1 /* $NetBSD: gzip.c,v 1.89 2006/11/13 21:57:59 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.
15 * 3. The name of the author may not be used to endorse or promote products
16 * derived from this software without specific prior written permission.
18 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
19 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
20 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
21 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
22 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
23 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
24 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
25 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
26 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 #include <sys/cdefs.h>
34 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green\n\
35 All rights reserved.\n");
40 * gzip.c -- GPL free gzip using zlib.
42 * RFC 1950 covers the zlib format
43 * RFC 1951 covers the deflate format
44 * RFC 1952 covers the gzip format
47 * - use mmap where possible
48 * - handle some signals better (remove outfile?)
49 * - make bzip2/compress -v/-t/-l support work as well as possible
52 #include <sys/param.h>
72 #define PRIdOFF PRId64
75 /* what type of file are we dealing with */
78 #ifndef NO_BZIP2_SUPPORT
81 #ifndef NO_COMPRESS_SUPPORT
88 #ifndef NO_BZIP2_SUPPORT
91 #define BZ2_SUFFIX ".bz2"
92 #define BZIP2_MAGIC "\102\132\150"
95 #ifndef NO_COMPRESS_SUPPORT
97 #define Z_MAGIC "\037\235"
100 #define GZ_SUFFIX ".gz"
102 #define BUFLEN (64 * 1024)
104 #define GZIP_MAGIC0 0x1F
105 #define GZIP_MAGIC1 0x8B
106 #define GZIP_OMAGIC1 0x9E
108 #define GZIP_TIMESTAMP (off_t)4
109 #define GZIP_ORIGNAME (off_t)10
111 #define HEAD_CRC 0x02
112 #define EXTRA_FIELD 0x04
113 #define ORIG_NAME 0x08
116 #define OS_CODE 3 /* Unix */
121 const char *normal; /* for unzip - must not be longer than zipped */
123 static suffixes_t suffixes[] = {
124 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N}
125 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */
127 SUFFIX(GZ_SUFFIX, ""),
132 SUFFIX(".taz", ".tar"),
133 SUFFIX(".tgz", ".tar"),
134 #ifndef NO_BZIP2_SUPPORT
135 SUFFIX(BZ2_SUFFIX, ""),
137 #ifndef NO_COMPRESS_SUPPORT
138 SUFFIX(Z_SUFFIX, ""),
140 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */
144 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0])
146 static const char gzip_version[] = "FreeBSD gzip 20070528";
149 static const char gzip_copyright[] = \
150 " Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green\n"
151 " All rights reserved.\n"
153 " Redistribution and use in source and binary forms, with or without\n"
154 " modification, are permitted provided that the following conditions\n"
156 " 1. Redistributions of source code must retain the above copyright\n"
157 " notice, this list of conditions and the following disclaimer.\n"
158 " 2. Redistributions in binary form must reproduce the above copyright\n"
159 " notice, this list of conditions and the following disclaimer in the\n"
160 " documentation and/or other materials provided with the distribution.\n"
161 " 3. The name of the author may not be used to endorse or promote products\n"
162 " derived from this software without specific prior written permission.\n"
164 " THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR\n"
165 " IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES\n"
166 " OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.\n"
167 " IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,\n"
168 " INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,\n"
169 " BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;\n"
170 " LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED\n"
171 " AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,\n"
172 " OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY\n"
173 " OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF\n"
177 static int cflag; /* stdout mode */
178 static int dflag; /* decompress mode */
179 static int lflag; /* list mode */
180 static int numflag = 6; /* gzip -1..-9 value */
183 static int fflag; /* force mode */
184 static int kflag; /* don't delete input files */
185 static int nflag; /* don't save name/timestamp */
186 static int Nflag; /* don't restore name/timestamp */
187 static int qflag; /* quiet mode */
188 static int rflag; /* recursive mode */
189 static int tflag; /* test */
190 static int vflag; /* verbose mode */
196 static int exit_value = 0; /* exit value */
198 static char *infile; /* name of file coming in */
200 static void maybe_err(const char *fmt, ...)
201 __attribute__((__format__(__printf__, 1, 2)));
202 #ifndef NO_BZIP2_SUPPORT
203 static void maybe_errx(const char *fmt, ...)
204 __attribute__((__format__(__printf__, 1, 2)));
206 static void maybe_warn(const char *fmt, ...)
207 __attribute__((__format__(__printf__, 1, 2)));
208 static void maybe_warnx(const char *fmt, ...)
209 __attribute__((__format__(__printf__, 1, 2)));
210 static enum filetype file_gettype(u_char *);
212 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz)
214 static off_t gz_compress(int, int, off_t *, const char *, uint32_t);
215 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *);
216 static off_t file_compress(char *, char *, size_t);
217 static off_t file_uncompress(char *, char *, size_t);
218 static void handle_pathname(char *);
219 static void handle_file(char *, struct stat *);
220 static void handle_stdin(void);
221 static void handle_stdout(void);
222 static void print_ratio(off_t, off_t, FILE *);
223 static void print_list(int fd, off_t, const char *, time_t);
224 static void usage(void);
225 static void display_version(void);
227 static void display_license(void);
229 static const suffixes_t *check_suffix(char *, int);
230 static ssize_t read_retry(int, void *, size_t);
233 #define unlink_input(f, sb) unlink(f)
235 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd);
236 static void prepend_gzip(char *, int *, char ***);
237 static void handle_dir(char *);
238 static void print_verbage(const char *, const char *, off_t, off_t);
239 static void print_test(const char *, int);
240 static void copymodes(int fd, const struct stat *, const char *file);
241 static int check_outfile(const char *outfile);
244 #ifndef NO_BZIP2_SUPPORT
245 static off_t unbzip2(int, int, char *, size_t, off_t *);
248 #ifndef NO_COMPRESS_SUPPORT
249 static FILE *zdopen(int);
250 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *);
253 int main(int, char **p);
256 #define getopt_long(a,b,c,d,e) getopt(a,b,c)
258 static const struct option longopts[] = {
259 { "stdout", no_argument, 0, 'c' },
260 { "to-stdout", no_argument, 0, 'c' },
261 { "decompress", no_argument, 0, 'd' },
262 { "uncompress", no_argument, 0, 'd' },
263 { "force", no_argument, 0, 'f' },
264 { "help", no_argument, 0, 'h' },
265 { "keep", no_argument, 0, 'k' },
266 { "list", no_argument, 0, 'l' },
267 { "no-name", no_argument, 0, 'n' },
268 { "name", no_argument, 0, 'N' },
269 { "quiet", no_argument, 0, 'q' },
270 { "recursive", no_argument, 0, 'r' },
271 { "suffix", required_argument, 0, 'S' },
272 { "test", no_argument, 0, 't' },
273 { "verbose", no_argument, 0, 'v' },
274 { "version", no_argument, 0, 'V' },
275 { "fast", no_argument, 0, '1' },
276 { "best", no_argument, 0, '9' },
277 { "ascii", no_argument, 0, 'a' },
278 { "license", no_argument, 0, 'L' },
279 { NULL, no_argument, 0, 0 },
284 main(int argc, char **argv)
286 const char *progname = getprogname();
293 /* XXX set up signals */
296 if ((gzip = getenv("GZIP")) != NULL)
297 prepend_gzip(gzip, &argc, &argv);
302 * handle being called `gunzip', `zcat' and `gzcat'
304 if (strcmp(progname, "gunzip") == 0)
306 else if (strcmp(progname, "zcat") == 0 ||
307 strcmp(progname, "gzcat") == 0)
311 #define OPT_LIST "123456789cdhltV"
313 #define OPT_LIST "123456789acdfhklLNnqrS:tVv"
316 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) {
318 case '1': case '2': case '3':
319 case '4': case '5': case '6':
320 case '7': case '8': case '9':
338 fprintf(stderr, "%s: option --ascii ignored on this system\n", progname);
364 len = strlen(optarg);
366 suffixes[0].zipped = optarg;
367 suffixes[0].ziplen = len;
369 suffixes[NUM_SUFFIXES - 1].zipped = "";
370 suffixes[NUM_SUFFIXES - 1].ziplen = 0;
391 if (dflag) /* stdin mode */
393 else /* stdout mode */
397 handle_pathname(argv[0]);
401 if (qflag == 0 && lflag && argc > 1)
402 print_list(-1, 0, "(totals)", 0);
407 /* maybe print a warning */
409 maybe_warn(const char *fmt, ...)
422 /* ... without an errno. */
424 maybe_warnx(const char *fmt, ...)
437 /* maybe print an error */
439 maybe_err(const char *fmt, ...)
451 #ifndef NO_BZIP2_SUPPORT
452 /* ... without an errno. */
454 maybe_errx(const char *fmt, ...)
468 /* split up $GZIP and prepend it to the argument list */
470 prepend_gzip(char *gzip, int *argc, char ***argv)
472 char *s, **nargv, **ac;
475 /* scan how many arguments there are */
477 while (*s == ' ' || *s == '\t')
482 while (*s != ' ' && *s != '\t')
494 nargv = (char **)malloc((*argc + 1) * sizeof(char *));
498 /* stash this away */
501 /* copy the program name first */
503 nargv[i++] = *(ac++);
505 /* take a copy of $GZIP and add it to the array */
510 /* Skip whitespaces. */
511 while (*s == ' ' || *s == '\t')
516 /* Find the end of this argument. */
517 while (*s != ' ' && *s != '\t')
519 /* Argument followed by NUL. */
521 /* Terminate by overwriting ' ' or '\t' with NUL. */
526 /* copy the original arguments and a NULL */
528 nargv[i++] = *(ac++);
533 /* compress input to output. Return bytes read, -1 on error */
535 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime)
538 char *outbufp, *inbufp;
539 off_t in_tot = 0, out_tot = 0;
544 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0,
549 outbufp = malloc(BUFLEN);
550 inbufp = malloc(BUFLEN);
551 if (outbufp == NULL || inbufp == NULL) {
552 maybe_err("malloc failed");
556 memset(&z, 0, sizeof z);
562 memcpy(outbufp, header, sizeof header);
570 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s",
571 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED,
572 *origname ? ORIG_NAME : 0,
575 (mtime >> 16) & 0xff,
576 (mtime >> 24) & 0xff,
577 numflag == 1 ? 4 : numflag == 9 ? 2 : 0,
580 /* this need PATH_MAX > BUFLEN ... */
581 maybe_err("snprintf");
586 z.next_out = (unsigned char *)outbufp + i;
587 z.avail_out = BUFLEN - i;
589 error = deflateInit2(&z, numflag, Z_DEFLATED,
590 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY);
592 maybe_warnx("deflateInit2 failed");
597 crc = crc32(0L, Z_NULL, 0);
599 if (z.avail_out == 0) {
600 if (write(out, outbufp, BUFLEN) != BUFLEN) {
607 z.next_out = (unsigned char *)outbufp;
608 z.avail_out = BUFLEN;
611 if (z.avail_in == 0) {
612 in_size = read(in, inbufp, BUFLEN);
621 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size);
623 z.next_in = (unsigned char *)inbufp;
624 z.avail_in = in_size;
627 error = deflate(&z, Z_NO_FLUSH);
628 if (error != Z_OK && error != Z_STREAM_END) {
629 maybe_warnx("deflate failed");
640 error = deflate(&z, Z_FINISH);
641 if (error != Z_OK && error != Z_STREAM_END) {
642 maybe_warnx("deflate failed");
647 len = (char *)z.next_out - outbufp;
649 w = write(out, outbufp, len);
650 if (w == -1 || (size_t)w != len) {
656 z.next_out = (unsigned char *)outbufp;
657 z.avail_out = BUFLEN;
659 if (error == Z_STREAM_END)
663 if (deflateEnd(&z) != Z_OK) {
664 maybe_warnx("deflateEnd failed");
669 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c",
671 (int)(crc >> 8) & 0xff,
672 (int)(crc >> 16) & 0xff,
673 (int)(crc >> 24) & 0xff,
675 (int)(in_tot >> 8) & 0xff,
676 (int)(in_tot >> 16) & 0xff,
677 (int)(in_tot >> 24) & 0xff);
679 maybe_err("snprintf");
680 if (write(out, outbufp, i) != i) {
697 * uncompress input to output then close the input. return the
698 * uncompressed size written, and put the compressed sized read
702 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep,
703 const char *filename)
706 char *outbufp, *inbufp;
707 off_t out_tot = -1, in_tot = 0;
708 uint32_t out_sub_tot = 0;
726 } state = GZSTATE_MAGIC0;
727 int flags = 0, skip_count = 0;
728 int error = Z_STREAM_ERROR, done_reading = 0;
733 #define ADVANCE() { z.next_in++; z.avail_in--; }
735 if ((outbufp = malloc(BUFLEN)) == NULL) {
736 maybe_err("malloc failed");
739 if ((inbufp = malloc(BUFLEN)) == NULL) {
740 maybe_err("malloc failed");
744 memset(&z, 0, sizeof z);
746 z.next_in = (unsigned char *)pre;
747 z.avail_out = BUFLEN;
748 z.next_out = (unsigned char *)outbufp;
757 if ((z.avail_in == 0 || needmore) && done_reading == 0) {
760 if (z.avail_in > 0) {
761 memmove(inbufp, z.next_in, z.avail_in);
763 z.next_in = (unsigned char *)inbufp;
764 in_size = read(in, z.next_in + z.avail_in,
765 BUFLEN - z.avail_in);
768 maybe_warn("failed to read stdin");
770 } else if (in_size == 0) {
774 z.avail_in += in_size;
779 if (z.avail_in == 0) {
780 if (done_reading && state != GZSTATE_MAGIC0) {
781 maybe_warnx("%s: unexpected end of file",
789 if (*z.next_in != GZIP_MAGIC0) {
791 maybe_warnx("%s: trailing garbage "
792 "ignored", filename);
795 maybe_warnx("input not gziped (MAGIC0)");
801 crc = crc32(0L, Z_NULL, 0);
805 if (*z.next_in != GZIP_MAGIC1 &&
806 *z.next_in != GZIP_OMAGIC1) {
807 maybe_warnx("input not gziped (MAGIC1)");
815 if (*z.next_in != Z_DEFLATED) {
816 maybe_warnx("unknown compression method");
830 case GZSTATE_SKIPPING:
831 if (skip_count > 0) {
839 if ((flags & EXTRA_FIELD) == 0) {
840 state = GZSTATE_ORIGNAME;
843 skip_count = *z.next_in;
849 skip_count |= ((*z.next_in) << 8);
855 if (skip_count > 0) {
862 case GZSTATE_ORIGNAME:
863 if ((flags & ORIG_NAME) == 0) {
872 case GZSTATE_COMMENT:
873 if ((flags & COMMENT) == 0) {
882 case GZSTATE_HEAD_CRC1:
883 if (flags & HEAD_CRC)
890 case GZSTATE_HEAD_CRC2:
891 if (skip_count > 0) {
899 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) {
900 maybe_warnx("failed to inflateInit");
907 error = inflate(&z, Z_FINISH);
909 /* Z_BUF_ERROR goes with Z_FINISH... */
916 maybe_warnx("Z_NEED_DICT error");
919 maybe_warnx("data stream error");
922 maybe_warnx("internal stream error");
925 maybe_warnx("memory allocation error");
929 maybe_warn("unknown error from inflate(): %d",
932 wr = BUFLEN - z.avail_out;
935 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr);
938 /* don't write anything with -t */
941 write(out, outbufp, wr) != wr) {
942 maybe_warn("error writing to output");
950 if (error == Z_STREAM_END) {
955 z.next_out = (unsigned char *)outbufp;
956 z.avail_out = BUFLEN;
963 if (z.avail_in < 4) {
968 maybe_warnx("truncated input");
971 origcrc = ((unsigned)z.next_in[0] & 0xff) |
972 ((unsigned)z.next_in[1] & 0xff) << 8 |
973 ((unsigned)z.next_in[2] & 0xff) << 16 |
974 ((unsigned)z.next_in[3] & 0xff) << 24;
975 if (origcrc != crc) {
976 maybe_warnx("invalid compressed"
985 if (!z.avail_in && done_reading) {
994 if (z.avail_in < 4) {
999 maybe_warnx("truncated input");
1002 origlen = ((unsigned)z.next_in[0] & 0xff) |
1003 ((unsigned)z.next_in[1] & 0xff) << 8 |
1004 ((unsigned)z.next_in[2] & 0xff) << 16 |
1005 ((unsigned)z.next_in[3] & 0xff) << 24;
1007 if (origlen != out_sub_tot) {
1008 maybe_warnx("invalid compressed"
1009 " data--length error");
1018 maybe_warnx("decompression error");
1021 state = GZSTATE_MAGIC0;
1030 if (state > GZSTATE_INIT)
1044 * set the owner, mode, flags & utimes using the given file descriptor.
1045 * file is only used in possible warning messages.
1048 copymodes(int fd, const struct stat *sbp, const char *file)
1050 struct timeval times[2];
1054 * If we have no info on the input, give this file some
1055 * default values and return..
1058 mode_t mask = umask(022);
1060 (void)fchmod(fd, DEFFILEMODE & ~mask);
1066 /* if the chown fails, remove set-id bits as-per compress(1) */
1067 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) {
1069 maybe_warn("couldn't fchown: %s", file);
1070 sb.st_mode &= ~(S_ISUID|S_ISGID);
1073 /* we only allow set-id and the 9 normal permission bits */
1074 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO;
1075 if (fchmod(fd, sb.st_mode) < 0)
1076 maybe_warn("couldn't fchmod: %s", file);
1078 /* only try flags if they exist already */
1079 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0)
1080 maybe_warn("couldn't fchflags: %s", file);
1082 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec);
1083 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec);
1084 if (futimes(fd, times) < 0)
1085 maybe_warn("couldn't utimes: %s", file);
1089 /* what sort of file is this? */
1090 static enum filetype
1091 file_gettype(u_char *buf)
1094 if (buf[0] == GZIP_MAGIC0 &&
1095 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1))
1098 #ifndef NO_BZIP2_SUPPORT
1099 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 &&
1100 buf[3] >= '0' && buf[3] <= '9')
1104 #ifndef NO_COMPRESS_SUPPORT
1105 if (memcmp(buf, Z_MAGIC, 2) == 0)
1113 /* check the outfile is OK. */
1115 check_outfile(const char *outfile)
1120 if (lflag == 0 && stat(outfile, &sb) == 0) {
1123 else if (isatty(STDIN_FILENO)) {
1124 char ans[10] = { 'n', '\0' }; /* default */
1126 fprintf(stderr, "%s already exists -- do you wish to "
1127 "overwrite (y or n)? " , outfile);
1128 (void)fgets(ans, sizeof(ans) - 1, stdin);
1129 if (ans[0] != 'y' && ans[0] != 'Y') {
1130 fprintf(stderr, "\tnot overwriting\n");
1135 maybe_warnx("%s already exists -- skipping", outfile);
1143 unlink_input(const char *file, const struct stat *sb)
1149 if (stat(file, &nsb) != 0)
1150 /* Must be gone alrady */
1152 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino)
1153 /* Definitely a different file */
1159 static const suffixes_t *
1160 check_suffix(char *file, int xlate)
1162 const suffixes_t *s;
1163 int len = strlen(file);
1166 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) {
1167 /* if it doesn't fit in "a.suf", don't bother */
1168 if (s->ziplen >= len)
1170 sp = file + len - s->ziplen;
1171 if (strcmp(s->zipped, sp) != 0)
1174 strcpy(sp, s->normal);
1181 * compress the given file: create a corresponding .gz file and remove the
1185 file_compress(char *file, char *outfile, size_t outsize)
1191 struct stat isb, osb;
1192 const suffixes_t *suff;
1195 in = open(file, O_RDONLY);
1197 maybe_warn("can't open %s", file);
1203 if (fstat(in, &isb) == 0) {
1204 if (isb.st_nlink > 1 && fflag == 0) {
1205 maybe_warnx("%s has %d other link%s -- "
1206 "skipping", file, isb.st_nlink - 1,
1207 isb.st_nlink == 1 ? "" : "s");
1213 if (fflag == 0 && (suff = check_suffix(file, 0))
1214 && suff->zipped[0] != 0) {
1215 maybe_warnx("%s already has %s suffix -- unchanged",
1216 file, suff->zipped);
1222 /* Add (usually) .gz to filename */
1223 if ((size_t)snprintf(outfile, outsize, "%s%s",
1224 file, suffixes[0].zipped) >= outsize)
1225 memcpy(outfile - suffixes[0].ziplen - 1,
1226 suffixes[0].zipped, suffixes[0].ziplen + 1);
1229 if (check_outfile(outfile) == 0) {
1237 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600);
1239 maybe_warn("could not create output: %s", outfile);
1244 out = STDOUT_FILENO;
1246 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime);
1251 * If there was an error, insize will be -1.
1252 * If we compressed to stdout, just return the size.
1253 * Otherwise stat the file and check it is the correct size.
1254 * We only blow away the file if we can stat the output and it
1255 * has the expected size.
1258 return insize == -1 ? -1 : size;
1261 if (fstat(out, &osb) != 0) {
1262 maybe_warn("couldn't stat: %s", outfile);
1266 if (osb.st_size != size) {
1267 maybe_warnx("output file: %s wrong size (%" PRIdOFF
1268 " != %" PRIdOFF "), deleting",
1269 outfile, osb.st_size, size);
1273 copymodes(out, &isb, outfile);
1275 if (close(out) == -1)
1276 maybe_warn("couldn't close output");
1278 /* output is good, ok to delete input */
1279 unlink_input(file, &isb);
1284 if (close(out) == -1)
1285 maybe_warn("couldn't close output");
1287 maybe_warnx("leaving original %s", file);
1293 /* uncompress the given file and remove the original */
1295 file_uncompress(char *file, char *outfile, size_t outsize)
1297 struct stat isb, osb;
1300 unsigned char header1[4];
1301 enum filetype method;
1302 int rv, fd, ofd, zfd = -1;
1304 time_t timestamp = 0;
1305 unsigned char name[PATH_MAX + 1];
1308 /* gather the old name info */
1310 fd = open(file, O_RDONLY);
1312 maybe_warn("can't open %s", file);
1316 strlcpy(outfile, file, outsize);
1317 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) {
1318 maybe_warnx("%s: unknown suffix -- ignored", file);
1322 rbytes = read(fd, header1, sizeof header1);
1323 if (rbytes != sizeof header1) {
1324 /* we don't want to fail here. */
1330 maybe_warn("can't read %s", file);
1332 goto unexpected_EOF;
1336 method = file_gettype(header1);
1339 if (fflag == 0 && method == FT_UNKNOWN) {
1340 maybe_warnx("%s: not in gzip format", file);
1347 if (method == FT_GZIP && Nflag) {
1348 unsigned char ts[4]; /* timestamp */
1350 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP);
1351 if (rv >= 0 && (size_t)rv < sizeof ts)
1352 goto unexpected_EOF;
1355 maybe_warn("can't read %s", file);
1358 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0];
1360 if (header1[3] & ORIG_NAME) {
1361 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME);
1363 maybe_warn("can't read %s", file);
1367 /* preserve original directory name */
1368 char *dp = strrchr(file, '/');
1373 snprintf(outfile, outsize, "%.*s%.*s",
1375 file, (int) rbytes, name);
1380 lseek(fd, 0, SEEK_SET);
1382 if (cflag == 0 || lflag) {
1383 if (fstat(fd, &isb) != 0)
1386 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) {
1387 maybe_warnx("%s has %d other links -- skipping",
1388 file, isb.st_nlink - 1);
1391 if (nflag == 0 && timestamp)
1392 isb.st_mtime = timestamp;
1393 if (check_outfile(outfile) == 0)
1398 if (cflag == 0 && lflag == 0) {
1399 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600);
1400 if (zfd == STDOUT_FILENO) {
1401 /* We won't close STDOUT_FILENO later... */
1403 close(STDOUT_FILENO);
1406 maybe_warn("can't open %s", outfile);
1410 zfd = STDOUT_FILENO;
1412 #ifndef NO_BZIP2_SUPPORT
1413 if (method == FT_BZIP2) {
1417 maybe_warnx("no -l with bzip2 files");
1421 size = unbzip2(fd, zfd, NULL, 0, NULL);
1425 #ifndef NO_COMPRESS_SUPPORT
1426 if (method == FT_Z) {
1431 maybe_warnx("no -l with Lempel-Ziv files");
1435 if ((in = zdopen(fd)) == NULL) {
1436 maybe_warn("zdopen for read: %s", file);
1440 out = fdopen(dup(zfd), "w");
1442 maybe_warn("fdopen for write: %s", outfile);
1447 size = zuncompress(in, out, NULL, 0, NULL);
1448 /* need to fclose() if ferror() is true... */
1449 if (ferror(in) | fclose(in)) {
1450 maybe_warn("failed infile fclose");
1454 if (fclose(out) != 0) {
1455 maybe_warn("failed outfile fclose");
1463 if (method == FT_UNKNOWN) {
1465 maybe_warnx("no -l for unknown filetypes");
1468 size = cat_fd(NULL, 0, NULL, fd);
1473 print_list(fd, isb.st_size, outfile, isb.st_mtime);
1475 return -1; /* XXX */
1478 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file);
1482 maybe_warn("couldn't close input");
1483 if (zfd != STDOUT_FILENO && close(zfd) != 0)
1484 maybe_warn("couldn't close output");
1489 maybe_warnx("%s: uncompress failed", file);
1493 /* if testing, or we uncompressed to stdout, this is all we need */
1498 /* if we are uncompressing to stdin, don't remove the file. */
1503 * if we create a file...
1506 * if we can't stat the file don't remove the file.
1509 ofd = open(outfile, O_RDWR, 0);
1511 maybe_warn("couldn't open (leaving original): %s",
1515 if (fstat(ofd, &osb) != 0) {
1516 maybe_warn("couldn't stat (leaving original): %s",
1521 if (osb.st_size != size) {
1522 maybe_warnx("stat gave different size: %" PRIdOFF
1523 " != %" PRIdOFF " (leaving original)",
1529 unlink_input(file, &isb);
1531 copymodes(ofd, &isb, outfile);
1537 maybe_warnx("%s: unexpected end of file", file);
1541 if (zfd != -1 && zfd != STDOUT_FILENO)
1548 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd)
1555 w = write(STDOUT_FILENO, prepend, count);
1556 if (w == -1 || (size_t)w != count) {
1557 maybe_warn("write to stdout");
1563 rv = read(fd, buf, sizeof buf);
1567 maybe_warn("read from fd %d", fd);
1571 if (write(STDOUT_FILENO, buf, rv) != rv) {
1572 maybe_warn("write to stdout");
1587 unsigned char header1[4];
1589 enum filetype method;
1591 #ifndef NO_COMPRESS_SUPPORT
1596 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) {
1597 maybe_warnx("standard input is a terminal -- ignoring");
1605 /* XXX could read the whole file, etc. */
1606 if (fstat(STDIN_FILENO, &isb) < 0) {
1607 maybe_warn("fstat");
1610 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime);
1614 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1);
1615 if (bytes_read == -1) {
1616 maybe_warn("can't read stdin");
1618 } else if (bytes_read != sizeof(header1)) {
1619 maybe_warnx("(stdin): unexpected end of file");
1623 method = file_gettype(header1);
1628 maybe_warnx("unknown compression format");
1631 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO);
1635 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO,
1636 (char *)header1, sizeof header1, &gsize, "(stdin)");
1638 #ifndef NO_BZIP2_SUPPORT
1640 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO,
1641 (char *)header1, sizeof header1, &gsize);
1644 #ifndef NO_COMPRESS_SUPPORT
1646 if ((in = zdopen(STDIN_FILENO)) == NULL) {
1647 maybe_warnx("zopen of stdin");
1651 usize = zuncompress(in, stdout, (char *)header1, sizeof header1, &gsize);
1658 if (vflag && !tflag && usize != -1 && gsize != -1)
1659 print_verbage(NULL, NULL, usize, gsize);
1661 print_test("(stdin)", usize != -1);
1676 if (fflag == 0 && isatty(STDOUT_FILENO)) {
1677 maybe_warnx("standard output is a terminal -- ignoring");
1681 /* If stdin is a file use it's mtime, otherwise use current time */
1682 ret = fstat(STDIN_FILENO, &sb);
1686 maybe_warn("Can't stat stdin");
1691 if (S_ISREG(sb.st_mode))
1692 mtime = (uint32_t)sb.st_mtime;
1694 systime = time(NULL);
1696 if (systime == -1) {
1701 mtime = (uint32_t)systime;
1704 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime);
1706 if (vflag && !tflag && usize != -1 && gsize != -1)
1707 print_verbage(NULL, NULL, usize, gsize);
1711 /* do what is asked for, for the path name */
1713 handle_pathname(char *path)
1715 char *opath = path, *s = NULL;
1720 /* check for stdout/stdin */
1721 if (path[0] == '-' && path[1] == '\0') {
1730 if (stat(path, &sb) != 0) {
1731 /* lets try <path>.gz if we're decompressing */
1732 if (dflag && s == NULL && errno == ENOENT) {
1734 slen = suffixes[0].ziplen;
1735 s = malloc(len + slen + 1);
1737 maybe_err("malloc");
1738 memcpy(s, path, len);
1739 memcpy(s + len, suffixes[0].zipped, slen + 1);
1743 maybe_warn("can't stat: %s", opath);
1747 if (S_ISDIR(sb.st_mode)) {
1753 maybe_warnx("%s is a directory", path);
1757 if (S_ISREG(sb.st_mode))
1758 handle_file(path, &sb);
1760 maybe_warnx("%s is not a regular file", path);
1767 /* compress/decompress a file */
1769 handle_file(char *file, struct stat *sbp)
1772 char outfile[PATH_MAX];
1776 usize = file_uncompress(file, outfile, sizeof(outfile));
1779 print_test(file, usize != -1);
1783 gsize = sbp->st_size;
1785 gsize = file_compress(file, outfile, sizeof(outfile));
1788 usize = sbp->st_size;
1793 if (vflag && !tflag)
1794 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize);
1799 /* this is used with -r to recursively descend directories */
1801 handle_dir(char *dir)
1809 fts = fts_open(path_argv, FTS_PHYSICAL, NULL);
1811 warn("couldn't fts_open %s", dir);
1815 while ((entry = fts_read(fts))) {
1816 switch(entry->fts_info) {
1824 maybe_warn("%s", entry->fts_path);
1827 handle_file(entry->fts_name, entry->fts_statp);
1830 (void)fts_close(fts);
1834 /* print a ratio - size reduction as a fraction of uncompressed size */
1836 print_ratio(off_t in, off_t out, FILE *where)
1838 int percent10; /* 10 * percent */
1846 * Output is more than double size of input! print -99.9%
1847 * Quite possibly we've failed to get the original size.
1852 * We only need 12 bits of result from the final division,
1853 * so reduce the values until a 32bit division will suffice.
1855 while (in > 0x100000) {
1860 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000;
1865 len = snprintf(buff, sizeof buff, "%2.2d.", percent10);
1866 /* Move the '.' to before the last digit */
1867 buff[len - 1] = buff[len - 2];
1868 buff[len - 2] = '.';
1869 fprintf(where, "%5s%%", buff);
1873 /* print compression statistics, and the new name (if there is one!) */
1875 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize)
1878 fprintf(stderr, "%s:%s ", file,
1879 strlen(file) < 7 ? "\t\t" : "\t");
1880 print_ratio(usize, gsize, stderr);
1882 fprintf(stderr, " -- replaced with %s", nfile);
1883 fprintf(stderr, "\n");
1887 /* print test results */
1889 print_test(const char *file, int ok)
1892 if (exit_value == 0 && ok == 0)
1894 fprintf(stderr, "%s:%s %s\n", file,
1895 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK");
1900 /* print a file's info ala --list */
1902 compressed uncompressed ratio uncompressed_name
1903 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar
1906 print_list(int fd, off_t out, const char *outfile, time_t ts)
1908 static int first = 1;
1910 static off_t in_tot, out_tot;
1918 printf("method crc date time ");
1921 printf(" compressed uncompressed "
1922 "ratio uncompressed_name\n");
1934 /* read the last 4 bytes - this is the uncompressed size */
1935 rv = lseek(fd, (off_t)(-8), SEEK_END);
1937 unsigned char buf[8];
1940 rv = read(fd, (char *)buf, sizeof(buf));
1942 maybe_warn("read of uncompressed size");
1943 else if (rv != sizeof(buf))
1944 maybe_warnx("read of uncompressed size");
1947 usize = buf[4] | buf[5] << 8 |
1948 buf[6] << 16 | buf[7] << 24;
1951 crc = buf[0] | buf[1] << 8 |
1952 buf[2] << 16 | buf[3] << 24;
1959 if (vflag && fd == -1)
1962 char *date = ctime(&ts);
1964 /* skip the day, 1/100th second, and year */
1967 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date);
1972 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in);
1973 print_ratio(in, out, stdout);
1974 printf(" %s\n", outfile);
1977 /* display the usage of NetBSD gzip */
1982 fprintf(stderr, "%s\n", gzip_version);
1984 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n"
1986 " -1 --fast fastest (worst) compression\n"
1987 " -2 .. -8 set compression level\n"
1988 " -9 --best best (slowest) compression\n"
1989 " -c --stdout write to stdout, keep original files\n"
1991 " -d --decompress uncompress files\n"
1993 " -f --force force overwriting & compress links\n"
1994 " -h --help display this help\n"
1995 " -k --keep don't delete input files during operation\n"
1996 " -l --list list compressed file contents\n"
1997 " -N --name save or restore original file name and time stamp\n"
1998 " -n --no-name don't save original file name or time stamp\n"
1999 " -q --quiet output no warnings\n"
2000 " -r --recursive recursively compress files in directories\n"
2001 " -S .suf use suffix .suf instead of .gz\n"
2003 " -t --test test compressed file\n"
2004 " -V --version display program version\n"
2005 " -v --verbose print extra statistics\n",
2014 /* display the license information of FreeBSD gzip */
2016 display_license(void)
2019 fprintf(stderr, "%s (based on NetBSD gzip 20060927)\n", gzip_version);
2020 fprintf(stderr, "%s\n", gzip_copyright);
2025 /* display the version of NetBSD gzip */
2027 display_version(void)
2030 fprintf(stderr, "%s\n", gzip_version);
2034 #ifndef NO_BZIP2_SUPPORT
2035 #include "unbzip2.c"
2037 #ifndef NO_COMPRESS_SUPPORT
2038 #include "zuncompress.c"
2042 read_retry(int fd, void *buf, size_t sz)
2045 size_t left = MIN(sz, (size_t) SSIZE_MAX);
2050 ret = read(fd, cp, left);
2053 } else if (ret == 0) {