gzip.c revision 1.39 1 /* $NetBSD: gzip.c,v 1.39 2004/04/27 01:23:35 mrg Exp $ */
2
3 /*
4 * Copyright (c) 1997, 1998, 2003, 2004 Matthew R. Green
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
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.
17 *
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
28 * SUCH DAMAGE.
29 */
30
31 #include <sys/cdefs.h>
32 #ifndef lint
33 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004 Matthew R. Green\n\
34 All rights reserved.\n");
35 __RCSID("$NetBSD: gzip.c,v 1.39 2004/04/27 01:23:35 mrg Exp $");
36 #endif /* not lint */
37
38 /*
39 * gzip.c -- GPL free gzip using zlib.
40 *
41 * TODO:
42 * - handle .taz/.tgz files?
43 * - use mmap where possible
44 * - handle some signals better (remove outfile?)
45 * - audit maybe_err()/maybe_warn() usage
46 */
47
48 #include <sys/param.h>
49 #include <sys/stat.h>
50 #include <sys/time.h>
51
52 #include <unistd.h>
53 #include <stdio.h>
54 #include <string.h>
55 #include <stdlib.h>
56 #include <err.h>
57 #include <errno.h>
58 #include <fcntl.h>
59 #include <zlib.h>
60 #include <fts.h>
61 #include <libgen.h>
62 #include <stdarg.h>
63 #include <getopt.h>
64
65 /* what type of file are we dealing with */
66 enum filetype {
67 FT_GZIP,
68 #ifndef NO_BZIP2_SUPPORT
69 FT_BZIP2,
70 #endif
71 #ifndef NO_COMPRESS_SUPPORT
72 FT_Z,
73 #endif
74 FT_LAST,
75 FT_UNKNOWN
76 };
77
78 #ifndef NO_BZIP2_SUPPORT
79 #include <bzlib.h>
80
81 #define BZ2_SUFFIX ".bz2"
82 #define BZIP2_MAGIC "\102\132\150"
83 #endif
84
85 #ifndef NO_COMPRESS_SUPPORT
86 #define Z_SUFFIX ".Z"
87 #define Z_MAGIC "\037\235"
88 #endif
89
90 #define GZ_SUFFIX ".gz"
91
92 #define BUFLEN (64 * 1024)
93
94 #define GZIP_MAGIC0 0x1F
95 #define GZIP_MAGIC1 0x8B
96 #define GZIP_OMAGIC1 0x9E
97
98 #define HEAD_CRC 0x02
99 #define EXTRA_FIELD 0x04
100 #define ORIG_NAME 0x08
101 #define COMMENT 0x10
102
103 #define OS_CODE 3 /* Unix */
104
105 static const char gzip_version[] = "NetBSD gzip 20040425";
106
107 static int cflag; /* stdout mode */
108 static int dflag; /* decompress mode */
109 static int lflag; /* list mode */
110 static int numflag = 5; /* gzip -1..-9 value */
111
112 #ifndef SMALL
113 static int fflag; /* force mode */
114 static int nflag; /* don't save name/timestamp */
115 static int Nflag; /* don't restore name/timestamp */
116 static int qflag; /* quiet mode */
117 static int rflag; /* recursive mode */
118 static int tflag; /* test */
119 static char *Sflag;
120 static int vflag; /* verbose mode */
121 #else
122 #define qflag 0
123 #endif
124
125 static char *suffix;
126 #define suffix_len (strlen(suffix) + 1) /* len + nul */
127 static char *newfile; /* name of newly created file */
128 static char *infile; /* name of file coming in */
129
130 static void maybe_err(int rv, const char *fmt, ...);
131 static void maybe_errx(int rv, const char *fmt, ...);
132 static void maybe_warn(const char *fmt, ...);
133 static void maybe_warnx(const char *fmt, ...);
134 static enum filetype file_gettype(u_char *);
135 static off_t gz_compress(FILE *, int, off_t *, const char *, time_t);
136 static off_t gz_uncompress(int, int, char *, size_t, off_t *);
137 static off_t file_compress(char *);
138 static off_t file_uncompress(char *);
139 static void handle_pathname(char *);
140 static void handle_file(char *, struct stat *);
141 static void handle_stdin(void);
142 static void handle_stdout(void);
143 static void print_ratio(off_t, off_t, FILE *);
144 static void print_list(int fd, off_t, const char *, time_t);
145 static void usage(void);
146 static void display_version(void);
147
148 #ifndef SMALL
149 static void prepend_gzip(char *, int *, char ***);
150 static void handle_dir(char *, struct stat *);
151 static void print_verbage(char *, char *, off_t, off_t);
152 static void print_test(char *, int);
153 static void copymodes(const char *, struct stat *);
154 #endif
155
156 #ifndef NO_BZIP2_SUPPORT
157 static off_t unbzip2(int, int, char *, size_t, off_t *);
158 #endif
159
160 #ifndef NO_COMPRESS_SUPPORT
161 static FILE *zopen(const char *, FILE *);
162 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *);
163 #endif
164
165 int main(int, char *p[]);
166
167 #ifdef SMALL
168 #define getopt_long(a,b,c,d,e) getopt(a,b,c)
169 #else
170 static const struct option longopts[] = {
171 { "stdout", no_argument, 0, 'c' },
172 { "to-stdout", no_argument, 0, 'c' },
173 { "decompress", no_argument, 0, 'd' },
174 { "uncompress", no_argument, 0, 'd' },
175 { "force", no_argument, 0, 'f' },
176 { "help", no_argument, 0, 'h' },
177 { "list", no_argument, 0, 'l' },
178 { "no-name", no_argument, 0, 'n' },
179 { "name", no_argument, 0, 'N' },
180 { "quiet", no_argument, 0, 'q' },
181 { "recursive", no_argument, 0, 'r' },
182 { "suffix", required_argument, 0, 'S' },
183 { "test", no_argument, 0, 't' },
184 { "verbose", no_argument, 0, 'v' },
185 { "version", no_argument, 0, 'V' },
186 { "fast", no_argument, 0, '1' },
187 { "best", no_argument, 0, '9' },
188 #if 0
189 /*
190 * This is what else GNU gzip implements. --ascii isn't useful
191 * on NetBSD, and I don't care to have a --license.
192 */
193 { "ascii", no_argument, 0, 'a' },
194 { "license", no_argument, 0, 'L' },
195 #endif
196 { NULL, no_argument, 0, 0 },
197 };
198 #endif
199
200 int
201 main(int argc, char **argv)
202 {
203 const char *progname = getprogname();
204 #ifndef SMALL
205 char *gzip;
206 #endif
207 int ch;
208
209 /* XXX set up signals */
210
211 suffix = GZ_SUFFIX;;
212
213 #ifndef SMALL
214 if ((gzip = getenv("GZIP")) != NULL)
215 prepend_gzip(gzip, &argc, &argv);
216 #endif
217
218 /*
219 * XXX
220 * handle being called `gunzip', `zcat' and `gzcat'
221 */
222 if (strcmp(progname, "gunzip") == 0)
223 dflag = 1;
224 else if (strcmp(progname, "zcat") == 0 ||
225 strcmp(progname, "gzcat") == 0)
226 dflag = cflag = 1;
227
228 #ifdef SMALL
229 #define OPT_LIST "cdhHltV123456789"
230 #else
231 #define OPT_LIST "cdfhHlnNqrS:tvV123456789"
232 #endif
233
234 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1)
235 switch (ch) {
236 case 'c':
237 cflag = 1;
238 break;
239 case 'd':
240 dflag = 1;
241 break;
242 case 'l':
243 lflag = 1;
244 dflag = 1;
245 break;
246 case 'V':
247 display_version();
248 /* NOTREACHED */
249 case '1': case '2': case '3':
250 case '4': case '5': case '6':
251 case '7': case '8': case '9':
252 numflag = ch - '0';
253 break;
254 #ifndef SMALL
255 case 'f':
256 fflag = 1;
257 break;
258 case 'n':
259 nflag = 1;
260 Nflag = 0;
261 break;
262 case 'N':
263 nflag = 0;
264 Nflag = 1;
265 break;
266 case 'q':
267 qflag = 1;
268 break;
269 case 'r':
270 rflag = 1;
271 break;
272 case 'S':
273 Sflag = optarg;
274 break;
275 case 't':
276 cflag = 1;
277 tflag = 1;
278 dflag = 1;
279 break;
280 case 'v':
281 vflag = 1;
282 break;
283 #endif
284 default:
285 usage();
286 /* NOTREACHED */
287 }
288 argv += optind;
289 argc -= optind;
290
291 if (argc == 0) {
292 if (dflag) /* stdin mode */
293 handle_stdin();
294 else /* stdout mode */
295 handle_stdout();
296 } else {
297 do {
298 handle_pathname(argv[0]);
299 } while (*++argv);
300 }
301 #ifndef SMALL
302 if (qflag == 0 && lflag && argc > 1)
303 print_list(-1, 0, "(totals)", 0);
304 #endif
305 exit(0);
306 }
307
308 /* maybe print a warning */
309 void
310 maybe_warn(const char *fmt, ...)
311 {
312 va_list ap;
313
314 if (qflag == 0) {
315 va_start(ap, fmt);
316 vwarn(fmt, ap);
317 va_end(ap);
318 }
319 }
320
321 void
322 maybe_warnx(const char *fmt, ...)
323 {
324 va_list ap;
325
326 if (qflag == 0) {
327 va_start(ap, fmt);
328 vwarnx(fmt, ap);
329 va_end(ap);
330 }
331 }
332
333 /* maybe print a warning */
334 void
335 maybe_err(int rv, const char *fmt, ...)
336 {
337 va_list ap;
338
339 if (qflag == 0) {
340 va_start(ap, fmt);
341 vwarn(fmt, ap);
342 va_end(ap);
343 }
344 exit(rv);
345 }
346
347 /* maybe print a warning */
348 void
349 maybe_errx(int rv, const char *fmt, ...)
350 {
351 va_list ap;
352
353 if (qflag == 0) {
354 va_start(ap, fmt);
355 vwarnx(fmt, ap);
356 va_end(ap);
357 }
358 exit(rv);
359 }
360
361 #ifndef SMALL
362 /* split up $GZIP and prepend it to the argument list */
363 static void
364 prepend_gzip(char *gzip, int *argc, char ***argv)
365 {
366 char *s, **nargv, **ac;
367 int nenvarg = 0, i;
368
369 /* scan how many arguments there are */
370 for (s = gzip; *s; s++) {
371 if (*s == ' ' || *s == '\t')
372 continue;
373 nenvarg++;
374 for (; *s; s++)
375 if (*s == ' ' || *s == '\t')
376 break;
377 if (*s == 0)
378 break;
379 }
380 /* punt early */
381 if (nenvarg == 0)
382 return;
383
384 *argc += nenvarg;
385 ac = *argv;
386
387 nargv = (char **)malloc((*argc + 1) * sizeof(char *));
388 if (nargv == NULL)
389 maybe_err(1, "malloc");
390
391 /* stash this away */
392 *argv = nargv;
393
394 /* copy the program name first */
395 i = 0;
396 nargv[i++] = *(ac++);
397
398 /* take a copy of $GZIP and add it to the array */
399 s = strdup(gzip);
400 if (s == NULL)
401 maybe_err(1, "strdup");
402 for (; *s; s++) {
403 if (*s == ' ' || *s == '\t')
404 continue;
405 nargv[i++] = s;
406 for (; *s; s++)
407 if (*s == ' ' || *s == '\t') {
408 *s = 0;
409 break;
410 }
411 }
412
413 /* copy the original arguments and a NULL */
414 while (*ac)
415 nargv[i++] = *(ac++);
416 nargv[i] = NULL;
417 }
418 #endif
419
420 /* compress input to output then close both files */
421 static off_t
422 gz_compress(FILE *in, int out, off_t *gsizep, const char *origname, time_t mtime)
423 {
424 z_stream z;
425 char inbuf[BUFLEN], outbuf[BUFLEN];
426 off_t in_tot = 0, out_tot = 0;
427 ssize_t in_size;
428 char *str;
429 int i, error;
430 uLong crc;
431
432 i = asprintf(&str, "%c%c%c%c%c%c%c%c%c%c%s",
433 GZIP_MAGIC0, GZIP_MAGIC1,
434 Z_DEFLATED, origname ? ORIG_NAME : 0,
435 (int)mtime & 0xff,
436 (int)(mtime >> 8) & 0xff,
437 (int)(mtime >> 16) & 0xff,
438 (int)(mtime >> 24) & 0xff,
439 0, OS_CODE, origname ? origname : "");
440 if (i == -1)
441 maybe_err(1, "asprintf");
442 if (origname)
443 i++;
444 if (write(out, str, i) != i)
445 maybe_err(1, "write");
446 free(str);
447
448 memset(&z, 0, sizeof z);
449 z.next_out = outbuf;
450 z.avail_out = sizeof outbuf;
451 z.zalloc = Z_NULL;
452 z.zfree = Z_NULL;
453 z.opaque = 0;
454
455 error = deflateInit2(&z, numflag, Z_DEFLATED,
456 -MAX_WBITS, 8, Z_DEFAULT_STRATEGY);
457 if (error != Z_OK)
458 maybe_errx(1, "deflateInit2 failed");
459
460 crc = crc32(0L, Z_NULL, 0);
461 for (;;) {
462 if (z.avail_out == 0) {
463 if (write(out, outbuf, sizeof outbuf) != sizeof outbuf)
464 maybe_err(1, "write");
465
466 out_tot += sizeof outbuf;
467 z.next_out = outbuf;
468 z.avail_out = sizeof outbuf;
469 }
470
471 if (z.avail_in == 0) {
472 in_size = fread(inbuf, 1, sizeof inbuf, in);
473 if (ferror(in))
474 maybe_err(1, "fread");
475 if (in_size == 0)
476 break;
477
478 crc = crc32(crc, (const Bytef *)inbuf, (unsigned)in_size);
479 in_tot += in_size;
480 z.next_in = inbuf;
481 z.avail_in = in_size;
482 }
483
484 error = deflate(&z, Z_NO_FLUSH);
485 if (error != Z_OK && error != Z_STREAM_END)
486 maybe_errx(1, "deflate failed");
487 }
488
489 /* clean up */
490 for (;;) {
491 size_t len;
492
493 error = deflate(&z, Z_FINISH);
494 if (error != Z_OK && error != Z_STREAM_END)
495 maybe_errx(1, "deflate failed");
496
497 len = sizeof outbuf - z.avail_out;
498
499 if (write(out, outbuf, len) != len)
500 maybe_err(1, "write");
501 out_tot += len;
502 z.next_out = outbuf;
503 z.avail_out = sizeof outbuf;
504
505 if (error == Z_STREAM_END)
506 break;
507 }
508
509 if (deflateEnd(&z) != Z_OK)
510 maybe_errx(1, "deflateEnd failed");
511
512 i = asprintf(&str, "%c%c%c%c%c%c%c%c",
513 (int)crc & 0xff,
514 (int)(crc >> 8) & 0xff,
515 (int)(crc >> 16) & 0xff,
516 (int)(crc >> 24) & 0xff,
517 (int)in_tot & 0xff,
518 (int)(in_tot >> 8) & 0xff,
519 (int)(in_tot >> 16) & 0xff,
520 (int)(in_tot >> 24) & 0xff);
521 if (i != 8)
522 maybe_err(1, "asprintf");
523 if (write(out, str, i) != i)
524 maybe_err(1, "write");
525 free(str);
526
527 if (fclose(in) < 0)
528 maybe_err(1, "failed fclose");
529
530 if (gsizep)
531 *gsizep = out_tot;
532 return in_tot;
533 }
534
535 /*
536 * uncompress input to output then close the input. return the
537 * uncompressed size written, and put the compressed sized read
538 * into `*gsizep'.
539 */
540 static off_t
541 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep)
542 {
543 z_stream z;
544 char outbuf[BUFLEN], inbuf[BUFLEN];
545 off_t out_tot, in_tot;
546 enum {
547 GZSTATE_MAGIC0,
548 GZSTATE_MAGIC1,
549 GZSTATE_METHOD,
550 GZSTATE_FLAGS,
551 GZSTATE_SKIPPING,
552 GZSTATE_EXTRA,
553 GZSTATE_EXTRA2,
554 GZSTATE_EXTRA3,
555 GZSTATE_ORIGNAME,
556 GZSTATE_COMMENT,
557 GZSTATE_HEAD_CRC1,
558 GZSTATE_HEAD_CRC2,
559 GZSTATE_INIT,
560 GZSTATE_READ
561 } state = GZSTATE_MAGIC0;
562 int flags = 0, skip_count = 0;
563 int error, done_reading = 0;
564
565 #define ADVANCE() { z.next_in++; z.avail_in--; }
566
567 memset(&z, 0, sizeof z);
568 z.avail_in = prelen;
569 z.next_in = pre;
570 z.avail_out = sizeof outbuf;
571 z.next_out = outbuf;
572 z.zalloc = NULL;
573 z.zfree = NULL;
574 z.opaque = 0;
575
576 in_tot = prelen;
577 out_tot = 0;
578
579 for (;;) {
580 if (z.avail_in == 0 && done_reading == 0) {
581 size_t in_size = read(in, inbuf, BUFLEN);
582
583 if (in_size == -1) {
584 #ifndef SMALL
585 if (tflag) {
586 print_test("(stdin)", 0);
587 return 0;
588 }
589 #endif
590 maybe_warn("failed to read stdin\n");
591 return -1;
592 } else if (in_size == 0)
593 done_reading = 1;
594
595 z.avail_in = in_size;
596 z.next_in = inbuf;
597
598 in_tot += in_size;
599 }
600 switch (state) {
601 case GZSTATE_MAGIC0:
602 if (*z.next_in != GZIP_MAGIC0)
603 maybe_err(1, "input not gziped\n");
604 ADVANCE();
605 state++;
606 break;
607
608 case GZSTATE_MAGIC1:
609 if (*z.next_in != GZIP_MAGIC1 &&
610 *z.next_in != GZIP_OMAGIC1)
611 maybe_err(1, "input not gziped\n");
612 ADVANCE();
613 state++;
614 break;
615
616 case GZSTATE_METHOD:
617 if (*z.next_in != Z_DEFLATED)
618 maybe_err(1, "unknown compression method\n");
619 ADVANCE();
620 state++;
621 break;
622
623 case GZSTATE_FLAGS:
624 flags = *z.next_in;
625 ADVANCE();
626 skip_count = 6;
627 state++;
628 break;
629
630 case GZSTATE_SKIPPING:
631 if (skip_count > 0) {
632 skip_count--;
633 ADVANCE();
634 } else
635 state++;
636 break;
637
638 case GZSTATE_EXTRA:
639 if ((flags & EXTRA_FIELD) == 0) {
640 state = GZSTATE_ORIGNAME;
641 break;
642 }
643 skip_count = *z.next_in;
644 ADVANCE();
645 state++;
646 break;
647
648 case GZSTATE_EXTRA2:
649 skip_count |= ((*z.next_in) << 8);
650 ADVANCE();
651 state++;
652 break;
653
654 case GZSTATE_EXTRA3:
655 if (skip_count > 0) {
656 skip_count--;
657 ADVANCE();
658 } else
659 state++;
660 break;
661
662 case GZSTATE_ORIGNAME:
663 if ((flags & ORIG_NAME) == 0) {
664 state++;
665 break;
666 }
667 if (*z.next_in == 0)
668 state++;
669 ADVANCE();
670 break;
671
672 case GZSTATE_COMMENT:
673 if ((flags & COMMENT) == 0) {
674 state++;
675 break;
676 }
677 if (*z.next_in == 0)
678 state++;
679 ADVANCE();
680 break;
681
682 case GZSTATE_HEAD_CRC1:
683 if (flags & HEAD_CRC)
684 skip_count = 2;
685 else
686 skip_count = 0;
687 state++;
688 break;
689
690 case GZSTATE_HEAD_CRC2:
691 if (skip_count > 0) {
692 skip_count--;
693 ADVANCE();
694 } else
695 state++;
696 break;
697
698 case GZSTATE_INIT:
699 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) {
700 maybe_err(1, "failed to inflateInit\n");
701 goto stop;
702 }
703 state++;
704 break;
705
706 case GZSTATE_READ:
707 error = inflate(&z, Z_FINISH);
708 if (error == Z_STREAM_END || error == Z_BUF_ERROR) {
709 size_t wr = BUFLEN - z.avail_out;
710
711 if (
712 #ifndef SMALL
713 /* don't write anything with -t */
714 tflag == 0 &&
715 #endif
716 write(out, outbuf, wr) != wr)
717 maybe_err(1, "error writing "
718 "to output\n");
719
720 out_tot += wr;
721
722 if (error == Z_STREAM_END)
723 goto stop;
724
725 z.next_out = outbuf;
726 z.avail_out = BUFLEN;
727
728 break;
729 }
730 if (error < 0) {
731 maybe_warnx("decompression error\n");
732 out_tot = -1;
733 goto stop;
734 }
735 break;
736 }
737 continue;
738 stop:
739 break;
740 }
741 if (state > GZSTATE_INIT)
742 inflateEnd(&z);
743
744 #ifndef SMALL
745 if (tflag) {
746 print_test("(stdin)", 1);
747 return 0;
748 }
749 #endif
750
751 if (gsizep)
752 *gsizep = in_tot;
753 return (out_tot);
754 }
755
756 #ifndef SMALL
757 /*
758 * set the owner, mode, flags & utimes for a file
759 */
760 static void
761 copymodes(const char *file, struct stat *sbp)
762 {
763 struct timeval times[2];
764
765 /*
766 * If we have no info on the input, give this file some
767 * default values and return..
768 */
769 if (sbp == NULL) {
770 mode_t mask = umask(022);
771
772 (void)chmod(file, DEFFILEMODE & ~mask);
773 (void)umask(mask);
774 return;
775 }
776
777 /* if the chown fails, remove set-id bits as-per compress(1) */
778 if (chown(file, sbp->st_uid, sbp->st_gid) < 0) {
779 if (errno != EPERM)
780 maybe_warn("couldn't chown: %s", file);
781 sbp->st_mode &= ~(S_ISUID|S_ISGID);
782 }
783
784 /* we only allow set-id and the 9 normal permission bits */
785 sbp->st_mode &= S_ISUID|S_ISGID|S_IRWXU|S_IRWXG|S_IRWXO;
786 if (chmod(file, sbp->st_mode) < 0)
787 maybe_warn("couldn't chmod: %s", file);
788
789 /* only try flags if they exist already */
790 if (sbp->st_flags != 0 && chflags(file, sbp->st_flags) < 0)
791 maybe_warn("couldn't chflags: %s", file);
792
793 TIMESPEC_TO_TIMEVAL(×[0], &sbp->st_atimespec);
794 TIMESPEC_TO_TIMEVAL(×[1], &sbp->st_mtimespec);
795 if (utimes(file, times) < 0)
796 maybe_warn("couldn't utimes: %s", file);
797 }
798 #endif
799
800 /* what sort of file is this? */
801 static enum filetype
802 file_gettype(u_char *buf)
803 {
804
805 if (buf[0] == GZIP_MAGIC0 &&
806 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1))
807 return FT_GZIP;
808 else
809 #ifndef NO_BZIP2_SUPPORT
810 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 &&
811 buf[3] >= '0' && buf[3] <= '9')
812 return FT_BZIP2;
813 else
814 #endif
815 #ifndef NO_COMPRESS_SUPPORT
816 if (memcmp(buf, Z_MAGIC, 2) == 0)
817 return FT_Z;
818 else
819 #endif
820 return FT_UNKNOWN;
821 }
822
823 /*
824 * compress the given file: create a corresponding .gz file and remove the
825 * original.
826 */
827 static off_t
828 file_compress(char *file)
829 {
830 FILE *in;
831 int out;
832 struct stat isb, osb;
833 char outfile[MAXPATHLEN];
834 off_t size;
835 #ifndef SMALL
836 u_int32_t mtime = 0;
837 char *savename;
838 #endif
839
840 if (cflag == 0) {
841 (void)strncpy(outfile, file, MAXPATHLEN - suffix_len);
842 outfile[MAXPATHLEN - suffix_len] = '\0';
843 (void)strlcat(outfile, suffix, sizeof(outfile));
844
845 #ifndef SMALL
846 if (fflag == 0) {
847 if (stat(outfile, &osb) == 0) {
848 maybe_warnx("%s already exists -- skipping",
849 outfile);
850 goto lose;
851 }
852 }
853 if (stat(file, &isb) == 0) {
854 if (isb.st_nlink > 1 && fflag == 0) {
855 maybe_warnx("%s has %d other link%s -- "
856 "skipping", file, isb.st_nlink - 1,
857 isb.st_nlink == 1 ? "" : "s");
858 goto lose;
859 }
860 if (nflag == 0)
861 mtime = (u_int32_t)isb.st_mtime;
862 }
863 #endif
864 }
865 in = fopen(file, "r");
866 if (in == 0)
867 maybe_err(1, "can't fopen %s", file);
868
869 if (cflag == 0) {
870 #ifndef SMALL
871 if (nflag == 0)
872 savename = basename(file);
873 else
874 savename = NULL;
875 #endif
876 out = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600);
877 if (out == -1) {
878 maybe_warn("could not create output: %s", outfile);
879 goto lose;
880 }
881 } else
882 out = STDOUT_FILENO;
883
884 #ifdef SMALL
885 gz_compress(in, out, NULL, NULL, 0);
886 #else
887 gz_compress(in, out, NULL, savename, mtime);
888 #endif
889
890 /*
891 * if we compressed to stdout, we don't know the size and
892 * we don't know the new file name, punt. if we can't stat
893 * the file, whine, otherwise set the size from the stat
894 * buffer. we only blow away the file if we can stat the
895 * output, just in case.
896 */
897 if (cflag == 0) {
898 if (close(out) == -1)
899 maybe_warn("couldn't close ouput");
900
901 if (stat(outfile, &osb) < 0) {
902 maybe_warn("couldn't stat: %s", outfile);
903 maybe_warnx("leaving original %s", file);
904 size = 0;
905 } else {
906 unlink(file);
907 size = osb.st_size;
908 }
909 newfile = outfile;
910 #ifndef SMALL
911 copymodes(outfile, &isb);
912 #endif
913 } else {
914 lose:
915 size = 0;
916 newfile = 0;
917 }
918
919 return (size);
920 }
921
922 /* uncompress the given file and remove the original */
923 static off_t
924 file_uncompress(char *file)
925 {
926 struct stat isb, osb;
927 char buf[PATH_MAX];
928 char *outfile = buf, *s;
929 off_t size;
930 ssize_t len = strlen(file);
931 int fd;
932 unsigned char header1[4], name[PATH_MAX + 1];
933 enum filetype method;
934
935 /* gather the old name info */
936
937 fd = open(file, O_RDONLY);
938 if (fd < 0)
939 maybe_err(1, "can't open %s", file);
940 if (read(fd, header1, sizeof header1) != sizeof header1) {
941 /* we don't want to fail here. */
942 #ifndef SMALL
943 if (fflag)
944 goto close_it;
945 #endif
946 maybe_err(1, "can't read %s", file);
947 }
948
949 method = file_gettype(header1);
950
951 #ifndef SMALL
952 if (Sflag == NULL) {
953 # ifndef NO_BZIP2_SUPPORT
954 if (method == FT_BZIP2)
955 suffix = BZ2_SUFFIX;
956 else
957 # endif
958 # ifndef NO_COMPRESS_SUPPORT
959 if (method == FT_Z)
960 suffix = Z_SUFFIX;
961 # endif
962 }
963
964 if (fflag == 0 && method == FT_UNKNOWN)
965 maybe_errx(1, "%s: not in gzip format", file);
966 #endif
967
968 if (cflag == 0 || lflag) {
969 s = &file[len - suffix_len + 1];
970 if (strncmp(s, suffix, suffix_len) == 0) {
971 (void)strncpy(outfile, file, len - suffix_len + 1);
972 outfile[len - suffix_len + 1] = '\0';
973 } else if (lflag == 0)
974 maybe_errx(1, "unknown suffix %s", s);
975 }
976
977 #ifdef SMALL
978 if (method == FT_GZIP && lflag)
979 #else
980 if (method == FT_GZIP && (Nflag || lflag))
981 #endif
982 {
983 if (header1[3] & ORIG_NAME) {
984 size_t rbytes;
985 int i;
986
987 rbytes = read(fd, name, PATH_MAX + 1);
988 if (rbytes < 0)
989 maybe_err(1, "can't read %s", file);
990 for (i = 0; i < rbytes && name[i]; i++)
991 ;
992 if (i < rbytes) {
993 name[i] = 0;
994 /* now maybe merge old dirname */
995 if (strchr(outfile, '/') == 0)
996 outfile = name;
997 else {
998 char *dir = dirname(outfile);
999 if (asprintf(&outfile, "%s/%s", dir,
1000 name) == -1)
1001 maybe_err(1, "malloc");
1002 }
1003 }
1004 }
1005 }
1006 #ifndef SMALL
1007 close_it:
1008 #endif
1009 close(fd);
1010
1011 if (cflag == 0 || lflag) {
1012 #ifndef SMALL
1013 if (fflag == 0 && lflag == 0 && stat(outfile, &osb) == 0) {
1014 maybe_warnx("%s already exists -- skipping", outfile);
1015 goto lose;
1016 }
1017 #endif
1018 if (stat(file, &isb) == 0) {
1019 #ifndef SMALL
1020 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) {
1021 maybe_warnx("%s has %d other links -- skipping",
1022 file, isb.st_nlink - 1);
1023 goto lose;
1024 }
1025 #endif
1026 } else
1027 goto lose;
1028 }
1029
1030 #ifndef NO_BZIP2_SUPPORT
1031 if (method == FT_BZIP2) {
1032 int in, out;
1033
1034 /* XXX */
1035 if (lflag)
1036 maybe_errx(1, "no -l with bzip2 files");
1037
1038 if ((in = open(file, O_RDONLY)) == -1)
1039 maybe_err(1, "open for read: %s", file);
1040 if (cflag == 1)
1041 out = STDOUT_FILENO;
1042 else
1043 out = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600);
1044 if (out == -1)
1045 maybe_err(1, "open for write: %s", outfile);
1046
1047 if ((size = unbzip2(in, out, NULL, 0, NULL)) == 0) {
1048 if (cflag == 0)
1049 unlink(outfile);
1050 goto lose;
1051 }
1052 } else
1053 #endif
1054
1055 #ifndef NO_COMPRESS_SUPPORT
1056 if (method == FT_Z) {
1057 FILE *in, *out;
1058 int fd;
1059
1060 /* XXX */
1061 if (lflag)
1062 maybe_errx(1, "no -l with Lempel-Ziv files");
1063
1064 if ((in = zopen(file, NULL)) == NULL)
1065 maybe_err(1, "open for read: %s", file);
1066
1067 if (cflag == 1)
1068 fd = STDOUT_FILENO;
1069 else {
1070 fd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600);
1071 if (fd == -1)
1072 maybe_err(1, "open for write: %s", outfile);
1073 }
1074 out = fdopen(fd, "w");
1075 if (out == NULL)
1076 maybe_err(1, "open for write: %s", outfile);
1077
1078 size = zuncompress(in, out, NULL, 0, NULL);
1079 if (cflag == 0) {
1080 if (size == 0) {
1081 unlink(outfile);
1082 goto lose;
1083 }
1084 if (ferror(in) || fclose(in)) {
1085 unlink(outfile);
1086 maybe_err(1, "failed infile fclose");
1087 }
1088 if (fclose(out)) {
1089 unlink(outfile);
1090 maybe_err(1, "failed outfile close");
1091 }
1092 }
1093 } else
1094 #endif
1095 {
1096 int fd, in;
1097
1098 if (lflag) {
1099 if ((fd = open(file, O_RDONLY)) == -1)
1100 maybe_err(1, "open");
1101 print_list(fd, isb.st_size, outfile, isb.st_mtime);
1102 return 0; /* XXX */
1103 }
1104
1105 in = open(file, O_RDONLY);
1106 if (in == -1)
1107 maybe_err(1, "can't open %s", file);
1108
1109 if (cflag == 0) {
1110 /* Use open(2) directly to get a safe file. */
1111 fd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600);
1112 if (fd < 0)
1113 maybe_err(1, "can't open %s", outfile);
1114 } else
1115 fd = STDOUT_FILENO;
1116
1117 size = gz_uncompress(in, fd, NULL, 0, NULL);
1118 if (cflag == 0) {
1119 if (size == -1) {
1120 unlink(outfile);
1121 goto lose;
1122 }
1123 if (close(fd))
1124 maybe_err(1, "failed close");
1125 }
1126 }
1127
1128 /* if testing, or we uncompressed to stdout, this is all we need */
1129 #ifndef SMALL
1130 if (tflag)
1131 return (size);
1132 #endif
1133 if (cflag)
1134 return (size);
1135
1136 /*
1137 * if we create a file...
1138 */
1139 if (cflag == 0) {
1140 /*
1141 * if we can't stat the file, or we are uncompressing to
1142 * stdin, don't remove the file.
1143 */
1144 if (stat(outfile, &osb) < 0) {
1145 maybe_warn("couldn't stat (leaving original): %s",
1146 outfile);
1147 goto lose;
1148 }
1149 if (osb.st_size != size) {
1150 maybe_warn("stat gave different size: %llu != %llu "
1151 "(leaving original)",
1152 (unsigned long long)size,
1153 (unsigned long long)osb.st_size);
1154 goto lose;
1155 }
1156 newfile = outfile;
1157 if (cflag == 0)
1158 unlink(file);
1159 size = osb.st_size;
1160 #ifndef SMALL
1161 copymodes(outfile, &isb);
1162 #endif
1163 }
1164 return (size);
1165
1166 lose:
1167 newfile = 0;
1168 return 0;
1169 }
1170
1171 #ifndef SMALL
1172 static off_t
1173 cat_stdin(unsigned char * prepend, size_t count, off_t *gsizep)
1174 {
1175 char buf[BUFLEN];
1176 size_t rv;
1177 off_t in_tot;
1178
1179 in_tot = count;
1180 if (write(STDOUT_FILENO, prepend, count) != count)
1181 maybe_err(1, "write to stdout");
1182 for (;;) {
1183 rv = read(STDIN_FILENO, buf, sizeof buf);
1184
1185 if (write(STDOUT_FILENO, buf, rv) != rv)
1186 maybe_err(1, "write to stdout");
1187 in_tot += rv;
1188 }
1189
1190 if (gsizep)
1191 *gsizep = in_tot;
1192 return (in_tot);
1193 }
1194 #endif
1195
1196 static void
1197 handle_stdin(void)
1198 {
1199 unsigned char header1[4];
1200 off_t usize, gsize;
1201 enum filetype method;
1202 #ifndef NO_COMPRESS_SUPPORT
1203 FILE *in;
1204 #endif
1205
1206 #ifndef SMALL
1207 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) {
1208 maybe_warnx("standard input is a terminal -- ignoring");
1209 return;
1210 }
1211 #endif
1212
1213 if (lflag) {
1214 struct stat isb;
1215
1216 /* XXX could read the whole file, etc. */
1217 if (fstat(STDIN_FILENO, &isb) < 0)
1218 maybe_err(1, "fstat");
1219 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime);
1220 return;
1221 }
1222
1223 if (read(STDIN_FILENO, header1, sizeof header1) != sizeof header1)
1224 maybe_err(1, "can't read stdin");
1225
1226 method = file_gettype(header1);
1227 switch (method) {
1228 default:
1229 #ifndef SMALL
1230 if (fflag == 0)
1231 maybe_errx(1, "unknown compression format");
1232 usize = cat_stdin(header1, sizeof header1, &gsize);
1233 break;
1234 #endif
1235 case FT_GZIP:
1236 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO,
1237 header1, sizeof header1, &gsize);
1238 break;
1239 #ifndef NO_BZIP2_SUPPORT
1240 case FT_BZIP2:
1241 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO,
1242 header1, sizeof header1, &gsize);
1243 break;
1244 #endif
1245 #ifndef NO_COMPRESS_SUPPORT
1246 case FT_Z:
1247 if ((in = zopen(NULL, stdin)) == NULL)
1248 maybe_err(1, "zopen of stdin");
1249
1250 usize = zuncompress(in, stdout, header1, sizeof header1, &gsize);
1251 break;
1252 #endif
1253 }
1254
1255 #ifndef SMALL
1256 if (vflag && !tflag && usize != -1 && gsize != -1)
1257 print_verbage(NULL, 0, usize, gsize);
1258 #endif
1259
1260 }
1261
1262 static void
1263 handle_stdout(void)
1264 {
1265 off_t gsize, usize;
1266
1267 #ifndef SMALL
1268 if (fflag == 0 && isatty(STDOUT_FILENO)) {
1269 maybe_warnx("standard output is a terminal -- ignoring");
1270 return;
1271 }
1272 #endif
1273 usize = gz_compress(stdin, STDOUT_FILENO, &gsize, NULL, 0);
1274
1275 #ifndef SMALL
1276 if (vflag && !tflag && usize != -1 && gsize != -1)
1277 print_verbage(NULL, 0, usize, gsize);
1278 #endif
1279 }
1280
1281 /* do what is asked for, for the path name */
1282 static void
1283 handle_pathname(char *path)
1284 {
1285 char *opath = path, *s = 0;
1286 ssize_t len;
1287 struct stat sb;
1288
1289 /* check for stdout/stdin */
1290 if (path[0] == '-' && path[1] == '\0') {
1291 if (dflag)
1292 handle_stdin();
1293 else
1294 handle_stdout();
1295 }
1296
1297 retry:
1298 if (stat(path, &sb) < 0) {
1299 /* lets try <path>.gz if we're decompressing */
1300 if (dflag && s == 0 && errno == ENOENT) {
1301 len = strlen(path);
1302 s = malloc(len + suffix_len);
1303 if (s == 0)
1304 maybe_err(1, "malloc");
1305 memmove(s, path, len);
1306 memmove(&s[len], suffix, suffix_len);
1307 path = s;
1308 goto retry;
1309 }
1310 maybe_warn("can't stat: %s", opath);
1311 goto out;
1312 }
1313
1314 if (S_ISDIR(sb.st_mode)) {
1315 #ifndef SMALL
1316 if (rflag)
1317 handle_dir(path, &sb);
1318 else
1319 #endif
1320 maybe_warn("%s is a directory", path);
1321 goto out;
1322 }
1323
1324 if (S_ISREG(sb.st_mode))
1325 handle_file(path, &sb);
1326
1327 out:
1328 if (s)
1329 free(s);
1330 }
1331
1332 /* compress/decompress a file */
1333 static void
1334 handle_file(char *file, struct stat *sbp)
1335 {
1336 off_t usize, gsize;
1337
1338 infile = file;
1339 if (dflag) {
1340 usize = file_uncompress(file);
1341 if (usize == 0)
1342 return;
1343 gsize = sbp->st_size;
1344 } else {
1345 gsize = file_compress(file);
1346 if (gsize == 0)
1347 return;
1348 usize = sbp->st_size;
1349 }
1350
1351
1352 #ifndef SMALL
1353 if (vflag && !tflag)
1354 print_verbage(file, cflag == 0 ? newfile : 0, usize, gsize);
1355 #endif
1356 }
1357
1358 #ifndef SMALL
1359 /* this is used with -r to recursively decend directories */
1360 static void
1361 handle_dir(char *dir, struct stat *sbp)
1362 {
1363 char *path_argv[2];
1364 FTS *fts;
1365 FTSENT *entry;
1366
1367 path_argv[0] = dir;
1368 path_argv[1] = 0;
1369 fts = fts_open(path_argv, FTS_PHYSICAL, NULL);
1370 if (fts == NULL) {
1371 warn("couldn't fts_open %s", dir);
1372 return;
1373 }
1374
1375 while ((entry = fts_read(fts))) {
1376 switch(entry->fts_info) {
1377 case FTS_D:
1378 case FTS_DP:
1379 continue;
1380
1381 case FTS_DNR:
1382 case FTS_ERR:
1383 case FTS_NS:
1384 maybe_warn("%s", entry->fts_path);
1385 continue;
1386 case FTS_F:
1387 handle_file(entry->fts_name, entry->fts_statp);
1388 }
1389 }
1390 (void)fts_close(fts);
1391 }
1392 #endif
1393
1394 /* print a ratio */
1395 static void
1396 print_ratio(off_t in, off_t out, FILE *where)
1397 {
1398 int64_t percent10; /* 10 * percent */
1399 off_t diff = in - out;
1400 char ch;
1401
1402 if (in == 0)
1403 percent10 = 0;
1404 else if (diff > 0x400000) /* anything with 22 or more bits */
1405 percent10 = diff / (in / 1000);
1406 else
1407 percent10 = (1000 * diff) / in;
1408
1409 if (percent10 < 0) {
1410 percent10 = -percent10;
1411 ch = '-';
1412 } else
1413 ch = ' ';
1414
1415 /*
1416 * ugh. for negative percentages < 10, we need to avoid printing a
1417 * a space between the "-" and the single number.
1418 */
1419 if (ch == '-' && percent10 / 10LL < 10)
1420 fprintf(where, " -%1d.%1u%%", (unsigned)(percent10 / 10LL),
1421 (unsigned)(percent10 % 10LL));
1422 else
1423 fprintf(where, "%c%2d.%1u%%", ch, (unsigned)(percent10 / 10LL),
1424 (unsigned)(percent10 % 10LL));
1425 }
1426
1427 #ifndef SMALL
1428 /* print compression statistics, and the new name (if there is one!) */
1429 static void
1430 print_verbage(char *file, char *nfile, off_t usize, off_t gsize)
1431 {
1432 if (file)
1433 fprintf(stderr, "%s:%s ", file,
1434 strlen(file) < 7 ? "\t\t" : "\t");
1435 print_ratio((off_t)usize, (off_t)gsize, stderr);
1436 if (nfile)
1437 fprintf(stderr, " -- replaced with %s", nfile);
1438 fprintf(stderr, "\n");
1439 fflush(stderr);
1440 }
1441
1442 /* print test results */
1443 static void
1444 print_test(char *file, int ok)
1445 {
1446
1447 fprintf(stderr, "%s:%s %s\n", file,
1448 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK");
1449 fflush(stderr);
1450 }
1451 #endif
1452
1453 /* print a file's info ala --list */
1454 /* eg:
1455 compressed uncompressed ratio uncompressed_name
1456 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar
1457 */
1458 static void
1459 print_list(int fd, off_t out, const char *outfile, time_t ts)
1460 {
1461 static int first = 1;
1462 #ifndef SMALL
1463 static off_t in_tot, out_tot;
1464 u_int32_t crc;
1465 #endif
1466 off_t in;
1467 int rv;
1468
1469 if (first) {
1470 #ifndef SMALL
1471 if (vflag)
1472 printf("method crc date time ");
1473 #endif
1474 if (qflag == 0)
1475 printf(" compressed uncompressed "
1476 "ratio uncompressed_name\n");
1477 }
1478 first = 0;
1479
1480 /* print totals? */
1481 #ifndef SMALL
1482 if (fd == -1) {
1483 in = in_tot;
1484 out = out_tot;
1485 } else
1486 #endif
1487 {
1488 /* read the last 4 bytes - this is the uncompressed size */
1489 rv = lseek(fd, (off_t)(-8), SEEK_END);
1490 if (rv != -1) {
1491 unsigned char buf[8];
1492 u_int32_t usize;
1493
1494 if (read(fd, (char *)buf, sizeof(buf)) != sizeof(buf))
1495 maybe_warn("read of uncompressed size");
1496 usize = buf[4] | buf[5] << 8 | buf[6] << 16 | buf[7] << 24;
1497 in = (off_t)usize;
1498 #ifndef SMALL
1499 crc = buf[0] | buf[1] << 8 | buf[2] << 16 | buf[3] << 24;
1500 #endif
1501 }
1502 }
1503
1504 #ifndef SMALL
1505 if (vflag && fd == -1)
1506 printf(" ");
1507 else if (vflag) {
1508 char *date = ctime(&ts);
1509
1510 /* skip the day, 1/100th second, and year */
1511 date += 4;
1512 date[12] = 0;
1513 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date);
1514 }
1515 in_tot += in;
1516 out_tot += out;
1517 #endif
1518 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in);
1519 print_ratio(in, out, stdout);
1520 printf(" %s\n", outfile);
1521 }
1522
1523 /* display the usage of NetBSD gzip */
1524 static void
1525 usage(void)
1526 {
1527
1528 fprintf(stderr, "%s\n", gzip_version);
1529 fprintf(stderr,
1530 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n"
1531 #ifndef SMALL
1532 " -c --stdout write to stdout, keep original files\n"
1533 " --to-stdout\n"
1534 " -d --decompress uncompress files\n"
1535 " --uncompress\n"
1536 " -f --force force overwriting & compress links\n"
1537 " -h --help display this help\n"
1538 " -n --no-name don't save original file name or time stamp\n"
1539 " -N --name save or restore original file name and time stamp\n"
1540 " -q --quiet output no warnings\n"
1541 " -r --recursive recursively compress files in directories\n"
1542 " -S .suf use suffix .suf instead of .gz\n"
1543 " --suffix .suf\n"
1544 " -t --test test compressed file\n"
1545 " -v --verbose print extra statistics\n"
1546 " -V --version display program version\n"
1547 " -1 --fast fastest (worst) compression\n"
1548 " -2 .. -8 set compression level\n"
1549 " -9 --best best (slowest) compression\n",
1550 #else
1551 ,
1552 #endif
1553 getprogname());
1554 exit(0);
1555 }
1556
1557 /* display the version of NetBSD gzip */
1558 static void
1559 display_version(void)
1560 {
1561
1562 fprintf(stderr, "%s\n", gzip_version);
1563 exit(0);
1564 }
1565
1566 #ifndef NO_BZIP2_SUPPORT
1567 #include "unbzip2.c"
1568 #endif
1569 #ifndef NO_COMPRESS_SUPPORT
1570 #include "zuncompress.c"
1571 #endif
1572