gzip.c revision 1.107 1 /* $NetBSD: gzip.c,v 1.107 2015/01/13 02:37:20 mrg Exp $ */
2
3 /*
4 * Copyright (c) 1997, 1998, 2003, 2004, 2006 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 *
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
26 * SUCH DAMAGE.
27 */
28
29 #include <sys/cdefs.h>
30 #ifndef lint
31 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004, 2006\
32 Matthew R. Green. All rights reserved.");
33 __RCSID("$NetBSD: gzip.c,v 1.107 2015/01/13 02:37:20 mrg Exp $");
34 #endif /* not lint */
35
36 /*
37 * gzip.c -- GPL free gzip using zlib.
38 *
39 * RFC 1950 covers the zlib format
40 * RFC 1951 covers the deflate format
41 * RFC 1952 covers the gzip format
42 *
43 * TODO:
44 * - use mmap where possible
45 * - handle some signals better (remove outfile?)
46 * - make bzip2/compress -v/-t/-l support work as well as possible
47 */
48
49 #include <sys/param.h>
50 #include <sys/stat.h>
51 #include <sys/time.h>
52
53 #include <inttypes.h>
54 #include <unistd.h>
55 #include <stdio.h>
56 #include <string.h>
57 #include <stdlib.h>
58 #include <err.h>
59 #include <errno.h>
60 #include <fcntl.h>
61 #include <zlib.h>
62 #include <fts.h>
63 #include <libgen.h>
64 #include <stdarg.h>
65 #include <getopt.h>
66 #include <time.h>
67
68 #ifndef PRIdOFF
69 #define PRIdOFF PRId64
70 #endif
71
72 /* what type of file are we dealing with */
73 enum filetype {
74 FT_GZIP,
75 #ifndef NO_BZIP2_SUPPORT
76 FT_BZIP2,
77 #endif
78 #ifndef NO_COMPRESS_SUPPORT
79 FT_Z,
80 #endif
81 #ifndef NO_PACK_SUPPORT
82 FT_PACK,
83 #endif
84 #ifndef NO_XZ_SUPPORT
85 FT_XZ,
86 #endif
87 FT_LAST,
88 FT_UNKNOWN
89 };
90
91 #ifndef NO_BZIP2_SUPPORT
92 #include <bzlib.h>
93
94 #define BZ2_SUFFIX ".bz2"
95 #define BZIP2_MAGIC "\102\132\150"
96 #endif
97
98 #ifndef NO_COMPRESS_SUPPORT
99 #define Z_SUFFIX ".Z"
100 #define Z_MAGIC "\037\235"
101 #endif
102
103 #ifndef NO_PACK_SUPPORT
104 #define PACK_MAGIC "\037\036"
105 #endif
106
107 #ifndef NO_XZ_SUPPORT
108 #include <lzma.h>
109 #define XZ_SUFFIX ".xz"
110 #define XZ_MAGIC "\3757zXZ"
111 #endif
112
113 #define GZ_SUFFIX ".gz"
114
115 #define BUFLEN (64 * 1024)
116
117 #define GZIP_MAGIC0 0x1F
118 #define GZIP_MAGIC1 0x8B
119 #define GZIP_OMAGIC1 0x9E
120
121 #define GZIP_TIMESTAMP (off_t)4
122 #define GZIP_ORIGNAME (off_t)10
123
124 #define HEAD_CRC 0x02
125 #define EXTRA_FIELD 0x04
126 #define ORIG_NAME 0x08
127 #define COMMENT 0x10
128
129 #define OS_CODE 3 /* Unix */
130
131 typedef struct {
132 const char *zipped;
133 int ziplen;
134 const char *normal; /* for unzip - must not be longer than zipped */
135 } suffixes_t;
136 static suffixes_t suffixes[] = {
137 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N}
138 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */
139 #ifndef SMALL
140 SUFFIX(GZ_SUFFIX, ""),
141 SUFFIX(".z", ""),
142 SUFFIX("-gz", ""),
143 SUFFIX("-z", ""),
144 SUFFIX("_z", ""),
145 SUFFIX(".taz", ".tar"),
146 SUFFIX(".tgz", ".tar"),
147 #ifndef NO_BZIP2_SUPPORT
148 SUFFIX(BZ2_SUFFIX, ""),
149 #endif
150 #ifndef NO_COMPRESS_SUPPORT
151 SUFFIX(Z_SUFFIX, ""),
152 #endif
153 #ifndef NO_XZ_SUPPORT
154 SUFFIX(XZ_SUFFIX, ""),
155 #endif
156 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */
157 #endif /* SMALL */
158 #undef SUFFIX
159 };
160 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0])
161 #define SUFFIX_MAXLEN 30
162
163 static const char gzip_version[] = "NetBSD gzip 20150113";
164
165 static int cflag; /* stdout mode */
166 static int dflag; /* decompress mode */
167 static int lflag; /* list mode */
168 static int numflag = 6; /* gzip -1..-9 value */
169
170 #ifndef SMALL
171 static int fflag; /* force mode */
172 static int kflag; /* don't delete input files */
173 static int nflag; /* don't save name/timestamp */
174 static int Nflag; /* don't restore name/timestamp */
175 static int qflag; /* quiet mode */
176 static int rflag; /* recursive mode */
177 static int tflag; /* test */
178 static int vflag; /* verbose mode */
179 #else
180 #define qflag 0
181 #define tflag 0
182 #endif
183
184 static int exit_value = 0; /* exit value */
185
186 static char *infile; /* name of file coming in */
187
188 static void maybe_err(const char *fmt, ...) __printflike(1, 2) __dead;
189 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \
190 !defined(NO_XZ_SUPPORT)
191 static void maybe_errx(const char *fmt, ...) __printflike(1, 2) __dead;
192 #endif
193 static void maybe_warn(const char *fmt, ...) __printflike(1, 2);
194 static void maybe_warnx(const char *fmt, ...) __printflike(1, 2);
195 static enum filetype file_gettype(u_char *);
196 #ifdef SMALL
197 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz)
198 #endif
199 static off_t gz_compress(int, int, off_t *, const char *, uint32_t);
200 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *);
201 static off_t file_compress(char *, char *, size_t);
202 static off_t file_uncompress(char *, char *, size_t);
203 static void handle_pathname(char *);
204 static void handle_file(char *, struct stat *);
205 static void handle_stdin(void);
206 static void handle_stdout(void);
207 static void print_ratio(off_t, off_t, FILE *);
208 static void print_list(int fd, off_t, const char *, time_t);
209 __dead static void usage(void);
210 __dead static void display_version(void);
211 static const suffixes_t *check_suffix(char *, int);
212 static ssize_t read_retry(int, void *, size_t);
213
214 #ifdef SMALL
215 #define unlink_input(f, sb) unlink(f)
216 #else
217 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd);
218 static void prepend_gzip(char *, int *, char ***);
219 static void handle_dir(char *);
220 static void print_verbage(const char *, const char *, off_t, off_t);
221 static void print_test(const char *, int);
222 static void copymodes(int fd, const struct stat *, const char *file);
223 static int check_outfile(const char *outfile);
224 #endif
225
226 #ifndef NO_BZIP2_SUPPORT
227 static off_t unbzip2(int, int, char *, size_t, off_t *);
228 #endif
229
230 #ifndef NO_COMPRESS_SUPPORT
231 static FILE *zdopen(int);
232 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *);
233 #endif
234
235 #ifndef NO_PACK_SUPPORT
236 static off_t unpack(int, int, char *, size_t, off_t *);
237 #endif
238
239 #ifndef NO_XZ_SUPPORT
240 static off_t unxz(int, int, char *, size_t, off_t *);
241 #endif
242
243 #ifdef SMALL
244 #define getopt_long(a,b,c,d,e) getopt(a,b,c)
245 #else
246 static const struct option longopts[] = {
247 { "stdout", no_argument, 0, 'c' },
248 { "to-stdout", no_argument, 0, 'c' },
249 { "decompress", no_argument, 0, 'd' },
250 { "uncompress", no_argument, 0, 'd' },
251 { "force", no_argument, 0, 'f' },
252 { "help", no_argument, 0, 'h' },
253 { "keep", no_argument, 0, 'k' },
254 { "list", no_argument, 0, 'l' },
255 { "no-name", no_argument, 0, 'n' },
256 { "name", no_argument, 0, 'N' },
257 { "quiet", no_argument, 0, 'q' },
258 { "recursive", no_argument, 0, 'r' },
259 { "suffix", required_argument, 0, 'S' },
260 { "test", no_argument, 0, 't' },
261 { "verbose", no_argument, 0, 'v' },
262 { "version", no_argument, 0, 'V' },
263 { "fast", no_argument, 0, '1' },
264 { "best", no_argument, 0, '9' },
265 #if 0
266 /*
267 * This is what else GNU gzip implements. --ascii isn't useful
268 * on NetBSD, and I don't care to have a --license.
269 */
270 { "ascii", no_argument, 0, 'a' },
271 { "license", no_argument, 0, 'L' },
272 #endif
273 { NULL, no_argument, 0, 0 },
274 };
275 #endif
276
277 int
278 main(int argc, char **argv)
279 {
280 const char *progname = getprogname();
281 #ifndef SMALL
282 char *gzip;
283 int len;
284 #endif
285 int ch;
286
287 /* XXX set up signals */
288
289 #ifndef SMALL
290 if ((gzip = getenv("GZIP")) != NULL)
291 prepend_gzip(gzip, &argc, &argv);
292 #endif
293
294 /*
295 * XXX
296 * handle being called `gunzip', `zcat' and `gzcat'
297 */
298 if (strcmp(progname, "gunzip") == 0)
299 dflag = 1;
300 else if (strcmp(progname, "zcat") == 0 ||
301 strcmp(progname, "gzcat") == 0)
302 dflag = cflag = 1;
303
304 #ifdef SMALL
305 #define OPT_LIST "123456789cdhlV"
306 #else
307 #define OPT_LIST "123456789cdfhklNnqrS:tVv"
308 #endif
309
310 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) {
311 switch (ch) {
312 case '1': case '2': case '3':
313 case '4': case '5': case '6':
314 case '7': case '8': case '9':
315 numflag = ch - '0';
316 break;
317 case 'c':
318 cflag = 1;
319 break;
320 case 'd':
321 dflag = 1;
322 break;
323 case 'l':
324 lflag = 1;
325 dflag = 1;
326 break;
327 case 'V':
328 display_version();
329 /* NOTREACHED */
330 #ifndef SMALL
331 case 'f':
332 fflag = 1;
333 break;
334 case 'k':
335 kflag = 1;
336 break;
337 case 'N':
338 nflag = 0;
339 Nflag = 1;
340 break;
341 case 'n':
342 nflag = 1;
343 Nflag = 0;
344 break;
345 case 'q':
346 qflag = 1;
347 break;
348 case 'r':
349 rflag = 1;
350 break;
351 case 'S':
352 len = strlen(optarg);
353 if (len != 0) {
354 if (len > SUFFIX_MAXLEN)
355 errx(1, "incorrect suffix: '%s'", optarg);
356 suffixes[0].zipped = optarg;
357 suffixes[0].ziplen = len;
358 } else {
359 suffixes[NUM_SUFFIXES - 1].zipped = "";
360 suffixes[NUM_SUFFIXES - 1].ziplen = 0;
361 }
362 break;
363 case 't':
364 cflag = 1;
365 tflag = 1;
366 dflag = 1;
367 break;
368 case 'v':
369 vflag = 1;
370 break;
371 #endif
372 default:
373 usage();
374 /* NOTREACHED */
375 }
376 }
377 argv += optind;
378 argc -= optind;
379
380 if (argc == 0) {
381 if (dflag) /* stdin mode */
382 handle_stdin();
383 else /* stdout mode */
384 handle_stdout();
385 } else {
386 do {
387 handle_pathname(argv[0]);
388 } while (*++argv);
389 }
390 #ifndef SMALL
391 if (qflag == 0 && lflag && argc > 1)
392 print_list(-1, 0, "(totals)", 0);
393 #endif
394 exit(exit_value);
395 }
396
397 /* maybe print a warning */
398 void
399 maybe_warn(const char *fmt, ...)
400 {
401 va_list ap;
402
403 if (qflag == 0) {
404 va_start(ap, fmt);
405 vwarn(fmt, ap);
406 va_end(ap);
407 }
408 if (exit_value == 0)
409 exit_value = 1;
410 }
411
412 /* ... without an errno. */
413 void
414 maybe_warnx(const char *fmt, ...)
415 {
416 va_list ap;
417
418 if (qflag == 0) {
419 va_start(ap, fmt);
420 vwarnx(fmt, ap);
421 va_end(ap);
422 }
423 if (exit_value == 0)
424 exit_value = 1;
425 }
426
427 /* maybe print an error */
428 void
429 maybe_err(const char *fmt, ...)
430 {
431 va_list ap;
432
433 if (qflag == 0) {
434 va_start(ap, fmt);
435 vwarn(fmt, ap);
436 va_end(ap);
437 }
438 exit(2);
439 }
440
441 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \
442 !defined(NO_XZ_SUPPORT)
443 /* ... without an errno. */
444 void
445 maybe_errx(const char *fmt, ...)
446 {
447 va_list ap;
448
449 if (qflag == 0) {
450 va_start(ap, fmt);
451 vwarnx(fmt, ap);
452 va_end(ap);
453 }
454 exit(2);
455 }
456 #endif
457
458 #ifndef SMALL
459 /* split up $GZIP and prepend it to the argument list */
460 static void
461 prepend_gzip(char *gzip, int *argc, char ***argv)
462 {
463 char *s, **nargv, **ac;
464 int nenvarg = 0, i;
465
466 /* scan how many arguments there are */
467 for (s = gzip;;) {
468 while (*s == ' ' || *s == '\t')
469 s++;
470 if (*s == 0)
471 goto count_done;
472 nenvarg++;
473 while (*s != ' ' && *s != '\t')
474 if (*s++ == 0)
475 goto count_done;
476 }
477 count_done:
478 /* punt early */
479 if (nenvarg == 0)
480 return;
481
482 *argc += nenvarg;
483 ac = *argv;
484
485 nargv = (char **)malloc((*argc + 1) * sizeof(char *));
486 if (nargv == NULL)
487 maybe_err("malloc");
488
489 /* stash this away */
490 *argv = nargv;
491
492 /* copy the program name first */
493 i = 0;
494 nargv[i++] = *(ac++);
495
496 /* take a copy of $GZIP and add it to the array */
497 s = strdup(gzip);
498 if (s == NULL)
499 maybe_err("strdup");
500 for (;;) {
501 /* Skip whitespaces. */
502 while (*s == ' ' || *s == '\t')
503 s++;
504 if (*s == 0)
505 goto copy_done;
506 nargv[i++] = s;
507 /* Find the end of this argument. */
508 while (*s != ' ' && *s != '\t')
509 if (*s++ == 0)
510 /* Argument followed by NUL. */
511 goto copy_done;
512 /* Terminate by overwriting ' ' or '\t' with NUL. */
513 *s++ = 0;
514 }
515 copy_done:
516
517 /* copy the original arguments and a NULL */
518 while (*ac)
519 nargv[i++] = *(ac++);
520 nargv[i] = NULL;
521 }
522 #endif
523
524 /* compress input to output. Return bytes read, -1 on error */
525 static off_t
526 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime)
527 {
528 z_stream z;
529 char *outbufp, *inbufp;
530 off_t in_tot = 0, out_tot = 0;
531 ssize_t in_size;
532 int i, error;
533 uLong crc;
534 #ifdef SMALL
535 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0,
536 0, 0, 0, 0,
537 0, OS_CODE };
538 #endif
539
540 outbufp = malloc(BUFLEN);
541 inbufp = malloc(BUFLEN);
542 if (outbufp == NULL || inbufp == NULL) {
543 maybe_err("malloc failed");
544 goto out;
545 }
546
547 memset(&z, 0, sizeof z);
548 z.zalloc = Z_NULL;
549 z.zfree = Z_NULL;
550 z.opaque = 0;
551
552 #ifdef SMALL
553 memcpy(outbufp, header, sizeof header);
554 i = sizeof header;
555 #else
556 if (nflag != 0) {
557 mtime = 0;
558 origname = "";
559 }
560
561 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s",
562 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED,
563 *origname ? ORIG_NAME : 0,
564 mtime & 0xff,
565 (mtime >> 8) & 0xff,
566 (mtime >> 16) & 0xff,
567 (mtime >> 24) & 0xff,
568 numflag == 1 ? 4 : numflag == 9 ? 2 : 0,
569 OS_CODE, origname);
570 if (i >= BUFLEN)
571 /* this need PATH_MAX > BUFLEN ... */
572 maybe_err("snprintf");
573 if (*origname)
574 i++;
575 #endif
576
577 z.next_out = (unsigned char *)outbufp + i;
578 z.avail_out = BUFLEN - i;
579
580 error = deflateInit2(&z, numflag, Z_DEFLATED,
581 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY);
582 if (error != Z_OK) {
583 maybe_warnx("deflateInit2 failed");
584 in_tot = -1;
585 goto out;
586 }
587
588 crc = crc32(0L, Z_NULL, 0);
589 for (;;) {
590 if (z.avail_out == 0) {
591 if (write(out, outbufp, BUFLEN) != BUFLEN) {
592 maybe_warn("write");
593 out_tot = -1;
594 goto out;
595 }
596
597 out_tot += BUFLEN;
598 z.next_out = (unsigned char *)outbufp;
599 z.avail_out = BUFLEN;
600 }
601
602 if (z.avail_in == 0) {
603 in_size = read(in, inbufp, BUFLEN);
604 if (in_size < 0) {
605 maybe_warn("read");
606 in_tot = -1;
607 goto out;
608 }
609 if (in_size == 0)
610 break;
611
612 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size);
613 in_tot += in_size;
614 z.next_in = (unsigned char *)inbufp;
615 z.avail_in = in_size;
616 }
617
618 error = deflate(&z, Z_NO_FLUSH);
619 if (error != Z_OK && error != Z_STREAM_END) {
620 maybe_warnx("deflate failed");
621 in_tot = -1;
622 goto out;
623 }
624 }
625
626 /* clean up */
627 for (;;) {
628 size_t len;
629 ssize_t w;
630
631 error = deflate(&z, Z_FINISH);
632 if (error != Z_OK && error != Z_STREAM_END) {
633 maybe_warnx("deflate failed");
634 in_tot = -1;
635 goto out;
636 }
637
638 len = (char *)z.next_out - outbufp;
639
640 w = write(out, outbufp, len);
641 if (w == -1 || (size_t)w != len) {
642 maybe_warn("write");
643 out_tot = -1;
644 goto out;
645 }
646 out_tot += len;
647 z.next_out = (unsigned char *)outbufp;
648 z.avail_out = BUFLEN;
649
650 if (error == Z_STREAM_END)
651 break;
652 }
653
654 if (deflateEnd(&z) != Z_OK) {
655 maybe_warnx("deflateEnd failed");
656 in_tot = -1;
657 goto out;
658 }
659
660 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c",
661 (int)crc & 0xff,
662 (int)(crc >> 8) & 0xff,
663 (int)(crc >> 16) & 0xff,
664 (int)(crc >> 24) & 0xff,
665 (int)in_tot & 0xff,
666 (int)(in_tot >> 8) & 0xff,
667 (int)(in_tot >> 16) & 0xff,
668 (int)(in_tot >> 24) & 0xff);
669 if (i != 8)
670 maybe_err("snprintf");
671 #if 0
672 if (in_tot > 0xffffffff)
673 maybe_warn("input file size >= 4GB cannot be saved");
674 #endif
675 if (write(out, outbufp, i) != i) {
676 maybe_warn("write");
677 in_tot = -1;
678 } else
679 out_tot += i;
680
681 out:
682 if (inbufp != NULL)
683 free(inbufp);
684 if (outbufp != NULL)
685 free(outbufp);
686 if (gsizep)
687 *gsizep = out_tot;
688 return in_tot;
689 }
690
691 /*
692 * uncompress input to output then close the input. return the
693 * uncompressed size written, and put the compressed sized read
694 * into `*gsizep'.
695 */
696 static off_t
697 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep,
698 const char *filename)
699 {
700 z_stream z;
701 char *outbufp, *inbufp;
702 off_t out_tot = -1, in_tot = 0;
703 uint32_t out_sub_tot = 0;
704 enum {
705 GZSTATE_MAGIC0,
706 GZSTATE_MAGIC1,
707 GZSTATE_METHOD,
708 GZSTATE_FLAGS,
709 GZSTATE_SKIPPING,
710 GZSTATE_EXTRA,
711 GZSTATE_EXTRA2,
712 GZSTATE_EXTRA3,
713 GZSTATE_ORIGNAME,
714 GZSTATE_COMMENT,
715 GZSTATE_HEAD_CRC1,
716 GZSTATE_HEAD_CRC2,
717 GZSTATE_INIT,
718 GZSTATE_READ,
719 GZSTATE_CRC,
720 GZSTATE_LEN,
721 } state = GZSTATE_MAGIC0;
722 int flags = 0, skip_count = 0;
723 int error = Z_STREAM_ERROR, done_reading = 0;
724 uLong crc = 0;
725 ssize_t wr;
726 int needmore = 0;
727
728 #define ADVANCE() { z.next_in++; z.avail_in--; }
729
730 if ((outbufp = malloc(BUFLEN)) == NULL) {
731 maybe_err("malloc failed");
732 goto out2;
733 }
734 if ((inbufp = malloc(BUFLEN)) == NULL) {
735 maybe_err("malloc failed");
736 goto out1;
737 }
738
739 memset(&z, 0, sizeof z);
740 z.avail_in = prelen;
741 z.next_in = (unsigned char *)pre;
742 z.avail_out = BUFLEN;
743 z.next_out = (unsigned char *)outbufp;
744 z.zalloc = NULL;
745 z.zfree = NULL;
746 z.opaque = 0;
747
748 in_tot = prelen;
749 out_tot = 0;
750
751 for (;;) {
752 if ((z.avail_in == 0 || needmore) && done_reading == 0) {
753 ssize_t in_size;
754
755 if (z.avail_in > 0) {
756 memmove(inbufp, z.next_in, z.avail_in);
757 }
758 z.next_in = (unsigned char *)inbufp;
759 in_size = read(in, z.next_in + z.avail_in,
760 BUFLEN - z.avail_in);
761
762 if (in_size == -1) {
763 maybe_warn("failed to read stdin");
764 goto stop_and_fail;
765 } else if (in_size == 0) {
766 done_reading = 1;
767 }
768
769 z.avail_in += in_size;
770 needmore = 0;
771
772 in_tot += in_size;
773 }
774 if (z.avail_in == 0) {
775 if (done_reading && state != GZSTATE_MAGIC0) {
776 maybe_warnx("%s: unexpected end of file",
777 filename);
778 goto stop_and_fail;
779 }
780 goto stop;
781 }
782 switch (state) {
783 case GZSTATE_MAGIC0:
784 if (*z.next_in != GZIP_MAGIC0) {
785 if (in_tot > 0) {
786 maybe_warnx("%s: trailing garbage "
787 "ignored", filename);
788 goto stop;
789 }
790 maybe_warnx("input not gziped (MAGIC0)");
791 goto stop_and_fail;
792 }
793 ADVANCE();
794 state++;
795 out_sub_tot = 0;
796 crc = crc32(0L, Z_NULL, 0);
797 break;
798
799 case GZSTATE_MAGIC1:
800 if (*z.next_in != GZIP_MAGIC1 &&
801 *z.next_in != GZIP_OMAGIC1) {
802 maybe_warnx("input not gziped (MAGIC1)");
803 goto stop_and_fail;
804 }
805 ADVANCE();
806 state++;
807 break;
808
809 case GZSTATE_METHOD:
810 if (*z.next_in != Z_DEFLATED) {
811 maybe_warnx("unknown compression method");
812 goto stop_and_fail;
813 }
814 ADVANCE();
815 state++;
816 break;
817
818 case GZSTATE_FLAGS:
819 flags = *z.next_in;
820 ADVANCE();
821 skip_count = 6;
822 state++;
823 break;
824
825 case GZSTATE_SKIPPING:
826 if (skip_count > 0) {
827 skip_count--;
828 ADVANCE();
829 } else
830 state++;
831 break;
832
833 case GZSTATE_EXTRA:
834 if ((flags & EXTRA_FIELD) == 0) {
835 state = GZSTATE_ORIGNAME;
836 break;
837 }
838 skip_count = *z.next_in;
839 ADVANCE();
840 state++;
841 break;
842
843 case GZSTATE_EXTRA2:
844 skip_count |= ((*z.next_in) << 8);
845 ADVANCE();
846 state++;
847 break;
848
849 case GZSTATE_EXTRA3:
850 if (skip_count > 0) {
851 skip_count--;
852 ADVANCE();
853 } else
854 state++;
855 break;
856
857 case GZSTATE_ORIGNAME:
858 if ((flags & ORIG_NAME) == 0) {
859 state++;
860 break;
861 }
862 if (*z.next_in == 0)
863 state++;
864 ADVANCE();
865 break;
866
867 case GZSTATE_COMMENT:
868 if ((flags & COMMENT) == 0) {
869 state++;
870 break;
871 }
872 if (*z.next_in == 0)
873 state++;
874 ADVANCE();
875 break;
876
877 case GZSTATE_HEAD_CRC1:
878 if (flags & HEAD_CRC)
879 skip_count = 2;
880 else
881 skip_count = 0;
882 state++;
883 break;
884
885 case GZSTATE_HEAD_CRC2:
886 if (skip_count > 0) {
887 skip_count--;
888 ADVANCE();
889 } else
890 state++;
891 break;
892
893 case GZSTATE_INIT:
894 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) {
895 maybe_warnx("failed to inflateInit");
896 goto stop_and_fail;
897 }
898 state++;
899 break;
900
901 case GZSTATE_READ:
902 error = inflate(&z, Z_FINISH);
903 switch (error) {
904 /* Z_BUF_ERROR goes with Z_FINISH... */
905 case Z_BUF_ERROR:
906 if (z.avail_out > 0 && !done_reading)
907 continue;
908
909 case Z_STREAM_END:
910 case Z_OK:
911 break;
912
913 case Z_NEED_DICT:
914 maybe_warnx("Z_NEED_DICT error");
915 goto stop_and_fail;
916 case Z_DATA_ERROR:
917 maybe_warnx("data stream error");
918 goto stop_and_fail;
919 case Z_STREAM_ERROR:
920 maybe_warnx("internal stream error");
921 goto stop_and_fail;
922 case Z_MEM_ERROR:
923 maybe_warnx("memory allocation error");
924 goto stop_and_fail;
925
926 default:
927 maybe_warn("unknown error from inflate(): %d",
928 error);
929 }
930 wr = BUFLEN - z.avail_out;
931
932 if (wr != 0) {
933 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr);
934 if (
935 #ifndef SMALL
936 /* don't write anything with -t */
937 tflag == 0 &&
938 #endif
939 write(out, outbufp, wr) != wr) {
940 maybe_warn("error writing to output");
941 goto stop_and_fail;
942 }
943
944 out_tot += wr;
945 out_sub_tot += wr;
946 }
947
948 if (error == Z_STREAM_END) {
949 inflateEnd(&z);
950 state++;
951 }
952
953 z.next_out = (unsigned char *)outbufp;
954 z.avail_out = BUFLEN;
955
956 break;
957 case GZSTATE_CRC:
958 {
959 uLong origcrc;
960
961 if (z.avail_in < 4) {
962 if (!done_reading) {
963 needmore = 1;
964 continue;
965 }
966 maybe_warnx("truncated input");
967 goto stop_and_fail;
968 }
969 origcrc = ((unsigned)z.next_in[0] & 0xff) |
970 ((unsigned)z.next_in[1] & 0xff) << 8 |
971 ((unsigned)z.next_in[2] & 0xff) << 16 |
972 ((unsigned)z.next_in[3] & 0xff) << 24;
973 if (origcrc != crc) {
974 maybe_warnx("invalid compressed"
975 " data--crc error");
976 goto stop_and_fail;
977 }
978 }
979
980 z.avail_in -= 4;
981 z.next_in += 4;
982
983 if (!z.avail_in && done_reading) {
984 goto stop;
985 }
986 state++;
987 break;
988 case GZSTATE_LEN:
989 {
990 uLong origlen;
991
992 if (z.avail_in < 4) {
993 if (!done_reading) {
994 needmore = 1;
995 continue;
996 }
997 maybe_warnx("truncated input");
998 goto stop_and_fail;
999 }
1000 origlen = ((unsigned)z.next_in[0] & 0xff) |
1001 ((unsigned)z.next_in[1] & 0xff) << 8 |
1002 ((unsigned)z.next_in[2] & 0xff) << 16 |
1003 ((unsigned)z.next_in[3] & 0xff) << 24;
1004
1005 if (origlen != out_sub_tot) {
1006 maybe_warnx("invalid compressed"
1007 " data--length error");
1008 goto stop_and_fail;
1009 }
1010 }
1011
1012 z.avail_in -= 4;
1013 z.next_in += 4;
1014
1015 if (error < 0) {
1016 maybe_warnx("decompression error");
1017 goto stop_and_fail;
1018 }
1019 state = GZSTATE_MAGIC0;
1020 break;
1021 }
1022 continue;
1023 stop_and_fail:
1024 out_tot = -1;
1025 stop:
1026 break;
1027 }
1028 if (state > GZSTATE_INIT)
1029 inflateEnd(&z);
1030
1031 free(inbufp);
1032 out1:
1033 free(outbufp);
1034 out2:
1035 if (gsizep)
1036 *gsizep = in_tot;
1037 return (out_tot);
1038 }
1039
1040 #ifndef SMALL
1041 /*
1042 * set the owner, mode, flags & utimes using the given file descriptor.
1043 * file is only used in possible warning messages.
1044 */
1045 static void
1046 copymodes(int fd, const struct stat *sbp, const char *file)
1047 {
1048 struct timeval times[2];
1049 struct stat sb;
1050
1051 /*
1052 * If we have no info on the input, give this file some
1053 * default values and return..
1054 */
1055 if (sbp == NULL) {
1056 mode_t mask = umask(022);
1057
1058 (void)fchmod(fd, DEFFILEMODE & ~mask);
1059 (void)umask(mask);
1060 return;
1061 }
1062 sb = *sbp;
1063
1064 /* if the chown fails, remove set-id bits as-per compress(1) */
1065 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) {
1066 if (errno != EPERM)
1067 maybe_warn("couldn't fchown: %s", file);
1068 sb.st_mode &= ~(S_ISUID|S_ISGID);
1069 }
1070
1071 /* we only allow set-id and the 9 normal permission bits */
1072 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO;
1073 if (fchmod(fd, sb.st_mode) < 0)
1074 maybe_warn("couldn't fchmod: %s", file);
1075
1076 /* only try flags if they exist already */
1077 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0)
1078 maybe_warn("couldn't fchflags: %s", file);
1079
1080 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec);
1081 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec);
1082 if (futimes(fd, times) < 0)
1083 maybe_warn("couldn't utimes: %s", file);
1084 }
1085 #endif
1086
1087 /* what sort of file is this? */
1088 static enum filetype
1089 file_gettype(u_char *buf)
1090 {
1091
1092 if (buf[0] == GZIP_MAGIC0 &&
1093 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1))
1094 return FT_GZIP;
1095 else
1096 #ifndef NO_BZIP2_SUPPORT
1097 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 &&
1098 buf[3] >= '0' && buf[3] <= '9')
1099 return FT_BZIP2;
1100 else
1101 #endif
1102 #ifndef NO_COMPRESS_SUPPORT
1103 if (memcmp(buf, Z_MAGIC, 2) == 0)
1104 return FT_Z;
1105 else
1106 #endif
1107 #ifndef NO_PACK_SUPPORT
1108 if (memcmp(buf, PACK_MAGIC, 2) == 0)
1109 return FT_PACK;
1110 else
1111 #endif
1112 #ifndef NO_XZ_SUPPORT
1113 if (memcmp(buf, XZ_MAGIC, 4) == 0) /* XXX: We only have 4 bytes */
1114 return FT_XZ;
1115 else
1116 #endif
1117 return FT_UNKNOWN;
1118 }
1119
1120 #ifndef SMALL
1121 /* check the outfile is OK. */
1122 static int
1123 check_outfile(const char *outfile)
1124 {
1125 struct stat sb;
1126 int ok = 1;
1127
1128 if (lflag == 0 && stat(outfile, &sb) == 0) {
1129 if (fflag)
1130 unlink(outfile);
1131 else if (isatty(STDIN_FILENO)) {
1132 char ans[10] = { 'n', '\0' }; /* default */
1133
1134 fprintf(stderr, "%s already exists -- do you wish to "
1135 "overwrite (y or n)? " , outfile);
1136 (void)fgets(ans, sizeof(ans) - 1, stdin);
1137 if (ans[0] != 'y' && ans[0] != 'Y') {
1138 fprintf(stderr, "\tnot overwriting\n");
1139 ok = 0;
1140 } else
1141 unlink(outfile);
1142 } else {
1143 maybe_warnx("%s already exists -- skipping", outfile);
1144 ok = 0;
1145 }
1146 }
1147 return ok;
1148 }
1149
1150 static void
1151 unlink_input(const char *file, const struct stat *sb)
1152 {
1153 struct stat nsb;
1154
1155 if (kflag)
1156 return;
1157 if (stat(file, &nsb) != 0)
1158 /* Must be gone already */
1159 return;
1160 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino)
1161 /* Definitely a different file */
1162 return;
1163 unlink(file);
1164 }
1165 #endif
1166
1167 static const suffixes_t *
1168 check_suffix(char *file, int xlate)
1169 {
1170 const suffixes_t *s;
1171 int len = strlen(file);
1172 char *sp;
1173
1174 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) {
1175 /* if it doesn't fit in "a.suf", don't bother */
1176 if (s->ziplen >= len)
1177 continue;
1178 sp = file + len - s->ziplen;
1179 if (strcmp(s->zipped, sp) != 0)
1180 continue;
1181 if (xlate)
1182 strcpy(sp, s->normal);
1183 return s;
1184 }
1185 return NULL;
1186 }
1187
1188 /*
1189 * compress the given file: create a corresponding .gz file and remove the
1190 * original.
1191 */
1192 static off_t
1193 file_compress(char *file, char *outfile, size_t outsize)
1194 {
1195 int in;
1196 int out;
1197 off_t size, insize;
1198 #ifndef SMALL
1199 struct stat isb, osb;
1200 const suffixes_t *suff;
1201 #endif
1202
1203 in = open(file, O_RDONLY);
1204 if (in == -1) {
1205 maybe_warn("can't open %s", file);
1206 return -1;
1207 }
1208
1209 if (cflag == 0) {
1210 #ifndef SMALL
1211 if (fstat(in, &isb) == 0) {
1212 if (isb.st_nlink > 1 && fflag == 0) {
1213 maybe_warnx("%s has %d other link%s -- "
1214 "skipping", file, isb.st_nlink - 1,
1215 isb.st_nlink == 1 ? "" : "s");
1216 close(in);
1217 return -1;
1218 }
1219 }
1220
1221 if (fflag == 0 && (suff = check_suffix(file, 0))
1222 && suff->zipped[0] != 0) {
1223 maybe_warnx("%s already has %s suffix -- unchanged",
1224 file, suff->zipped);
1225 close(in);
1226 return -1;
1227 }
1228 #endif
1229
1230 /* Add (usually) .gz to filename */
1231 if ((size_t)snprintf(outfile, outsize, "%s%s",
1232 file, suffixes[0].zipped) >= outsize)
1233 memcpy(outfile + outsize - suffixes[0].ziplen - 1,
1234 suffixes[0].zipped, suffixes[0].ziplen + 1);
1235
1236 #ifndef SMALL
1237 if (check_outfile(outfile) == 0) {
1238 close(in);
1239 return -1;
1240 }
1241 #endif
1242 }
1243
1244 if (cflag == 0) {
1245 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600);
1246 if (out == -1) {
1247 maybe_warn("could not create output: %s", outfile);
1248 fclose(stdin);
1249 return -1;
1250 }
1251 } else
1252 out = STDOUT_FILENO;
1253
1254 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime);
1255
1256 (void)close(in);
1257
1258 /*
1259 * If there was an error, insize will be -1.
1260 * If we compressed to stdout, just return the size.
1261 * Otherwise stat the file and check it is the correct size.
1262 * We only blow away the file if we can stat the output and it
1263 * has the expected size.
1264 */
1265 if (cflag != 0)
1266 return insize == -1 ? -1 : size;
1267
1268 #ifndef SMALL
1269 if (fstat(out, &osb) != 0) {
1270 maybe_warn("couldn't stat: %s", outfile);
1271 goto bad_outfile;
1272 }
1273
1274 if (osb.st_size != size) {
1275 maybe_warnx("output file: %s wrong size (%" PRIdOFF
1276 " != %" PRIdOFF "), deleting",
1277 outfile, osb.st_size, size);
1278 goto bad_outfile;
1279 }
1280
1281 copymodes(out, &isb, outfile);
1282 #endif
1283 if (close(out) == -1)
1284 maybe_warn("couldn't close output");
1285
1286 /* output is good, ok to delete input */
1287 unlink_input(file, &isb);
1288 return size;
1289
1290 #ifndef SMALL
1291 bad_outfile:
1292 if (close(out) == -1)
1293 maybe_warn("couldn't close output");
1294
1295 maybe_warnx("leaving original %s", file);
1296 unlink(outfile);
1297 return size;
1298 #endif
1299 }
1300
1301 /* uncompress the given file and remove the original */
1302 static off_t
1303 file_uncompress(char *file, char *outfile, size_t outsize)
1304 {
1305 struct stat isb, osb;
1306 off_t size;
1307 ssize_t rbytes;
1308 unsigned char header1[4];
1309 enum filetype method;
1310 int fd, ofd, zfd = -1;
1311 #ifndef SMALL
1312 ssize_t rv;
1313 time_t timestamp = 0;
1314 char name[PATH_MAX + 1];
1315 #endif
1316
1317 /* gather the old name info */
1318
1319 fd = open(file, O_RDONLY);
1320 if (fd < 0) {
1321 maybe_warn("can't open %s", file);
1322 goto lose;
1323 }
1324
1325 strlcpy(outfile, file, outsize);
1326 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) {
1327 maybe_warnx("%s: unknown suffix -- ignored", file);
1328 goto lose;
1329 }
1330
1331 rbytes = read(fd, header1, sizeof header1);
1332 if (rbytes != sizeof header1) {
1333 /* we don't want to fail here. */
1334 #ifndef SMALL
1335 if (fflag)
1336 goto lose;
1337 #endif
1338 if (rbytes == -1)
1339 maybe_warn("can't read %s", file);
1340 else
1341 goto unexpected_EOF;
1342 goto lose;
1343 }
1344
1345 method = file_gettype(header1);
1346 #ifndef SMALL
1347 if (fflag == 0 && method == FT_UNKNOWN) {
1348 maybe_warnx("%s: not in gzip format", file);
1349 goto lose;
1350 }
1351
1352 #endif
1353
1354 #ifndef SMALL
1355 if (method == FT_GZIP && Nflag) {
1356 unsigned char ts[4]; /* timestamp */
1357
1358 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP);
1359 if (rv >= 0 && rv < (ssize_t)(sizeof ts))
1360 goto unexpected_EOF;
1361 if (rv == -1) {
1362 if (!fflag)
1363 maybe_warn("can't read %s", file);
1364 goto lose;
1365 }
1366 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0];
1367
1368 if (header1[3] & ORIG_NAME) {
1369 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME);
1370 if (rbytes < 0) {
1371 maybe_warn("can't read %s", file);
1372 goto lose;
1373 }
1374 if (name[0] != 0) {
1375 char *dp, *nf;
1376
1377 /* strip saved directory name */
1378 nf = strrchr(name, '/');
1379 if (nf == NULL)
1380 nf = name;
1381 else
1382 nf++;
1383
1384 /* preserve original directory name */
1385 dp = strrchr(file, '/');
1386 if (dp == NULL)
1387 dp = file;
1388 else
1389 dp++;
1390 snprintf(outfile, outsize, "%.*s%.*s",
1391 (int) (dp - file),
1392 file, (int) rbytes, nf);
1393 }
1394 }
1395 }
1396 #endif
1397 lseek(fd, 0, SEEK_SET);
1398
1399 if (cflag == 0 || lflag) {
1400 if (fstat(fd, &isb) != 0)
1401 goto lose;
1402 #ifndef SMALL
1403 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) {
1404 maybe_warnx("%s has %d other links -- skipping",
1405 file, isb.st_nlink - 1);
1406 goto lose;
1407 }
1408 if (nflag == 0 && timestamp)
1409 isb.st_mtime = timestamp;
1410 if (check_outfile(outfile) == 0)
1411 goto lose;
1412 #endif
1413 }
1414
1415 if (cflag == 0 && lflag == 0) {
1416 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600);
1417 if (zfd == STDOUT_FILENO) {
1418 /* We won't close STDOUT_FILENO later... */
1419 zfd = dup(zfd);
1420 close(STDOUT_FILENO);
1421 }
1422 if (zfd == -1) {
1423 maybe_warn("can't open %s", outfile);
1424 goto lose;
1425 }
1426 } else
1427 zfd = STDOUT_FILENO;
1428
1429 switch (method) {
1430 #ifndef NO_BZIP2_SUPPORT
1431 case FT_BZIP2:
1432 /* XXX */
1433 if (lflag) {
1434 maybe_warnx("no -l with bzip2 files");
1435 goto lose;
1436 }
1437
1438 size = unbzip2(fd, zfd, NULL, 0, NULL);
1439 break;
1440 #endif
1441
1442 #ifndef NO_COMPRESS_SUPPORT
1443 case FT_Z: {
1444 FILE *in, *out;
1445
1446 /* XXX */
1447 if (lflag) {
1448 maybe_warnx("no -l with Lempel-Ziv files");
1449 goto lose;
1450 }
1451
1452 if ((in = zdopen(fd)) == NULL) {
1453 maybe_warn("zdopen for read: %s", file);
1454 goto lose;
1455 }
1456
1457 out = fdopen(dup(zfd), "w");
1458 if (out == NULL) {
1459 maybe_warn("fdopen for write: %s", outfile);
1460 fclose(in);
1461 goto lose;
1462 }
1463
1464 size = zuncompress(in, out, NULL, 0, NULL);
1465 /* need to fclose() if ferror() is true... */
1466 if (ferror(in) | fclose(in)) {
1467 maybe_warn("failed infile fclose");
1468 unlink(outfile);
1469 (void)fclose(out);
1470 }
1471 if (fclose(out) != 0) {
1472 maybe_warn("failed outfile fclose");
1473 unlink(outfile);
1474 goto lose;
1475 }
1476 break;
1477 }
1478 #endif
1479
1480 #ifndef NO_PACK_SUPPORT
1481 case FT_PACK:
1482 if (lflag) {
1483 maybe_warnx("no -l with packed files");
1484 goto lose;
1485 }
1486
1487 size = unpack(fd, zfd, NULL, 0, NULL);
1488 break;
1489 #endif
1490
1491 #ifndef NO_XZ_SUPPORT
1492 case FT_XZ:
1493 if (lflag) {
1494 maybe_warnx("no -l with xz files");
1495 goto lose;
1496 }
1497
1498 size = unxz(fd, zfd, NULL, 0, NULL);
1499 break;
1500 #endif
1501
1502 #ifndef SMALL
1503 case FT_UNKNOWN:
1504 if (lflag) {
1505 maybe_warnx("no -l for unknown filetypes");
1506 goto lose;
1507 }
1508 size = cat_fd(NULL, 0, NULL, fd);
1509 break;
1510 #endif
1511 default:
1512 if (lflag) {
1513 print_list(fd, isb.st_size, outfile, isb.st_mtime);
1514 close(fd);
1515 return -1; /* XXX */
1516 }
1517
1518 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file);
1519 break;
1520 }
1521
1522 if (close(fd) != 0)
1523 maybe_warn("couldn't close input");
1524 if (zfd != STDOUT_FILENO && close(zfd) != 0)
1525 maybe_warn("couldn't close output");
1526
1527 if (size == -1) {
1528 if (cflag == 0)
1529 unlink(outfile);
1530 maybe_warnx("%s: uncompress failed", file);
1531 return -1;
1532 }
1533
1534 /* if testing, or we uncompressed to stdout, this is all we need */
1535 #ifndef SMALL
1536 if (tflag)
1537 return size;
1538 #endif
1539 /* if we are uncompressing to stdin, don't remove the file. */
1540 if (cflag)
1541 return size;
1542
1543 /*
1544 * if we create a file...
1545 */
1546 /*
1547 * if we can't stat the file don't remove the file.
1548 */
1549
1550 ofd = open(outfile, O_RDWR, 0);
1551 if (ofd == -1) {
1552 maybe_warn("couldn't open (leaving original): %s",
1553 outfile);
1554 return -1;
1555 }
1556 if (fstat(ofd, &osb) != 0) {
1557 maybe_warn("couldn't stat (leaving original): %s",
1558 outfile);
1559 close(ofd);
1560 return -1;
1561 }
1562 if (osb.st_size != size) {
1563 maybe_warnx("stat gave different size: %" PRIdOFF
1564 " != %" PRIdOFF " (leaving original)",
1565 size, osb.st_size);
1566 close(ofd);
1567 unlink(outfile);
1568 return -1;
1569 }
1570 unlink_input(file, &isb);
1571 #ifndef SMALL
1572 copymodes(ofd, &isb, outfile);
1573 #endif
1574 close(ofd);
1575 return size;
1576
1577 unexpected_EOF:
1578 maybe_warnx("%s: unexpected end of file", file);
1579 lose:
1580 if (fd != -1)
1581 close(fd);
1582 if (zfd != -1 && zfd != STDOUT_FILENO)
1583 close(fd);
1584 return -1;
1585 }
1586
1587 #ifndef SMALL
1588 static off_t
1589 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd)
1590 {
1591 char buf[BUFLEN];
1592 off_t in_tot;
1593 ssize_t w;
1594
1595 in_tot = count;
1596 w = write(STDOUT_FILENO, prepend, count);
1597 if (w == -1 || (size_t)w != count) {
1598 maybe_warn("write to stdout");
1599 return -1;
1600 }
1601 for (;;) {
1602 ssize_t rv;
1603
1604 rv = read(fd, buf, sizeof buf);
1605 if (rv == 0)
1606 break;
1607 if (rv < 0) {
1608 maybe_warn("read from fd %d", fd);
1609 break;
1610 }
1611
1612 if (write(STDOUT_FILENO, buf, rv) != rv) {
1613 maybe_warn("write to stdout");
1614 break;
1615 }
1616 in_tot += rv;
1617 }
1618
1619 if (gsizep)
1620 *gsizep = in_tot;
1621 return (in_tot);
1622 }
1623 #endif
1624
1625 static void
1626 handle_stdin(void)
1627 {
1628 unsigned char header1[4];
1629 off_t usize, gsize;
1630 enum filetype method;
1631 ssize_t bytes_read;
1632 #ifndef NO_COMPRESS_SUPPORT
1633 FILE *in;
1634 #endif
1635
1636 #ifndef SMALL
1637 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) {
1638 maybe_warnx("standard input is a terminal -- ignoring");
1639 return;
1640 }
1641 #endif
1642
1643 if (lflag) {
1644 struct stat isb;
1645
1646 /* XXX could read the whole file, etc. */
1647 if (fstat(STDIN_FILENO, &isb) < 0) {
1648 maybe_warn("fstat");
1649 return;
1650 }
1651 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime);
1652 return;
1653 }
1654
1655 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1);
1656 if (bytes_read == -1) {
1657 maybe_warn("can't read stdin");
1658 return;
1659 } else if (bytes_read != sizeof(header1)) {
1660 maybe_warnx("(stdin): unexpected end of file");
1661 return;
1662 }
1663
1664 method = file_gettype(header1);
1665 switch (method) {
1666 default:
1667 #ifndef SMALL
1668 if (fflag == 0) {
1669 maybe_warnx("unknown compression format");
1670 return;
1671 }
1672 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO);
1673 break;
1674 #endif
1675 case FT_GZIP:
1676 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO,
1677 (char *)header1, sizeof header1, &gsize, "(stdin)");
1678 break;
1679 #ifndef NO_BZIP2_SUPPORT
1680 case FT_BZIP2:
1681 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO,
1682 (char *)header1, sizeof header1, &gsize);
1683 break;
1684 #endif
1685 #ifndef NO_COMPRESS_SUPPORT
1686 case FT_Z:
1687 if ((in = zdopen(STDIN_FILENO)) == NULL) {
1688 maybe_warnx("zopen of stdin");
1689 return;
1690 }
1691
1692 usize = zuncompress(in, stdout, (char *)header1,
1693 sizeof header1, &gsize);
1694 fclose(in);
1695 break;
1696 #endif
1697 #ifndef NO_PACK_SUPPORT
1698 case FT_PACK:
1699 usize = unpack(STDIN_FILENO, STDOUT_FILENO,
1700 (char *)header1, sizeof header1, &gsize);
1701 break;
1702 #endif
1703 #ifndef NO_XZ_SUPPORT
1704 case FT_XZ:
1705 usize = unxz(STDIN_FILENO, STDOUT_FILENO,
1706 (char *)header1, sizeof header1, &gsize);
1707 break;
1708 #endif
1709 }
1710
1711 #ifndef SMALL
1712 if (vflag && !tflag && usize != -1 && gsize != -1)
1713 print_verbage(NULL, NULL, usize, gsize);
1714 if (vflag && tflag)
1715 print_test("(stdin)", usize != -1);
1716 #endif
1717
1718 }
1719
1720 static void
1721 handle_stdout(void)
1722 {
1723 off_t gsize, usize;
1724 struct stat sb;
1725 time_t systime;
1726 uint32_t mtime;
1727 int ret;
1728
1729 #ifndef SMALL
1730 if (fflag == 0 && isatty(STDOUT_FILENO)) {
1731 maybe_warnx("standard output is a terminal -- ignoring");
1732 return;
1733 }
1734 #endif
1735 /* If stdin is a file use its mtime, otherwise use current time */
1736 ret = fstat(STDIN_FILENO, &sb);
1737
1738 #ifndef SMALL
1739 if (ret < 0) {
1740 maybe_warn("Can't stat stdin");
1741 return;
1742 }
1743 #endif
1744
1745 if (S_ISREG(sb.st_mode))
1746 mtime = (uint32_t)sb.st_mtime;
1747 else {
1748 systime = time(NULL);
1749 #ifndef SMALL
1750 if (systime == -1) {
1751 maybe_warn("time");
1752 return;
1753 }
1754 #endif
1755 mtime = (uint32_t)systime;
1756 }
1757
1758 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime);
1759 #ifndef SMALL
1760 if (vflag && !tflag && usize != -1 && gsize != -1)
1761 print_verbage(NULL, NULL, usize, gsize);
1762 #endif
1763 }
1764
1765 /* do what is asked for, for the path name */
1766 static void
1767 handle_pathname(char *path)
1768 {
1769 char *opath = path, *s = NULL;
1770 ssize_t len;
1771 int slen;
1772 struct stat sb;
1773
1774 /* check for stdout/stdin */
1775 if (path[0] == '-' && path[1] == '\0') {
1776 if (dflag)
1777 handle_stdin();
1778 else
1779 handle_stdout();
1780 return;
1781 }
1782
1783 retry:
1784 if (stat(path, &sb) != 0) {
1785 /* lets try <path>.gz if we're decompressing */
1786 if (dflag && s == NULL && errno == ENOENT) {
1787 len = strlen(path);
1788 slen = suffixes[0].ziplen;
1789 s = malloc(len + slen + 1);
1790 if (s == NULL)
1791 maybe_err("malloc");
1792 memcpy(s, path, len);
1793 memcpy(s + len, suffixes[0].zipped, slen + 1);
1794 path = s;
1795 goto retry;
1796 }
1797 maybe_warn("can't stat: %s", opath);
1798 goto out;
1799 }
1800
1801 if (S_ISDIR(sb.st_mode)) {
1802 #ifndef SMALL
1803 if (rflag)
1804 handle_dir(path);
1805 else
1806 #endif
1807 maybe_warnx("%s is a directory", path);
1808 goto out;
1809 }
1810
1811 if (S_ISREG(sb.st_mode))
1812 handle_file(path, &sb);
1813 else
1814 maybe_warnx("%s is not a regular file", path);
1815
1816 out:
1817 if (s)
1818 free(s);
1819 }
1820
1821 /* compress/decompress a file */
1822 static void
1823 handle_file(char *file, struct stat *sbp)
1824 {
1825 off_t usize, gsize;
1826 char outfile[PATH_MAX];
1827
1828 infile = file;
1829 if (dflag) {
1830 usize = file_uncompress(file, outfile, sizeof(outfile));
1831 #ifndef SMALL
1832 if (vflag && tflag)
1833 print_test(file, usize != -1);
1834 #endif
1835 if (usize == -1)
1836 return;
1837 gsize = sbp->st_size;
1838 } else {
1839 gsize = file_compress(file, outfile, sizeof(outfile));
1840 if (gsize == -1)
1841 return;
1842 usize = sbp->st_size;
1843 }
1844
1845
1846 #ifndef SMALL
1847 if (vflag && !tflag)
1848 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize);
1849 #endif
1850 }
1851
1852 #ifndef SMALL
1853 /* this is used with -r to recursively descend directories */
1854 static void
1855 handle_dir(char *dir)
1856 {
1857 char *path_argv[2];
1858 FTS *fts;
1859 FTSENT *entry;
1860
1861 path_argv[0] = dir;
1862 path_argv[1] = 0;
1863 fts = fts_open(path_argv, FTS_PHYSICAL, NULL);
1864 if (fts == NULL) {
1865 warn("couldn't fts_open %s", dir);
1866 return;
1867 }
1868
1869 while ((entry = fts_read(fts))) {
1870 switch(entry->fts_info) {
1871 case FTS_D:
1872 case FTS_DP:
1873 continue;
1874
1875 case FTS_DNR:
1876 case FTS_ERR:
1877 case FTS_NS:
1878 maybe_warn("%s", entry->fts_path);
1879 continue;
1880 case FTS_F:
1881 handle_file(entry->fts_name, entry->fts_statp);
1882 }
1883 }
1884 (void)fts_close(fts);
1885 }
1886 #endif
1887
1888 /* print a ratio - size reduction as a fraction of uncompressed size */
1889 static void
1890 print_ratio(off_t in, off_t out, FILE *where)
1891 {
1892 int percent10; /* 10 * percent */
1893 off_t diff;
1894 char buff[8];
1895 int len;
1896
1897 diff = in - out/2;
1898 if (diff <= 0)
1899 /*
1900 * Output is more than double size of input! print -99.9%
1901 * Quite possibly we've failed to get the original size.
1902 */
1903 percent10 = -999;
1904 else {
1905 /*
1906 * We only need 12 bits of result from the final division,
1907 * so reduce the values until a 32bit division will suffice.
1908 */
1909 while (in > 0x100000) {
1910 diff >>= 1;
1911 in >>= 1;
1912 }
1913 if (in != 0)
1914 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000;
1915 else
1916 percent10 = 0;
1917 }
1918
1919 len = snprintf(buff, sizeof buff, "%2.2d.", percent10);
1920 /* Move the '.' to before the last digit */
1921 buff[len - 1] = buff[len - 2];
1922 buff[len - 2] = '.';
1923 fprintf(where, "%5s%%", buff);
1924 }
1925
1926 #ifndef SMALL
1927 /* print compression statistics, and the new name (if there is one!) */
1928 static void
1929 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize)
1930 {
1931 if (file)
1932 fprintf(stderr, "%s:%s ", file,
1933 strlen(file) < 7 ? "\t\t" : "\t");
1934 print_ratio(usize, gsize, stderr);
1935 if (nfile)
1936 fprintf(stderr, " -- replaced with %s", nfile);
1937 fprintf(stderr, "\n");
1938 fflush(stderr);
1939 }
1940
1941 /* print test results */
1942 static void
1943 print_test(const char *file, int ok)
1944 {
1945
1946 if (exit_value == 0 && ok == 0)
1947 exit_value = 1;
1948 fprintf(stderr, "%s:%s %s\n", file,
1949 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK");
1950 fflush(stderr);
1951 }
1952 #endif
1953
1954 /* print a file's info ala --list */
1955 /* eg:
1956 compressed uncompressed ratio uncompressed_name
1957 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar
1958 */
1959 static void
1960 print_list(int fd, off_t out, const char *outfile, time_t ts)
1961 {
1962 static int first = 1;
1963 #ifndef SMALL
1964 static off_t in_tot, out_tot;
1965 uint32_t crc = 0;
1966 #endif
1967 off_t in = 0, rv;
1968
1969 if (first) {
1970 #ifndef SMALL
1971 if (vflag)
1972 printf("method crc date time ");
1973 #endif
1974 if (qflag == 0)
1975 printf(" compressed uncompressed "
1976 "ratio uncompressed_name\n");
1977 }
1978 first = 0;
1979
1980 /* print totals? */
1981 #ifndef SMALL
1982 if (fd == -1) {
1983 in = in_tot;
1984 out = out_tot;
1985 } else
1986 #endif
1987 {
1988 /* read the last 4 bytes - this is the uncompressed size */
1989 rv = lseek(fd, (off_t)(-8), SEEK_END);
1990 if (rv != -1) {
1991 unsigned char buf[8];
1992 uint32_t usize;
1993
1994 rv = read(fd, (char *)buf, sizeof(buf));
1995 if (rv == -1)
1996 maybe_warn("read of uncompressed size");
1997 else if (rv != sizeof(buf))
1998 maybe_warnx("read of uncompressed size");
1999
2000 else {
2001 usize = buf[4] | buf[5] << 8 |
2002 buf[6] << 16 | buf[7] << 24;
2003 in = (off_t)usize;
2004 #ifndef SMALL
2005 crc = buf[0] | buf[1] << 8 |
2006 buf[2] << 16 | buf[3] << 24;
2007 #endif
2008 }
2009 }
2010 }
2011
2012 #ifndef SMALL
2013 if (vflag && fd == -1)
2014 printf(" ");
2015 else if (vflag) {
2016 char *date = ctime(&ts);
2017
2018 /* skip the day, 1/100th second, and year */
2019 date += 4;
2020 date[12] = 0;
2021 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date);
2022 }
2023 in_tot += in;
2024 out_tot += out;
2025 #endif
2026 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in);
2027 print_ratio(in, out, stdout);
2028 printf(" %s\n", outfile);
2029 }
2030
2031 /* display the usage of NetBSD gzip */
2032 static void
2033 usage(void)
2034 {
2035
2036 fprintf(stderr, "%s\n", gzip_version);
2037 fprintf(stderr,
2038 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n"
2039 #ifndef SMALL
2040 " -1 --fast fastest (worst) compression\n"
2041 " -2 .. -8 set compression level\n"
2042 " -9 --best best (slowest) compression\n"
2043 " -c --stdout write to stdout, keep original files\n"
2044 " --to-stdout\n"
2045 " -d --decompress uncompress files\n"
2046 " --uncompress\n"
2047 " -f --force force overwriting & compress links\n"
2048 " -h --help display this help\n"
2049 " -k --keep don't delete input files during operation\n"
2050 " -l --list list compressed file contents\n"
2051 " -N --name save or restore original file name and time stamp\n"
2052 " -n --no-name don't save original file name or time stamp\n"
2053 " -q --quiet output no warnings\n"
2054 " -r --recursive recursively compress files in directories\n"
2055 " -S .suf use suffix .suf instead of .gz\n"
2056 " --suffix .suf\n"
2057 " -t --test test compressed file\n"
2058 " -V --version display program version\n"
2059 " -v --verbose print extra statistics\n",
2060 #else
2061 ,
2062 #endif
2063 getprogname());
2064 exit(0);
2065 }
2066
2067 /* display the version of NetBSD gzip */
2068 static void
2069 display_version(void)
2070 {
2071
2072 fprintf(stderr, "%s\n", gzip_version);
2073 exit(0);
2074 }
2075
2076 #ifndef NO_BZIP2_SUPPORT
2077 #include "unbzip2.c"
2078 #endif
2079 #ifndef NO_COMPRESS_SUPPORT
2080 #include "zuncompress.c"
2081 #endif
2082 #ifndef NO_PACK_SUPPORT
2083 #include "unpack.c"
2084 #endif
2085 #ifndef NO_XZ_SUPPORT
2086 #include "unxz.c"
2087 #endif
2088
2089 static ssize_t
2090 read_retry(int fd, void *buf, size_t sz)
2091 {
2092 char *cp = buf;
2093 size_t left = MIN(sz, (size_t) SSIZE_MAX);
2094
2095 while (left > 0) {
2096 ssize_t ret;
2097
2098 ret = read(fd, cp, left);
2099 if (ret == -1) {
2100 return ret;
2101 } else if (ret == 0) {
2102 break; /* EOF */
2103 }
2104 cp += ret;
2105 left -= ret;
2106 }
2107
2108 return sz - left;
2109 }
2110