1 /* $NetBSD: gzip.c,v 1.97 2009/10/11 09:17:21 mrg Exp $ */
4 * Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
19 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
21 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
22 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
23 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
24 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 #include <sys/cdefs.h>
32 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004, 2006\
33 Matthew R. Green. All rights reserved.");
38 * gzip.c -- GPL free gzip using zlib.
40 * RFC 1950 covers the zlib format
41 * RFC 1951 covers the deflate format
42 * RFC 1952 covers the gzip format
45 * - use mmap where possible
46 * - handle some signals better (remove outfile?)
47 * - make bzip2/compress -v/-t/-l support work as well as possible
50 #include <sys/param.h>
70 #define PRIdOFF PRId64
73 /* what type of file are we dealing with */
76 #ifndef NO_BZIP2_SUPPORT
79 #ifndef NO_COMPRESS_SUPPORT
82 #ifndef NO_PACK_SUPPORT
89 #ifndef NO_BZIP2_SUPPORT
92 #define BZ2_SUFFIX ".bz2"
93 #define BZIP2_MAGIC "\102\132\150"
96 #ifndef NO_COMPRESS_SUPPORT
98 #define Z_MAGIC "\037\235"
101 #ifndef NO_PACK_SUPPORT
102 #define PACK_MAGIC "\037\036"
105 #define GZ_SUFFIX ".gz"
107 #define BUFLEN (64 * 1024)
109 #define GZIP_MAGIC0 0x1F
110 #define GZIP_MAGIC1 0x8B
111 #define GZIP_OMAGIC1 0x9E
113 #define GZIP_TIMESTAMP (off_t)4
114 #define GZIP_ORIGNAME (off_t)10
116 #define HEAD_CRC 0x02
117 #define EXTRA_FIELD 0x04
118 #define ORIG_NAME 0x08
121 #define OS_CODE 3 /* Unix */
126 const char *normal; /* for unzip - must not be longer than zipped */
128 static suffixes_t suffixes[] = {
129 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N}
130 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */
132 SUFFIX(GZ_SUFFIX, ""),
137 SUFFIX(".taz", ".tar"),
138 SUFFIX(".tgz", ".tar"),
139 #ifndef NO_BZIP2_SUPPORT
140 SUFFIX(BZ2_SUFFIX, ""),
141 SUFFIX(".tbz", ".tar"),
142 SUFFIX(".tbz2", ".tar"),
144 #ifndef NO_COMPRESS_SUPPORT
145 SUFFIX(Z_SUFFIX, ""),
147 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */
151 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0])
152 #define SUFFIX_MAXLEN 30
154 static const char gzip_version[] = "FreeBSD gzip 20100407";
157 static const char gzip_copyright[] = \
158 " Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green\n"
159 " All rights reserved.\n"
161 " Redistribution and use in source and binary forms, with or without\n"
162 " modification, are permitted provided that the following conditions\n"
164 " 1. Redistributions of source code must retain the above copyright\n"
165 " notice, this list of conditions and the following disclaimer.\n"
166 " 2. Redistributions in binary form must reproduce the above copyright\n"
167 " notice, this list of conditions and the following disclaimer in the\n"
168 " documentation and/or other materials provided with the distribution.\n"
170 " THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR\n"
171 " IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES\n"
172 " OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.\n"
173 " IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,\n"
174 " INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,\n"
175 " BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;\n"
176 " LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED\n"
177 " AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,\n"
178 " OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY\n"
179 " OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF\n"
183 static int cflag; /* stdout mode */
184 static int dflag; /* decompress mode */
185 static int lflag; /* list mode */
186 static int numflag = 6; /* gzip -1..-9 value */
189 static int fflag; /* force mode */
190 static int kflag; /* don't delete input files */
191 static int nflag; /* don't save name/timestamp */
192 static int Nflag; /* don't restore name/timestamp */
193 static int qflag; /* quiet mode */
194 static int rflag; /* recursive mode */
195 static int tflag; /* test */
196 static int vflag; /* verbose mode */
202 static int exit_value = 0; /* exit value */
204 static char *infile; /* name of file coming in */
206 static void maybe_err(const char *fmt, ...) __dead2
207 __attribute__((__format__(__printf__, 1, 2)));
208 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT)
209 static void maybe_errx(const char *fmt, ...) __dead2
210 __attribute__((__format__(__printf__, 1, 2)));
212 static void maybe_warn(const char *fmt, ...)
213 __attribute__((__format__(__printf__, 1, 2)));
214 static void maybe_warnx(const char *fmt, ...)
215 __attribute__((__format__(__printf__, 1, 2)));
216 static enum filetype file_gettype(u_char *);
218 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz)
220 static off_t gz_compress(int, int, off_t *, const char *, uint32_t);
221 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *);
222 static off_t file_compress(char *, char *, size_t);
223 static off_t file_uncompress(char *, char *, size_t);
224 static void handle_pathname(char *);
225 static void handle_file(char *, struct stat *);
226 static void handle_stdin(void);
227 static void handle_stdout(void);
228 static void print_ratio(off_t, off_t, FILE *);
229 static void print_list(int fd, off_t, const char *, time_t);
230 static void usage(void);
231 static void display_version(void);
233 static void display_license(void);
235 static const suffixes_t *check_suffix(char *, int);
236 static ssize_t read_retry(int, void *, size_t);
239 #define unlink_input(f, sb) unlink(f)
241 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd);
242 static void prepend_gzip(char *, int *, char ***);
243 static void handle_dir(char *);
244 static void print_verbage(const char *, const char *, off_t, off_t);
245 static void print_test(const char *, int);
246 static void copymodes(int fd, const struct stat *, const char *file);
247 static int check_outfile(const char *outfile);
250 #ifndef NO_BZIP2_SUPPORT
251 static off_t unbzip2(int, int, char *, size_t, off_t *);
254 #ifndef NO_COMPRESS_SUPPORT
255 static FILE *zdopen(int);
256 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *);
259 #ifndef NO_PACK_SUPPORT
260 static off_t unpack(int, int, char *, size_t, off_t *);
263 int main(int, char **p);
266 #define getopt_long(a,b,c,d,e) getopt(a,b,c)
268 static const struct option longopts[] = {
269 { "stdout", no_argument, 0, 'c' },
270 { "to-stdout", no_argument, 0, 'c' },
271 { "decompress", no_argument, 0, 'd' },
272 { "uncompress", no_argument, 0, 'd' },
273 { "force", no_argument, 0, 'f' },
274 { "help", no_argument, 0, 'h' },
275 { "keep", no_argument, 0, 'k' },
276 { "list", no_argument, 0, 'l' },
277 { "no-name", no_argument, 0, 'n' },
278 { "name", no_argument, 0, 'N' },
279 { "quiet", no_argument, 0, 'q' },
280 { "recursive", no_argument, 0, 'r' },
281 { "suffix", required_argument, 0, 'S' },
282 { "test", no_argument, 0, 't' },
283 { "verbose", no_argument, 0, 'v' },
284 { "version", no_argument, 0, 'V' },
285 { "fast", no_argument, 0, '1' },
286 { "best", no_argument, 0, '9' },
287 { "ascii", no_argument, 0, 'a' },
288 { "license", no_argument, 0, 'L' },
289 { NULL, no_argument, 0, 0 },
294 main(int argc, char **argv)
296 const char *progname = getprogname();
303 /* XXX set up signals */
306 if ((gzip = getenv("GZIP")) != NULL)
307 prepend_gzip(gzip, &argc, &argv);
312 * handle being called `gunzip', `zcat' and `gzcat'
314 if (strcmp(progname, "gunzip") == 0)
316 else if (strcmp(progname, "zcat") == 0 ||
317 strcmp(progname, "gzcat") == 0)
321 #define OPT_LIST "123456789cdhltV"
323 #define OPT_LIST "123456789acdfhklLNnqrS:tVv"
326 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) {
328 case '1': case '2': case '3':
329 case '4': case '5': case '6':
330 case '7': case '8': case '9':
348 fprintf(stderr, "%s: option --ascii ignored on this system\n", progname);
374 len = strlen(optarg);
376 if (len > SUFFIX_MAXLEN)
377 errx(1, "incorrect suffix: '%s': too long", optarg);
378 suffixes[0].zipped = optarg;
379 suffixes[0].ziplen = len;
381 suffixes[NUM_SUFFIXES - 1].zipped = "";
382 suffixes[NUM_SUFFIXES - 1].ziplen = 0;
403 if (dflag) /* stdin mode */
405 else /* stdout mode */
409 handle_pathname(argv[0]);
413 if (qflag == 0 && lflag && argc > 1)
414 print_list(-1, 0, "(totals)", 0);
419 /* maybe print a warning */
421 maybe_warn(const char *fmt, ...)
434 /* ... without an errno. */
436 maybe_warnx(const char *fmt, ...)
449 /* maybe print an error */
451 maybe_err(const char *fmt, ...)
463 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT)
464 /* ... without an errno. */
466 maybe_errx(const char *fmt, ...)
480 /* split up $GZIP and prepend it to the argument list */
482 prepend_gzip(char *gzip, int *argc, char ***argv)
484 char *s, **nargv, **ac;
487 /* scan how many arguments there are */
489 while (*s == ' ' || *s == '\t')
494 while (*s != ' ' && *s != '\t')
506 nargv = (char **)malloc((*argc + 1) * sizeof(char *));
510 /* stash this away */
513 /* copy the program name first */
515 nargv[i++] = *(ac++);
517 /* take a copy of $GZIP and add it to the array */
522 /* Skip whitespaces. */
523 while (*s == ' ' || *s == '\t')
528 /* Find the end of this argument. */
529 while (*s != ' ' && *s != '\t')
531 /* Argument followed by NUL. */
533 /* Terminate by overwriting ' ' or '\t' with NUL. */
538 /* copy the original arguments and a NULL */
540 nargv[i++] = *(ac++);
545 /* compress input to output. Return bytes read, -1 on error */
547 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime)
550 char *outbufp, *inbufp;
551 off_t in_tot = 0, out_tot = 0;
556 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0,
561 outbufp = malloc(BUFLEN);
562 inbufp = malloc(BUFLEN);
563 if (outbufp == NULL || inbufp == NULL) {
564 maybe_err("malloc failed");
568 memset(&z, 0, sizeof z);
574 memcpy(outbufp, header, sizeof header);
582 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s",
583 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED,
584 *origname ? ORIG_NAME : 0,
587 (mtime >> 16) & 0xff,
588 (mtime >> 24) & 0xff,
589 numflag == 1 ? 4 : numflag == 9 ? 2 : 0,
592 /* this need PATH_MAX > BUFLEN ... */
593 maybe_err("snprintf");
598 z.next_out = (unsigned char *)outbufp + i;
599 z.avail_out = BUFLEN - i;
601 error = deflateInit2(&z, numflag, Z_DEFLATED,
602 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY);
604 maybe_warnx("deflateInit2 failed");
609 crc = crc32(0L, Z_NULL, 0);
611 if (z.avail_out == 0) {
612 if (write(out, outbufp, BUFLEN) != BUFLEN) {
619 z.next_out = (unsigned char *)outbufp;
620 z.avail_out = BUFLEN;
623 if (z.avail_in == 0) {
624 in_size = read(in, inbufp, BUFLEN);
633 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size);
635 z.next_in = (unsigned char *)inbufp;
636 z.avail_in = in_size;
639 error = deflate(&z, Z_NO_FLUSH);
640 if (error != Z_OK && error != Z_STREAM_END) {
641 maybe_warnx("deflate failed");
652 error = deflate(&z, Z_FINISH);
653 if (error != Z_OK && error != Z_STREAM_END) {
654 maybe_warnx("deflate failed");
659 len = (char *)z.next_out - outbufp;
661 w = write(out, outbufp, len);
662 if (w == -1 || (size_t)w != len) {
668 z.next_out = (unsigned char *)outbufp;
669 z.avail_out = BUFLEN;
671 if (error == Z_STREAM_END)
675 if (deflateEnd(&z) != Z_OK) {
676 maybe_warnx("deflateEnd failed");
681 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c",
683 (int)(crc >> 8) & 0xff,
684 (int)(crc >> 16) & 0xff,
685 (int)(crc >> 24) & 0xff,
687 (int)(in_tot >> 8) & 0xff,
688 (int)(in_tot >> 16) & 0xff,
689 (int)(in_tot >> 24) & 0xff);
691 maybe_err("snprintf");
692 if (write(out, outbufp, i) != i) {
709 * uncompress input to output then close the input. return the
710 * uncompressed size written, and put the compressed sized read
714 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep,
715 const char *filename)
718 char *outbufp, *inbufp;
719 off_t out_tot = -1, in_tot = 0;
720 uint32_t out_sub_tot = 0;
738 } state = GZSTATE_MAGIC0;
739 int flags = 0, skip_count = 0;
740 int error = Z_STREAM_ERROR, done_reading = 0;
745 #define ADVANCE() { z.next_in++; z.avail_in--; }
747 if ((outbufp = malloc(BUFLEN)) == NULL) {
748 maybe_err("malloc failed");
751 if ((inbufp = malloc(BUFLEN)) == NULL) {
752 maybe_err("malloc failed");
756 memset(&z, 0, sizeof z);
758 z.next_in = (unsigned char *)pre;
759 z.avail_out = BUFLEN;
760 z.next_out = (unsigned char *)outbufp;
769 if ((z.avail_in == 0 || needmore) && done_reading == 0) {
772 if (z.avail_in > 0) {
773 memmove(inbufp, z.next_in, z.avail_in);
775 z.next_in = (unsigned char *)inbufp;
776 in_size = read(in, z.next_in + z.avail_in,
777 BUFLEN - z.avail_in);
780 maybe_warn("failed to read stdin");
782 } else if (in_size == 0) {
786 z.avail_in += in_size;
791 if (z.avail_in == 0) {
792 if (done_reading && state != GZSTATE_MAGIC0) {
793 maybe_warnx("%s: unexpected end of file",
801 if (*z.next_in != GZIP_MAGIC0) {
803 maybe_warnx("%s: trailing garbage "
804 "ignored", filename);
807 maybe_warnx("input not gziped (MAGIC0)");
813 crc = crc32(0L, Z_NULL, 0);
817 if (*z.next_in != GZIP_MAGIC1 &&
818 *z.next_in != GZIP_OMAGIC1) {
819 maybe_warnx("input not gziped (MAGIC1)");
827 if (*z.next_in != Z_DEFLATED) {
828 maybe_warnx("unknown compression method");
842 case GZSTATE_SKIPPING:
843 if (skip_count > 0) {
851 if ((flags & EXTRA_FIELD) == 0) {
852 state = GZSTATE_ORIGNAME;
855 skip_count = *z.next_in;
861 skip_count |= ((*z.next_in) << 8);
867 if (skip_count > 0) {
874 case GZSTATE_ORIGNAME:
875 if ((flags & ORIG_NAME) == 0) {
884 case GZSTATE_COMMENT:
885 if ((flags & COMMENT) == 0) {
894 case GZSTATE_HEAD_CRC1:
895 if (flags & HEAD_CRC)
902 case GZSTATE_HEAD_CRC2:
903 if (skip_count > 0) {
911 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) {
912 maybe_warnx("failed to inflateInit");
919 error = inflate(&z, Z_FINISH);
921 /* Z_BUF_ERROR goes with Z_FINISH... */
928 maybe_warnx("Z_NEED_DICT error");
931 maybe_warnx("data stream error");
934 maybe_warnx("internal stream error");
937 maybe_warnx("memory allocation error");
941 maybe_warn("unknown error from inflate(): %d",
944 wr = BUFLEN - z.avail_out;
947 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr);
950 /* don't write anything with -t */
953 write(out, outbufp, wr) != wr) {
954 maybe_warn("error writing to output");
962 if (error == Z_STREAM_END) {
967 z.next_out = (unsigned char *)outbufp;
968 z.avail_out = BUFLEN;
975 if (z.avail_in < 4) {
980 maybe_warnx("truncated input");
983 origcrc = ((unsigned)z.next_in[0] & 0xff) |
984 ((unsigned)z.next_in[1] & 0xff) << 8 |
985 ((unsigned)z.next_in[2] & 0xff) << 16 |
986 ((unsigned)z.next_in[3] & 0xff) << 24;
987 if (origcrc != crc) {
988 maybe_warnx("invalid compressed"
997 if (!z.avail_in && done_reading) {
1006 if (z.avail_in < 4) {
1007 if (!done_reading) {
1011 maybe_warnx("truncated input");
1014 origlen = ((unsigned)z.next_in[0] & 0xff) |
1015 ((unsigned)z.next_in[1] & 0xff) << 8 |
1016 ((unsigned)z.next_in[2] & 0xff) << 16 |
1017 ((unsigned)z.next_in[3] & 0xff) << 24;
1019 if (origlen != out_sub_tot) {
1020 maybe_warnx("invalid compressed"
1021 " data--length error");
1030 maybe_warnx("decompression error");
1033 state = GZSTATE_MAGIC0;
1042 if (state > GZSTATE_INIT)
1056 * set the owner, mode, flags & utimes using the given file descriptor.
1057 * file is only used in possible warning messages.
1060 copymodes(int fd, const struct stat *sbp, const char *file)
1062 struct timeval times[2];
1066 * If we have no info on the input, give this file some
1067 * default values and return..
1070 mode_t mask = umask(022);
1072 (void)fchmod(fd, DEFFILEMODE & ~mask);
1078 /* if the chown fails, remove set-id bits as-per compress(1) */
1079 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) {
1081 maybe_warn("couldn't fchown: %s", file);
1082 sb.st_mode &= ~(S_ISUID|S_ISGID);
1085 /* we only allow set-id and the 9 normal permission bits */
1086 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO;
1087 if (fchmod(fd, sb.st_mode) < 0)
1088 maybe_warn("couldn't fchmod: %s", file);
1090 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec);
1091 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec);
1092 if (futimes(fd, times) < 0)
1093 maybe_warn("couldn't utimes: %s", file);
1095 /* only try flags if they exist already */
1096 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0)
1097 maybe_warn("couldn't fchflags: %s", file);
1101 /* what sort of file is this? */
1102 static enum filetype
1103 file_gettype(u_char *buf)
1106 if (buf[0] == GZIP_MAGIC0 &&
1107 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1))
1110 #ifndef NO_BZIP2_SUPPORT
1111 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 &&
1112 buf[3] >= '0' && buf[3] <= '9')
1116 #ifndef NO_COMPRESS_SUPPORT
1117 if (memcmp(buf, Z_MAGIC, 2) == 0)
1121 #ifndef NO_PACK_SUPPORT
1122 if (memcmp(buf, PACK_MAGIC, 2) == 0)
1130 /* check the outfile is OK. */
1132 check_outfile(const char *outfile)
1137 if (lflag == 0 && stat(outfile, &sb) == 0) {
1140 else if (isatty(STDIN_FILENO)) {
1141 char ans[10] = { 'n', '\0' }; /* default */
1143 fprintf(stderr, "%s already exists -- do you wish to "
1144 "overwrite (y or n)? " , outfile);
1145 (void)fgets(ans, sizeof(ans) - 1, stdin);
1146 if (ans[0] != 'y' && ans[0] != 'Y') {
1147 fprintf(stderr, "\tnot overwriting\n");
1152 maybe_warnx("%s already exists -- skipping", outfile);
1160 unlink_input(const char *file, const struct stat *sb)
1166 if (stat(file, &nsb) != 0)
1167 /* Must be gone alrady */
1169 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino)
1170 /* Definitely a different file */
1176 static const suffixes_t *
1177 check_suffix(char *file, int xlate)
1179 const suffixes_t *s;
1180 int len = strlen(file);
1183 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) {
1184 /* if it doesn't fit in "a.suf", don't bother */
1185 if (s->ziplen >= len)
1187 sp = file + len - s->ziplen;
1188 if (strcmp(s->zipped, sp) != 0)
1191 strcpy(sp, s->normal);
1198 * compress the given file: create a corresponding .gz file and remove the
1202 file_compress(char *file, char *outfile, size_t outsize)
1208 struct stat isb, osb;
1209 const suffixes_t *suff;
1212 in = open(file, O_RDONLY);
1214 maybe_warn("can't open %s", file);
1220 if (fstat(in, &isb) == 0) {
1221 if (isb.st_nlink > 1 && fflag == 0) {
1222 maybe_warnx("%s has %d other link%s -- "
1223 "skipping", file, isb.st_nlink - 1,
1224 isb.st_nlink == 1 ? "" : "s");
1230 if (fflag == 0 && (suff = check_suffix(file, 0))
1231 && suff->zipped[0] != 0) {
1232 maybe_warnx("%s already has %s suffix -- unchanged",
1233 file, suff->zipped);
1239 /* Add (usually) .gz to filename */
1240 if ((size_t)snprintf(outfile, outsize, "%s%s",
1241 file, suffixes[0].zipped) >= outsize)
1242 memcpy(outfile + outsize - suffixes[0].ziplen - 1,
1243 suffixes[0].zipped, suffixes[0].ziplen + 1);
1246 if (check_outfile(outfile) == 0) {
1254 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600);
1256 maybe_warn("could not create output: %s", outfile);
1261 out = STDOUT_FILENO;
1263 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime);
1268 * If there was an error, insize will be -1.
1269 * If we compressed to stdout, just return the size.
1270 * Otherwise stat the file and check it is the correct size.
1271 * We only blow away the file if we can stat the output and it
1272 * has the expected size.
1275 return insize == -1 ? -1 : size;
1278 if (fstat(out, &osb) != 0) {
1279 maybe_warn("couldn't stat: %s", outfile);
1283 if (osb.st_size != size) {
1284 maybe_warnx("output file: %s wrong size (%" PRIdOFF
1285 " != %" PRIdOFF "), deleting",
1286 outfile, osb.st_size, size);
1290 copymodes(out, &isb, outfile);
1292 if (close(out) == -1)
1293 maybe_warn("couldn't close output");
1295 /* output is good, ok to delete input */
1296 unlink_input(file, &isb);
1301 if (close(out) == -1)
1302 maybe_warn("couldn't close output");
1304 maybe_warnx("leaving original %s", file);
1310 /* uncompress the given file and remove the original */
1312 file_uncompress(char *file, char *outfile, size_t outsize)
1314 struct stat isb, osb;
1317 unsigned char header1[4];
1318 enum filetype method;
1319 int fd, ofd, zfd = -1;
1322 time_t timestamp = 0;
1323 unsigned char name[PATH_MAX + 1];
1326 /* gather the old name info */
1328 fd = open(file, O_RDONLY);
1330 maybe_warn("can't open %s", file);
1334 strlcpy(outfile, file, outsize);
1335 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) {
1336 maybe_warnx("%s: unknown suffix -- ignored", file);
1340 rbytes = read(fd, header1, sizeof header1);
1341 if (rbytes != sizeof header1) {
1342 /* we don't want to fail here. */
1348 maybe_warn("can't read %s", file);
1350 goto unexpected_EOF;
1354 method = file_gettype(header1);
1357 if (fflag == 0 && method == FT_UNKNOWN) {
1358 maybe_warnx("%s: not in gzip format", file);
1365 if (method == FT_GZIP && Nflag) {
1366 unsigned char ts[4]; /* timestamp */
1368 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP);
1369 if (rv >= 0 && rv < (ssize_t)(sizeof ts))
1370 goto unexpected_EOF;
1373 maybe_warn("can't read %s", file);
1376 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0];
1378 if (header1[3] & ORIG_NAME) {
1379 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME);
1381 maybe_warn("can't read %s", file);
1385 /* preserve original directory name */
1386 char *dp = strrchr(file, '/');
1391 snprintf(outfile, outsize, "%.*s%.*s",
1393 file, (int) rbytes, name);
1398 lseek(fd, 0, SEEK_SET);
1400 if (cflag == 0 || lflag) {
1401 if (fstat(fd, &isb) != 0)
1404 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) {
1405 maybe_warnx("%s has %d other links -- skipping",
1406 file, isb.st_nlink - 1);
1409 if (nflag == 0 && timestamp)
1410 isb.st_mtime = timestamp;
1411 if (check_outfile(outfile) == 0)
1416 if (cflag == 0 && lflag == 0) {
1417 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600);
1418 if (zfd == STDOUT_FILENO) {
1419 /* We won't close STDOUT_FILENO later... */
1421 close(STDOUT_FILENO);
1424 maybe_warn("can't open %s", outfile);
1428 zfd = STDOUT_FILENO;
1430 #ifndef NO_BZIP2_SUPPORT
1431 if (method == FT_BZIP2) {
1435 maybe_warnx("no -l with bzip2 files");
1439 size = unbzip2(fd, zfd, NULL, 0, NULL);
1443 #ifndef NO_COMPRESS_SUPPORT
1444 if (method == FT_Z) {
1449 maybe_warnx("no -l with Lempel-Ziv files");
1453 if ((in = zdopen(fd)) == NULL) {
1454 maybe_warn("zdopen for read: %s", file);
1458 out = fdopen(dup(zfd), "w");
1460 maybe_warn("fdopen for write: %s", outfile);
1465 size = zuncompress(in, out, NULL, 0, NULL);
1466 /* need to fclose() if ferror() is true... */
1467 if (ferror(in) | fclose(in)) {
1468 maybe_warn("failed infile fclose");
1472 if (fclose(out) != 0) {
1473 maybe_warn("failed outfile fclose");
1480 #ifndef NO_PACK_SUPPORT
1481 if (method == FT_PACK) {
1483 maybe_warnx("no -l with packed files");
1487 size = unpack(fd, zfd, NULL, 0, NULL);
1492 if (method == FT_UNKNOWN) {
1494 maybe_warnx("no -l for unknown filetypes");
1497 size = cat_fd(NULL, 0, NULL, fd);
1502 print_list(fd, isb.st_size, outfile, isb.st_mtime);
1504 return -1; /* XXX */
1507 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file);
1511 maybe_warn("couldn't close input");
1512 if (zfd != STDOUT_FILENO && close(zfd) != 0)
1513 maybe_warn("couldn't close output");
1518 maybe_warnx("%s: uncompress failed", file);
1522 /* if testing, or we uncompressed to stdout, this is all we need */
1527 /* if we are uncompressing to stdin, don't remove the file. */
1532 * if we create a file...
1535 * if we can't stat the file don't remove the file.
1538 ofd = open(outfile, O_RDWR, 0);
1540 maybe_warn("couldn't open (leaving original): %s",
1544 if (fstat(ofd, &osb) != 0) {
1545 maybe_warn("couldn't stat (leaving original): %s",
1550 if (osb.st_size != size) {
1551 maybe_warnx("stat gave different size: %" PRIdOFF
1552 " != %" PRIdOFF " (leaving original)",
1558 unlink_input(file, &isb);
1560 copymodes(ofd, &isb, outfile);
1566 maybe_warnx("%s: unexpected end of file", file);
1570 if (zfd != -1 && zfd != STDOUT_FILENO)
1577 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd)
1584 w = write(STDOUT_FILENO, prepend, count);
1585 if (w == -1 || (size_t)w != count) {
1586 maybe_warn("write to stdout");
1592 rv = read(fd, buf, sizeof buf);
1596 maybe_warn("read from fd %d", fd);
1600 if (write(STDOUT_FILENO, buf, rv) != rv) {
1601 maybe_warn("write to stdout");
1616 unsigned char header1[4];
1618 enum filetype method;
1620 #ifndef NO_COMPRESS_SUPPORT
1625 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) {
1626 maybe_warnx("standard input is a terminal -- ignoring");
1634 /* XXX could read the whole file, etc. */
1635 if (fstat(STDIN_FILENO, &isb) < 0) {
1636 maybe_warn("fstat");
1639 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime);
1643 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1);
1644 if (bytes_read == -1) {
1645 maybe_warn("can't read stdin");
1647 } else if (bytes_read != sizeof(header1)) {
1648 maybe_warnx("(stdin): unexpected end of file");
1652 method = file_gettype(header1);
1657 maybe_warnx("unknown compression format");
1660 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO);
1664 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO,
1665 (char *)header1, sizeof header1, &gsize, "(stdin)");
1667 #ifndef NO_BZIP2_SUPPORT
1669 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO,
1670 (char *)header1, sizeof header1, &gsize);
1673 #ifndef NO_COMPRESS_SUPPORT
1675 if ((in = zdopen(STDIN_FILENO)) == NULL) {
1676 maybe_warnx("zopen of stdin");
1680 usize = zuncompress(in, stdout, (char *)header1, sizeof header1, &gsize);
1684 #ifndef NO_PACK_SUPPORT
1686 usize = unpack(STDIN_FILENO, STDOUT_FILENO,
1687 (char *)header1, sizeof header1, &gsize);
1693 if (vflag && !tflag && usize != -1 && gsize != -1)
1694 print_verbage(NULL, NULL, usize, gsize);
1696 print_test("(stdin)", usize != -1);
1711 if (fflag == 0 && isatty(STDOUT_FILENO)) {
1712 maybe_warnx("standard output is a terminal -- ignoring");
1716 /* If stdin is a file use it's mtime, otherwise use current time */
1717 ret = fstat(STDIN_FILENO, &sb);
1721 maybe_warn("Can't stat stdin");
1726 if (S_ISREG(sb.st_mode))
1727 mtime = (uint32_t)sb.st_mtime;
1729 systime = time(NULL);
1731 if (systime == -1) {
1736 mtime = (uint32_t)systime;
1739 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime);
1741 if (vflag && !tflag && usize != -1 && gsize != -1)
1742 print_verbage(NULL, NULL, usize, gsize);
1746 /* do what is asked for, for the path name */
1748 handle_pathname(char *path)
1750 char *opath = path, *s = NULL;
1755 /* check for stdout/stdin */
1756 if (path[0] == '-' && path[1] == '\0') {
1765 if (stat(path, &sb) != 0) {
1766 /* lets try <path>.gz if we're decompressing */
1767 if (dflag && s == NULL && errno == ENOENT) {
1769 slen = suffixes[0].ziplen;
1770 s = malloc(len + slen + 1);
1772 maybe_err("malloc");
1773 memcpy(s, path, len);
1774 memcpy(s + len, suffixes[0].zipped, slen + 1);
1778 maybe_warn("can't stat: %s", opath);
1782 if (S_ISDIR(sb.st_mode)) {
1788 maybe_warnx("%s is a directory", path);
1792 if (S_ISREG(sb.st_mode))
1793 handle_file(path, &sb);
1795 maybe_warnx("%s is not a regular file", path);
1802 /* compress/decompress a file */
1804 handle_file(char *file, struct stat *sbp)
1807 char outfile[PATH_MAX];
1811 usize = file_uncompress(file, outfile, sizeof(outfile));
1814 print_test(file, usize != -1);
1818 gsize = sbp->st_size;
1820 gsize = file_compress(file, outfile, sizeof(outfile));
1823 usize = sbp->st_size;
1828 if (vflag && !tflag)
1829 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize);
1834 /* this is used with -r to recursively descend directories */
1836 handle_dir(char *dir)
1844 fts = fts_open(path_argv, FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1846 warn("couldn't fts_open %s", dir);
1850 while ((entry = fts_read(fts))) {
1851 switch(entry->fts_info) {
1859 maybe_warn("%s", entry->fts_path);
1862 handle_file(entry->fts_path, entry->fts_statp);
1865 (void)fts_close(fts);
1869 /* print a ratio - size reduction as a fraction of uncompressed size */
1871 print_ratio(off_t in, off_t out, FILE *where)
1873 int percent10; /* 10 * percent */
1881 * Output is more than double size of input! print -99.9%
1882 * Quite possibly we've failed to get the original size.
1887 * We only need 12 bits of result from the final division,
1888 * so reduce the values until a 32bit division will suffice.
1890 while (in > 0x100000) {
1895 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000;
1900 len = snprintf(buff, sizeof buff, "%2.2d.", percent10);
1901 /* Move the '.' to before the last digit */
1902 buff[len - 1] = buff[len - 2];
1903 buff[len - 2] = '.';
1904 fprintf(where, "%5s%%", buff);
1908 /* print compression statistics, and the new name (if there is one!) */
1910 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize)
1913 fprintf(stderr, "%s:%s ", file,
1914 strlen(file) < 7 ? "\t\t" : "\t");
1915 print_ratio(usize, gsize, stderr);
1917 fprintf(stderr, " -- replaced with %s", nfile);
1918 fprintf(stderr, "\n");
1922 /* print test results */
1924 print_test(const char *file, int ok)
1927 if (exit_value == 0 && ok == 0)
1929 fprintf(stderr, "%s:%s %s\n", file,
1930 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK");
1935 /* print a file's info ala --list */
1937 compressed uncompressed ratio uncompressed_name
1938 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar
1941 print_list(int fd, off_t out, const char *outfile, time_t ts)
1943 static int first = 1;
1945 static off_t in_tot, out_tot;
1953 printf("method crc date time ");
1956 printf(" compressed uncompressed "
1957 "ratio uncompressed_name\n");
1969 /* read the last 4 bytes - this is the uncompressed size */
1970 rv = lseek(fd, (off_t)(-8), SEEK_END);
1972 unsigned char buf[8];
1975 rv = read(fd, (char *)buf, sizeof(buf));
1977 maybe_warn("read of uncompressed size");
1978 else if (rv != sizeof(buf))
1979 maybe_warnx("read of uncompressed size");
1982 usize = buf[4] | buf[5] << 8 |
1983 buf[6] << 16 | buf[7] << 24;
1986 crc = buf[0] | buf[1] << 8 |
1987 buf[2] << 16 | buf[3] << 24;
1994 if (vflag && fd == -1)
1997 char *date = ctime(&ts);
1999 /* skip the day, 1/100th second, and year */
2002 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date);
2007 (void)&ts; /* XXX */
2009 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in);
2010 print_ratio(in, out, stdout);
2011 printf(" %s\n", outfile);
2014 /* display the usage of NetBSD gzip */
2019 fprintf(stderr, "%s\n", gzip_version);
2022 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n",
2024 "usage: %s [-123456789acdfhklLNnqrtVv] [-S .suffix] [<file> [<file> ...]]\n"
2025 " -1 --fast fastest (worst) compression\n"
2026 " -2 .. -8 set compression level\n"
2027 " -9 --best best (slowest) compression\n"
2028 " -c --stdout write to stdout, keep original files\n"
2030 " -d --decompress uncompress files\n"
2032 " -f --force force overwriting & compress links\n"
2033 " -h --help display this help\n"
2034 " -k --keep don't delete input files during operation\n"
2035 " -l --list list compressed file contents\n"
2036 " -N --name save or restore original file name and time stamp\n"
2037 " -n --no-name don't save original file name or time stamp\n"
2038 " -q --quiet output no warnings\n"
2039 " -r --recursive recursively compress files in directories\n"
2040 " -S .suf use suffix .suf instead of .gz\n"
2042 " -t --test test compressed file\n"
2043 " -V --version display program version\n"
2044 " -v --verbose print extra statistics\n",
2051 /* display the license information of FreeBSD gzip */
2053 display_license(void)
2056 fprintf(stderr, "%s (based on NetBSD gzip 20091011)\n", gzip_version);
2057 fprintf(stderr, "%s\n", gzip_copyright);
2062 /* display the version of NetBSD gzip */
2064 display_version(void)
2067 fprintf(stderr, "%s\n", gzip_version);
2071 #ifndef NO_BZIP2_SUPPORT
2072 #include "unbzip2.c"
2074 #ifndef NO_COMPRESS_SUPPORT
2075 #include "zuncompress.c"
2077 #ifndef NO_PACK_SUPPORT
2082 read_retry(int fd, void *buf, size_t sz)
2085 size_t left = MIN(sz, (size_t) SSIZE_MAX);
2090 ret = read(fd, cp, left);
2093 } else if (ret == 0) {