If 'opt' takes an argument, parse that. */ if (option == 'p' || option == 'r') { if (*p != '\0') option_arg = p; else if (*argv == NULL) { fprintf(stderr, "Option -%c requires argument.\n", option); usage(progname); } else option_arg = *argv++; p = ""; /* End of this option word. */ } /* Now, handle the option. */ switch (option) { case 'd': dump_on_failure = 1; break; case 'k': keep_temp_files = 1; break; case 'p': #ifdef PROGRAM testprogfile = option_arg; #else fprintf(stderr, "-p option not permitted\n"); usage(progname); #endif break; case 'q': if (!vlevel) verbosity--; break; case 'r': refdir = option_arg; break; case 'u': until_failure++; break; case 'v': if (!vlevel) verbosity++; break; default: fprintf(stderr, "Unrecognized option '%c'\n", option); usage(progname); } } } /* * Sanity-check that our options make sense. */ #ifdef PROGRAM if (testprogfile == NULL) { if ((tmp2 = (char *)malloc(strlen(testprogdir) + 1 + strlen(PROGRAM) + 1)) == NULL) { fprintf(stderr, "ERROR: Out of memory."); exit(1); } strcpy(tmp2, testprogdir); strcat(tmp2, "/"); strcat(tmp2, PROGRAM); testprogfile = tmp2; } { char *testprg; #if defined(_WIN32) && !defined(__CYGWIN__) /* Command.com sometimes rejects '/' separators. */ testprg = strdup(testprogfile); for (i = 0; testprg[i] != '\0'; i++) { if (testprg[i] == '/') testprg[i] = '\\'; } testprogfile = testprg; #endif /* Quote the name that gets put into shell command lines. */ testprg = malloc(strlen(testprogfile) + 3); strcpy(testprg, "\""); strcat(testprg, testprogfile); strcat(testprg, "\""); testprog = testprg; } #endif #if !defined(_WIN32) && defined(SIGPIPE) { /* Ignore SIGPIPE signals */ struct sigaction sa; sa.sa_handler = SIG_IGN; sigemptyset(&sa.sa_mask); sa.sa_flags = 0; sigaction(SIGPIPE, &sa, NULL); } #endif /* * Create a temp directory for the following tests. * Include the time the tests started as part of the name, * to make it easier to track the results of multiple tests. */ now = time(NULL); for (i = 0; ; i++) { strftime(tmpdir_timestamp, sizeof(tmpdir_timestamp), "%Y-%m-%dT%H.%M.%S", localtime(&now)); sprintf(tmpdir, "%s/%s.%s-%03d", tmp, progname, tmpdir_timestamp, i); if (assertMakeDir(tmpdir,0755)) break; if (i >= 999) { fprintf(stderr, "ERROR: Unable to create temp directory %s\n", tmpdir); exit(1); } } /* * If the user didn't specify a directory for locating * reference files, try to find the reference files in * the "usual places." */ refdir = refdir_alloc = get_refdir(refdir); /* * Banner with basic information. */ printf("\n"); printf("If tests fail or crash, details will be in:\n"); printf(" %s\n", tmpdir); printf("\n"); if (verbosity > VERBOSITY_SUMMARY_ONLY) { printf("Reference files will be read from: %s\n", refdir); #ifdef PROGRAM printf("Running tests on: %s\n", testprog); #endif printf("Exercising: "); fflush(stdout); printf("%s\n", EXTRA_VERSION); } else { printf("Running "); fflush(stdout); } /* * Run some or all of the individual tests. */ saved_argv = argv; do { argv = saved_argv; do { int test_num; test_num = get_test_set(test_set, limit, *argv, tests); if (test_num < 0) { printf("*** INVALID Test %s\n", *argv); free(refdir_alloc); free(testprogdir); usage(progname); return (1); } for (i = 0; i < test_num; i++) { tests_run++; if (test_run(test_set[i], tmpdir)) { tests_failed++; if (until_failure) goto finish; } } if (*argv != NULL) argv++; } while (*argv != NULL); } while (until_failure); finish: /* Must be freed after all tests run */ free(tmp2); free(testprogdir); free(pwd); /* * Report summary statistics. */ if (verbosity > VERBOSITY_SUMMARY_ONLY) { printf("\n"); printf("Totals:\n"); printf(" Tests run: %8d\n", tests_run); printf(" Tests failed: %8d\n", tests_failed); printf(" Assertions checked:%8d\n", assertions); printf(" Assertions failed: %8d\n", failures); printf(" Skips reported: %8d\n", skips); } if (failures) { printf("\n"); printf("Failing tests:\n"); for (i = 0; i < limit; ++i) { if (tests[i].failures) printf(" %d: %s (%d failures)\n", i, tests[i].name, tests[i].failures); } printf("\n"); printf("Details for failing tests: %s\n", tmpdir); printf("\n"); } else { if (verbosity == VERBOSITY_SUMMARY_ONLY) printf("\n"); printf("%d tests passed, no failures\n", tests_run); } free(refdir_alloc); /* If the final tmpdir is empty, we can remove it. */ /* This should be the usual case when all tests succeed. */ assertChdir(".."); rmdir(tmpdir); return (tests_failed ? 1 : 0); } Index: head/contrib/libarchive/libarchive/test/test_archive_cmdline.c =================================================================== --- head/contrib/libarchive/libarchive/test/test_archive_cmdline.c (revision 309589) +++ head/contrib/libarchive/libarchive/test/test_archive_cmdline.c (revision 309590) @@ -1,141 +1,141 @@ /*- * Copyright (c) 2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" __FBSDID("$FreeBSD$"); #define __LIBARCHIVE_TEST #include "archive_cmdline_private.h" DEFINE_TEST(test_archive_cmdline) { struct archive_cmdline *cl; /* Command name only. */ assert((cl = __archive_cmdline_allocate()) != NULL); if (cl == NULL) return; assertEqualInt(ARCHIVE_OK, __archive_cmdline_parse(cl, "gzip")); assertEqualInt(1, cl->argc); assertEqualString("gzip", cl->path); assertEqualString("gzip", cl->argv[0]); assertEqualInt(ARCHIVE_OK, __archive_cmdline_free(cl)); assert((cl = __archive_cmdline_allocate()) != NULL); if (cl == NULL) return; assertEqualInt(ARCHIVE_OK, __archive_cmdline_parse(cl, "gzip ")); assertEqualInt(1, cl->argc); failure("path should not include a space character"); assertEqualString("gzip", cl->path); failure("arg0 should not include a space character"); assertEqualString("gzip", cl->argv[0]); assertEqualInt(ARCHIVE_OK, __archive_cmdline_free(cl)); assert((cl = __archive_cmdline_allocate()) != NULL); if (cl == NULL) return; assertEqualInt(ARCHIVE_OK, __archive_cmdline_parse(cl, "/usr/bin/gzip ")); assertEqualInt(1, cl->argc); failure("path should be a full path"); assertEqualString("/usr/bin/gzip", cl->path); failure("arg0 should not be a full path"); assertEqualString("gzip", cl->argv[0]); assertEqualInt(ARCHIVE_OK, __archive_cmdline_free(cl)); - /* A command line includes space characer. */ + /* A command line includes space character. */ assert((cl = __archive_cmdline_allocate()) != NULL); if (cl == NULL) return; assertEqualInt(ARCHIVE_OK, __archive_cmdline_parse(cl, "\"gzip \"")); assertEqualInt(1, cl->argc); failure("path should include a space character"); assertEqualString("gzip ", cl->path); failure("arg0 should include a space character"); assertEqualString("gzip ", cl->argv[0]); assertEqualInt(ARCHIVE_OK, __archive_cmdline_free(cl)); - /* A command line includes space characer: pattern 2.*/ + /* A command line includes space character: pattern 2.*/ assert((cl = __archive_cmdline_allocate()) != NULL); if (cl == NULL) return; assertEqualInt(ARCHIVE_OK, __archive_cmdline_parse(cl, "\"gzip \"x")); assertEqualInt(1, cl->argc); failure("path should include a space character"); assertEqualString("gzip x", cl->path); failure("arg0 should include a space character"); assertEqualString("gzip x", cl->argv[0]); assertEqualInt(ARCHIVE_OK, __archive_cmdline_free(cl)); - /* A command line includes space characer: pattern 3.*/ + /* A command line includes space character: pattern 3.*/ assert((cl = __archive_cmdline_allocate()) != NULL); if (cl == NULL) return; assertEqualInt(ARCHIVE_OK, __archive_cmdline_parse(cl, "\"gzip \"x\" s \"")); assertEqualInt(1, cl->argc); failure("path should include a space character"); assertEqualString("gzip x s ", cl->path); failure("arg0 should include a space character"); assertEqualString("gzip x s ", cl->argv[0]); assertEqualInt(ARCHIVE_OK, __archive_cmdline_free(cl)); - /* A command line includes space characer: pattern 4.*/ + /* A command line includes space character: pattern 4.*/ assert((cl = __archive_cmdline_allocate()) != NULL); if (cl == NULL) return; assertEqualInt(ARCHIVE_OK, __archive_cmdline_parse(cl, "\"gzip\\\" \"")); assertEqualInt(1, cl->argc); failure("path should include a space character"); assertEqualString("gzip\" ", cl->path); failure("arg0 should include a space character"); assertEqualString("gzip\" ", cl->argv[0]); assertEqualInt(ARCHIVE_OK, __archive_cmdline_free(cl)); /* A command name with a argument. */ assert((cl = __archive_cmdline_allocate()) != NULL); if (cl == NULL) return; assertEqualInt(ARCHIVE_OK, __archive_cmdline_parse(cl, "gzip -d")); assertEqualInt(2, cl->argc); assertEqualString("gzip", cl->path); assertEqualString("gzip", cl->argv[0]); assertEqualString("-d", cl->argv[1]); assertEqualInt(ARCHIVE_OK, __archive_cmdline_free(cl)); /* A command name with two arguments. */ assert((cl = __archive_cmdline_allocate()) != NULL); if (cl == NULL) return; assertEqualInt(ARCHIVE_OK, __archive_cmdline_parse(cl, "gzip -d -q")); assertEqualInt(3, cl->argc); assertEqualString("gzip", cl->path); assertEqualString("gzip", cl->argv[0]); assertEqualString("-d", cl->argv[1]); assertEqualString("-q", cl->argv[2]); assertEqualInt(ARCHIVE_OK, __archive_cmdline_free(cl)); } Index: head/contrib/libarchive/libarchive/test/test_compat_plexus_archiver_tar.c =================================================================== --- head/contrib/libarchive/libarchive/test/test_compat_plexus_archiver_tar.c (nonexistent) +++ head/contrib/libarchive/libarchive/test/test_compat_plexus_archiver_tar.c (revision 309590) @@ -0,0 +1,69 @@ +/*- + * 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 "test.h" +__FBSDID("$FreeBSD"); + +/* + * Verify our ability to read sample files created by plexus-archiver version + * 2.6.2 and lower (project switched to Apache Commons Compress with 2.6.3). + * + * These files may have tar entries with uid and gid header fields filled with + * spaces without any octal digit. + */ + +DEFINE_TEST(test_compat_plexus_archiver_tar) +{ + char name[] = "test_compat_plexus_archiver_tar.tar"; + struct archive_entry *ae; + struct archive *a; + int r; + + assert((a = archive_read_new()) != NULL); + assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); + extract_reference_file(name); + assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, name, + 10240)); + + /* Read first entry. */ + assertEqualIntA(a, ARCHIVE_OK, r = archive_read_next_header(a, &ae)); + if (r != ARCHIVE_OK) { + archive_read_free(a); + return; + } + assertEqualString("commons-logging-1.2/NOTICE.txt", + archive_entry_pathname(ae)); + assertEqualInt(1404583896, archive_entry_mtime(ae)); + assertEqualInt(0100664, archive_entry_mode(ae)); + assertEqualInt(0, archive_entry_uid(ae)); + assertEqualInt(0, archive_entry_gid(ae)); + + /* Verify that the format detection worked. */ + assertEqualInt(archive_filter_code(a, 0), ARCHIVE_FILTER_NONE); + assertEqualInt(archive_format(a), ARCHIVE_FORMAT_TAR_USTAR); + + assertEqualInt(ARCHIVE_OK, archive_read_close(a)); + assertEqualInt(ARCHIVE_OK, archive_read_free(a)); +} Property changes on: head/contrib/libarchive/libarchive/test/test_compat_plexus_archiver_tar.c ___________________________________________________________________ Added: svn:eol-style ## -0,0 +1 ## +native \ No newline at end of property Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Added: svn:mime-type ## -0,0 +1 ## +text/plain \ No newline at end of property Index: head/contrib/libarchive/libarchive/test/test_compat_plexus_archiver_tar.tar.uu =================================================================== --- head/contrib/libarchive/libarchive/test/test_compat_plexus_archiver_tar.tar.uu (nonexistent) +++ head/contrib/libarchive/libarchive/test/test_compat_plexus_archiver_tar.tar.uu (revision 309590) @@ -0,0 +1,49 @@ +begin 644 test_compat_plexus_archiver_tar.tar +M8V]M;6]N ${.TARGET}.tmp @mv ${.TARGET}.tmp ${.TARGET} CLEANTESTS+= list.h list.h.tmp ${PACKAGE}FILES+= README ${PACKAGE}FILES+= test_acl_pax.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_pax_libarchive_2x.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_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_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_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_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_length_at_end.zip.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_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_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_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_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