1 /* $NetBSD: gzip.c,v 1.94 2009/04/12 10:31:14 lukem 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 * gzip.c -- GPL free gzip using zlib.
32 * RFC 1950 covers the zlib format
33 * RFC 1951 covers the deflate format
34 * RFC 1952 covers the gzip format
37 * - use mmap where possible
38 * - handle some signals better (remove outfile?)
39 * - make bzip2/compress -v/-t/-l support work as well as possible
42 #include <sys/param.h>
62 #define PRIdOFF PRId64
69 /* what type of file are we dealing with */
72 #ifndef NO_BZIP2_SUPPORT
75 #ifndef NO_COMPRESS_SUPPORT
78 #ifndef NO_PACK_SUPPORT
85 #ifndef NO_BZIP2_SUPPORT
88 #define BZ2_SUFFIX ".bz2"
89 #define BZIP2_MAGIC "\102\132\150"
92 #ifndef NO_COMPRESS_SUPPORT
94 #define Z_MAGIC "\037\235"
97 #ifndef NO_PACK_SUPPORT
98 #define PACK_MAGIC "\037\036"
101 #define GZ_SUFFIX ".gz"
103 #define BUFLEN (64 * 1024)
105 #define GZIP_MAGIC0 0x1F
106 #define GZIP_MAGIC1 0x8B
107 #define GZIP_OMAGIC1 0x9E
109 #define GZIP_TIMESTAMP (off_t)4
110 #define GZIP_ORIGNAME (off_t)10
112 #define HEAD_CRC 0x02
113 #define EXTRA_FIELD 0x04
114 #define ORIG_NAME 0x08
117 #define OS_CODE 3 /* Unix */
122 const char *normal; /* for unzip - must not be longer than zipped */
124 static suffixes_t suffixes[] = {
125 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N}
126 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */
128 SUFFIX(GZ_SUFFIX, ""),
133 SUFFIX(".taz", ".tar"),
134 SUFFIX(".tgz", ".tar"),
135 #ifndef NO_BZIP2_SUPPORT
136 SUFFIX(BZ2_SUFFIX, ""),
138 #ifndef NO_COMPRESS_SUPPORT
139 SUFFIX(Z_SUFFIX, ""),
141 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */
145 #define NUM_SUFFIXES NELEM(suffixes)
147 #define SUFFIX_MAXLEN 30
149 static const char gzip_version[] = "NetBSD gzip 20060927";
151 static int cflag; /* stdout mode */
152 static int dflag; /* decompress mode */
153 static int lflag; /* list mode */
154 static int numflag = 6; /* gzip -1..-9 value */
157 static int fflag; /* force mode */
158 static int kflag; /* don't delete input files */
159 static int nflag; /* don't save name/timestamp */
160 static int Nflag; /* don't restore name/timestamp */
161 static int qflag; /* quiet mode */
162 static int rflag; /* recursive mode */
163 static int tflag; /* test */
164 static int vflag; /* verbose mode */
170 static int exit_value = 0; /* exit value */
172 static char *infile; /* name of file coming in */
174 static void maybe_err(const char *fmt, ...)
175 __attribute__((__format__(__printf__, 1, 2)));
176 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT)
177 static void maybe_errx(const char *fmt, ...)
178 __attribute__((__format__(__printf__, 1, 2)));
180 static void maybe_warn(const char *fmt, ...)
181 __attribute__((__format__(__printf__, 1, 2)));
182 static void maybe_warnx(const char *fmt, ...)
183 __attribute__((__format__(__printf__, 1, 2)));
184 static enum filetype file_gettype(u_char *);
186 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz)
188 static off_t gz_compress(int, int, off_t *, const char *, uint32_t);
189 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *);
190 static off_t file_compress(char *, char *, size_t);
191 static off_t file_uncompress(char *, char *, size_t);
192 static void handle_pathname(char *);
193 static void handle_file(char *, struct stat *);
194 static void handle_stdin(void);
195 static void handle_stdout(void);
196 static void print_ratio(off_t, off_t, FILE *);
197 static void print_list(int fd, off_t, const char *, time_t);
198 static void usage(void) __dead2;
199 static void display_version(void) __dead2;
200 static const suffixes_t *check_suffix(char *, int);
201 static ssize_t read_retry(int, void *, size_t);
204 #define unlink_input(f, sb) unlink(f)
206 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd);
207 static void prepend_gzip(char *, int *, char ***);
208 static void handle_dir(char *);
209 static void print_verbage(const char *, const char *, off_t, off_t);
210 static void print_test(const char *, int);
211 static void copymodes(int fd, const struct stat *, const char *file);
212 static int check_outfile(const char *outfile);
215 #ifndef NO_BZIP2_SUPPORT
216 static off_t unbzip2(int, int, char *, size_t, off_t *);
219 #ifndef NO_COMPRESS_SUPPORT
220 static FILE *zdopen(int);
221 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *);
224 #ifndef NO_PACK_SUPPORT
225 static off_t unpack(int, int, char *, size_t, off_t *);
229 #define getopt_long(a,b,c,d,e) getopt(a,b,c)
231 static const struct option longopts[] = {
232 { "stdout", no_argument, 0, 'c' },
233 { "to-stdout", no_argument, 0, 'c' },
234 { "decompress", no_argument, 0, 'd' },
235 { "uncompress", no_argument, 0, 'd' },
236 { "force", no_argument, 0, 'f' },
237 { "help", no_argument, 0, 'h' },
238 { "keep", no_argument, 0, 'k' },
239 { "list", no_argument, 0, 'l' },
240 { "no-name", no_argument, 0, 'n' },
241 { "name", no_argument, 0, 'N' },
242 { "quiet", no_argument, 0, 'q' },
243 { "recursive", no_argument, 0, 'r' },
244 { "suffix", required_argument, 0, 'S' },
245 { "test", no_argument, 0, 't' },
246 { "verbose", no_argument, 0, 'v' },
247 { "version", no_argument, 0, 'V' },
248 { "fast", no_argument, 0, '1' },
249 { "best", no_argument, 0, '9' },
252 * This is what else GNU gzip implements. --ascii isn't useful
253 * on NetBSD, and I don't care to have a --license.
255 { "ascii", no_argument, 0, 'a' },
256 { "license", no_argument, 0, 'L' },
258 { NULL, no_argument, 0, 0 },
263 main(int argc, char **argv)
265 const char *progname = getprogname();
272 /* XXX set up signals */
275 if ((gzip = getenv("GZIP")) != NULL)
276 prepend_gzip(gzip, &argc, &argv);
281 * handle being called `gunzip', `zcat' and `gzcat'
283 if (strcmp(progname, "gunzip") == 0)
285 else if (strcmp(progname, "zcat") == 0 ||
286 strcmp(progname, "gzcat") == 0)
290 #define OPT_LIST "123456789cdhltV"
292 #define OPT_LIST "123456789cdfhklNnqrS:tVv"
295 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) {
297 case '1': case '2': case '3':
298 case '4': case '5': case '6':
299 case '7': case '8': case '9':
337 len = strlen(optarg);
339 if (len >= SUFFIX_MAXLEN)
340 errx(1, "incorrect suffix: '%s'", optarg);
341 suffixes[0].zipped = optarg;
342 suffixes[0].ziplen = len;
344 suffixes[NUM_SUFFIXES - 1].zipped = "";
345 suffixes[NUM_SUFFIXES - 1].ziplen = 0;
366 if (dflag) /* stdin mode */
368 else /* stdout mode */
372 handle_pathname(argv[0]);
376 if (qflag == 0 && lflag && argc > 1)
377 print_list(-1, 0, "(totals)", 0);
382 /* maybe print a warning */
384 maybe_warn(const char *fmt, ...)
397 /* ... without an errno. */
399 maybe_warnx(const char *fmt, ...)
412 /* maybe print an error */
414 maybe_err(const char *fmt, ...)
426 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT)
427 /* ... without an errno. */
429 maybe_errx(const char *fmt, ...)
443 /* split up $GZIP and prepend it to the argument list */
445 prepend_gzip(char *gzip, int *argc, char ***argv)
447 char *s, **nargv, **ac;
450 /* scan how many arguments there are */
452 while (*s == ' ' || *s == '\t')
457 while (*s != ' ' && *s != '\t')
469 nargv = (char **)malloc((*argc + 1) * sizeof(char *));
473 /* stash this away */
476 /* copy the program name first */
478 nargv[i++] = *(ac++);
480 /* take a copy of $GZIP and add it to the array */
485 /* Skip whitespaces. */
486 while (*s == ' ' || *s == '\t')
491 /* Find the end of this argument. */
492 while (*s != ' ' && *s != '\t')
494 /* Argument followed by NUL. */
496 /* Terminate by overwriting ' ' or '\t' with NUL. */
501 /* copy the original arguments and a NULL */
503 nargv[i++] = *(ac++);
508 /* compress input to output. Return bytes read, -1 on error */
510 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime)
513 char *outbufp, *inbufp;
514 off_t in_tot = 0, out_tot = 0;
519 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0,
524 outbufp = malloc(BUFLEN);
525 inbufp = malloc(BUFLEN);
526 if (outbufp == NULL || inbufp == NULL) {
527 maybe_err("malloc failed");
531 memset(&z, 0, sizeof z);
537 memcpy(outbufp, header, sizeof header);
545 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s",
546 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED,
547 *origname ? ORIG_NAME : 0,
550 (mtime >> 16) & 0xff,
551 (mtime >> 24) & 0xff,
552 numflag == 1 ? 4 : numflag == 9 ? 2 : 0,
555 /* this need PATH_MAX > BUFLEN ... */
556 maybe_err("snprintf");
561 z.next_out = outbufp + i;
562 z.avail_out = BUFLEN - i;
564 error = deflateInit2(&z, numflag, Z_DEFLATED,
565 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY);
567 maybe_warnx("deflateInit2 failed");
572 crc = crc32(0L, Z_NULL, 0);
574 if (z.avail_out == 0) {
575 if (write(out, outbufp, BUFLEN) != BUFLEN) {
582 z.next_out = outbufp;
583 z.avail_out = BUFLEN;
586 if (z.avail_in == 0) {
587 in_size = read(in, inbufp, BUFLEN);
596 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size);
599 z.avail_in = in_size;
602 error = deflate(&z, Z_NO_FLUSH);
603 if (error != Z_OK && error != Z_STREAM_END) {
604 maybe_warnx("deflate failed");
615 error = deflate(&z, Z_FINISH);
616 if (error != Z_OK && error != Z_STREAM_END) {
617 maybe_warnx("deflate failed");
622 len = (char *)z.next_out - outbufp;
624 w = write(out, outbufp, len);
625 if (w == -1 || (size_t)w != len) {
631 z.next_out = outbufp;
632 z.avail_out = BUFLEN;
634 if (error == Z_STREAM_END)
638 if (deflateEnd(&z) != Z_OK) {
639 maybe_warnx("deflateEnd failed");
644 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c",
646 (int)(crc >> 8) & 0xff,
647 (int)(crc >> 16) & 0xff,
648 (int)(crc >> 24) & 0xff,
650 (int)(in_tot >> 8) & 0xff,
651 (int)(in_tot >> 16) & 0xff,
652 (int)(in_tot >> 24) & 0xff);
654 maybe_err("snprintf");
656 if (in_tot > 0xffffffff)
657 maybe_warn("input file size >= 4GB cannot be saved");
659 if (write(out, outbufp, i) != i) {
676 * uncompress input to output then close the input. return the
677 * uncompressed size written, and put the compressed sized read
681 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep,
682 const char *filename)
685 char *outbufp, *inbufp;
686 off_t out_tot = -1, in_tot = 0;
687 uint32_t out_sub_tot = 0;
705 } state = GZSTATE_MAGIC0;
706 int flags = 0, skip_count = 0;
707 int error = Z_STREAM_ERROR, done_reading = 0;
712 #define ADVANCE() { z.next_in++; z.avail_in--; }
714 if ((outbufp = malloc(BUFLEN)) == NULL) {
715 maybe_err("malloc failed");
718 if ((inbufp = malloc(BUFLEN)) == NULL) {
719 maybe_err("malloc failed");
723 memset(&z, 0, sizeof z);
726 z.avail_out = BUFLEN;
727 z.next_out = outbufp;
736 if ((z.avail_in == 0 || needmore) && done_reading == 0) {
739 if (z.avail_in > 0) {
740 memmove(inbufp, z.next_in, z.avail_in);
743 in_size = read(in, z.next_in + z.avail_in,
744 BUFLEN - z.avail_in);
747 maybe_warn("failed to read stdin");
749 } else if (in_size == 0) {
753 z.avail_in += in_size;
758 if (z.avail_in == 0) {
759 if (done_reading && state != GZSTATE_MAGIC0) {
760 maybe_warnx("%s: unexpected end of file",
768 if (*z.next_in != GZIP_MAGIC0) {
770 maybe_warnx("%s: trailing garbage "
771 "ignored", filename);
774 maybe_warnx("input not gziped (MAGIC0)");
780 crc = crc32(0L, Z_NULL, 0);
784 if (*z.next_in != GZIP_MAGIC1 &&
785 *z.next_in != GZIP_OMAGIC1) {
786 maybe_warnx("input not gziped (MAGIC1)");
794 if (*z.next_in != Z_DEFLATED) {
795 maybe_warnx("unknown compression method");
809 case GZSTATE_SKIPPING:
810 if (skip_count > 0) {
818 if ((flags & EXTRA_FIELD) == 0) {
819 state = GZSTATE_ORIGNAME;
822 skip_count = *z.next_in;
828 skip_count |= ((*z.next_in) << 8);
834 if (skip_count > 0) {
841 case GZSTATE_ORIGNAME:
842 if ((flags & ORIG_NAME) == 0) {
851 case GZSTATE_COMMENT:
852 if ((flags & COMMENT) == 0) {
861 case GZSTATE_HEAD_CRC1:
862 if (flags & HEAD_CRC)
869 case GZSTATE_HEAD_CRC2:
870 if (skip_count > 0) {
878 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) {
879 maybe_warnx("failed to inflateInit");
886 error = inflate(&z, Z_FINISH);
888 /* Z_BUF_ERROR goes with Z_FINISH... */
890 if (z.avail_out > 0 && !done_reading)
897 maybe_warnx("Z_NEED_DICT error");
900 maybe_warnx("data stream error");
903 maybe_warnx("internal stream error");
906 maybe_warnx("memory allocation error");
910 maybe_warn("unknown error from inflate(): %d",
913 wr = BUFLEN - z.avail_out;
916 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr);
919 /* don't write anything with -t */
922 write(out, outbufp, wr) != wr) {
923 maybe_warn("error writing to output");
931 if (error == Z_STREAM_END) {
936 z.next_out = outbufp;
937 z.avail_out = BUFLEN;
944 if (z.avail_in < 4) {
949 maybe_warnx("truncated input");
952 origcrc = ((unsigned)z.next_in[0] & 0xff) |
953 ((unsigned)z.next_in[1] & 0xff) << 8 |
954 ((unsigned)z.next_in[2] & 0xff) << 16 |
955 ((unsigned)z.next_in[3] & 0xff) << 24;
956 if (origcrc != crc) {
957 maybe_warnx("invalid compressed"
966 if (!z.avail_in && done_reading) {
975 if (z.avail_in < 4) {
980 maybe_warnx("truncated input");
983 origlen = ((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;
988 if (origlen != out_sub_tot) {
989 maybe_warnx("invalid compressed"
990 " data--length error");
999 maybe_warnx("decompression error");
1002 state = GZSTATE_MAGIC0;
1011 if (state > GZSTATE_INIT)
1025 * set the owner, mode, flags & utimes using the given file descriptor.
1026 * file is only used in possible warning messages.
1029 copymodes(int fd, const struct stat *sbp, const char *file)
1031 struct timeval times[2];
1035 * If we have no info on the input, give this file some
1036 * default values and return..
1039 mode_t mask = umask(022);
1041 (void)fchmod(fd, DEFFILEMODE & ~mask);
1047 /* if the chown fails, remove set-id bits as-per compress(1) */
1048 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) {
1050 maybe_warn("couldn't fchown: %s", file);
1051 sb.st_mode &= ~(S_ISUID|S_ISGID);
1054 /* we only allow set-id and the 9 normal permission bits */
1055 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO;
1056 if (fchmod(fd, sb.st_mode) < 0)
1057 maybe_warn("couldn't fchmod: %s", file);
1059 /* only try flags if they exist already */
1060 #ifdef _ST_FLAGS_PRESENT_
1061 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0)
1062 maybe_warn("couldn't fchflags: %s", file);
1065 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec);
1066 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec);
1067 if (futimes(fd, times) < 0)
1068 maybe_warn("couldn't utimes: %s", file);
1072 /* what sort of file is this? */
1073 static enum filetype
1074 file_gettype(u_char *buf)
1077 if (buf[0] == GZIP_MAGIC0 &&
1078 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1))
1081 #ifndef NO_BZIP2_SUPPORT
1082 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 &&
1083 buf[3] >= '0' && buf[3] <= '9')
1087 #ifndef NO_COMPRESS_SUPPORT
1088 if (memcmp(buf, Z_MAGIC, 2) == 0)
1092 #ifndef NO_PACK_SUPPORT
1093 if (memcmp(buf, PACK_MAGIC, 2) == 0)
1101 /* check the outfile is OK. */
1103 check_outfile(const char *outfile)
1108 if (lflag == 0 && stat(outfile, &sb) == 0) {
1111 else if (isatty(STDIN_FILENO)) {
1112 char ans[10] = { 'n', '\0' }; /* default */
1114 fprintf(stderr, "%s already exists -- do you wish to "
1115 "overwrite (y or n)? " , outfile);
1116 (void)fgets(ans, sizeof(ans) - 1, stdin);
1117 if (ans[0] != 'y' && ans[0] != 'Y') {
1118 fprintf(stderr, "\tnot overwriting\n");
1123 maybe_warnx("%s already exists -- skipping", outfile);
1131 unlink_input(const char *file, const struct stat *sb)
1137 if (stat(file, &nsb) != 0)
1138 /* Must be gone alrady */
1140 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino)
1141 /* Definitely a different file */
1147 static const suffixes_t *
1148 check_suffix(char *file, int xlate)
1150 const suffixes_t *s;
1151 int len = strlen(file);
1154 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) {
1155 /* if it doesn't fit in "a.suf", don't bother */
1156 if (s->ziplen >= len)
1158 sp = file + len - s->ziplen;
1159 if (strcmp(s->zipped, sp) != 0)
1162 strcpy(sp, s->normal);
1169 * compress the given file: create a corresponding .gz file and remove the
1173 file_compress(char *file, char *outfile, size_t outsize)
1179 struct stat isb, osb;
1180 const suffixes_t *suff;
1183 in = open(file, O_RDONLY);
1185 maybe_warn("can't open %s", file);
1191 if (fstat(in, &isb) == 0) {
1192 if (isb.st_nlink > 1 && fflag == 0) {
1193 maybe_warnx("%s has %d other link%s -- "
1194 "skipping", file, isb.st_nlink - 1,
1195 isb.st_nlink == 1 ? "" : "s");
1201 if (fflag == 0 && (suff = check_suffix(file, 0))
1202 && suff->zipped[0] != 0) {
1203 maybe_warnx("%s already has %s suffix -- unchanged",
1204 file, suff->zipped);
1210 /* Add (usually) .gz to filename */
1211 if ((size_t)snprintf(outfile, outsize, "%s%s",
1212 file, suffixes[0].zipped) >= outsize) {
1213 errx(1, "file path too long: %s", file);
1216 if (check_outfile(outfile) == 0) {
1224 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600);
1226 maybe_warn("could not create output: %s", outfile);
1231 out = STDOUT_FILENO;
1233 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime);
1238 * If there was an error, insize will be -1.
1239 * If we compressed to stdout, just return the size.
1240 * Otherwise stat the file and check it is the correct size.
1241 * We only blow away the file if we can stat the output and it
1242 * has the expected size.
1245 return insize == -1 ? -1 : size;
1248 if (fstat(out, &osb) != 0) {
1249 maybe_warn("couldn't stat: %s", outfile);
1253 if (osb.st_size != size) {
1254 maybe_warnx("output file: %s wrong size (%" PRIdOFF
1255 " != %" PRIdOFF "), deleting",
1256 outfile, osb.st_size, size);
1260 copymodes(out, &isb, outfile);
1262 if (close(out) == -1)
1263 maybe_warn("couldn't close output");
1265 /* output is good, ok to delete input */
1266 unlink_input(file, &isb);
1271 if (close(out) == -1)
1272 maybe_warn("couldn't close output");
1274 maybe_warnx("leaving original %s", file);
1280 /* uncompress the given file and remove the original */
1282 file_uncompress(char *file, char *outfile, size_t outsize)
1284 struct stat isb, osb;
1287 unsigned char header1[4];
1288 enum filetype method;
1289 int fd, ofd, zfd = -1;
1292 time_t timestamp = 0;
1293 unsigned char name[PATH_MAX + 1];
1296 /* gather the old name info */
1298 fd = open(file, O_RDONLY);
1300 maybe_warn("can't open %s", file);
1304 if ((size_t)snprintf(outfile, outsize, "%s", file) >= outsize)
1305 errx(1, "file path too long: %s", file);
1306 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) {
1307 maybe_warnx("%s: unknown suffix -- ignored", file);
1311 rbytes = read(fd, header1, sizeof header1);
1312 if (rbytes != sizeof header1) {
1313 /* we don't want to fail here. */
1319 maybe_warn("can't read %s", file);
1321 goto unexpected_EOF;
1325 method = file_gettype(header1);
1328 if (fflag == 0 && method == FT_UNKNOWN) {
1329 maybe_warnx("%s: not in gzip format", file);
1336 if (method == FT_GZIP && Nflag) {
1337 unsigned char ts[4]; /* timestamp */
1339 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP);
1340 if (rv >= 0 && rv < (ssize_t)(sizeof ts))
1341 goto unexpected_EOF;
1344 maybe_warn("can't read %s", file);
1347 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0];
1349 if (header1[3] & ORIG_NAME) {
1350 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME);
1352 maybe_warn("can't read %s", file);
1356 /* preserve original directory name */
1357 char *dp = strrchr(file, '/');
1362 snprintf(outfile, outsize, "%.*s%.*s",
1364 file, (int) rbytes, name);
1369 lseek(fd, 0, SEEK_SET);
1371 if (cflag == 0 || lflag) {
1372 if (fstat(fd, &isb) != 0)
1375 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) {
1376 maybe_warnx("%s has %d other links -- skipping",
1377 file, isb.st_nlink - 1);
1380 if (nflag == 0 && timestamp)
1381 isb.st_mtime = timestamp;
1382 if (check_outfile(outfile) == 0)
1387 if (cflag == 0 && lflag == 0) {
1388 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600);
1389 if (zfd == STDOUT_FILENO) {
1390 /* We won't close STDOUT_FILENO later... */
1392 close(STDOUT_FILENO);
1395 maybe_warn("can't open %s", outfile);
1399 zfd = STDOUT_FILENO;
1401 #ifndef NO_BZIP2_SUPPORT
1402 if (method == FT_BZIP2) {
1406 maybe_warnx("no -l with bzip2 files");
1410 size = unbzip2(fd, zfd, NULL, 0, NULL);
1414 #ifndef NO_COMPRESS_SUPPORT
1415 if (method == FT_Z) {
1420 maybe_warnx("no -l with Lempel-Ziv files");
1424 if ((in = zdopen(fd)) == NULL) {
1425 maybe_warn("zdopen for read: %s", file);
1429 out = fdopen(dup(zfd), "w");
1431 maybe_warn("fdopen for write: %s", outfile);
1436 size = zuncompress(in, out, NULL, 0, NULL);
1437 /* need to fclose() if ferror() is true... */
1438 if (ferror(in) | fclose(in)) {
1439 maybe_warn("failed infile fclose");
1443 if (fclose(out) != 0) {
1444 maybe_warn("failed outfile fclose");
1451 #ifndef NO_PACK_SUPPORT
1452 if (method == FT_PACK) {
1454 maybe_warnx("no -l with packed files");
1458 size = unpack(fd, zfd, NULL, 0, NULL);
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 header1, sizeof header1, &gsize, "(stdin)");
1638 #ifndef NO_BZIP2_SUPPORT
1640 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO,
1641 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, header1, sizeof header1, &gsize);
1655 #ifndef NO_PACK_SUPPORT
1657 usize = unpack(STDIN_FILENO, STDOUT_FILENO,
1658 (char *)header1, sizeof header1, &gsize);
1664 if (vflag && !tflag && usize != -1 && gsize != -1)
1665 print_verbage(NULL, NULL, usize, gsize);
1667 print_test("(stdin)", usize != -1);
1682 if (fflag == 0 && isatty(STDOUT_FILENO)) {
1683 maybe_warnx("standard output is a terminal -- ignoring");
1687 /* If stdin is a file use it's mtime, otherwise use current time */
1688 ret = fstat(STDIN_FILENO, &sb);
1692 maybe_warn("Can't stat stdin");
1697 if (S_ISREG(sb.st_mode))
1698 mtime = (uint32_t)sb.st_mtime;
1700 systime = time(NULL);
1702 if (systime == -1) {
1707 mtime = (uint32_t)systime;
1710 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime);
1712 if (vflag && !tflag && usize != -1 && gsize != -1)
1713 print_verbage(NULL, NULL, usize, gsize);
1717 /* do what is asked for, for the path name */
1719 handle_pathname(char *path)
1721 char *opath = path, *s = NULL;
1726 /* check for stdout/stdin */
1727 if (path[0] == '-' && path[1] == '\0') {
1736 if (stat(path, &sb) != 0) {
1737 /* lets try <path>.gz if we're decompressing */
1738 if (dflag && s == NULL && errno == ENOENT) {
1740 slen = suffixes[0].ziplen;
1741 s = malloc(len + slen + 1);
1743 maybe_err("malloc");
1744 memcpy(s, path, len);
1745 memcpy(s + len, suffixes[0].zipped, slen + 1);
1749 maybe_warn("can't stat: %s", opath);
1753 if (S_ISDIR(sb.st_mode)) {
1759 maybe_warnx("%s is a directory", path);
1763 if (S_ISREG(sb.st_mode))
1764 handle_file(path, &sb);
1766 maybe_warnx("%s is not a regular file", path);
1773 /* compress/decompress a file */
1775 handle_file(char *file, struct stat *sbp)
1778 char outfile[PATH_MAX];
1782 usize = file_uncompress(file, outfile, sizeof(outfile));
1785 print_test(file, usize != -1);
1789 gsize = sbp->st_size;
1791 gsize = file_compress(file, outfile, sizeof(outfile));
1794 usize = sbp->st_size;
1799 if (vflag && !tflag)
1800 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize);
1805 /* this is used with -r to recursively descend directories */
1807 handle_dir(char *dir)
1814 path_argv[1] = NULL;
1815 fts = fts_open(path_argv, FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1817 warn("couldn't fts_open %s", dir);
1821 while ((entry = fts_read(fts))) {
1822 switch(entry->fts_info) {
1830 maybe_warn("%s", entry->fts_path);
1833 handle_file(entry->fts_path, entry->fts_statp);
1836 (void)fts_close(fts);
1840 /* print a ratio - size reduction as a fraction of uncompressed size */
1842 print_ratio(off_t in, off_t out, FILE *where)
1844 int percent10; /* 10 * percent */
1852 * Output is more than double size of input! print -99.9%
1853 * Quite possibly we've failed to get the original size.
1858 * We only need 12 bits of result from the final division,
1859 * so reduce the values until a 32bit division will suffice.
1861 while (in > 0x100000) {
1866 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000;
1871 len = snprintf(buff, sizeof buff, "%2.2d.", percent10);
1872 /* Move the '.' to before the last digit */
1873 buff[len - 1] = buff[len - 2];
1874 buff[len - 2] = '.';
1875 fprintf(where, "%5s%%", buff);
1879 /* print compression statistics, and the new name (if there is one!) */
1881 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize)
1884 fprintf(stderr, "%s:%s ", file,
1885 strlen(file) < 7 ? "\t\t" : "\t");
1886 print_ratio(usize, gsize, stderr);
1888 fprintf(stderr, " -- replaced with %s", nfile);
1889 fprintf(stderr, "\n");
1893 /* print test results */
1895 print_test(const char *file, int ok)
1898 if (exit_value == 0 && ok == 0)
1900 fprintf(stderr, "%s:%s %s\n", file,
1901 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK");
1906 /* print a file's info ala --list */
1908 compressed uncompressed ratio uncompressed_name
1909 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar
1912 print_list(int fd, off_t out, const char *outfile, time_t ts)
1914 static int first = 1;
1916 static off_t in_tot, out_tot;
1924 printf("method crc date time ");
1927 printf(" compressed uncompressed "
1928 "ratio uncompressed_name\n");
1940 /* read the last 4 bytes - this is the uncompressed size */
1941 rv = lseek(fd, (off_t)(-8), SEEK_END);
1943 unsigned char buf[8];
1946 rv = read(fd, (char *)buf, sizeof(buf));
1948 maybe_warn("read of uncompressed size");
1949 else if (rv != sizeof(buf))
1950 maybe_warnx("read of uncompressed size");
1953 usize = buf[4] | buf[5] << 8 |
1954 buf[6] << 16 | buf[7] << 24;
1957 crc = buf[0] | buf[1] << 8 |
1958 buf[2] << 16 | buf[3] << 24;
1965 if (vflag && fd == -1)
1968 char *date = ctime(&ts);
1970 /* skip the day, 1/100th second, and year */
1973 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date);
1978 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in);
1979 print_ratio(in, out, stdout);
1980 printf(" %s\n", outfile);
1983 /* display the usage of NetBSD gzip */
1988 fprintf(stderr, "%s\n", gzip_version);
1991 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n",
1993 "usage: %s [-123456789acdfhklLNnqrtVv] [-S .suffix] [<file> [<file> ...]]\n"
1994 " -1 --fast fastest (worst) compression\n"
1995 " -2 .. -8 set compression level\n"
1996 " -9 --best best (slowest) compression\n"
1997 " -c --stdout write to stdout, keep original files\n"
1999 " -d --decompress uncompress files\n"
2001 " -f --force force overwriting & compress links\n"
2002 " -h --help display this help\n"
2003 " -k --keep don't delete input files during operation\n"
2004 " -l --list list compressed file contents\n"
2005 " -N --name save or restore original file name and time stamp\n"
2006 " -n --no-name don't save original file name or time stamp\n"
2007 " -q --quiet output no warnings\n"
2008 " -r --recursive recursively compress files in directories\n"
2009 " -S .suf use suffix .suf instead of .gz\n"
2011 " -t --test test compressed file\n"
2012 " -V --version display program version\n"
2013 " -v --verbose print extra statistics\n",
2019 /* display the version of NetBSD gzip */
2021 display_version(void)
2024 fprintf(stderr, "%s\n", gzip_version);
2028 #ifndef NO_BZIP2_SUPPORT
2029 #include "unbzip2.c"
2031 #ifndef NO_COMPRESS_SUPPORT
2032 #include "zuncompress.c"
2034 #ifndef NO_PACK_SUPPORT
2039 read_retry(int fd, void *buf, size_t sz)
2042 size_t left = MIN(sz, (size_t) SSIZE_MAX);
2047 ret = read(fd, cp, left);
2050 } else if (ret == 0) {