gzip.c revision 1.114 1 /* $NetBSD: gzip.c,v 1.114 2018/10/06 16:36:45 martin Exp $ */
2
3 /*
4 * Copyright (c) 1997, 1998, 2003, 2004, 2006, 2008, 2009, 2010, 2011, 2015, 2017
5 * Matthew R. Green
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
18 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
19 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
20 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
21 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
22 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
23 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
24 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
25 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 */
29
30 #include <sys/cdefs.h>
31 #ifndef lint
32 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004, 2006, 2008,\
33 2009, 2010, 2011, 2015, 2017 Matthew R. Green. All rights reserved.");
34 __RCSID("$NetBSD: gzip.c,v 1.114 2018/10/06 16:36:45 martin Exp $");
35 #endif /* not lint */
36
37 /*
38 * gzip.c -- GPL free gzip using zlib.
39 *
40 * RFC 1950 covers the zlib format
41 * RFC 1951 covers the deflate format
42 * RFC 1952 covers the gzip format
43 *
44 * TODO:
45 * - use mmap where possible
46 * - handle some signals better (remove outfile?)
47 * - make bzip2/compress -v/-t/-l support work as well as possible
48 */
49
50 #include <sys/param.h>
51 #include <sys/stat.h>
52 #include <sys/time.h>
53
54 #include <inttypes.h>
55 #include <unistd.h>
56 #include <stdio.h>
57 #include <string.h>
58 #include <stdlib.h>
59 #include <err.h>
60 #include <errno.h>
61 #include <fcntl.h>
62 #include <zlib.h>
63 #include <fts.h>
64 #include <libgen.h>
65 #include <stdarg.h>
66 #include <getopt.h>
67 #include <time.h>
68
69 #ifndef PRIdOFF
70 #define PRIdOFF PRId64
71 #endif
72
73 /* what type of file are we dealing with */
74 enum filetype {
75 FT_GZIP,
76 #ifndef NO_BZIP2_SUPPORT
77 FT_BZIP2,
78 #endif
79 #ifndef NO_COMPRESS_SUPPORT
80 FT_Z,
81 #endif
82 #ifndef NO_PACK_SUPPORT
83 FT_PACK,
84 #endif
85 #ifndef NO_XZ_SUPPORT
86 FT_XZ,
87 #endif
88 FT_LAST,
89 FT_UNKNOWN
90 };
91
92 #ifndef NO_BZIP2_SUPPORT
93 #include <bzlib.h>
94
95 #define BZ2_SUFFIX ".bz2"
96 #define BZIP2_MAGIC "\102\132\150"
97 #endif
98
99 #ifndef NO_COMPRESS_SUPPORT
100 #define Z_SUFFIX ".Z"
101 #define Z_MAGIC "\037\235"
102 #endif
103
104 #ifndef NO_PACK_SUPPORT
105 #define PACK_MAGIC "\037\036"
106 #endif
107
108 #ifndef NO_XZ_SUPPORT
109 #include <lzma.h>
110 #define XZ_SUFFIX ".xz"
111 #define XZ_MAGIC "\3757zXZ"
112 #endif
113
114 #define GZ_SUFFIX ".gz"
115
116 #define BUFLEN (64 * 1024)
117
118 #define GZIP_MAGIC0 0x1F
119 #define GZIP_MAGIC1 0x8B
120 #define GZIP_OMAGIC1 0x9E
121
122 #define GZIP_TIMESTAMP (off_t)4
123 #define GZIP_ORIGNAME (off_t)10
124
125 #define HEAD_CRC 0x02
126 #define EXTRA_FIELD 0x04
127 #define ORIG_NAME 0x08
128 #define COMMENT 0x10
129
130 #define OS_CODE 3 /* Unix */
131
132 typedef struct {
133 const char *zipped;
134 int ziplen;
135 const char *normal; /* for unzip - must not be longer than zipped */
136 } suffixes_t;
137 static suffixes_t suffixes[] = {
138 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N}
139 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */
140 #ifndef SMALL
141 SUFFIX(GZ_SUFFIX, ""),
142 SUFFIX(".z", ""),
143 SUFFIX("-gz", ""),
144 SUFFIX("-z", ""),
145 SUFFIX("_z", ""),
146 SUFFIX(".taz", ".tar"),
147 SUFFIX(".tgz", ".tar"),
148 #ifndef NO_BZIP2_SUPPORT
149 SUFFIX(BZ2_SUFFIX, ""),
150 #endif
151 #ifndef NO_COMPRESS_SUPPORT
152 SUFFIX(Z_SUFFIX, ""),
153 #endif
154 #ifndef NO_XZ_SUPPORT
155 SUFFIX(XZ_SUFFIX, ""),
156 #endif
157 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */
158 #endif /* SMALL */
159 #undef SUFFIX
160 };
161 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0])
162 #define SUFFIX_MAXLEN 30
163
164 static const char gzip_version[] = "NetBSD gzip 20170803";
165
166 static int cflag; /* stdout mode */
167 static int dflag; /* decompress mode */
168 static int lflag; /* list mode */
169 static int numflag = 6; /* gzip -1..-9 value */
170
171 #ifndef SMALL
172 static int fflag; /* force mode */
173 static int kflag; /* don't delete input files */
174 static int nflag; /* don't save name/timestamp */
175 static int Nflag; /* don't restore name/timestamp */
176 static int qflag; /* quiet mode */
177 static int rflag; /* recursive mode */
178 static int tflag; /* test */
179 static int vflag; /* verbose mode */
180 static sig_atomic_t print_info = 0;
181 #else
182 #define qflag 0
183 #define tflag 0
184 #endif
185
186 static int exit_value = 0; /* exit value */
187
188 static const char *infile; /* name of file coming in */
189
190 static void maybe_err(const char *fmt, ...) __printflike(1, 2) __dead;
191 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \
192 !defined(NO_XZ_SUPPORT)
193 static void maybe_errx(const char *fmt, ...) __printflike(1, 2) __dead;
194 #endif
195 static void maybe_warn(const char *fmt, ...) __printflike(1, 2);
196 static void maybe_warnx(const char *fmt, ...) __printflike(1, 2);
197 static enum filetype file_gettype(u_char *);
198 #ifdef SMALL
199 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz)
200 #endif
201 static off_t gz_compress(int, int, off_t *, const char *, uint32_t);
202 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *);
203 static off_t file_compress(char *, char *, size_t);
204 static off_t file_uncompress(char *, char *, size_t);
205 static void handle_pathname(char *);
206 static void handle_file(char *, struct stat *);
207 static void handle_stdin(void);
208 static void handle_stdout(void);
209 static void print_ratio(off_t, off_t, FILE *);
210 static void print_list(int fd, off_t, const char *, time_t);
211 __dead static void usage(void);
212 __dead static void display_version(void);
213 static const suffixes_t *check_suffix(char *, int);
214 static ssize_t read_retry(int, void *, size_t);
215 static ssize_t write_retry(int, const void *, size_t);
216 static void print_list_out(off_t, off_t, const char*);
217
218 #ifdef SMALL
219 #define infile_set(f,t) infile_set(f)
220 #endif
221 static void infile_set(const char *newinfile, off_t total);
222
223 #ifdef SMALL
224 #define unlink_input(f, sb) unlink(f)
225 #define check_siginfo() /* nothing */
226 #define setup_signals() /* nothing */
227 #define infile_newdata(t) /* nothing */
228 #else
229 static off_t infile_total; /* total expected to read/write */
230 static off_t infile_current; /* current read/write */
231
232 static void check_siginfo(void);
233 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd);
234 static void prepend_gzip(char *, int *, char ***);
235 static void handle_dir(char *);
236 static void print_verbage(const char *, const char *, off_t, off_t);
237 static void print_test(const char *, int);
238 static void copymodes(int fd, const struct stat *, const char *file);
239 static int check_outfile(const char *outfile);
240 static void setup_signals(void);
241 static void infile_newdata(size_t newdata);
242 static void infile_clear(void);
243 #endif
244
245 #ifndef NO_BZIP2_SUPPORT
246 static off_t unbzip2(int, int, char *, size_t, off_t *);
247 #endif
248
249 #ifndef NO_COMPRESS_SUPPORT
250 static FILE *zdopen(int);
251 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *);
252 #endif
253
254 #ifndef NO_PACK_SUPPORT
255 static off_t unpack(int, int, char *, size_t, off_t *);
256 #endif
257
258 #ifndef NO_XZ_SUPPORT
259 static off_t unxz(int, int, char *, size_t, off_t *);
260 static off_t unxz_len(int);
261 #endif
262
263 #ifdef SMALL
264 #define getopt_long(a,b,c,d,e) getopt(a,b,c)
265 #else
266 static const struct option longopts[] = {
267 { "stdout", no_argument, 0, 'c' },
268 { "to-stdout", no_argument, 0, 'c' },
269 { "decompress", no_argument, 0, 'd' },
270 { "uncompress", no_argument, 0, 'd' },
271 { "force", no_argument, 0, 'f' },
272 { "help", no_argument, 0, 'h' },
273 { "keep", no_argument, 0, 'k' },
274 { "list", no_argument, 0, 'l' },
275 { "no-name", no_argument, 0, 'n' },
276 { "name", no_argument, 0, 'N' },
277 { "quiet", no_argument, 0, 'q' },
278 { "recursive", no_argument, 0, 'r' },
279 { "suffix", required_argument, 0, 'S' },
280 { "test", no_argument, 0, 't' },
281 { "verbose", no_argument, 0, 'v' },
282 { "version", no_argument, 0, 'V' },
283 { "fast", no_argument, 0, '1' },
284 { "best", no_argument, 0, '9' },
285 #if 0
286 /*
287 * This is what else GNU gzip implements. --ascii isn't useful
288 * on NetBSD, and I don't care to have a --license.
289 */
290 { "ascii", no_argument, 0, 'a' },
291 { "license", no_argument, 0, 'L' },
292 #endif
293 { NULL, no_argument, 0, 0 },
294 };
295 #endif
296
297 int
298 main(int argc, char **argv)
299 {
300 const char *progname = getprogname();
301 #ifndef SMALL
302 char *gzip;
303 int len;
304 #endif
305 int ch;
306
307 setup_signals();
308
309 #ifndef SMALL
310 if ((gzip = getenv("GZIP")) != NULL)
311 prepend_gzip(gzip, &argc, &argv);
312 #endif
313
314 /*
315 * XXX
316 * handle being called `gunzip', `zcat' and `gzcat'
317 */
318 if (strcmp(progname, "gunzip") == 0)
319 dflag = 1;
320 else if (strcmp(progname, "zcat") == 0 ||
321 strcmp(progname, "gzcat") == 0)
322 dflag = cflag = 1;
323
324 #ifdef SMALL
325 #define OPT_LIST "123456789cdhlV"
326 #else
327 #define OPT_LIST "123456789cdfhklNnqrS:tVv"
328 #endif
329
330 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) {
331 switch (ch) {
332 case '1': case '2': case '3':
333 case '4': case '5': case '6':
334 case '7': case '8': case '9':
335 numflag = ch - '0';
336 break;
337 case 'c':
338 cflag = 1;
339 break;
340 case 'd':
341 dflag = 1;
342 break;
343 case 'l':
344 lflag = 1;
345 dflag = 1;
346 break;
347 case 'V':
348 display_version();
349 /* NOTREACHED */
350 #ifndef SMALL
351 case 'f':
352 fflag = 1;
353 break;
354 case 'k':
355 kflag = 1;
356 break;
357 case 'N':
358 nflag = 0;
359 Nflag = 1;
360 break;
361 case 'n':
362 nflag = 1;
363 Nflag = 0;
364 break;
365 case 'q':
366 qflag = 1;
367 break;
368 case 'r':
369 rflag = 1;
370 break;
371 case 'S':
372 len = strlen(optarg);
373 if (len != 0) {
374 if (len > SUFFIX_MAXLEN)
375 errx(1, "incorrect suffix: '%s'", optarg);
376 suffixes[0].zipped = optarg;
377 suffixes[0].ziplen = len;
378 } else {
379 suffixes[NUM_SUFFIXES - 1].zipped = "";
380 suffixes[NUM_SUFFIXES - 1].ziplen = 0;
381 }
382 break;
383 case 't':
384 cflag = 1;
385 tflag = 1;
386 dflag = 1;
387 break;
388 case 'v':
389 vflag = 1;
390 break;
391 #endif
392 default:
393 usage();
394 /* NOTREACHED */
395 }
396 }
397 argv += optind;
398 argc -= optind;
399
400 if (argc == 0) {
401 if (dflag) /* stdin mode */
402 handle_stdin();
403 else /* stdout mode */
404 handle_stdout();
405 } else {
406 do {
407 handle_pathname(argv[0]);
408 } while (*++argv);
409 }
410 #ifndef SMALL
411 if (qflag == 0 && lflag && argc > 1)
412 print_list(-1, 0, "(totals)", 0);
413 #endif
414 exit(exit_value);
415 }
416
417 /* maybe print a warning */
418 void
419 maybe_warn(const char *fmt, ...)
420 {
421 va_list ap;
422
423 if (qflag == 0) {
424 va_start(ap, fmt);
425 vwarn(fmt, ap);
426 va_end(ap);
427 }
428 if (exit_value == 0)
429 exit_value = 1;
430 }
431
432 /* ... without an errno. */
433 void
434 maybe_warnx(const char *fmt, ...)
435 {
436 va_list ap;
437
438 if (qflag == 0) {
439 va_start(ap, fmt);
440 vwarnx(fmt, ap);
441 va_end(ap);
442 }
443 if (exit_value == 0)
444 exit_value = 1;
445 }
446
447 /* maybe print an error */
448 void
449 maybe_err(const char *fmt, ...)
450 {
451 va_list ap;
452
453 if (qflag == 0) {
454 va_start(ap, fmt);
455 vwarn(fmt, ap);
456 va_end(ap);
457 }
458 exit(2);
459 }
460
461 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \
462 !defined(NO_XZ_SUPPORT)
463 /* ... without an errno. */
464 void
465 maybe_errx(const char *fmt, ...)
466 {
467 va_list ap;
468
469 if (qflag == 0) {
470 va_start(ap, fmt);
471 vwarnx(fmt, ap);
472 va_end(ap);
473 }
474 exit(2);
475 }
476 #endif
477
478 #ifndef SMALL
479 /* split up $GZIP and prepend it to the argument list */
480 static void
481 prepend_gzip(char *gzip, int *argc, char ***argv)
482 {
483 char *s, **nargv, **ac;
484 int nenvarg = 0, i;
485
486 /* scan how many arguments there are */
487 for (s = gzip;;) {
488 while (*s == ' ' || *s == '\t')
489 s++;
490 if (*s == 0)
491 goto count_done;
492 nenvarg++;
493 while (*s != ' ' && *s != '\t')
494 if (*s++ == 0)
495 goto count_done;
496 }
497 count_done:
498 /* punt early */
499 if (nenvarg == 0)
500 return;
501
502 *argc += nenvarg;
503 ac = *argv;
504
505 nargv = (char **)malloc((*argc + 1) * sizeof(char *));
506 if (nargv == NULL)
507 maybe_err("malloc");
508
509 /* stash this away */
510 *argv = nargv;
511
512 /* copy the program name first */
513 i = 0;
514 nargv[i++] = *(ac++);
515
516 /* take a copy of $GZIP and add it to the array */
517 s = strdup(gzip);
518 if (s == NULL)
519 maybe_err("strdup");
520 for (;;) {
521 /* Skip whitespaces. */
522 while (*s == ' ' || *s == '\t')
523 s++;
524 if (*s == 0)
525 goto copy_done;
526 nargv[i++] = s;
527 /* Find the end of this argument. */
528 while (*s != ' ' && *s != '\t')
529 if (*s++ == 0)
530 /* Argument followed by NUL. */
531 goto copy_done;
532 /* Terminate by overwriting ' ' or '\t' with NUL. */
533 *s++ = 0;
534 }
535 copy_done:
536
537 /* copy the original arguments and a NULL */
538 while (*ac)
539 nargv[i++] = *(ac++);
540 nargv[i] = NULL;
541 }
542 #endif
543
544 /* compress input to output. Return bytes read, -1 on error */
545 static off_t
546 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime)
547 {
548 z_stream z;
549 char *outbufp, *inbufp;
550 off_t in_tot = 0, out_tot = 0;
551 ssize_t in_size;
552 int i, error;
553 uLong crc;
554 #ifdef SMALL
555 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0,
556 0, 0, 0, 0,
557 0, OS_CODE };
558 #endif
559
560 outbufp = malloc(BUFLEN);
561 inbufp = malloc(BUFLEN);
562 if (outbufp == NULL || inbufp == NULL) {
563 maybe_err("malloc failed");
564 goto out;
565 }
566
567 memset(&z, 0, sizeof z);
568 z.zalloc = Z_NULL;
569 z.zfree = Z_NULL;
570 z.opaque = 0;
571
572 #ifdef SMALL
573 memcpy(outbufp, header, sizeof header);
574 i = sizeof header;
575 #else
576 if (nflag != 0) {
577 mtime = 0;
578 origname = "";
579 }
580
581 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s",
582 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED,
583 *origname ? ORIG_NAME : 0,
584 mtime & 0xff,
585 (mtime >> 8) & 0xff,
586 (mtime >> 16) & 0xff,
587 (mtime >> 24) & 0xff,
588 numflag == 1 ? 4 : numflag == 9 ? 2 : 0,
589 OS_CODE, origname);
590 if (i >= BUFLEN)
591 /* this need PATH_MAX > BUFLEN ... */
592 maybe_err("snprintf");
593 if (*origname)
594 i++;
595 #endif
596
597 z.next_out = (unsigned char *)outbufp + i;
598 z.avail_out = BUFLEN - i;
599
600 error = deflateInit2(&z, numflag, Z_DEFLATED,
601 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY);
602 if (error != Z_OK) {
603 maybe_warnx("deflateInit2 failed");
604 in_tot = -1;
605 goto out;
606 }
607
608 crc = crc32(0L, Z_NULL, 0);
609 for (;;) {
610 if (z.avail_out == 0) {
611 if (write_retry(out, outbufp, BUFLEN) != BUFLEN) {
612 maybe_warn("write");
613 out_tot = -1;
614 goto out;
615 }
616
617 out_tot += BUFLEN;
618 z.next_out = (unsigned char *)outbufp;
619 z.avail_out = BUFLEN;
620 }
621
622 if (z.avail_in == 0) {
623 in_size = read(in, inbufp, BUFLEN);
624 if (in_size < 0) {
625 maybe_warn("read");
626 in_tot = -1;
627 goto out;
628 }
629 if (in_size == 0)
630 break;
631 infile_newdata(in_size);
632
633 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size);
634 in_tot += in_size;
635 z.next_in = (unsigned char *)inbufp;
636 z.avail_in = in_size;
637 }
638
639 error = deflate(&z, Z_NO_FLUSH);
640 if (error != Z_OK && error != Z_STREAM_END) {
641 maybe_warnx("deflate failed");
642 in_tot = -1;
643 goto out;
644 }
645 }
646
647 /* clean up */
648 for (;;) {
649 size_t len;
650 ssize_t w;
651
652 error = deflate(&z, Z_FINISH);
653 if (error != Z_OK && error != Z_STREAM_END) {
654 maybe_warnx("deflate failed");
655 in_tot = -1;
656 goto out;
657 }
658
659 len = (char *)z.next_out - outbufp;
660
661 w = write_retry(out, outbufp, len);
662 if (w == -1 || (size_t)w != len) {
663 maybe_warn("write");
664 out_tot = -1;
665 goto out;
666 }
667 out_tot += len;
668 z.next_out = (unsigned char *)outbufp;
669 z.avail_out = BUFLEN;
670
671 if (error == Z_STREAM_END)
672 break;
673 }
674
675 if (deflateEnd(&z) != Z_OK) {
676 maybe_warnx("deflateEnd failed");
677 in_tot = -1;
678 goto out;
679 }
680
681 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c",
682 (int)crc & 0xff,
683 (int)(crc >> 8) & 0xff,
684 (int)(crc >> 16) & 0xff,
685 (int)(crc >> 24) & 0xff,
686 (int)in_tot & 0xff,
687 (int)(in_tot >> 8) & 0xff,
688 (int)(in_tot >> 16) & 0xff,
689 (int)(in_tot >> 24) & 0xff);
690 if (i != 8)
691 maybe_err("snprintf");
692 #if 0
693 if (in_tot > 0xffffffff)
694 maybe_warn("input file size >= 4GB cannot be saved");
695 #endif
696 if (write_retry(out, outbufp, i) != i) {
697 maybe_warn("write");
698 in_tot = -1;
699 } else
700 out_tot += i;
701
702 out:
703 if (inbufp != NULL)
704 free(inbufp);
705 if (outbufp != NULL)
706 free(outbufp);
707 if (gsizep)
708 *gsizep = out_tot;
709 return in_tot;
710 }
711
712 /*
713 * uncompress input to output then close the input. return the
714 * uncompressed size written, and put the compressed sized read
715 * into `*gsizep'.
716 */
717 static off_t
718 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep,
719 const char *filename)
720 {
721 z_stream z;
722 char *outbufp, *inbufp;
723 off_t out_tot = -1, in_tot = 0;
724 uint32_t out_sub_tot = 0;
725 enum {
726 GZSTATE_MAGIC0,
727 GZSTATE_MAGIC1,
728 GZSTATE_METHOD,
729 GZSTATE_FLAGS,
730 GZSTATE_SKIPPING,
731 GZSTATE_EXTRA,
732 GZSTATE_EXTRA2,
733 GZSTATE_EXTRA3,
734 GZSTATE_ORIGNAME,
735 GZSTATE_COMMENT,
736 GZSTATE_HEAD_CRC1,
737 GZSTATE_HEAD_CRC2,
738 GZSTATE_INIT,
739 GZSTATE_READ,
740 GZSTATE_CRC,
741 GZSTATE_LEN,
742 } state = GZSTATE_MAGIC0;
743 int flags = 0, skip_count = 0;
744 int error = Z_STREAM_ERROR, done_reading = 0;
745 uLong crc = 0;
746 ssize_t wr;
747 int needmore = 0;
748
749 #define ADVANCE() { z.next_in++; z.avail_in--; }
750
751 if ((outbufp = malloc(BUFLEN)) == NULL) {
752 maybe_err("malloc failed");
753 goto out2;
754 }
755 if ((inbufp = malloc(BUFLEN)) == NULL) {
756 maybe_err("malloc failed");
757 goto out1;
758 }
759
760 memset(&z, 0, sizeof z);
761 z.avail_in = prelen;
762 z.next_in = (unsigned char *)pre;
763 z.avail_out = BUFLEN;
764 z.next_out = (unsigned char *)outbufp;
765 z.zalloc = NULL;
766 z.zfree = NULL;
767 z.opaque = 0;
768
769 in_tot = prelen;
770 out_tot = 0;
771
772 for (;;) {
773 check_siginfo();
774 if ((z.avail_in == 0 || needmore) && done_reading == 0) {
775 ssize_t in_size;
776
777 if (z.avail_in > 0) {
778 memmove(inbufp, z.next_in, z.avail_in);
779 }
780 z.next_in = (unsigned char *)inbufp;
781 in_size = read(in, z.next_in + z.avail_in,
782 BUFLEN - z.avail_in);
783
784 if (in_size == -1) {
785 maybe_warn("failed to read stdin");
786 goto stop_and_fail;
787 } else if (in_size == 0) {
788 done_reading = 1;
789 }
790 infile_newdata(in_size);
791
792 z.avail_in += in_size;
793 needmore = 0;
794
795 in_tot += in_size;
796 }
797 if (z.avail_in == 0) {
798 if (done_reading && state != GZSTATE_MAGIC0) {
799 maybe_warnx("%s: unexpected end of file",
800 filename);
801 goto stop_and_fail;
802 }
803 goto stop;
804 }
805 switch (state) {
806 case GZSTATE_MAGIC0:
807 if (*z.next_in != GZIP_MAGIC0) {
808 if (in_tot > 0) {
809 maybe_warnx("%s: trailing garbage "
810 "ignored", filename);
811 goto stop;
812 }
813 maybe_warnx("input not gziped (MAGIC0)");
814 exit_value = 2;
815 goto stop_and_fail;
816 }
817 ADVANCE();
818 state++;
819 out_sub_tot = 0;
820 crc = crc32(0L, Z_NULL, 0);
821 break;
822
823 case GZSTATE_MAGIC1:
824 if (*z.next_in != GZIP_MAGIC1 &&
825 *z.next_in != GZIP_OMAGIC1) {
826 maybe_warnx("input not gziped (MAGIC1)");
827 goto stop_and_fail;
828 }
829 ADVANCE();
830 state++;
831 break;
832
833 case GZSTATE_METHOD:
834 if (*z.next_in != Z_DEFLATED) {
835 maybe_warnx("unknown compression method");
836 goto stop_and_fail;
837 }
838 ADVANCE();
839 state++;
840 break;
841
842 case GZSTATE_FLAGS:
843 flags = *z.next_in;
844 ADVANCE();
845 skip_count = 6;
846 state++;
847 break;
848
849 case GZSTATE_SKIPPING:
850 if (skip_count > 0) {
851 skip_count--;
852 ADVANCE();
853 } else
854 state++;
855 break;
856
857 case GZSTATE_EXTRA:
858 if ((flags & EXTRA_FIELD) == 0) {
859 state = GZSTATE_ORIGNAME;
860 break;
861 }
862 skip_count = *z.next_in;
863 ADVANCE();
864 state++;
865 break;
866
867 case GZSTATE_EXTRA2:
868 skip_count |= ((*z.next_in) << 8);
869 ADVANCE();
870 state++;
871 break;
872
873 case GZSTATE_EXTRA3:
874 if (skip_count > 0) {
875 skip_count--;
876 ADVANCE();
877 } else
878 state++;
879 break;
880
881 case GZSTATE_ORIGNAME:
882 if ((flags & ORIG_NAME) == 0) {
883 state++;
884 break;
885 }
886 if (*z.next_in == 0)
887 state++;
888 ADVANCE();
889 break;
890
891 case GZSTATE_COMMENT:
892 if ((flags & COMMENT) == 0) {
893 state++;
894 break;
895 }
896 if (*z.next_in == 0)
897 state++;
898 ADVANCE();
899 break;
900
901 case GZSTATE_HEAD_CRC1:
902 if (flags & HEAD_CRC)
903 skip_count = 2;
904 else
905 skip_count = 0;
906 state++;
907 break;
908
909 case GZSTATE_HEAD_CRC2:
910 if (skip_count > 0) {
911 skip_count--;
912 ADVANCE();
913 } else
914 state++;
915 break;
916
917 case GZSTATE_INIT:
918 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) {
919 maybe_warnx("failed to inflateInit");
920 goto stop_and_fail;
921 }
922 state++;
923 break;
924
925 case GZSTATE_READ:
926 error = inflate(&z, Z_FINISH);
927 switch (error) {
928 /* Z_BUF_ERROR goes with Z_FINISH... */
929 case Z_BUF_ERROR:
930 if (z.avail_out > 0 && !done_reading)
931 continue;
932
933 case Z_STREAM_END:
934 case Z_OK:
935 break;
936
937 case Z_NEED_DICT:
938 maybe_warnx("Z_NEED_DICT error");
939 goto stop_and_fail;
940 case Z_DATA_ERROR:
941 maybe_warnx("data stream error");
942 goto stop_and_fail;
943 case Z_STREAM_ERROR:
944 maybe_warnx("internal stream error");
945 goto stop_and_fail;
946 case Z_MEM_ERROR:
947 maybe_warnx("memory allocation error");
948 goto stop_and_fail;
949
950 default:
951 maybe_warn("unknown error from inflate(): %d",
952 error);
953 }
954 wr = BUFLEN - z.avail_out;
955
956 if (wr != 0) {
957 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr);
958 if (
959 #ifndef SMALL
960 /* don't write anything with -t */
961 tflag == 0 &&
962 #endif
963 write_retry(out, outbufp, wr) != wr) {
964 maybe_warn("error writing to output");
965 goto stop_and_fail;
966 }
967
968 out_tot += wr;
969 out_sub_tot += wr;
970 }
971
972 if (error == Z_STREAM_END) {
973 inflateEnd(&z);
974 state++;
975 }
976
977 z.next_out = (unsigned char *)outbufp;
978 z.avail_out = BUFLEN;
979
980 break;
981 case GZSTATE_CRC:
982 {
983 uLong origcrc;
984
985 if (z.avail_in < 4) {
986 if (!done_reading) {
987 needmore = 1;
988 continue;
989 }
990 maybe_warnx("truncated input");
991 goto stop_and_fail;
992 }
993 origcrc = ((unsigned)z.next_in[0] & 0xff) |
994 ((unsigned)z.next_in[1] & 0xff) << 8 |
995 ((unsigned)z.next_in[2] & 0xff) << 16 |
996 ((unsigned)z.next_in[3] & 0xff) << 24;
997 if (origcrc != crc) {
998 maybe_warnx("invalid compressed"
999 " data--crc error");
1000 goto stop_and_fail;
1001 }
1002 }
1003
1004 z.avail_in -= 4;
1005 z.next_in += 4;
1006
1007 if (!z.avail_in && done_reading) {
1008 goto stop;
1009 }
1010 state++;
1011 break;
1012 case GZSTATE_LEN:
1013 {
1014 uLong origlen;
1015
1016 if (z.avail_in < 4) {
1017 if (!done_reading) {
1018 needmore = 1;
1019 continue;
1020 }
1021 maybe_warnx("truncated input");
1022 goto stop_and_fail;
1023 }
1024 origlen = ((unsigned)z.next_in[0] & 0xff) |
1025 ((unsigned)z.next_in[1] & 0xff) << 8 |
1026 ((unsigned)z.next_in[2] & 0xff) << 16 |
1027 ((unsigned)z.next_in[3] & 0xff) << 24;
1028
1029 if (origlen != out_sub_tot) {
1030 maybe_warnx("invalid compressed"
1031 " data--length error");
1032 goto stop_and_fail;
1033 }
1034 }
1035
1036 z.avail_in -= 4;
1037 z.next_in += 4;
1038
1039 if (error < 0) {
1040 maybe_warnx("decompression error");
1041 goto stop_and_fail;
1042 }
1043 state = GZSTATE_MAGIC0;
1044 break;
1045 }
1046 continue;
1047 stop_and_fail:
1048 out_tot = -1;
1049 stop:
1050 break;
1051 }
1052 if (state > GZSTATE_INIT)
1053 inflateEnd(&z);
1054
1055 free(inbufp);
1056 out1:
1057 free(outbufp);
1058 out2:
1059 if (gsizep)
1060 *gsizep = in_tot;
1061 return (out_tot);
1062 }
1063
1064 #ifndef SMALL
1065 /*
1066 * set the owner, mode, flags & utimes using the given file descriptor.
1067 * file is only used in possible warning messages.
1068 */
1069 static void
1070 copymodes(int fd, const struct stat *sbp, const char *file)
1071 {
1072 struct timeval times[2];
1073 struct stat sb;
1074
1075 /*
1076 * If we have no info on the input, give this file some
1077 * default values and return..
1078 */
1079 if (sbp == NULL) {
1080 mode_t mask = umask(022);
1081
1082 (void)fchmod(fd, DEFFILEMODE & ~mask);
1083 (void)umask(mask);
1084 return;
1085 }
1086 sb = *sbp;
1087
1088 /* if the chown fails, remove set-id bits as-per compress(1) */
1089 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) {
1090 if (errno != EPERM)
1091 maybe_warn("couldn't fchown: %s", file);
1092 sb.st_mode &= ~(S_ISUID|S_ISGID);
1093 }
1094
1095 /* we only allow set-id and the 9 normal permission bits */
1096 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO;
1097 if (fchmod(fd, sb.st_mode) < 0)
1098 maybe_warn("couldn't fchmod: %s", file);
1099
1100 /* only try flags if they exist already */
1101 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0)
1102 maybe_warn("couldn't fchflags: %s", file);
1103
1104 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec);
1105 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec);
1106 if (futimes(fd, times) < 0)
1107 maybe_warn("couldn't utimes: %s", file);
1108 }
1109 #endif
1110
1111 /* what sort of file is this? */
1112 static enum filetype
1113 file_gettype(u_char *buf)
1114 {
1115
1116 if (buf[0] == GZIP_MAGIC0 &&
1117 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1))
1118 return FT_GZIP;
1119 else
1120 #ifndef NO_BZIP2_SUPPORT
1121 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 &&
1122 buf[3] >= '0' && buf[3] <= '9')
1123 return FT_BZIP2;
1124 else
1125 #endif
1126 #ifndef NO_COMPRESS_SUPPORT
1127 if (memcmp(buf, Z_MAGIC, 2) == 0)
1128 return FT_Z;
1129 else
1130 #endif
1131 #ifndef NO_PACK_SUPPORT
1132 if (memcmp(buf, PACK_MAGIC, 2) == 0)
1133 return FT_PACK;
1134 else
1135 #endif
1136 #ifndef NO_XZ_SUPPORT
1137 if (memcmp(buf, XZ_MAGIC, 4) == 0) /* XXX: We only have 4 bytes */
1138 return FT_XZ;
1139 else
1140 #endif
1141 return FT_UNKNOWN;
1142 }
1143
1144 #ifndef SMALL
1145 /* check the outfile is OK. */
1146 static int
1147 check_outfile(const char *outfile)
1148 {
1149 struct stat sb;
1150 int ok = 1;
1151
1152 if (lflag == 0 && stat(outfile, &sb) == 0) {
1153 if (fflag)
1154 unlink(outfile);
1155 else if (isatty(STDIN_FILENO)) {
1156 char ans[10] = { 'n', '\0' }; /* default */
1157
1158 fprintf(stderr, "%s already exists -- do you wish to "
1159 "overwrite (y or n)? " , outfile);
1160 (void)fgets(ans, sizeof(ans) - 1, stdin);
1161 if (ans[0] != 'y' && ans[0] != 'Y') {
1162 fprintf(stderr, "\tnot overwriting\n");
1163 ok = 0;
1164 } else
1165 unlink(outfile);
1166 } else {
1167 maybe_warnx("%s already exists -- skipping", outfile);
1168 ok = 0;
1169 }
1170 }
1171 return ok;
1172 }
1173
1174 static void
1175 unlink_input(const char *file, const struct stat *sb)
1176 {
1177 struct stat nsb;
1178
1179 if (kflag)
1180 return;
1181 if (stat(file, &nsb) != 0)
1182 /* Must be gone already */
1183 return;
1184 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino)
1185 /* Definitely a different file */
1186 return;
1187 unlink(file);
1188 }
1189
1190 static void
1191 got_siginfo(int signo)
1192 {
1193
1194 print_info = 1;
1195 }
1196
1197 static void
1198 setup_signals(void)
1199 {
1200
1201 signal(SIGINFO, got_siginfo);
1202 }
1203
1204 static void
1205 infile_newdata(size_t newdata)
1206 {
1207
1208 infile_current += newdata;
1209 }
1210 #endif
1211
1212 static void
1213 infile_set(const char *newinfile, off_t total)
1214 {
1215
1216 if (newinfile)
1217 infile = newinfile;
1218 #ifndef SMALL
1219 infile_total = total;
1220 #endif
1221 }
1222
1223 static void
1224 infile_clear(void)
1225 {
1226
1227 infile = NULL;
1228 #ifndef SMALL
1229 infile_total = infile_current = 0;
1230 #endif
1231 }
1232
1233 static const suffixes_t *
1234 check_suffix(char *file, int xlate)
1235 {
1236 const suffixes_t *s;
1237 int len = strlen(file);
1238 char *sp;
1239
1240 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) {
1241 /* if it doesn't fit in "a.suf", don't bother */
1242 if (s->ziplen >= len)
1243 continue;
1244 sp = file + len - s->ziplen;
1245 if (strcmp(s->zipped, sp) != 0)
1246 continue;
1247 if (xlate)
1248 strcpy(sp, s->normal);
1249 return s;
1250 }
1251 return NULL;
1252 }
1253
1254 /*
1255 * compress the given file: create a corresponding .gz file and remove the
1256 * original.
1257 */
1258 static off_t
1259 file_compress(char *file, char *outfile, size_t outsize)
1260 {
1261 int in;
1262 int out;
1263 off_t size, in_size;
1264 #ifndef SMALL
1265 struct stat isb, osb;
1266 const suffixes_t *suff;
1267 #endif
1268
1269 in = open(file, O_RDONLY);
1270 if (in == -1) {
1271 maybe_warn("can't open %s", file);
1272 return -1;
1273 }
1274
1275 #ifndef SMALL
1276 if (fstat(in, &isb) != 0) {
1277 close(in);
1278 maybe_warn("can't stat %s", file);
1279 return -1;
1280 }
1281 infile_set(file, isb.st_size);
1282 #endif
1283
1284 if (cflag == 0) {
1285 #ifndef SMALL
1286 if (isb.st_nlink > 1 && fflag == 0) {
1287 maybe_warnx("%s has %d other link%s -- "
1288 "skipping", file, isb.st_nlink - 1,
1289 isb.st_nlink == 1 ? "" : "s");
1290 close(in);
1291 return -1;
1292 }
1293
1294 if (fflag == 0 && (suff = check_suffix(file, 0))
1295 && suff->zipped[0] != 0) {
1296 maybe_warnx("%s already has %s suffix -- unchanged",
1297 file, suff->zipped);
1298 close(in);
1299 return -1;
1300 }
1301 #endif
1302
1303 /* Add (usually) .gz to filename */
1304 if ((size_t)snprintf(outfile, outsize, "%s%s",
1305 file, suffixes[0].zipped) >= outsize)
1306 memcpy(outfile + outsize - suffixes[0].ziplen - 1,
1307 suffixes[0].zipped, suffixes[0].ziplen + 1);
1308
1309 #ifndef SMALL
1310 if (check_outfile(outfile) == 0) {
1311 close(in);
1312 return -1;
1313 }
1314 #endif
1315 }
1316
1317 if (cflag == 0) {
1318 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600);
1319 if (out == -1) {
1320 maybe_warn("could not create output: %s", outfile);
1321 fclose(stdin);
1322 return -1;
1323 }
1324 } else
1325 out = STDOUT_FILENO;
1326
1327 in_size = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime);
1328
1329 (void)close(in);
1330
1331 /*
1332 * If there was an error, in_size will be -1.
1333 * If we compressed to stdout, just return the size.
1334 * Otherwise stat the file and check it is the correct size.
1335 * We only blow away the file if we can stat the output and it
1336 * has the expected size.
1337 */
1338 if (cflag != 0)
1339 return in_size == -1 ? -1 : size;
1340
1341 #ifndef SMALL
1342 if (fstat(out, &osb) != 0) {
1343 maybe_warn("couldn't stat: %s", outfile);
1344 goto bad_outfile;
1345 }
1346
1347 if (osb.st_size != size) {
1348 maybe_warnx("output file: %s wrong size (%" PRIdOFF
1349 " != %" PRIdOFF "), deleting",
1350 outfile, osb.st_size, size);
1351 goto bad_outfile;
1352 }
1353
1354 copymodes(out, &isb, outfile);
1355 #endif
1356 if (close(out) == -1)
1357 maybe_warn("couldn't close output");
1358
1359 /* output is good, ok to delete input */
1360 unlink_input(file, &isb);
1361 return size;
1362
1363 #ifndef SMALL
1364 bad_outfile:
1365 if (close(out) == -1)
1366 maybe_warn("couldn't close output");
1367
1368 maybe_warnx("leaving original %s", file);
1369 unlink(outfile);
1370 return size;
1371 #endif
1372 }
1373
1374 /* uncompress the given file and remove the original */
1375 static off_t
1376 file_uncompress(char *file, char *outfile, size_t outsize)
1377 {
1378 struct stat isb, osb;
1379 off_t size;
1380 ssize_t rbytes;
1381 unsigned char header1[4];
1382 enum filetype method;
1383 int fd, ofd, zfd = -1;
1384 size_t in_size;
1385 #ifndef SMALL
1386 ssize_t rv;
1387 time_t timestamp = 0;
1388 char name[PATH_MAX + 1];
1389 #endif
1390
1391 /* gather the old name info */
1392
1393 fd = open(file, O_RDONLY);
1394 if (fd < 0) {
1395 maybe_warn("can't open %s", file);
1396 goto lose;
1397 }
1398 if (fstat(fd, &isb) != 0) {
1399 close(fd);
1400 maybe_warn("can't stat %s", file);
1401 goto lose;
1402 }
1403 if (S_ISREG(isb.st_mode))
1404 in_size = isb.st_size;
1405 else
1406 in_size = 0;
1407 infile_set(file, in_size);
1408
1409 strlcpy(outfile, file, outsize);
1410 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) {
1411 maybe_warnx("%s: unknown suffix -- ignored", file);
1412 goto lose;
1413 }
1414
1415 rbytes = read(fd, header1, sizeof header1);
1416 if (rbytes != sizeof header1) {
1417 /* we don't want to fail here. */
1418 #ifndef SMALL
1419 if (fflag)
1420 goto lose;
1421 #endif
1422 if (rbytes == -1)
1423 maybe_warn("can't read %s", file);
1424 else
1425 goto unexpected_EOF;
1426 goto lose;
1427 }
1428 infile_newdata(rbytes);
1429
1430 method = file_gettype(header1);
1431 #ifndef SMALL
1432 if (fflag == 0 && method == FT_UNKNOWN) {
1433 maybe_warnx("%s: not in gzip format", file);
1434 goto lose;
1435 }
1436
1437 #endif
1438
1439 #ifndef SMALL
1440 if (method == FT_GZIP && Nflag) {
1441 unsigned char ts[4]; /* timestamp */
1442
1443 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP);
1444 if (rv >= 0 && rv < (ssize_t)(sizeof ts))
1445 goto unexpected_EOF;
1446 if (rv == -1) {
1447 if (!fflag)
1448 maybe_warn("can't read %s", file);
1449 goto lose;
1450 }
1451 infile_newdata(rv);
1452 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0];
1453
1454 if (header1[3] & ORIG_NAME) {
1455 rbytes = pread(fd, name, sizeof(name) - 1, GZIP_ORIGNAME);
1456 if (rbytes < 0) {
1457 maybe_warn("can't read %s", file);
1458 goto lose;
1459 }
1460 if (name[0] != '\0') {
1461 char *dp, *nf;
1462
1463 /* Make sure that name is NUL-terminated */
1464 name[rbytes] = '\0';
1465
1466 /* strip saved directory name */
1467 nf = strrchr(name, '/');
1468 if (nf == NULL)
1469 nf = name;
1470 else
1471 nf++;
1472
1473 /* preserve original directory name */
1474 dp = strrchr(file, '/');
1475 if (dp == NULL)
1476 dp = file;
1477 else
1478 dp++;
1479 snprintf(outfile, outsize, "%.*s%.*s",
1480 (int) (dp - file),
1481 file, (int) rbytes, nf);
1482 }
1483 }
1484 }
1485 #endif
1486 lseek(fd, 0, SEEK_SET);
1487
1488 if (cflag == 0 || lflag) {
1489 #ifndef SMALL
1490 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) {
1491 maybe_warnx("%s has %d other links -- skipping",
1492 file, isb.st_nlink - 1);
1493 goto lose;
1494 }
1495 if (nflag == 0 && timestamp)
1496 isb.st_mtime = timestamp;
1497 if (check_outfile(outfile) == 0)
1498 goto lose;
1499 #endif
1500 }
1501
1502 if (cflag)
1503 zfd = STDOUT_FILENO;
1504 else if (lflag)
1505 zfd = -1;
1506 else {
1507 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600);
1508 if (zfd == STDOUT_FILENO) {
1509 /* We won't close STDOUT_FILENO later... */
1510 zfd = dup(zfd);
1511 close(STDOUT_FILENO);
1512 }
1513 if (zfd == -1) {
1514 maybe_warn("can't open %s", outfile);
1515 goto lose;
1516 }
1517 }
1518
1519 switch (method) {
1520 #ifndef NO_BZIP2_SUPPORT
1521 case FT_BZIP2:
1522 /* XXX */
1523 if (lflag) {
1524 maybe_warnx("no -l with bzip2 files");
1525 goto lose;
1526 }
1527
1528 size = unbzip2(fd, zfd, NULL, 0, NULL);
1529 break;
1530 #endif
1531
1532 #ifndef NO_COMPRESS_SUPPORT
1533 case FT_Z: {
1534 FILE *in, *out;
1535
1536 /* XXX */
1537 if (lflag) {
1538 maybe_warnx("no -l with Lempel-Ziv files");
1539 goto lose;
1540 }
1541
1542 if ((in = zdopen(fd)) == NULL) {
1543 maybe_warn("zdopen for read: %s", file);
1544 goto lose;
1545 }
1546
1547 out = fdopen(dup(zfd), "w");
1548 if (out == NULL) {
1549 maybe_warn("fdopen for write: %s", outfile);
1550 fclose(in);
1551 goto lose;
1552 }
1553
1554 size = zuncompress(in, out, NULL, 0, NULL);
1555 /* need to fclose() if ferror() is true... */
1556 if (ferror(in) | fclose(in)) {
1557 maybe_warn("failed infile fclose");
1558 unlink(outfile);
1559 (void)fclose(out);
1560 }
1561 if (fclose(out) != 0) {
1562 maybe_warn("failed outfile fclose");
1563 unlink(outfile);
1564 goto lose;
1565 }
1566 break;
1567 }
1568 #endif
1569
1570 #ifndef NO_PACK_SUPPORT
1571 case FT_PACK:
1572 if (lflag) {
1573 maybe_warnx("no -l with packed files");
1574 goto lose;
1575 }
1576
1577 size = unpack(fd, zfd, NULL, 0, NULL);
1578 break;
1579 #endif
1580
1581 #ifndef NO_XZ_SUPPORT
1582 case FT_XZ:
1583 if (lflag) {
1584 size = unxz_len(fd);
1585 print_list_out(in_size, size, file);
1586 return -1;
1587 }
1588 size = unxz(fd, zfd, NULL, 0, NULL);
1589 break;
1590 #endif
1591
1592 #ifndef SMALL
1593 case FT_UNKNOWN:
1594 if (lflag) {
1595 maybe_warnx("no -l for unknown filetypes");
1596 goto lose;
1597 }
1598 size = cat_fd(NULL, 0, NULL, fd);
1599 break;
1600 #endif
1601 default:
1602 if (lflag) {
1603 print_list(fd, in_size, outfile, isb.st_mtime);
1604 close(fd);
1605 return -1; /* XXX */
1606 }
1607
1608 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file);
1609 break;
1610 }
1611
1612 if (close(fd) != 0)
1613 maybe_warn("couldn't close input");
1614 if (zfd != STDOUT_FILENO && close(zfd) != 0)
1615 maybe_warn("couldn't close output");
1616
1617 if (size == -1) {
1618 if (cflag == 0)
1619 unlink(outfile);
1620 maybe_warnx("%s: uncompress failed", file);
1621 return -1;
1622 }
1623
1624 /* if testing, or we uncompressed to stdout, this is all we need */
1625 #ifndef SMALL
1626 if (tflag)
1627 return size;
1628 #endif
1629 /* if we are uncompressing to stdin, don't remove the file. */
1630 if (cflag)
1631 return size;
1632
1633 /*
1634 * if we create a file...
1635 */
1636 /*
1637 * if we can't stat the file don't remove the file.
1638 */
1639
1640 ofd = open(outfile, O_RDWR, 0);
1641 if (ofd == -1) {
1642 maybe_warn("couldn't open (leaving original): %s",
1643 outfile);
1644 return -1;
1645 }
1646 if (fstat(ofd, &osb) != 0) {
1647 maybe_warn("couldn't stat (leaving original): %s",
1648 outfile);
1649 close(ofd);
1650 return -1;
1651 }
1652 if (osb.st_size != size) {
1653 maybe_warnx("stat gave different size: %" PRIdOFF
1654 " != %" PRIdOFF " (leaving original)",
1655 size, osb.st_size);
1656 close(ofd);
1657 unlink(outfile);
1658 return -1;
1659 }
1660 unlink_input(file, &isb);
1661 #ifndef SMALL
1662 copymodes(ofd, &isb, outfile);
1663 #endif
1664 close(ofd);
1665 return size;
1666
1667 unexpected_EOF:
1668 maybe_warnx("%s: unexpected end of file", file);
1669 lose:
1670 if (fd != -1)
1671 close(fd);
1672 if (zfd != -1 && zfd != STDOUT_FILENO)
1673 close(fd);
1674 return -1;
1675 }
1676
1677 #ifndef SMALL
1678 static void
1679 check_siginfo(void)
1680 {
1681 if (print_info == 0)
1682 return;
1683 if (infile) {
1684 if (infile_total) {
1685 int pcent = (int)((100.0 * infile_current) / infile_total);
1686
1687 fprintf(stderr, "%s: done %llu/%llu bytes %d%%\n",
1688 infile, (unsigned long long)infile_current,
1689 (unsigned long long)infile_total, pcent);
1690 } else
1691 fprintf(stderr, "%s: done %llu bytes\n",
1692 infile, (unsigned long long)infile_current);
1693 }
1694 print_info = 0;
1695 }
1696
1697 static off_t
1698 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd)
1699 {
1700 char buf[BUFLEN];
1701 off_t in_tot;
1702 ssize_t w;
1703
1704 in_tot = count;
1705 w = write_retry(STDOUT_FILENO, prepend, count);
1706 if (w == -1 || (size_t)w != count) {
1707 maybe_warn("write to stdout");
1708 return -1;
1709 }
1710 for (;;) {
1711 ssize_t rv;
1712
1713 rv = read(fd, buf, sizeof buf);
1714 if (rv == 0)
1715 break;
1716 if (rv < 0) {
1717 maybe_warn("read from fd %d", fd);
1718 break;
1719 }
1720 infile_newdata(rv);
1721
1722 if (write_retry(STDOUT_FILENO, buf, rv) != rv) {
1723 maybe_warn("write to stdout");
1724 break;
1725 }
1726 in_tot += rv;
1727 }
1728
1729 if (gsizep)
1730 *gsizep = in_tot;
1731 return (in_tot);
1732 }
1733 #endif
1734
1735 static void
1736 handle_stdin(void)
1737 {
1738 struct stat isb;
1739 unsigned char header1[4];
1740 size_t in_size;
1741 off_t usize, gsize;
1742 enum filetype method;
1743 ssize_t bytes_read;
1744 #ifndef NO_COMPRESS_SUPPORT
1745 FILE *in;
1746 #endif
1747
1748 #ifndef SMALL
1749 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) {
1750 maybe_warnx("standard input is a terminal -- ignoring");
1751 goto out;
1752 }
1753 #endif
1754
1755 if (fstat(STDIN_FILENO, &isb) < 0) {
1756 maybe_warn("fstat");
1757 goto out;
1758 }
1759 if (S_ISREG(isb.st_mode))
1760 in_size = isb.st_size;
1761 else
1762 in_size = 0;
1763 infile_set("(stdin)", in_size);
1764
1765 if (lflag) {
1766 print_list(STDIN_FILENO, in_size, infile, isb.st_mtime);
1767 goto out;
1768 }
1769
1770 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1);
1771 if (bytes_read == -1) {
1772 maybe_warn("can't read stdin");
1773 goto out;
1774 } else if (bytes_read != sizeof(header1)) {
1775 maybe_warnx("(stdin): unexpected end of file");
1776 goto out;
1777 }
1778
1779 method = file_gettype(header1);
1780 switch (method) {
1781 default:
1782 #ifndef SMALL
1783 if (fflag == 0) {
1784 maybe_warnx("unknown compression format");
1785 goto out;
1786 }
1787 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO);
1788 break;
1789 #endif
1790 case FT_GZIP:
1791 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO,
1792 (char *)header1, sizeof header1, &gsize, "(stdin)");
1793 break;
1794 #ifndef NO_BZIP2_SUPPORT
1795 case FT_BZIP2:
1796 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO,
1797 (char *)header1, sizeof header1, &gsize);
1798 break;
1799 #endif
1800 #ifndef NO_COMPRESS_SUPPORT
1801 case FT_Z:
1802 if ((in = zdopen(STDIN_FILENO)) == NULL) {
1803 maybe_warnx("zopen of stdin");
1804 goto out;
1805 }
1806
1807 usize = zuncompress(in, stdout, (char *)header1,
1808 sizeof header1, &gsize);
1809 fclose(in);
1810 break;
1811 #endif
1812 #ifndef NO_PACK_SUPPORT
1813 case FT_PACK:
1814 usize = unpack(STDIN_FILENO, STDOUT_FILENO,
1815 (char *)header1, sizeof header1, &gsize);
1816 break;
1817 #endif
1818 #ifndef NO_XZ_SUPPORT
1819 case FT_XZ:
1820 usize = unxz(STDIN_FILENO, STDOUT_FILENO,
1821 (char *)header1, sizeof header1, &gsize);
1822 break;
1823 #endif
1824 }
1825
1826 #ifndef SMALL
1827 if (vflag && !tflag && usize != -1 && gsize != -1)
1828 print_verbage(NULL, NULL, usize, gsize);
1829 if (vflag && tflag)
1830 print_test("(stdin)", usize != -1);
1831 #else
1832 (void)&usize;
1833 #endif
1834
1835 out:
1836 infile_clear();
1837 }
1838
1839 static void
1840 handle_stdout(void)
1841 {
1842 off_t gsize;
1843 #ifndef SMALL
1844 off_t usize;
1845 struct stat sb;
1846 time_t systime;
1847 uint32_t mtime;
1848 int ret;
1849
1850 infile_set("(stdout)", 0);
1851
1852 if (fflag == 0 && isatty(STDOUT_FILENO)) {
1853 maybe_warnx("standard output is a terminal -- ignoring");
1854 return;
1855 }
1856
1857 /* If stdin is a file use its mtime, otherwise use current time */
1858 ret = fstat(STDIN_FILENO, &sb);
1859 if (ret < 0) {
1860 maybe_warn("Can't stat stdin");
1861 return;
1862 }
1863
1864 if (S_ISREG(sb.st_mode)) {
1865 infile_set("(stdout)", sb.st_size);
1866 mtime = (uint32_t)sb.st_mtime;
1867 } else {
1868 systime = time(NULL);
1869 if (systime == -1) {
1870 maybe_warn("time");
1871 return;
1872 }
1873 mtime = (uint32_t)systime;
1874 }
1875
1876 usize =
1877 #endif
1878 gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime);
1879 #ifndef SMALL
1880 if (vflag && !tflag && usize != -1 && gsize != -1)
1881 print_verbage(NULL, NULL, usize, gsize);
1882 #endif
1883 }
1884
1885 /* do what is asked for, for the path name */
1886 static void
1887 handle_pathname(char *path)
1888 {
1889 char *opath = path, *s = NULL;
1890 ssize_t len;
1891 int slen;
1892 struct stat sb;
1893
1894 /* check for stdout/stdin */
1895 if (path[0] == '-' && path[1] == '\0') {
1896 if (dflag)
1897 handle_stdin();
1898 else
1899 handle_stdout();
1900 return;
1901 }
1902
1903 retry:
1904 if (stat(path, &sb) != 0) {
1905 /* lets try <path>.gz if we're decompressing */
1906 if (dflag && s == NULL && errno == ENOENT) {
1907 len = strlen(path);
1908 slen = suffixes[0].ziplen;
1909 s = malloc(len + slen + 1);
1910 if (s == NULL)
1911 maybe_err("malloc");
1912 memcpy(s, path, len);
1913 memcpy(s + len, suffixes[0].zipped, slen + 1);
1914 path = s;
1915 goto retry;
1916 }
1917 maybe_warn("can't stat: %s", opath);
1918 goto out;
1919 }
1920
1921 if (S_ISDIR(sb.st_mode)) {
1922 #ifndef SMALL
1923 if (rflag)
1924 handle_dir(path);
1925 else
1926 #endif
1927 maybe_warnx("%s is a directory", path);
1928 goto out;
1929 }
1930
1931 if (S_ISREG(sb.st_mode))
1932 handle_file(path, &sb);
1933 else
1934 maybe_warnx("%s is not a regular file", path);
1935
1936 out:
1937 if (s)
1938 free(s);
1939 }
1940
1941 /* compress/decompress a file */
1942 static void
1943 handle_file(char *file, struct stat *sbp)
1944 {
1945 off_t usize, gsize;
1946 char outfile[PATH_MAX];
1947
1948 infile_set(file, sbp->st_size);
1949 if (dflag) {
1950 usize = file_uncompress(file, outfile, sizeof(outfile));
1951 #ifndef SMALL
1952 if (vflag && tflag)
1953 print_test(file, usize != -1);
1954 #endif
1955 if (usize == -1)
1956 return;
1957 gsize = sbp->st_size;
1958 } else {
1959 gsize = file_compress(file, outfile, sizeof(outfile));
1960 if (gsize == -1)
1961 return;
1962 usize = sbp->st_size;
1963 }
1964 infile_clear();
1965
1966 #ifndef SMALL
1967 if (vflag && !tflag)
1968 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize);
1969 #endif
1970 }
1971
1972 #ifndef SMALL
1973 /* this is used with -r to recursively descend directories */
1974 static void
1975 handle_dir(char *dir)
1976 {
1977 char *path_argv[2];
1978 FTS *fts;
1979 FTSENT *entry;
1980
1981 path_argv[0] = dir;
1982 path_argv[1] = 0;
1983 fts = fts_open(path_argv, FTS_PHYSICAL, NULL);
1984 if (fts == NULL) {
1985 warn("couldn't fts_open %s", dir);
1986 return;
1987 }
1988
1989 while ((entry = fts_read(fts))) {
1990 switch(entry->fts_info) {
1991 case FTS_D:
1992 case FTS_DP:
1993 continue;
1994
1995 case FTS_DNR:
1996 case FTS_ERR:
1997 case FTS_NS:
1998 maybe_warn("%s", entry->fts_path);
1999 continue;
2000 case FTS_F:
2001 handle_file(entry->fts_name, entry->fts_statp);
2002 }
2003 }
2004 (void)fts_close(fts);
2005 }
2006 #endif
2007
2008 /* print a ratio - size reduction as a fraction of uncompressed size */
2009 static void
2010 print_ratio(off_t in, off_t out, FILE *where)
2011 {
2012 int percent10; /* 10 * percent */
2013 off_t diff;
2014 char buff[8];
2015 int len;
2016
2017 diff = in - out/2;
2018 if (in == 0 && out == 0)
2019 percent10 = 0;
2020 else if (diff < 0)
2021 /*
2022 * Output is more than double size of input! print -99.9%
2023 * Quite possibly we've failed to get the original size.
2024 */
2025 percent10 = -999;
2026 else {
2027 /*
2028 * We only need 12 bits of result from the final division,
2029 * so reduce the values until a 32bit division will suffice.
2030 */
2031 while (in > 0x100000) {
2032 diff >>= 1;
2033 in >>= 1;
2034 }
2035 if (in != 0)
2036 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000;
2037 else
2038 percent10 = 0;
2039 }
2040
2041 len = snprintf(buff, sizeof buff, "%2.2d.", percent10);
2042 /* Move the '.' to before the last digit */
2043 buff[len - 1] = buff[len - 2];
2044 buff[len - 2] = '.';
2045 fprintf(where, "%5s%%", buff);
2046 }
2047
2048 #ifndef SMALL
2049 /* print compression statistics, and the new name (if there is one!) */
2050 static void
2051 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize)
2052 {
2053 if (file)
2054 fprintf(stderr, "%s:%s ", file,
2055 strlen(file) < 7 ? "\t\t" : "\t");
2056 print_ratio(usize, gsize, stderr);
2057 if (nfile)
2058 fprintf(stderr, " -- replaced with %s", nfile);
2059 fprintf(stderr, "\n");
2060 fflush(stderr);
2061 }
2062
2063 /* print test results */
2064 static void
2065 print_test(const char *file, int ok)
2066 {
2067
2068 if (exit_value == 0 && ok == 0)
2069 exit_value = 1;
2070 fprintf(stderr, "%s:%s %s\n", file,
2071 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK");
2072 fflush(stderr);
2073 }
2074 #endif
2075
2076 /* print a file's info ala --list */
2077 /* eg:
2078 compressed uncompressed ratio uncompressed_name
2079 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar
2080 */
2081 static void
2082 print_list(int fd, off_t out, const char *outfile, time_t ts)
2083 {
2084 static int first = 1;
2085 #ifndef SMALL
2086 static off_t in_tot, out_tot;
2087 uint32_t crc = 0;
2088 #endif
2089 off_t in = 0, rv;
2090
2091 if (first) {
2092 #ifndef SMALL
2093 if (vflag)
2094 printf("method crc date time ");
2095 #endif
2096 if (qflag == 0)
2097 printf(" compressed uncompressed "
2098 "ratio uncompressed_name\n");
2099 }
2100 first = 0;
2101
2102 /* print totals? */
2103 #ifndef SMALL
2104 if (fd == -1) {
2105 in = in_tot;
2106 out = out_tot;
2107 } else
2108 #endif
2109 {
2110 /* read the last 4 bytes - this is the uncompressed size */
2111 rv = lseek(fd, (off_t)(-8), SEEK_END);
2112 if (rv != -1) {
2113 unsigned char buf[8];
2114 uint32_t usize;
2115
2116 rv = read(fd, (char *)buf, sizeof(buf));
2117 if (rv == -1)
2118 maybe_warn("read of uncompressed size");
2119 else if (rv != sizeof(buf))
2120 maybe_warnx("read of uncompressed size");
2121
2122 else {
2123 usize = buf[4];
2124 usize |= (unsigned int)buf[5] << 8;
2125 usize |= (unsigned int)buf[6] << 16;
2126 usize |= (unsigned int)buf[7] << 24;
2127 in = (off_t)usize;
2128 #ifndef SMALL
2129 crc = buf[0];
2130 crc |= (unsigned int)buf[1] << 8;
2131 crc |= (unsigned int)buf[2] << 16;
2132 crc |= (unsigned int)buf[3] << 24;
2133 #endif
2134 }
2135 }
2136 }
2137
2138 #ifndef SMALL
2139 if (vflag && fd == -1)
2140 printf(" ");
2141 else if (vflag) {
2142 char *date = ctime(&ts);
2143
2144 /* skip the day, 1/100th second, and year */
2145 date += 4;
2146 date[12] = 0;
2147 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date);
2148 }
2149 in_tot += in;
2150 out_tot += out;
2151 #endif
2152 print_list_out(out, in, outfile);
2153 }
2154
2155 static void
2156 print_list_out(off_t out, off_t in, const char *outfile)
2157 {
2158 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in);
2159 print_ratio(in, out, stdout);
2160 printf(" %s\n", outfile);
2161 }
2162
2163 /* display the usage of NetBSD gzip */
2164 static void
2165 usage(void)
2166 {
2167
2168 fprintf(stderr, "%s\n", gzip_version);
2169 fprintf(stderr,
2170 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n"
2171 #ifndef SMALL
2172 " -1 --fast fastest (worst) compression\n"
2173 " -2 .. -8 set compression level\n"
2174 " -9 --best best (slowest) compression\n"
2175 " -c --stdout write to stdout, keep original files\n"
2176 " --to-stdout\n"
2177 " -d --decompress uncompress files\n"
2178 " --uncompress\n"
2179 " -f --force force overwriting & compress links\n"
2180 " -h --help display this help\n"
2181 " -k --keep don't delete input files during operation\n"
2182 " -l --list list compressed file contents\n"
2183 " -N --name save or restore original file name and time stamp\n"
2184 " -n --no-name don't save original file name or time stamp\n"
2185 " -q --quiet output no warnings\n"
2186 " -r --recursive recursively compress files in directories\n"
2187 " -S .suf use suffix .suf instead of .gz\n"
2188 " --suffix .suf\n"
2189 " -t --test test compressed file\n"
2190 " -V --version display program version\n"
2191 " -v --verbose print extra statistics\n",
2192 #else
2193 ,
2194 #endif
2195 getprogname());
2196 exit(0);
2197 }
2198
2199 /* display the version of NetBSD gzip */
2200 static void
2201 display_version(void)
2202 {
2203
2204 fprintf(stderr, "%s\n", gzip_version);
2205 exit(0);
2206 }
2207
2208 #ifndef NO_BZIP2_SUPPORT
2209 #include "unbzip2.c"
2210 #endif
2211 #ifndef NO_COMPRESS_SUPPORT
2212 #include "zuncompress.c"
2213 #endif
2214 #ifndef NO_PACK_SUPPORT
2215 #include "unpack.c"
2216 #endif
2217 #ifndef NO_XZ_SUPPORT
2218 #include "unxz.c"
2219 #endif
2220
2221 static ssize_t
2222 read_retry(int fd, void *buf, size_t sz)
2223 {
2224 char *cp = buf;
2225 size_t left = MIN(sz, (size_t) SSIZE_MAX);
2226
2227 while (left > 0) {
2228 ssize_t ret;
2229
2230 ret = read(fd, cp, left);
2231 if (ret == -1) {
2232 return ret;
2233 } else if (ret == 0) {
2234 break; /* EOF */
2235 }
2236 cp += ret;
2237 left -= ret;
2238 }
2239
2240 return sz - left;
2241 }
2242
2243 static ssize_t
2244 write_retry(int fd, const void *buf, size_t sz)
2245 {
2246 const char *cp = buf;
2247 size_t left = MIN(sz, (size_t) SSIZE_MAX);
2248
2249 while (left > 0) {
2250 ssize_t ret;
2251
2252 ret = write(fd, cp, left);
2253 if (ret == -1) {
2254 return ret;
2255 } else if (ret == 0) {
2256 abort(); /* Can't happen */
2257 }
2258 cp += ret;
2259 left -= ret;
2260 }
2261
2262 return sz - left;
2263 }
2264