gzip.c revision 1.112 1 /* $NetBSD: gzip.c,v 1.112 2017/08/23 13:04:17 christos 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 1.112 2017/08/23 13:04:17 christos 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
217 #ifdef SMALL
218 #define infile_set(f,t) infile_set(f)
219 #endif
220 static void infile_set(const char *newinfile, off_t total);
221
222 #ifdef SMALL
223 #define unlink_input(f, sb) unlink(f)
224 #define check_siginfo() /* nothing */
225 #define setup_signals() /* nothing */
226 #define infile_newdata(t) /* nothing */
227 #else
228 static off_t infile_total; /* total expected to read/write */
229 static off_t infile_current; /* current read/write */
230
231 static void check_siginfo(void);
232 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd);
233 static void prepend_gzip(char *, int *, char ***);
234 static void handle_dir(char *);
235 static void print_verbage(const char *, const char *, off_t, off_t);
236 static void print_test(const char *, int);
237 static void copymodes(int fd, const struct stat *, const char *file);
238 static int check_outfile(const char *outfile);
239 static void setup_signals(void);
240 static void infile_newdata(size_t newdata);
241 static void infile_clear(void);
242 #endif
243
244 #ifndef NO_BZIP2_SUPPORT
245 static off_t unbzip2(int, int, char *, size_t, off_t *);
246 #endif
247
248 #ifndef NO_COMPRESS_SUPPORT
249 static FILE *zdopen(int);
250 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *);
251 #endif
252
253 #ifndef NO_PACK_SUPPORT
254 static off_t unpack(int, int, char *, size_t, off_t *);
255 #endif
256
257 #ifndef NO_XZ_SUPPORT
258 static off_t unxz(int, int, char *, size_t, off_t *);
259 #endif
260
261 #ifdef SMALL
262 #define getopt_long(a,b,c,d,e) getopt(a,b,c)
263 #else
264 static const struct option longopts[] = {
265 { "stdout", no_argument, 0, 'c' },
266 { "to-stdout", no_argument, 0, 'c' },
267 { "decompress", no_argument, 0, 'd' },
268 { "uncompress", no_argument, 0, 'd' },
269 { "force", no_argument, 0, 'f' },
270 { "help", no_argument, 0, 'h' },
271 { "keep", no_argument, 0, 'k' },
272 { "list", no_argument, 0, 'l' },
273 { "no-name", no_argument, 0, 'n' },
274 { "name", no_argument, 0, 'N' },
275 { "quiet", no_argument, 0, 'q' },
276 { "recursive", no_argument, 0, 'r' },
277 { "suffix", required_argument, 0, 'S' },
278 { "test", no_argument, 0, 't' },
279 { "verbose", no_argument, 0, 'v' },
280 { "version", no_argument, 0, 'V' },
281 { "fast", no_argument, 0, '1' },
282 { "best", no_argument, 0, '9' },
283 #if 0
284 /*
285 * This is what else GNU gzip implements. --ascii isn't useful
286 * on NetBSD, and I don't care to have a --license.
287 */
288 { "ascii", no_argument, 0, 'a' },
289 { "license", no_argument, 0, 'L' },
290 #endif
291 { NULL, no_argument, 0, 0 },
292 };
293 #endif
294
295 int
296 main(int argc, char **argv)
297 {
298 const char *progname = getprogname();
299 #ifndef SMALL
300 char *gzip;
301 int len;
302 #endif
303 int ch;
304
305 setup_signals();
306
307 #ifndef SMALL
308 if ((gzip = getenv("GZIP")) != NULL)
309 prepend_gzip(gzip, &argc, &argv);
310 #endif
311
312 /*
313 * XXX
314 * handle being called `gunzip', `zcat' and `gzcat'
315 */
316 if (strcmp(progname, "gunzip") == 0)
317 dflag = 1;
318 else if (strcmp(progname, "zcat") == 0 ||
319 strcmp(progname, "gzcat") == 0)
320 dflag = cflag = 1;
321
322 #ifdef SMALL
323 #define OPT_LIST "123456789cdhlV"
324 #else
325 #define OPT_LIST "123456789cdfhklNnqrS:tVv"
326 #endif
327
328 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) {
329 switch (ch) {
330 case '1': case '2': case '3':
331 case '4': case '5': case '6':
332 case '7': case '8': case '9':
333 numflag = ch - '0';
334 break;
335 case 'c':
336 cflag = 1;
337 break;
338 case 'd':
339 dflag = 1;
340 break;
341 case 'l':
342 lflag = 1;
343 dflag = 1;
344 break;
345 case 'V':
346 display_version();
347 /* NOTREACHED */
348 #ifndef SMALL
349 case 'f':
350 fflag = 1;
351 break;
352 case 'k':
353 kflag = 1;
354 break;
355 case 'N':
356 nflag = 0;
357 Nflag = 1;
358 break;
359 case 'n':
360 nflag = 1;
361 Nflag = 0;
362 break;
363 case 'q':
364 qflag = 1;
365 break;
366 case 'r':
367 rflag = 1;
368 break;
369 case 'S':
370 len = strlen(optarg);
371 if (len != 0) {
372 if (len > SUFFIX_MAXLEN)
373 errx(1, "incorrect suffix: '%s'", optarg);
374 suffixes[0].zipped = optarg;
375 suffixes[0].ziplen = len;
376 } else {
377 suffixes[NUM_SUFFIXES - 1].zipped = "";
378 suffixes[NUM_SUFFIXES - 1].ziplen = 0;
379 }
380 break;
381 case 't':
382 cflag = 1;
383 tflag = 1;
384 dflag = 1;
385 break;
386 case 'v':
387 vflag = 1;
388 break;
389 #endif
390 default:
391 usage();
392 /* NOTREACHED */
393 }
394 }
395 argv += optind;
396 argc -= optind;
397
398 if (argc == 0) {
399 if (dflag) /* stdin mode */
400 handle_stdin();
401 else /* stdout mode */
402 handle_stdout();
403 } else {
404 do {
405 handle_pathname(argv[0]);
406 } while (*++argv);
407 }
408 #ifndef SMALL
409 if (qflag == 0 && lflag && argc > 1)
410 print_list(-1, 0, "(totals)", 0);
411 #endif
412 exit(exit_value);
413 }
414
415 /* maybe print a warning */
416 void
417 maybe_warn(const char *fmt, ...)
418 {
419 va_list ap;
420
421 if (qflag == 0) {
422 va_start(ap, fmt);
423 vwarn(fmt, ap);
424 va_end(ap);
425 }
426 if (exit_value == 0)
427 exit_value = 1;
428 }
429
430 /* ... without an errno. */
431 void
432 maybe_warnx(const char *fmt, ...)
433 {
434 va_list ap;
435
436 if (qflag == 0) {
437 va_start(ap, fmt);
438 vwarnx(fmt, ap);
439 va_end(ap);
440 }
441 if (exit_value == 0)
442 exit_value = 1;
443 }
444
445 /* maybe print an error */
446 void
447 maybe_err(const char *fmt, ...)
448 {
449 va_list ap;
450
451 if (qflag == 0) {
452 va_start(ap, fmt);
453 vwarn(fmt, ap);
454 va_end(ap);
455 }
456 exit(2);
457 }
458
459 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \
460 !defined(NO_XZ_SUPPORT)
461 /* ... without an errno. */
462 void
463 maybe_errx(const char *fmt, ...)
464 {
465 va_list ap;
466
467 if (qflag == 0) {
468 va_start(ap, fmt);
469 vwarnx(fmt, ap);
470 va_end(ap);
471 }
472 exit(2);
473 }
474 #endif
475
476 #ifndef SMALL
477 /* split up $GZIP and prepend it to the argument list */
478 static void
479 prepend_gzip(char *gzip, int *argc, char ***argv)
480 {
481 char *s, **nargv, **ac;
482 int nenvarg = 0, i;
483
484 /* scan how many arguments there are */
485 for (s = gzip;;) {
486 while (*s == ' ' || *s == '\t')
487 s++;
488 if (*s == 0)
489 goto count_done;
490 nenvarg++;
491 while (*s != ' ' && *s != '\t')
492 if (*s++ == 0)
493 goto count_done;
494 }
495 count_done:
496 /* punt early */
497 if (nenvarg == 0)
498 return;
499
500 *argc += nenvarg;
501 ac = *argv;
502
503 nargv = (char **)malloc((*argc + 1) * sizeof(char *));
504 if (nargv == NULL)
505 maybe_err("malloc");
506
507 /* stash this away */
508 *argv = nargv;
509
510 /* copy the program name first */
511 i = 0;
512 nargv[i++] = *(ac++);
513
514 /* take a copy of $GZIP and add it to the array */
515 s = strdup(gzip);
516 if (s == NULL)
517 maybe_err("strdup");
518 for (;;) {
519 /* Skip whitespaces. */
520 while (*s == ' ' || *s == '\t')
521 s++;
522 if (*s == 0)
523 goto copy_done;
524 nargv[i++] = s;
525 /* Find the end of this argument. */
526 while (*s != ' ' && *s != '\t')
527 if (*s++ == 0)
528 /* Argument followed by NUL. */
529 goto copy_done;
530 /* Terminate by overwriting ' ' or '\t' with NUL. */
531 *s++ = 0;
532 }
533 copy_done:
534
535 /* copy the original arguments and a NULL */
536 while (*ac)
537 nargv[i++] = *(ac++);
538 nargv[i] = NULL;
539 }
540 #endif
541
542 /* compress input to output. Return bytes read, -1 on error */
543 static off_t
544 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime)
545 {
546 z_stream z;
547 char *outbufp, *inbufp;
548 off_t in_tot = 0, out_tot = 0;
549 ssize_t in_size;
550 int i, error;
551 uLong crc;
552 #ifdef SMALL
553 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0,
554 0, 0, 0, 0,
555 0, OS_CODE };
556 #endif
557
558 outbufp = malloc(BUFLEN);
559 inbufp = malloc(BUFLEN);
560 if (outbufp == NULL || inbufp == NULL) {
561 maybe_err("malloc failed");
562 goto out;
563 }
564
565 memset(&z, 0, sizeof z);
566 z.zalloc = Z_NULL;
567 z.zfree = Z_NULL;
568 z.opaque = 0;
569
570 #ifdef SMALL
571 memcpy(outbufp, header, sizeof header);
572 i = sizeof header;
573 #else
574 if (nflag != 0) {
575 mtime = 0;
576 origname = "";
577 }
578
579 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s",
580 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED,
581 *origname ? ORIG_NAME : 0,
582 mtime & 0xff,
583 (mtime >> 8) & 0xff,
584 (mtime >> 16) & 0xff,
585 (mtime >> 24) & 0xff,
586 numflag == 1 ? 4 : numflag == 9 ? 2 : 0,
587 OS_CODE, origname);
588 if (i >= BUFLEN)
589 /* this need PATH_MAX > BUFLEN ... */
590 maybe_err("snprintf");
591 if (*origname)
592 i++;
593 #endif
594
595 z.next_out = (unsigned char *)outbufp + i;
596 z.avail_out = BUFLEN - i;
597
598 error = deflateInit2(&z, numflag, Z_DEFLATED,
599 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY);
600 if (error != Z_OK) {
601 maybe_warnx("deflateInit2 failed");
602 in_tot = -1;
603 goto out;
604 }
605
606 crc = crc32(0L, Z_NULL, 0);
607 for (;;) {
608 if (z.avail_out == 0) {
609 if (write_retry(out, outbufp, BUFLEN) != BUFLEN) {
610 maybe_warn("write");
611 out_tot = -1;
612 goto out;
613 }
614
615 out_tot += BUFLEN;
616 z.next_out = (unsigned char *)outbufp;
617 z.avail_out = BUFLEN;
618 }
619
620 if (z.avail_in == 0) {
621 in_size = read(in, inbufp, BUFLEN);
622 if (in_size < 0) {
623 maybe_warn("read");
624 in_tot = -1;
625 goto out;
626 }
627 if (in_size == 0)
628 break;
629 infile_newdata(in_size);
630
631 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size);
632 in_tot += in_size;
633 z.next_in = (unsigned char *)inbufp;
634 z.avail_in = in_size;
635 }
636
637 error = deflate(&z, Z_NO_FLUSH);
638 if (error != Z_OK && error != Z_STREAM_END) {
639 maybe_warnx("deflate failed");
640 in_tot = -1;
641 goto out;
642 }
643 }
644
645 /* clean up */
646 for (;;) {
647 size_t len;
648 ssize_t w;
649
650 error = deflate(&z, Z_FINISH);
651 if (error != Z_OK && error != Z_STREAM_END) {
652 maybe_warnx("deflate failed");
653 in_tot = -1;
654 goto out;
655 }
656
657 len = (char *)z.next_out - outbufp;
658
659 w = write_retry(out, outbufp, len);
660 if (w == -1 || (size_t)w != len) {
661 maybe_warn("write");
662 out_tot = -1;
663 goto out;
664 }
665 out_tot += len;
666 z.next_out = (unsigned char *)outbufp;
667 z.avail_out = BUFLEN;
668
669 if (error == Z_STREAM_END)
670 break;
671 }
672
673 if (deflateEnd(&z) != Z_OK) {
674 maybe_warnx("deflateEnd failed");
675 in_tot = -1;
676 goto out;
677 }
678
679 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c",
680 (int)crc & 0xff,
681 (int)(crc >> 8) & 0xff,
682 (int)(crc >> 16) & 0xff,
683 (int)(crc >> 24) & 0xff,
684 (int)in_tot & 0xff,
685 (int)(in_tot >> 8) & 0xff,
686 (int)(in_tot >> 16) & 0xff,
687 (int)(in_tot >> 24) & 0xff);
688 if (i != 8)
689 maybe_err("snprintf");
690 #if 0
691 if (in_tot > 0xffffffff)
692 maybe_warn("input file size >= 4GB cannot be saved");
693 #endif
694 if (write_retry(out, outbufp, i) != i) {
695 maybe_warn("write");
696 in_tot = -1;
697 } else
698 out_tot += i;
699
700 out:
701 if (inbufp != NULL)
702 free(inbufp);
703 if (outbufp != NULL)
704 free(outbufp);
705 if (gsizep)
706 *gsizep = out_tot;
707 return in_tot;
708 }
709
710 /*
711 * uncompress input to output then close the input. return the
712 * uncompressed size written, and put the compressed sized read
713 * into `*gsizep'.
714 */
715 static off_t
716 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep,
717 const char *filename)
718 {
719 z_stream z;
720 char *outbufp, *inbufp;
721 off_t out_tot = -1, in_tot = 0;
722 uint32_t out_sub_tot = 0;
723 enum {
724 GZSTATE_MAGIC0,
725 GZSTATE_MAGIC1,
726 GZSTATE_METHOD,
727 GZSTATE_FLAGS,
728 GZSTATE_SKIPPING,
729 GZSTATE_EXTRA,
730 GZSTATE_EXTRA2,
731 GZSTATE_EXTRA3,
732 GZSTATE_ORIGNAME,
733 GZSTATE_COMMENT,
734 GZSTATE_HEAD_CRC1,
735 GZSTATE_HEAD_CRC2,
736 GZSTATE_INIT,
737 GZSTATE_READ,
738 GZSTATE_CRC,
739 GZSTATE_LEN,
740 } state = GZSTATE_MAGIC0;
741 int flags = 0, skip_count = 0;
742 int error = Z_STREAM_ERROR, done_reading = 0;
743 uLong crc = 0;
744 ssize_t wr;
745 int needmore = 0;
746
747 #define ADVANCE() { z.next_in++; z.avail_in--; }
748
749 if ((outbufp = malloc(BUFLEN)) == NULL) {
750 maybe_err("malloc failed");
751 goto out2;
752 }
753 if ((inbufp = malloc(BUFLEN)) == NULL) {
754 maybe_err("malloc failed");
755 goto out1;
756 }
757
758 memset(&z, 0, sizeof z);
759 z.avail_in = prelen;
760 z.next_in = (unsigned char *)pre;
761 z.avail_out = BUFLEN;
762 z.next_out = (unsigned char *)outbufp;
763 z.zalloc = NULL;
764 z.zfree = NULL;
765 z.opaque = 0;
766
767 in_tot = prelen;
768 out_tot = 0;
769
770 for (;;) {
771 check_siginfo();
772 if ((z.avail_in == 0 || needmore) && done_reading == 0) {
773 ssize_t in_size;
774
775 if (z.avail_in > 0) {
776 memmove(inbufp, z.next_in, z.avail_in);
777 }
778 z.next_in = (unsigned char *)inbufp;
779 in_size = read(in, z.next_in + z.avail_in,
780 BUFLEN - z.avail_in);
781
782 if (in_size == -1) {
783 maybe_warn("failed to read stdin");
784 goto stop_and_fail;
785 } else if (in_size == 0) {
786 done_reading = 1;
787 }
788 infile_newdata(in_size);
789
790 z.avail_in += in_size;
791 needmore = 0;
792
793 in_tot += in_size;
794 }
795 if (z.avail_in == 0) {
796 if (done_reading && state != GZSTATE_MAGIC0) {
797 maybe_warnx("%s: unexpected end of file",
798 filename);
799 goto stop_and_fail;
800 }
801 goto stop;
802 }
803 switch (state) {
804 case GZSTATE_MAGIC0:
805 if (*z.next_in != GZIP_MAGIC0) {
806 if (in_tot > 0) {
807 maybe_warnx("%s: trailing garbage "
808 "ignored", filename);
809 goto stop;
810 }
811 maybe_warnx("input not gziped (MAGIC0)");
812 exit_value = 2;
813 goto stop_and_fail;
814 }
815 ADVANCE();
816 state++;
817 out_sub_tot = 0;
818 crc = crc32(0L, Z_NULL, 0);
819 break;
820
821 case GZSTATE_MAGIC1:
822 if (*z.next_in != GZIP_MAGIC1 &&
823 *z.next_in != GZIP_OMAGIC1) {
824 maybe_warnx("input not gziped (MAGIC1)");
825 goto stop_and_fail;
826 }
827 ADVANCE();
828 state++;
829 break;
830
831 case GZSTATE_METHOD:
832 if (*z.next_in != Z_DEFLATED) {
833 maybe_warnx("unknown compression method");
834 goto stop_and_fail;
835 }
836 ADVANCE();
837 state++;
838 break;
839
840 case GZSTATE_FLAGS:
841 flags = *z.next_in;
842 ADVANCE();
843 skip_count = 6;
844 state++;
845 break;
846
847 case GZSTATE_SKIPPING:
848 if (skip_count > 0) {
849 skip_count--;
850 ADVANCE();
851 } else
852 state++;
853 break;
854
855 case GZSTATE_EXTRA:
856 if ((flags & EXTRA_FIELD) == 0) {
857 state = GZSTATE_ORIGNAME;
858 break;
859 }
860 skip_count = *z.next_in;
861 ADVANCE();
862 state++;
863 break;
864
865 case GZSTATE_EXTRA2:
866 skip_count |= ((*z.next_in) << 8);
867 ADVANCE();
868 state++;
869 break;
870
871 case GZSTATE_EXTRA3:
872 if (skip_count > 0) {
873 skip_count--;
874 ADVANCE();
875 } else
876 state++;
877 break;
878
879 case GZSTATE_ORIGNAME:
880 if ((flags & ORIG_NAME) == 0) {
881 state++;
882 break;
883 }
884 if (*z.next_in == 0)
885 state++;
886 ADVANCE();
887 break;
888
889 case GZSTATE_COMMENT:
890 if ((flags & COMMENT) == 0) {
891 state++;
892 break;
893 }
894 if (*z.next_in == 0)
895 state++;
896 ADVANCE();
897 break;
898
899 case GZSTATE_HEAD_CRC1:
900 if (flags & HEAD_CRC)
901 skip_count = 2;
902 else
903 skip_count = 0;
904 state++;
905 break;
906
907 case GZSTATE_HEAD_CRC2:
908 if (skip_count > 0) {
909 skip_count--;
910 ADVANCE();
911 } else
912 state++;
913 break;
914
915 case GZSTATE_INIT:
916 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) {
917 maybe_warnx("failed to inflateInit");
918 goto stop_and_fail;
919 }
920 state++;
921 break;
922
923 case GZSTATE_READ:
924 error = inflate(&z, Z_FINISH);
925 switch (error) {
926 /* Z_BUF_ERROR goes with Z_FINISH... */
927 case Z_BUF_ERROR:
928 if (z.avail_out > 0 && !done_reading)
929 continue;
930
931 case Z_STREAM_END:
932 case Z_OK:
933 break;
934
935 case Z_NEED_DICT:
936 maybe_warnx("Z_NEED_DICT error");
937 goto stop_and_fail;
938 case Z_DATA_ERROR:
939 maybe_warnx("data stream error");
940 goto stop_and_fail;
941 case Z_STREAM_ERROR:
942 maybe_warnx("internal stream error");
943 goto stop_and_fail;
944 case Z_MEM_ERROR:
945 maybe_warnx("memory allocation error");
946 goto stop_and_fail;
947
948 default:
949 maybe_warn("unknown error from inflate(): %d",
950 error);
951 }
952 wr = BUFLEN - z.avail_out;
953
954 if (wr != 0) {
955 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr);
956 if (
957 #ifndef SMALL
958 /* don't write anything with -t */
959 tflag == 0 &&
960 #endif
961 write_retry(out, outbufp, wr) != wr) {
962 maybe_warn("error writing to output");
963 goto stop_and_fail;
964 }
965
966 out_tot += wr;
967 out_sub_tot += wr;
968 }
969
970 if (error == Z_STREAM_END) {
971 inflateEnd(&z);
972 state++;
973 }
974
975 z.next_out = (unsigned char *)outbufp;
976 z.avail_out = BUFLEN;
977
978 break;
979 case GZSTATE_CRC:
980 {
981 uLong origcrc;
982
983 if (z.avail_in < 4) {
984 if (!done_reading) {
985 needmore = 1;
986 continue;
987 }
988 maybe_warnx("truncated input");
989 goto stop_and_fail;
990 }
991 origcrc = ((unsigned)z.next_in[0] & 0xff) |
992 ((unsigned)z.next_in[1] & 0xff) << 8 |
993 ((unsigned)z.next_in[2] & 0xff) << 16 |
994 ((unsigned)z.next_in[3] & 0xff) << 24;
995 if (origcrc != crc) {
996 maybe_warnx("invalid compressed"
997 " data--crc error");
998 goto stop_and_fail;
999 }
1000 }
1001
1002 z.avail_in -= 4;
1003 z.next_in += 4;
1004
1005 if (!z.avail_in && done_reading) {
1006 goto stop;
1007 }
1008 state++;
1009 break;
1010 case GZSTATE_LEN:
1011 {
1012 uLong origlen;
1013
1014 if (z.avail_in < 4) {
1015 if (!done_reading) {
1016 needmore = 1;
1017 continue;
1018 }
1019 maybe_warnx("truncated input");
1020 goto stop_and_fail;
1021 }
1022 origlen = ((unsigned)z.next_in[0] & 0xff) |
1023 ((unsigned)z.next_in[1] & 0xff) << 8 |
1024 ((unsigned)z.next_in[2] & 0xff) << 16 |
1025 ((unsigned)z.next_in[3] & 0xff) << 24;
1026
1027 if (origlen != out_sub_tot) {
1028 maybe_warnx("invalid compressed"
1029 " data--length error");
1030 goto stop_and_fail;
1031 }
1032 }
1033
1034 z.avail_in -= 4;
1035 z.next_in += 4;
1036
1037 if (error < 0) {
1038 maybe_warnx("decompression error");
1039 goto stop_and_fail;
1040 }
1041 state = GZSTATE_MAGIC0;
1042 break;
1043 }
1044 continue;
1045 stop_and_fail:
1046 out_tot = -1;
1047 stop:
1048 break;
1049 }
1050 if (state > GZSTATE_INIT)
1051 inflateEnd(&z);
1052
1053 free(inbufp);
1054 out1:
1055 free(outbufp);
1056 out2:
1057 if (gsizep)
1058 *gsizep = in_tot;
1059 return (out_tot);
1060 }
1061
1062 #ifndef SMALL
1063 /*
1064 * set the owner, mode, flags & utimes using the given file descriptor.
1065 * file is only used in possible warning messages.
1066 */
1067 static void
1068 copymodes(int fd, const struct stat *sbp, const char *file)
1069 {
1070 struct timeval times[2];
1071 struct stat sb;
1072
1073 /*
1074 * If we have no info on the input, give this file some
1075 * default values and return..
1076 */
1077 if (sbp == NULL) {
1078 mode_t mask = umask(022);
1079
1080 (void)fchmod(fd, DEFFILEMODE & ~mask);
1081 (void)umask(mask);
1082 return;
1083 }
1084 sb = *sbp;
1085
1086 /* if the chown fails, remove set-id bits as-per compress(1) */
1087 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) {
1088 if (errno != EPERM)
1089 maybe_warn("couldn't fchown: %s", file);
1090 sb.st_mode &= ~(S_ISUID|S_ISGID);
1091 }
1092
1093 /* we only allow set-id and the 9 normal permission bits */
1094 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO;
1095 if (fchmod(fd, sb.st_mode) < 0)
1096 maybe_warn("couldn't fchmod: %s", file);
1097
1098 /* only try flags if they exist already */
1099 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0)
1100 maybe_warn("couldn't fchflags: %s", file);
1101
1102 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec);
1103 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec);
1104 if (futimes(fd, times) < 0)
1105 maybe_warn("couldn't utimes: %s", file);
1106 }
1107 #endif
1108
1109 /* what sort of file is this? */
1110 static enum filetype
1111 file_gettype(u_char *buf)
1112 {
1113
1114 if (buf[0] == GZIP_MAGIC0 &&
1115 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1))
1116 return FT_GZIP;
1117 else
1118 #ifndef NO_BZIP2_SUPPORT
1119 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 &&
1120 buf[3] >= '0' && buf[3] <= '9')
1121 return FT_BZIP2;
1122 else
1123 #endif
1124 #ifndef NO_COMPRESS_SUPPORT
1125 if (memcmp(buf, Z_MAGIC, 2) == 0)
1126 return FT_Z;
1127 else
1128 #endif
1129 #ifndef NO_PACK_SUPPORT
1130 if (memcmp(buf, PACK_MAGIC, 2) == 0)
1131 return FT_PACK;
1132 else
1133 #endif
1134 #ifndef NO_XZ_SUPPORT
1135 if (memcmp(buf, XZ_MAGIC, 4) == 0) /* XXX: We only have 4 bytes */
1136 return FT_XZ;
1137 else
1138 #endif
1139 return FT_UNKNOWN;
1140 }
1141
1142 #ifndef SMALL
1143 /* check the outfile is OK. */
1144 static int
1145 check_outfile(const char *outfile)
1146 {
1147 struct stat sb;
1148 int ok = 1;
1149
1150 if (lflag == 0 && stat(outfile, &sb) == 0) {
1151 if (fflag)
1152 unlink(outfile);
1153 else if (isatty(STDIN_FILENO)) {
1154 char ans[10] = { 'n', '\0' }; /* default */
1155
1156 fprintf(stderr, "%s already exists -- do you wish to "
1157 "overwrite (y or n)? " , outfile);
1158 (void)fgets(ans, sizeof(ans) - 1, stdin);
1159 if (ans[0] != 'y' && ans[0] != 'Y') {
1160 fprintf(stderr, "\tnot overwriting\n");
1161 ok = 0;
1162 } else
1163 unlink(outfile);
1164 } else {
1165 maybe_warnx("%s already exists -- skipping", outfile);
1166 ok = 0;
1167 }
1168 }
1169 return ok;
1170 }
1171
1172 static void
1173 unlink_input(const char *file, const struct stat *sb)
1174 {
1175 struct stat nsb;
1176
1177 if (kflag)
1178 return;
1179 if (stat(file, &nsb) != 0)
1180 /* Must be gone already */
1181 return;
1182 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino)
1183 /* Definitely a different file */
1184 return;
1185 unlink(file);
1186 }
1187
1188 static void
1189 got_siginfo(int signo)
1190 {
1191
1192 print_info = 1;
1193 }
1194
1195 static void
1196 setup_signals(void)
1197 {
1198
1199 signal(SIGINFO, got_siginfo);
1200 }
1201
1202 static void
1203 infile_newdata(size_t newdata)
1204 {
1205
1206 infile_current += newdata;
1207 }
1208 #endif
1209
1210 static void
1211 infile_set(const char *newinfile, off_t total)
1212 {
1213
1214 if (newinfile)
1215 infile = newinfile;
1216 #ifndef SMALL
1217 infile_total = total;
1218 #endif
1219 }
1220
1221 static void
1222 infile_clear(void)
1223 {
1224
1225 infile = NULL;
1226 #ifndef SMALL
1227 infile_total = infile_current = 0;
1228 #endif
1229 }
1230
1231 static const suffixes_t *
1232 check_suffix(char *file, int xlate)
1233 {
1234 const suffixes_t *s;
1235 int len = strlen(file);
1236 char *sp;
1237
1238 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) {
1239 /* if it doesn't fit in "a.suf", don't bother */
1240 if (s->ziplen >= len)
1241 continue;
1242 sp = file + len - s->ziplen;
1243 if (strcmp(s->zipped, sp) != 0)
1244 continue;
1245 if (xlate)
1246 strcpy(sp, s->normal);
1247 return s;
1248 }
1249 return NULL;
1250 }
1251
1252 /*
1253 * compress the given file: create a corresponding .gz file and remove the
1254 * original.
1255 */
1256 static off_t
1257 file_compress(char *file, char *outfile, size_t outsize)
1258 {
1259 int in;
1260 int out;
1261 off_t size, in_size;
1262 #ifndef SMALL
1263 struct stat isb, osb;
1264 const suffixes_t *suff;
1265 #endif
1266
1267 in = open(file, O_RDONLY);
1268 if (in == -1) {
1269 maybe_warn("can't open %s", file);
1270 return -1;
1271 }
1272
1273 #ifndef SMALL
1274 if (fstat(in, &isb) != 0) {
1275 close(in);
1276 maybe_warn("can't stat %s", file);
1277 return -1;
1278 }
1279 infile_set(file, isb.st_size);
1280 #endif
1281
1282 if (cflag == 0) {
1283 #ifndef SMALL
1284 if (isb.st_nlink > 1 && fflag == 0) {
1285 maybe_warnx("%s has %d other link%s -- "
1286 "skipping", file, isb.st_nlink - 1,
1287 isb.st_nlink == 1 ? "" : "s");
1288 close(in);
1289 return -1;
1290 }
1291
1292 if (fflag == 0 && (suff = check_suffix(file, 0))
1293 && suff->zipped[0] != 0) {
1294 maybe_warnx("%s already has %s suffix -- unchanged",
1295 file, suff->zipped);
1296 close(in);
1297 return -1;
1298 }
1299 #endif
1300
1301 /* Add (usually) .gz to filename */
1302 if ((size_t)snprintf(outfile, outsize, "%s%s",
1303 file, suffixes[0].zipped) >= outsize)
1304 memcpy(outfile + outsize - suffixes[0].ziplen - 1,
1305 suffixes[0].zipped, suffixes[0].ziplen + 1);
1306
1307 #ifndef SMALL
1308 if (check_outfile(outfile) == 0) {
1309 close(in);
1310 return -1;
1311 }
1312 #endif
1313 }
1314
1315 if (cflag == 0) {
1316 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600);
1317 if (out == -1) {
1318 maybe_warn("could not create output: %s", outfile);
1319 fclose(stdin);
1320 return -1;
1321 }
1322 } else
1323 out = STDOUT_FILENO;
1324
1325 in_size = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime);
1326
1327 (void)close(in);
1328
1329 /*
1330 * If there was an error, in_size will be -1.
1331 * If we compressed to stdout, just return the size.
1332 * Otherwise stat the file and check it is the correct size.
1333 * We only blow away the file if we can stat the output and it
1334 * has the expected size.
1335 */
1336 if (cflag != 0)
1337 return in_size == -1 ? -1 : size;
1338
1339 #ifndef SMALL
1340 if (fstat(out, &osb) != 0) {
1341 maybe_warn("couldn't stat: %s", outfile);
1342 goto bad_outfile;
1343 }
1344
1345 if (osb.st_size != size) {
1346 maybe_warnx("output file: %s wrong size (%" PRIdOFF
1347 " != %" PRIdOFF "), deleting",
1348 outfile, osb.st_size, size);
1349 goto bad_outfile;
1350 }
1351
1352 copymodes(out, &isb, outfile);
1353 #endif
1354 if (close(out) == -1)
1355 maybe_warn("couldn't close output");
1356
1357 /* output is good, ok to delete input */
1358 unlink_input(file, &isb);
1359 return size;
1360
1361 #ifndef SMALL
1362 bad_outfile:
1363 if (close(out) == -1)
1364 maybe_warn("couldn't close output");
1365
1366 maybe_warnx("leaving original %s", file);
1367 unlink(outfile);
1368 return size;
1369 #endif
1370 }
1371
1372 /* uncompress the given file and remove the original */
1373 static off_t
1374 file_uncompress(char *file, char *outfile, size_t outsize)
1375 {
1376 struct stat isb, osb;
1377 off_t size;
1378 ssize_t rbytes;
1379 unsigned char header1[4];
1380 enum filetype method;
1381 int fd, ofd, zfd = -1;
1382 size_t in_size;
1383 #ifndef SMALL
1384 ssize_t rv;
1385 time_t timestamp = 0;
1386 char name[PATH_MAX + 1];
1387 #endif
1388
1389 /* gather the old name info */
1390
1391 fd = open(file, O_RDONLY);
1392 if (fd < 0) {
1393 maybe_warn("can't open %s", file);
1394 goto lose;
1395 }
1396 if (fstat(fd, &isb) != 0) {
1397 close(fd);
1398 maybe_warn("can't stat %s", file);
1399 goto lose;
1400 }
1401 if (S_ISREG(isb.st_mode))
1402 in_size = isb.st_size;
1403 else
1404 in_size = 0;
1405 infile_set(file, in_size);
1406
1407 strlcpy(outfile, file, outsize);
1408 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) {
1409 maybe_warnx("%s: unknown suffix -- ignored", file);
1410 goto lose;
1411 }
1412
1413 rbytes = read(fd, header1, sizeof header1);
1414 if (rbytes != sizeof header1) {
1415 /* we don't want to fail here. */
1416 #ifndef SMALL
1417 if (fflag)
1418 goto lose;
1419 #endif
1420 if (rbytes == -1)
1421 maybe_warn("can't read %s", file);
1422 else
1423 goto unexpected_EOF;
1424 goto lose;
1425 }
1426 infile_newdata(rbytes);
1427
1428 method = file_gettype(header1);
1429 #ifndef SMALL
1430 if (fflag == 0 && method == FT_UNKNOWN) {
1431 maybe_warnx("%s: not in gzip format", file);
1432 goto lose;
1433 }
1434
1435 #endif
1436
1437 #ifndef SMALL
1438 if (method == FT_GZIP && Nflag) {
1439 unsigned char ts[4]; /* timestamp */
1440
1441 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP);
1442 if (rv >= 0 && rv < (ssize_t)(sizeof ts))
1443 goto unexpected_EOF;
1444 if (rv == -1) {
1445 if (!fflag)
1446 maybe_warn("can't read %s", file);
1447 goto lose;
1448 }
1449 infile_newdata(rv);
1450 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0];
1451
1452 if (header1[3] & ORIG_NAME) {
1453 rbytes = pread(fd, name, sizeof(name) - 1, GZIP_ORIGNAME);
1454 if (rbytes < 0) {
1455 maybe_warn("can't read %s", file);
1456 goto lose;
1457 }
1458 if (name[0] != '\0') {
1459 char *dp, *nf;
1460
1461 /* Make sure that name is NUL-terminated */
1462 name[rbytes] = '\0';
1463
1464 /* strip saved directory name */
1465 nf = strrchr(name, '/');
1466 if (nf == NULL)
1467 nf = name;
1468 else
1469 nf++;
1470
1471 /* preserve original directory name */
1472 dp = strrchr(file, '/');
1473 if (dp == NULL)
1474 dp = file;
1475 else
1476 dp++;
1477 snprintf(outfile, outsize, "%.*s%.*s",
1478 (int) (dp - file),
1479 file, (int) rbytes, nf);
1480 }
1481 }
1482 }
1483 #endif
1484 lseek(fd, 0, SEEK_SET);
1485
1486 if (cflag == 0 || lflag) {
1487 #ifndef SMALL
1488 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) {
1489 maybe_warnx("%s has %d other links -- skipping",
1490 file, isb.st_nlink - 1);
1491 goto lose;
1492 }
1493 if (nflag == 0 && timestamp)
1494 isb.st_mtime = timestamp;
1495 if (check_outfile(outfile) == 0)
1496 goto lose;
1497 #endif
1498 }
1499
1500 if (cflag)
1501 zfd = STDOUT_FILENO;
1502 else if (lflag)
1503 zfd = -1;
1504 else {
1505 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600);
1506 if (zfd == STDOUT_FILENO) {
1507 /* We won't close STDOUT_FILENO later... */
1508 zfd = dup(zfd);
1509 close(STDOUT_FILENO);
1510 }
1511 if (zfd == -1) {
1512 maybe_warn("can't open %s", outfile);
1513 goto lose;
1514 }
1515 }
1516
1517 switch (method) {
1518 #ifndef NO_BZIP2_SUPPORT
1519 case FT_BZIP2:
1520 /* XXX */
1521 if (lflag) {
1522 maybe_warnx("no -l with bzip2 files");
1523 goto lose;
1524 }
1525
1526 size = unbzip2(fd, zfd, NULL, 0, NULL);
1527 break;
1528 #endif
1529
1530 #ifndef NO_COMPRESS_SUPPORT
1531 case FT_Z: {
1532 FILE *in, *out;
1533
1534 /* XXX */
1535 if (lflag) {
1536 maybe_warnx("no -l with Lempel-Ziv files");
1537 goto lose;
1538 }
1539
1540 if ((in = zdopen(fd)) == NULL) {
1541 maybe_warn("zdopen for read: %s", file);
1542 goto lose;
1543 }
1544
1545 out = fdopen(dup(zfd), "w");
1546 if (out == NULL) {
1547 maybe_warn("fdopen for write: %s", outfile);
1548 fclose(in);
1549 goto lose;
1550 }
1551
1552 size = zuncompress(in, out, NULL, 0, NULL);
1553 /* need to fclose() if ferror() is true... */
1554 if (ferror(in) | fclose(in)) {
1555 maybe_warn("failed infile fclose");
1556 unlink(outfile);
1557 (void)fclose(out);
1558 }
1559 if (fclose(out) != 0) {
1560 maybe_warn("failed outfile fclose");
1561 unlink(outfile);
1562 goto lose;
1563 }
1564 break;
1565 }
1566 #endif
1567
1568 #ifndef NO_PACK_SUPPORT
1569 case FT_PACK:
1570 if (lflag) {
1571 maybe_warnx("no -l with packed files");
1572 goto lose;
1573 }
1574
1575 size = unpack(fd, zfd, NULL, 0, NULL);
1576 break;
1577 #endif
1578
1579 #ifndef NO_XZ_SUPPORT
1580 case FT_XZ:
1581 if (lflag) {
1582 maybe_warnx("no -l with xz files");
1583 goto lose;
1584 }
1585
1586 size = unxz(fd, zfd, NULL, 0, NULL);
1587 break;
1588 #endif
1589
1590 #ifndef SMALL
1591 case FT_UNKNOWN:
1592 if (lflag) {
1593 maybe_warnx("no -l for unknown filetypes");
1594 goto lose;
1595 }
1596 size = cat_fd(NULL, 0, NULL, fd);
1597 break;
1598 #endif
1599 default:
1600 if (lflag) {
1601 print_list(fd, in_size, outfile, isb.st_mtime);
1602 close(fd);
1603 return -1; /* XXX */
1604 }
1605
1606 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file);
1607 break;
1608 }
1609
1610 if (close(fd) != 0)
1611 maybe_warn("couldn't close input");
1612 if (zfd != STDOUT_FILENO && close(zfd) != 0)
1613 maybe_warn("couldn't close output");
1614
1615 if (size == -1) {
1616 if (cflag == 0)
1617 unlink(outfile);
1618 maybe_warnx("%s: uncompress failed", file);
1619 return -1;
1620 }
1621
1622 /* if testing, or we uncompressed to stdout, this is all we need */
1623 #ifndef SMALL
1624 if (tflag)
1625 return size;
1626 #endif
1627 /* if we are uncompressing to stdin, don't remove the file. */
1628 if (cflag)
1629 return size;
1630
1631 /*
1632 * if we create a file...
1633 */
1634 /*
1635 * if we can't stat the file don't remove the file.
1636 */
1637
1638 ofd = open(outfile, O_RDWR, 0);
1639 if (ofd == -1) {
1640 maybe_warn("couldn't open (leaving original): %s",
1641 outfile);
1642 return -1;
1643 }
1644 if (fstat(ofd, &osb) != 0) {
1645 maybe_warn("couldn't stat (leaving original): %s",
1646 outfile);
1647 close(ofd);
1648 return -1;
1649 }
1650 if (osb.st_size != size) {
1651 maybe_warnx("stat gave different size: %" PRIdOFF
1652 " != %" PRIdOFF " (leaving original)",
1653 size, osb.st_size);
1654 close(ofd);
1655 unlink(outfile);
1656 return -1;
1657 }
1658 unlink_input(file, &isb);
1659 #ifndef SMALL
1660 copymodes(ofd, &isb, outfile);
1661 #endif
1662 close(ofd);
1663 return size;
1664
1665 unexpected_EOF:
1666 maybe_warnx("%s: unexpected end of file", file);
1667 lose:
1668 if (fd != -1)
1669 close(fd);
1670 if (zfd != -1 && zfd != STDOUT_FILENO)
1671 close(fd);
1672 return -1;
1673 }
1674
1675 #ifndef SMALL
1676 static void
1677 check_siginfo(void)
1678 {
1679 if (print_info == 0)
1680 return;
1681 if (infile) {
1682 if (infile_total) {
1683 int pcent = (int)((100.0 * infile_current) / infile_total);
1684
1685 fprintf(stderr, "%s: done %llu/%llu bytes %d%%\n",
1686 infile, (unsigned long long)infile_current,
1687 (unsigned long long)infile_total, pcent);
1688 } else
1689 fprintf(stderr, "%s: done %llu bytes\n",
1690 infile, (unsigned long long)infile_current);
1691 }
1692 print_info = 0;
1693 }
1694
1695 static off_t
1696 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd)
1697 {
1698 char buf[BUFLEN];
1699 off_t in_tot;
1700 ssize_t w;
1701
1702 in_tot = count;
1703 w = write_retry(STDOUT_FILENO, prepend, count);
1704 if (w == -1 || (size_t)w != count) {
1705 maybe_warn("write to stdout");
1706 return -1;
1707 }
1708 for (;;) {
1709 ssize_t rv;
1710
1711 rv = read(fd, buf, sizeof buf);
1712 if (rv == 0)
1713 break;
1714 if (rv < 0) {
1715 maybe_warn("read from fd %d", fd);
1716 break;
1717 }
1718 infile_newdata(rv);
1719
1720 if (write_retry(STDOUT_FILENO, buf, rv) != rv) {
1721 maybe_warn("write to stdout");
1722 break;
1723 }
1724 in_tot += rv;
1725 }
1726
1727 if (gsizep)
1728 *gsizep = in_tot;
1729 return (in_tot);
1730 }
1731 #endif
1732
1733 static void
1734 handle_stdin(void)
1735 {
1736 struct stat isb;
1737 unsigned char header1[4];
1738 size_t in_size;
1739 off_t usize, gsize;
1740 enum filetype method;
1741 ssize_t bytes_read;
1742 #ifndef NO_COMPRESS_SUPPORT
1743 FILE *in;
1744 #endif
1745
1746 #ifndef SMALL
1747 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) {
1748 maybe_warnx("standard input is a terminal -- ignoring");
1749 goto out;
1750 }
1751 #endif
1752
1753 if (fstat(STDIN_FILENO, &isb) < 0) {
1754 maybe_warn("fstat");
1755 goto out;
1756 }
1757 if (S_ISREG(isb.st_mode))
1758 in_size = isb.st_size;
1759 else
1760 in_size = 0;
1761 infile_set("(stdin)", in_size);
1762
1763 if (lflag) {
1764 print_list(STDIN_FILENO, in_size, infile, isb.st_mtime);
1765 goto out;
1766 }
1767
1768 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1);
1769 if (bytes_read == -1) {
1770 maybe_warn("can't read stdin");
1771 goto out;
1772 } else if (bytes_read != sizeof(header1)) {
1773 maybe_warnx("(stdin): unexpected end of file");
1774 goto out;
1775 }
1776
1777 method = file_gettype(header1);
1778 switch (method) {
1779 default:
1780 #ifndef SMALL
1781 if (fflag == 0) {
1782 maybe_warnx("unknown compression format");
1783 goto out;
1784 }
1785 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO);
1786 break;
1787 #endif
1788 case FT_GZIP:
1789 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO,
1790 (char *)header1, sizeof header1, &gsize, "(stdin)");
1791 break;
1792 #ifndef NO_BZIP2_SUPPORT
1793 case FT_BZIP2:
1794 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO,
1795 (char *)header1, sizeof header1, &gsize);
1796 break;
1797 #endif
1798 #ifndef NO_COMPRESS_SUPPORT
1799 case FT_Z:
1800 if ((in = zdopen(STDIN_FILENO)) == NULL) {
1801 maybe_warnx("zopen of stdin");
1802 goto out;
1803 }
1804
1805 usize = zuncompress(in, stdout, (char *)header1,
1806 sizeof header1, &gsize);
1807 fclose(in);
1808 break;
1809 #endif
1810 #ifndef NO_PACK_SUPPORT
1811 case FT_PACK:
1812 usize = unpack(STDIN_FILENO, STDOUT_FILENO,
1813 (char *)header1, sizeof header1, &gsize);
1814 break;
1815 #endif
1816 #ifndef NO_XZ_SUPPORT
1817 case FT_XZ:
1818 usize = unxz(STDIN_FILENO, STDOUT_FILENO,
1819 (char *)header1, sizeof header1, &gsize);
1820 break;
1821 #endif
1822 }
1823
1824 #ifndef SMALL
1825 if (vflag && !tflag && usize != -1 && gsize != -1)
1826 print_verbage(NULL, NULL, usize, gsize);
1827 if (vflag && tflag)
1828 print_test("(stdin)", usize != -1);
1829 #else
1830 (void)&usize;
1831 #endif
1832
1833 out:
1834 infile_clear();
1835 }
1836
1837 static void
1838 handle_stdout(void)
1839 {
1840 off_t gsize;
1841 #ifndef SMALL
1842 off_t usize;
1843 struct stat sb;
1844 time_t systime;
1845 uint32_t mtime;
1846 int ret;
1847
1848 infile_set("(stdout)", 0);
1849
1850 if (fflag == 0 && isatty(STDOUT_FILENO)) {
1851 maybe_warnx("standard output is a terminal -- ignoring");
1852 return;
1853 }
1854
1855 /* If stdin is a file use its mtime, otherwise use current time */
1856 ret = fstat(STDIN_FILENO, &sb);
1857 if (ret < 0) {
1858 maybe_warn("Can't stat stdin");
1859 return;
1860 }
1861
1862 if (S_ISREG(sb.st_mode)) {
1863 infile_set("(stdout)", sb.st_size);
1864 mtime = (uint32_t)sb.st_mtime;
1865 } else {
1866 systime = time(NULL);
1867 if (systime == -1) {
1868 maybe_warn("time");
1869 return;
1870 }
1871 mtime = (uint32_t)systime;
1872 }
1873
1874 usize =
1875 #endif
1876 gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime);
1877 #ifndef SMALL
1878 if (vflag && !tflag && usize != -1 && gsize != -1)
1879 print_verbage(NULL, NULL, usize, gsize);
1880 #endif
1881 }
1882
1883 /* do what is asked for, for the path name */
1884 static void
1885 handle_pathname(char *path)
1886 {
1887 char *opath = path, *s = NULL;
1888 ssize_t len;
1889 int slen;
1890 struct stat sb;
1891
1892 /* check for stdout/stdin */
1893 if (path[0] == '-' && path[1] == '\0') {
1894 if (dflag)
1895 handle_stdin();
1896 else
1897 handle_stdout();
1898 return;
1899 }
1900
1901 retry:
1902 if (stat(path, &sb) != 0) {
1903 /* lets try <path>.gz if we're decompressing */
1904 if (dflag && s == NULL && errno == ENOENT) {
1905 len = strlen(path);
1906 slen = suffixes[0].ziplen;
1907 s = malloc(len + slen + 1);
1908 if (s == NULL)
1909 maybe_err("malloc");
1910 memcpy(s, path, len);
1911 memcpy(s + len, suffixes[0].zipped, slen + 1);
1912 path = s;
1913 goto retry;
1914 }
1915 maybe_warn("can't stat: %s", opath);
1916 goto out;
1917 }
1918
1919 if (S_ISDIR(sb.st_mode)) {
1920 #ifndef SMALL
1921 if (rflag)
1922 handle_dir(path);
1923 else
1924 #endif
1925 maybe_warnx("%s is a directory", path);
1926 goto out;
1927 }
1928
1929 if (S_ISREG(sb.st_mode))
1930 handle_file(path, &sb);
1931 else
1932 maybe_warnx("%s is not a regular file", path);
1933
1934 out:
1935 if (s)
1936 free(s);
1937 }
1938
1939 /* compress/decompress a file */
1940 static void
1941 handle_file(char *file, struct stat *sbp)
1942 {
1943 off_t usize, gsize;
1944 char outfile[PATH_MAX];
1945
1946 infile_set(file, sbp->st_size);
1947 if (dflag) {
1948 usize = file_uncompress(file, outfile, sizeof(outfile));
1949 #ifndef SMALL
1950 if (vflag && tflag)
1951 print_test(file, usize != -1);
1952 #endif
1953 if (usize == -1)
1954 return;
1955 gsize = sbp->st_size;
1956 } else {
1957 gsize = file_compress(file, outfile, sizeof(outfile));
1958 if (gsize == -1)
1959 return;
1960 usize = sbp->st_size;
1961 }
1962 infile_clear();
1963
1964 #ifndef SMALL
1965 if (vflag && !tflag)
1966 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize);
1967 #endif
1968 }
1969
1970 #ifndef SMALL
1971 /* this is used with -r to recursively descend directories */
1972 static void
1973 handle_dir(char *dir)
1974 {
1975 char *path_argv[2];
1976 FTS *fts;
1977 FTSENT *entry;
1978
1979 path_argv[0] = dir;
1980 path_argv[1] = 0;
1981 fts = fts_open(path_argv, FTS_PHYSICAL, NULL);
1982 if (fts == NULL) {
1983 warn("couldn't fts_open %s", dir);
1984 return;
1985 }
1986
1987 while ((entry = fts_read(fts))) {
1988 switch(entry->fts_info) {
1989 case FTS_D:
1990 case FTS_DP:
1991 continue;
1992
1993 case FTS_DNR:
1994 case FTS_ERR:
1995 case FTS_NS:
1996 maybe_warn("%s", entry->fts_path);
1997 continue;
1998 case FTS_F:
1999 handle_file(entry->fts_name, entry->fts_statp);
2000 }
2001 }
2002 (void)fts_close(fts);
2003 }
2004 #endif
2005
2006 /* print a ratio - size reduction as a fraction of uncompressed size */
2007 static void
2008 print_ratio(off_t in, off_t out, FILE *where)
2009 {
2010 int percent10; /* 10 * percent */
2011 off_t diff;
2012 char buff[8];
2013 int len;
2014
2015 diff = in - out/2;
2016 if (in == 0 && out == 0)
2017 percent10 = 0;
2018 else if (diff < 0)
2019 /*
2020 * Output is more than double size of input! print -99.9%
2021 * Quite possibly we've failed to get the original size.
2022 */
2023 percent10 = -999;
2024 else {
2025 /*
2026 * We only need 12 bits of result from the final division,
2027 * so reduce the values until a 32bit division will suffice.
2028 */
2029 while (in > 0x100000) {
2030 diff >>= 1;
2031 in >>= 1;
2032 }
2033 if (in != 0)
2034 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000;
2035 else
2036 percent10 = 0;
2037 }
2038
2039 len = snprintf(buff, sizeof buff, "%2.2d.", percent10);
2040 /* Move the '.' to before the last digit */
2041 buff[len - 1] = buff[len - 2];
2042 buff[len - 2] = '.';
2043 fprintf(where, "%5s%%", buff);
2044 }
2045
2046 #ifndef SMALL
2047 /* print compression statistics, and the new name (if there is one!) */
2048 static void
2049 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize)
2050 {
2051 if (file)
2052 fprintf(stderr, "%s:%s ", file,
2053 strlen(file) < 7 ? "\t\t" : "\t");
2054 print_ratio(usize, gsize, stderr);
2055 if (nfile)
2056 fprintf(stderr, " -- replaced with %s", nfile);
2057 fprintf(stderr, "\n");
2058 fflush(stderr);
2059 }
2060
2061 /* print test results */
2062 static void
2063 print_test(const char *file, int ok)
2064 {
2065
2066 if (exit_value == 0 && ok == 0)
2067 exit_value = 1;
2068 fprintf(stderr, "%s:%s %s\n", file,
2069 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK");
2070 fflush(stderr);
2071 }
2072 #endif
2073
2074 /* print a file's info ala --list */
2075 /* eg:
2076 compressed uncompressed ratio uncompressed_name
2077 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar
2078 */
2079 static void
2080 print_list(int fd, off_t out, const char *outfile, time_t ts)
2081 {
2082 static int first = 1;
2083 #ifndef SMALL
2084 static off_t in_tot, out_tot;
2085 uint32_t crc = 0;
2086 #endif
2087 off_t in = 0, rv;
2088
2089 if (first) {
2090 #ifndef SMALL
2091 if (vflag)
2092 printf("method crc date time ");
2093 #endif
2094 if (qflag == 0)
2095 printf(" compressed uncompressed "
2096 "ratio uncompressed_name\n");
2097 }
2098 first = 0;
2099
2100 /* print totals? */
2101 #ifndef SMALL
2102 if (fd == -1) {
2103 in = in_tot;
2104 out = out_tot;
2105 } else
2106 #endif
2107 {
2108 /* read the last 4 bytes - this is the uncompressed size */
2109 rv = lseek(fd, (off_t)(-8), SEEK_END);
2110 if (rv != -1) {
2111 unsigned char buf[8];
2112 uint32_t usize;
2113
2114 rv = read(fd, (char *)buf, sizeof(buf));
2115 if (rv == -1)
2116 maybe_warn("read of uncompressed size");
2117 else if (rv != sizeof(buf))
2118 maybe_warnx("read of uncompressed size");
2119
2120 else {
2121 usize = buf[4] | buf[5] << 8 |
2122 buf[6] << 16 | buf[7] << 24;
2123 in = (off_t)usize;
2124 #ifndef SMALL
2125 crc = buf[0] | buf[1] << 8 |
2126 buf[2] << 16 | buf[3] << 24;
2127 #endif
2128 }
2129 }
2130 }
2131
2132 #ifndef SMALL
2133 if (vflag && fd == -1)
2134 printf(" ");
2135 else if (vflag) {
2136 char *date = ctime(&ts);
2137
2138 /* skip the day, 1/100th second, and year */
2139 date += 4;
2140 date[12] = 0;
2141 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date);
2142 }
2143 in_tot += in;
2144 out_tot += out;
2145 #endif
2146 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in);
2147 print_ratio(in, out, stdout);
2148 printf(" %s\n", outfile);
2149 }
2150
2151 /* display the usage of NetBSD gzip */
2152 static void
2153 usage(void)
2154 {
2155
2156 fprintf(stderr, "%s\n", gzip_version);
2157 fprintf(stderr,
2158 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n"
2159 #ifndef SMALL
2160 " -1 --fast fastest (worst) compression\n"
2161 " -2 .. -8 set compression level\n"
2162 " -9 --best best (slowest) compression\n"
2163 " -c --stdout write to stdout, keep original files\n"
2164 " --to-stdout\n"
2165 " -d --decompress uncompress files\n"
2166 " --uncompress\n"
2167 " -f --force force overwriting & compress links\n"
2168 " -h --help display this help\n"
2169 " -k --keep don't delete input files during operation\n"
2170 " -l --list list compressed file contents\n"
2171 " -N --name save or restore original file name and time stamp\n"
2172 " -n --no-name don't save original file name or time stamp\n"
2173 " -q --quiet output no warnings\n"
2174 " -r --recursive recursively compress files in directories\n"
2175 " -S .suf use suffix .suf instead of .gz\n"
2176 " --suffix .suf\n"
2177 " -t --test test compressed file\n"
2178 " -V --version display program version\n"
2179 " -v --verbose print extra statistics\n",
2180 #else
2181 ,
2182 #endif
2183 getprogname());
2184 exit(0);
2185 }
2186
2187 /* display the version of NetBSD gzip */
2188 static void
2189 display_version(void)
2190 {
2191
2192 fprintf(stderr, "%s\n", gzip_version);
2193 exit(0);
2194 }
2195
2196 #ifndef NO_BZIP2_SUPPORT
2197 #include "unbzip2.c"
2198 #endif
2199 #ifndef NO_COMPRESS_SUPPORT
2200 #include "zuncompress.c"
2201 #endif
2202 #ifndef NO_PACK_SUPPORT
2203 #include "unpack.c"
2204 #endif
2205 #ifndef NO_XZ_SUPPORT
2206 #include "unxz.c"
2207 #endif
2208
2209 static ssize_t
2210 read_retry(int fd, void *buf, size_t sz)
2211 {
2212 char *cp = buf;
2213 size_t left = MIN(sz, (size_t) SSIZE_MAX);
2214
2215 while (left > 0) {
2216 ssize_t ret;
2217
2218 ret = read(fd, cp, left);
2219 if (ret == -1) {
2220 return ret;
2221 } else if (ret == 0) {
2222 break; /* EOF */
2223 }
2224 cp += ret;
2225 left -= ret;
2226 }
2227
2228 return sz - left;
2229 }
2230
2231 static ssize_t
2232 write_retry(int fd, const void *buf, size_t sz)
2233 {
2234 const char *cp = buf;
2235 size_t left = MIN(sz, (size_t) SSIZE_MAX);
2236
2237 while (left > 0) {
2238 ssize_t ret;
2239
2240 ret = write(fd, cp, left);
2241 if (ret == -1) {
2242 return ret;
2243 } else if (ret == 0) {
2244 abort(); /* Can't happen */
2245 }
2246 cp += ret;
2247 left -= ret;
2248 }
2249
2250 return sz - left;
2251 }
2252