gzip.c revision 1.116 1 /* $NetBSD: gzip.c,v 1.116 2018/10/27 11:39:12 skrll 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 1.116 2018/10/27 11:39:12 skrll 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 #ifndef NO_LZ_SUPPORT
89 FT_LZ,
90 #endif
91 FT_LAST,
92 FT_UNKNOWN
93 };
94
95 #ifndef NO_BZIP2_SUPPORT
96 #include <bzlib.h>
97
98 #define BZ2_SUFFIX ".bz2"
99 #define BZIP2_MAGIC "\102\132\150"
100 #endif
101
102 #ifndef NO_COMPRESS_SUPPORT
103 #define Z_SUFFIX ".Z"
104 #define Z_MAGIC "\037\235"
105 #endif
106
107 #ifndef NO_PACK_SUPPORT
108 #define PACK_MAGIC "\037\036"
109 #endif
110
111 #ifndef NO_XZ_SUPPORT
112 #include <lzma.h>
113 #define XZ_SUFFIX ".xz"
114 #define XZ_MAGIC "\3757zXZ"
115 #endif
116
117 #ifndef NO_LZ_SUPPORT
118 #define LZ_SUFFIX ".lz"
119 #define LZ_MAGIC "LZIP"
120 #endif
121
122 #define GZ_SUFFIX ".gz"
123
124 #define BUFLEN (64 * 1024)
125
126 #define GZIP_MAGIC0 0x1F
127 #define GZIP_MAGIC1 0x8B
128 #define GZIP_OMAGIC1 0x9E
129
130 #define GZIP_TIMESTAMP (off_t)4
131 #define GZIP_ORIGNAME (off_t)10
132
133 #define HEAD_CRC 0x02
134 #define EXTRA_FIELD 0x04
135 #define ORIG_NAME 0x08
136 #define COMMENT 0x10
137
138 #define OS_CODE 3 /* Unix */
139
140 typedef struct {
141 const char *zipped;
142 int ziplen;
143 const char *normal; /* for unzip - must not be longer than zipped */
144 } suffixes_t;
145 static suffixes_t suffixes[] = {
146 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N}
147 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */
148 #ifndef SMALL
149 SUFFIX(GZ_SUFFIX, ""),
150 SUFFIX(".z", ""),
151 SUFFIX("-gz", ""),
152 SUFFIX("-z", ""),
153 SUFFIX("_z", ""),
154 SUFFIX(".taz", ".tar"),
155 SUFFIX(".tgz", ".tar"),
156 #ifndef NO_BZIP2_SUPPORT
157 SUFFIX(BZ2_SUFFIX, ""),
158 #endif
159 #ifndef NO_COMPRESS_SUPPORT
160 SUFFIX(Z_SUFFIX, ""),
161 #endif
162 #ifndef NO_XZ_SUPPORT
163 SUFFIX(XZ_SUFFIX, ""),
164 #endif
165 #ifndef NO_LZ_SUPPORT
166 SUFFIX(LZ_SUFFIX, ""),
167 #endif
168 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */
169 #endif /* SMALL */
170 #undef SUFFIX
171 };
172 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0])
173 #define SUFFIX_MAXLEN 30
174
175 static const char gzip_version[] = "NetBSD gzip 20170803";
176
177 static int cflag; /* stdout mode */
178 static int dflag; /* decompress mode */
179 static int lflag; /* list mode */
180 static int numflag = 6; /* gzip -1..-9 value */
181
182 #ifndef SMALL
183 static int fflag; /* force mode */
184 static int kflag; /* don't delete input files */
185 static int nflag; /* don't save name/timestamp */
186 static int Nflag; /* don't restore name/timestamp */
187 static int qflag; /* quiet mode */
188 static int rflag; /* recursive mode */
189 static int tflag; /* test */
190 static int vflag; /* verbose mode */
191 static sig_atomic_t print_info = 0;
192 #else
193 #define qflag 0
194 #define tflag 0
195 #endif
196
197 static int exit_value = 0; /* exit value */
198
199 static const char *infile; /* name of file coming in */
200
201 static void maybe_err(const char *fmt, ...) __printflike(1, 2) __dead;
202 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \
203 !defined(NO_XZ_SUPPORT)
204 static void maybe_errx(const char *fmt, ...) __printflike(1, 2) __dead;
205 #endif
206 static void maybe_warn(const char *fmt, ...) __printflike(1, 2);
207 static void maybe_warnx(const char *fmt, ...) __printflike(1, 2);
208 static enum filetype file_gettype(u_char *);
209 #ifdef SMALL
210 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz)
211 #endif
212 static off_t gz_compress(int, int, off_t *, const char *, uint32_t);
213 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *);
214 static off_t file_compress(char *, char *, size_t);
215 static off_t file_uncompress(char *, char *, size_t);
216 static void handle_pathname(char *);
217 static void handle_file(char *, struct stat *);
218 static void handle_stdin(void);
219 static void handle_stdout(void);
220 static void print_ratio(off_t, off_t, FILE *);
221 static void print_list(int fd, off_t, const char *, time_t);
222 __dead static void usage(void);
223 __dead static void display_version(void);
224 static const suffixes_t *check_suffix(char *, int);
225 static ssize_t read_retry(int, void *, size_t);
226 static ssize_t write_retry(int, const void *, size_t);
227 static void print_list_out(off_t, off_t, const char*);
228
229 #ifdef SMALL
230 #define infile_set(f,t) infile_set(f)
231 #endif
232 static void infile_set(const char *newinfile, off_t total);
233
234 #ifdef SMALL
235 #define unlink_input(f, sb) unlink(f)
236 #define check_siginfo() /* nothing */
237 #define setup_signals() /* nothing */
238 #define infile_newdata(t) /* nothing */
239 #else
240 static off_t infile_total; /* total expected to read/write */
241 static off_t infile_current; /* current read/write */
242
243 static void check_siginfo(void);
244 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd);
245 static void prepend_gzip(char *, int *, char ***);
246 static void handle_dir(char *);
247 static void print_verbage(const char *, const char *, off_t, off_t);
248 static void print_test(const char *, int);
249 static void copymodes(int fd, const struct stat *, const char *file);
250 static int check_outfile(const char *outfile);
251 static void setup_signals(void);
252 static void infile_newdata(size_t newdata);
253 static void infile_clear(void);
254 #endif
255
256 #ifndef NO_BZIP2_SUPPORT
257 static off_t unbzip2(int, int, char *, size_t, off_t *);
258 #endif
259
260 #ifndef NO_COMPRESS_SUPPORT
261 static FILE *zdopen(int);
262 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *);
263 #endif
264
265 #ifndef NO_PACK_SUPPORT
266 static off_t unpack(int, int, char *, size_t, off_t *);
267 #endif
268
269 #ifndef NO_XZ_SUPPORT
270 static off_t unxz(int, int, char *, size_t, off_t *);
271 static off_t unxz_len(int);
272 #endif
273
274 #ifndef NO_LZ_SUPPORT
275 static off_t unlz(int, int, char *, size_t, off_t *);
276 #endif
277
278 #ifdef SMALL
279 #define getopt_long(a,b,c,d,e) getopt(a,b,c)
280 #else
281 static const struct option longopts[] = {
282 { "stdout", no_argument, 0, 'c' },
283 { "to-stdout", no_argument, 0, 'c' },
284 { "decompress", no_argument, 0, 'd' },
285 { "uncompress", no_argument, 0, 'd' },
286 { "force", no_argument, 0, 'f' },
287 { "help", no_argument, 0, 'h' },
288 { "keep", no_argument, 0, 'k' },
289 { "list", no_argument, 0, 'l' },
290 { "no-name", no_argument, 0, 'n' },
291 { "name", no_argument, 0, 'N' },
292 { "quiet", no_argument, 0, 'q' },
293 { "recursive", no_argument, 0, 'r' },
294 { "suffix", required_argument, 0, 'S' },
295 { "test", no_argument, 0, 't' },
296 { "verbose", no_argument, 0, 'v' },
297 { "version", no_argument, 0, 'V' },
298 { "fast", no_argument, 0, '1' },
299 { "best", no_argument, 0, '9' },
300 #if 0
301 /*
302 * This is what else GNU gzip implements. --ascii isn't useful
303 * on NetBSD, and I don't care to have a --license.
304 */
305 { "ascii", no_argument, 0, 'a' },
306 { "license", no_argument, 0, 'L' },
307 #endif
308 { NULL, no_argument, 0, 0 },
309 };
310 #endif
311
312 int
313 main(int argc, char **argv)
314 {
315 const char *progname = getprogname();
316 #ifndef SMALL
317 char *gzip;
318 int len;
319 #endif
320 int ch;
321
322 setup_signals();
323
324 #ifndef SMALL
325 if ((gzip = getenv("GZIP")) != NULL)
326 prepend_gzip(gzip, &argc, &argv);
327 #endif
328
329 /*
330 * XXX
331 * handle being called `gunzip', `zcat' and `gzcat'
332 */
333 if (strcmp(progname, "gunzip") == 0)
334 dflag = 1;
335 else if (strcmp(progname, "zcat") == 0 ||
336 strcmp(progname, "gzcat") == 0)
337 dflag = cflag = 1;
338
339 #ifdef SMALL
340 #define OPT_LIST "123456789cdhlV"
341 #else
342 #define OPT_LIST "123456789cdfhklNnqrS:tVv"
343 #endif
344
345 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) {
346 switch (ch) {
347 case '1': case '2': case '3':
348 case '4': case '5': case '6':
349 case '7': case '8': case '9':
350 numflag = ch - '0';
351 break;
352 case 'c':
353 cflag = 1;
354 break;
355 case 'd':
356 dflag = 1;
357 break;
358 case 'l':
359 lflag = 1;
360 dflag = 1;
361 break;
362 case 'V':
363 display_version();
364 /* NOTREACHED */
365 #ifndef SMALL
366 case 'f':
367 fflag = 1;
368 break;
369 case 'k':
370 kflag = 1;
371 break;
372 case 'N':
373 nflag = 0;
374 Nflag = 1;
375 break;
376 case 'n':
377 nflag = 1;
378 Nflag = 0;
379 break;
380 case 'q':
381 qflag = 1;
382 break;
383 case 'r':
384 rflag = 1;
385 break;
386 case 'S':
387 len = strlen(optarg);
388 if (len != 0) {
389 if (len > SUFFIX_MAXLEN)
390 errx(1, "incorrect suffix: '%s'", optarg);
391 suffixes[0].zipped = optarg;
392 suffixes[0].ziplen = len;
393 } else {
394 suffixes[NUM_SUFFIXES - 1].zipped = "";
395 suffixes[NUM_SUFFIXES - 1].ziplen = 0;
396 }
397 break;
398 case 't':
399 cflag = 1;
400 tflag = 1;
401 dflag = 1;
402 break;
403 case 'v':
404 vflag = 1;
405 break;
406 #endif
407 default:
408 usage();
409 /* NOTREACHED */
410 }
411 }
412 argv += optind;
413 argc -= optind;
414
415 if (argc == 0) {
416 if (dflag) /* stdin mode */
417 handle_stdin();
418 else /* stdout mode */
419 handle_stdout();
420 } else {
421 do {
422 handle_pathname(argv[0]);
423 } while (*++argv);
424 }
425 #ifndef SMALL
426 if (qflag == 0 && lflag && argc > 1)
427 print_list(-1, 0, "(totals)", 0);
428 #endif
429 exit(exit_value);
430 }
431
432 /* maybe print a warning */
433 void
434 maybe_warn(const char *fmt, ...)
435 {
436 va_list ap;
437
438 if (qflag == 0) {
439 va_start(ap, fmt);
440 vwarn(fmt, ap);
441 va_end(ap);
442 }
443 if (exit_value == 0)
444 exit_value = 1;
445 }
446
447 /* ... without an errno. */
448 void
449 maybe_warnx(const char *fmt, ...)
450 {
451 va_list ap;
452
453 if (qflag == 0) {
454 va_start(ap, fmt);
455 vwarnx(fmt, ap);
456 va_end(ap);
457 }
458 if (exit_value == 0)
459 exit_value = 1;
460 }
461
462 /* maybe print an error */
463 void
464 maybe_err(const char *fmt, ...)
465 {
466 va_list ap;
467
468 if (qflag == 0) {
469 va_start(ap, fmt);
470 vwarn(fmt, ap);
471 va_end(ap);
472 }
473 exit(2);
474 }
475
476 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \
477 !defined(NO_XZ_SUPPORT)
478 /* ... without an errno. */
479 void
480 maybe_errx(const char *fmt, ...)
481 {
482 va_list ap;
483
484 if (qflag == 0) {
485 va_start(ap, fmt);
486 vwarnx(fmt, ap);
487 va_end(ap);
488 }
489 exit(2);
490 }
491 #endif
492
493 #ifndef SMALL
494 /* split up $GZIP and prepend it to the argument list */
495 static void
496 prepend_gzip(char *gzip, int *argc, char ***argv)
497 {
498 char *s, **nargv, **ac;
499 int nenvarg = 0, i;
500
501 /* scan how many arguments there are */
502 for (s = gzip;;) {
503 while (*s == ' ' || *s == '\t')
504 s++;
505 if (*s == 0)
506 goto count_done;
507 nenvarg++;
508 while (*s != ' ' && *s != '\t')
509 if (*s++ == 0)
510 goto count_done;
511 }
512 count_done:
513 /* punt early */
514 if (nenvarg == 0)
515 return;
516
517 *argc += nenvarg;
518 ac = *argv;
519
520 nargv = (char **)malloc((*argc + 1) * sizeof(char *));
521 if (nargv == NULL)
522 maybe_err("malloc");
523
524 /* stash this away */
525 *argv = nargv;
526
527 /* copy the program name first */
528 i = 0;
529 nargv[i++] = *(ac++);
530
531 /* take a copy of $GZIP and add it to the array */
532 s = strdup(gzip);
533 if (s == NULL)
534 maybe_err("strdup");
535 for (;;) {
536 /* Skip whitespaces. */
537 while (*s == ' ' || *s == '\t')
538 s++;
539 if (*s == 0)
540 goto copy_done;
541 nargv[i++] = s;
542 /* Find the end of this argument. */
543 while (*s != ' ' && *s != '\t')
544 if (*s++ == 0)
545 /* Argument followed by NUL. */
546 goto copy_done;
547 /* Terminate by overwriting ' ' or '\t' with NUL. */
548 *s++ = 0;
549 }
550 copy_done:
551
552 /* copy the original arguments and a NULL */
553 while (*ac)
554 nargv[i++] = *(ac++);
555 nargv[i] = NULL;
556 }
557 #endif
558
559 /* compress input to output. Return bytes read, -1 on error */
560 static off_t
561 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime)
562 {
563 z_stream z;
564 char *outbufp, *inbufp;
565 off_t in_tot = 0, out_tot = 0;
566 ssize_t in_size;
567 int i, error;
568 uLong crc;
569 #ifdef SMALL
570 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0,
571 0, 0, 0, 0,
572 0, OS_CODE };
573 #endif
574
575 outbufp = malloc(BUFLEN);
576 inbufp = malloc(BUFLEN);
577 if (outbufp == NULL || inbufp == NULL) {
578 maybe_err("malloc failed");
579 goto out;
580 }
581
582 memset(&z, 0, sizeof z);
583 z.zalloc = Z_NULL;
584 z.zfree = Z_NULL;
585 z.opaque = 0;
586
587 #ifdef SMALL
588 memcpy(outbufp, header, sizeof header);
589 i = sizeof header;
590 #else
591 if (nflag != 0) {
592 mtime = 0;
593 origname = "";
594 }
595
596 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s",
597 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED,
598 *origname ? ORIG_NAME : 0,
599 mtime & 0xff,
600 (mtime >> 8) & 0xff,
601 (mtime >> 16) & 0xff,
602 (mtime >> 24) & 0xff,
603 numflag == 1 ? 4 : numflag == 9 ? 2 : 0,
604 OS_CODE, origname);
605 if (i >= BUFLEN)
606 /* this need PATH_MAX > BUFLEN ... */
607 maybe_err("snprintf");
608 if (*origname)
609 i++;
610 #endif
611
612 z.next_out = (unsigned char *)outbufp + i;
613 z.avail_out = BUFLEN - i;
614
615 error = deflateInit2(&z, numflag, Z_DEFLATED,
616 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY);
617 if (error != Z_OK) {
618 maybe_warnx("deflateInit2 failed");
619 in_tot = -1;
620 goto out;
621 }
622
623 crc = crc32(0L, Z_NULL, 0);
624 for (;;) {
625 if (z.avail_out == 0) {
626 if (write_retry(out, outbufp, BUFLEN) != BUFLEN) {
627 maybe_warn("write");
628 out_tot = -1;
629 goto out;
630 }
631
632 out_tot += BUFLEN;
633 z.next_out = (unsigned char *)outbufp;
634 z.avail_out = BUFLEN;
635 }
636
637 if (z.avail_in == 0) {
638 in_size = read(in, inbufp, BUFLEN);
639 if (in_size < 0) {
640 maybe_warn("read");
641 in_tot = -1;
642 goto out;
643 }
644 if (in_size == 0)
645 break;
646 infile_newdata(in_size);
647
648 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size);
649 in_tot += in_size;
650 z.next_in = (unsigned char *)inbufp;
651 z.avail_in = in_size;
652 }
653
654 error = deflate(&z, Z_NO_FLUSH);
655 if (error != Z_OK && error != Z_STREAM_END) {
656 maybe_warnx("deflate failed");
657 in_tot = -1;
658 goto out;
659 }
660 }
661
662 /* clean up */
663 for (;;) {
664 size_t len;
665 ssize_t w;
666
667 error = deflate(&z, Z_FINISH);
668 if (error != Z_OK && error != Z_STREAM_END) {
669 maybe_warnx("deflate failed");
670 in_tot = -1;
671 goto out;
672 }
673
674 len = (char *)z.next_out - outbufp;
675
676 w = write_retry(out, outbufp, len);
677 if (w == -1 || (size_t)w != len) {
678 maybe_warn("write");
679 out_tot = -1;
680 goto out;
681 }
682 out_tot += len;
683 z.next_out = (unsigned char *)outbufp;
684 z.avail_out = BUFLEN;
685
686 if (error == Z_STREAM_END)
687 break;
688 }
689
690 if (deflateEnd(&z) != Z_OK) {
691 maybe_warnx("deflateEnd failed");
692 in_tot = -1;
693 goto out;
694 }
695
696 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c",
697 (int)crc & 0xff,
698 (int)(crc >> 8) & 0xff,
699 (int)(crc >> 16) & 0xff,
700 (int)(crc >> 24) & 0xff,
701 (int)in_tot & 0xff,
702 (int)(in_tot >> 8) & 0xff,
703 (int)(in_tot >> 16) & 0xff,
704 (int)(in_tot >> 24) & 0xff);
705 if (i != 8)
706 maybe_err("snprintf");
707 #if 0
708 if (in_tot > 0xffffffff)
709 maybe_warn("input file size >= 4GB cannot be saved");
710 #endif
711 if (write_retry(out, outbufp, i) != i) {
712 maybe_warn("write");
713 in_tot = -1;
714 } else
715 out_tot += i;
716
717 out:
718 if (inbufp != NULL)
719 free(inbufp);
720 if (outbufp != NULL)
721 free(outbufp);
722 if (gsizep)
723 *gsizep = out_tot;
724 return in_tot;
725 }
726
727 /*
728 * uncompress input to output then close the input. return the
729 * uncompressed size written, and put the compressed sized read
730 * into `*gsizep'.
731 */
732 static off_t
733 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep,
734 const char *filename)
735 {
736 z_stream z;
737 char *outbufp, *inbufp;
738 off_t out_tot = -1, in_tot = 0;
739 uint32_t out_sub_tot = 0;
740 enum {
741 GZSTATE_MAGIC0,
742 GZSTATE_MAGIC1,
743 GZSTATE_METHOD,
744 GZSTATE_FLAGS,
745 GZSTATE_SKIPPING,
746 GZSTATE_EXTRA,
747 GZSTATE_EXTRA2,
748 GZSTATE_EXTRA3,
749 GZSTATE_ORIGNAME,
750 GZSTATE_COMMENT,
751 GZSTATE_HEAD_CRC1,
752 GZSTATE_HEAD_CRC2,
753 GZSTATE_INIT,
754 GZSTATE_READ,
755 GZSTATE_CRC,
756 GZSTATE_LEN,
757 } state = GZSTATE_MAGIC0;
758 int flags = 0, skip_count = 0;
759 int error = Z_STREAM_ERROR, done_reading = 0;
760 uLong crc = 0;
761 ssize_t wr;
762 int needmore = 0;
763
764 #define ADVANCE() { z.next_in++; z.avail_in--; }
765
766 if ((outbufp = malloc(BUFLEN)) == NULL) {
767 maybe_err("malloc failed");
768 goto out2;
769 }
770 if ((inbufp = malloc(BUFLEN)) == NULL) {
771 maybe_err("malloc failed");
772 goto out1;
773 }
774
775 memset(&z, 0, sizeof z);
776 z.avail_in = prelen;
777 z.next_in = (unsigned char *)pre;
778 z.avail_out = BUFLEN;
779 z.next_out = (unsigned char *)outbufp;
780 z.zalloc = NULL;
781 z.zfree = NULL;
782 z.opaque = 0;
783
784 in_tot = prelen;
785 out_tot = 0;
786
787 for (;;) {
788 check_siginfo();
789 if ((z.avail_in == 0 || needmore) && done_reading == 0) {
790 ssize_t in_size;
791
792 if (z.avail_in > 0) {
793 memmove(inbufp, z.next_in, z.avail_in);
794 }
795 z.next_in = (unsigned char *)inbufp;
796 in_size = read(in, z.next_in + z.avail_in,
797 BUFLEN - z.avail_in);
798
799 if (in_size == -1) {
800 maybe_warn("failed to read stdin");
801 goto stop_and_fail;
802 } else if (in_size == 0) {
803 done_reading = 1;
804 }
805 infile_newdata(in_size);
806
807 z.avail_in += in_size;
808 needmore = 0;
809
810 in_tot += in_size;
811 }
812 if (z.avail_in == 0) {
813 if (done_reading && state != GZSTATE_MAGIC0) {
814 maybe_warnx("%s: unexpected end of file",
815 filename);
816 goto stop_and_fail;
817 }
818 goto stop;
819 }
820 switch (state) {
821 case GZSTATE_MAGIC0:
822 if (*z.next_in != GZIP_MAGIC0) {
823 if (in_tot > 0) {
824 maybe_warnx("%s: trailing garbage "
825 "ignored", filename);
826 goto stop;
827 }
828 maybe_warnx("input not gziped (MAGIC0)");
829 exit_value = 2;
830 goto stop_and_fail;
831 }
832 ADVANCE();
833 state++;
834 out_sub_tot = 0;
835 crc = crc32(0L, Z_NULL, 0);
836 break;
837
838 case GZSTATE_MAGIC1:
839 if (*z.next_in != GZIP_MAGIC1 &&
840 *z.next_in != GZIP_OMAGIC1) {
841 maybe_warnx("input not gziped (MAGIC1)");
842 goto stop_and_fail;
843 }
844 ADVANCE();
845 state++;
846 break;
847
848 case GZSTATE_METHOD:
849 if (*z.next_in != Z_DEFLATED) {
850 maybe_warnx("unknown compression method");
851 goto stop_and_fail;
852 }
853 ADVANCE();
854 state++;
855 break;
856
857 case GZSTATE_FLAGS:
858 flags = *z.next_in;
859 ADVANCE();
860 skip_count = 6;
861 state++;
862 break;
863
864 case GZSTATE_SKIPPING:
865 if (skip_count > 0) {
866 skip_count--;
867 ADVANCE();
868 } else
869 state++;
870 break;
871
872 case GZSTATE_EXTRA:
873 if ((flags & EXTRA_FIELD) == 0) {
874 state = GZSTATE_ORIGNAME;
875 break;
876 }
877 skip_count = *z.next_in;
878 ADVANCE();
879 state++;
880 break;
881
882 case GZSTATE_EXTRA2:
883 skip_count |= ((*z.next_in) << 8);
884 ADVANCE();
885 state++;
886 break;
887
888 case GZSTATE_EXTRA3:
889 if (skip_count > 0) {
890 skip_count--;
891 ADVANCE();
892 } else
893 state++;
894 break;
895
896 case GZSTATE_ORIGNAME:
897 if ((flags & ORIG_NAME) == 0) {
898 state++;
899 break;
900 }
901 if (*z.next_in == 0)
902 state++;
903 ADVANCE();
904 break;
905
906 case GZSTATE_COMMENT:
907 if ((flags & COMMENT) == 0) {
908 state++;
909 break;
910 }
911 if (*z.next_in == 0)
912 state++;
913 ADVANCE();
914 break;
915
916 case GZSTATE_HEAD_CRC1:
917 if (flags & HEAD_CRC)
918 skip_count = 2;
919 else
920 skip_count = 0;
921 state++;
922 break;
923
924 case GZSTATE_HEAD_CRC2:
925 if (skip_count > 0) {
926 skip_count--;
927 ADVANCE();
928 } else
929 state++;
930 break;
931
932 case GZSTATE_INIT:
933 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) {
934 maybe_warnx("failed to inflateInit");
935 goto stop_and_fail;
936 }
937 state++;
938 break;
939
940 case GZSTATE_READ:
941 error = inflate(&z, Z_FINISH);
942 switch (error) {
943 /* Z_BUF_ERROR goes with Z_FINISH... */
944 case Z_BUF_ERROR:
945 if (z.avail_out > 0 && !done_reading)
946 continue;
947
948 case Z_STREAM_END:
949 case Z_OK:
950 break;
951
952 case Z_NEED_DICT:
953 maybe_warnx("Z_NEED_DICT error");
954 goto stop_and_fail;
955 case Z_DATA_ERROR:
956 maybe_warnx("data stream error");
957 goto stop_and_fail;
958 case Z_STREAM_ERROR:
959 maybe_warnx("internal stream error");
960 goto stop_and_fail;
961 case Z_MEM_ERROR:
962 maybe_warnx("memory allocation error");
963 goto stop_and_fail;
964
965 default:
966 maybe_warn("unknown error from inflate(): %d",
967 error);
968 }
969 wr = BUFLEN - z.avail_out;
970
971 if (wr != 0) {
972 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr);
973 if (
974 #ifndef SMALL
975 /* don't write anything with -t */
976 tflag == 0 &&
977 #endif
978 write_retry(out, outbufp, wr) != wr) {
979 maybe_warn("error writing to output");
980 goto stop_and_fail;
981 }
982
983 out_tot += wr;
984 out_sub_tot += wr;
985 }
986
987 if (error == Z_STREAM_END) {
988 inflateEnd(&z);
989 state++;
990 }
991
992 z.next_out = (unsigned char *)outbufp;
993 z.avail_out = BUFLEN;
994
995 break;
996 case GZSTATE_CRC:
997 {
998 uLong origcrc;
999
1000 if (z.avail_in < 4) {
1001 if (!done_reading) {
1002 needmore = 1;
1003 continue;
1004 }
1005 maybe_warnx("truncated input");
1006 goto stop_and_fail;
1007 }
1008 origcrc = ((unsigned)z.next_in[0] & 0xff) |
1009 ((unsigned)z.next_in[1] & 0xff) << 8 |
1010 ((unsigned)z.next_in[2] & 0xff) << 16 |
1011 ((unsigned)z.next_in[3] & 0xff) << 24;
1012 if (origcrc != crc) {
1013 maybe_warnx("invalid compressed"
1014 " data--crc error");
1015 goto stop_and_fail;
1016 }
1017 }
1018
1019 z.avail_in -= 4;
1020 z.next_in += 4;
1021
1022 if (!z.avail_in && done_reading) {
1023 goto stop;
1024 }
1025 state++;
1026 break;
1027 case GZSTATE_LEN:
1028 {
1029 uLong origlen;
1030
1031 if (z.avail_in < 4) {
1032 if (!done_reading) {
1033 needmore = 1;
1034 continue;
1035 }
1036 maybe_warnx("truncated input");
1037 goto stop_and_fail;
1038 }
1039 origlen = ((unsigned)z.next_in[0] & 0xff) |
1040 ((unsigned)z.next_in[1] & 0xff) << 8 |
1041 ((unsigned)z.next_in[2] & 0xff) << 16 |
1042 ((unsigned)z.next_in[3] & 0xff) << 24;
1043
1044 if (origlen != out_sub_tot) {
1045 maybe_warnx("invalid compressed"
1046 " data--length error");
1047 goto stop_and_fail;
1048 }
1049 }
1050
1051 z.avail_in -= 4;
1052 z.next_in += 4;
1053
1054 if (error < 0) {
1055 maybe_warnx("decompression error");
1056 goto stop_and_fail;
1057 }
1058 state = GZSTATE_MAGIC0;
1059 break;
1060 }
1061 continue;
1062 stop_and_fail:
1063 out_tot = -1;
1064 stop:
1065 break;
1066 }
1067 if (state > GZSTATE_INIT)
1068 inflateEnd(&z);
1069
1070 free(inbufp);
1071 out1:
1072 free(outbufp);
1073 out2:
1074 if (gsizep)
1075 *gsizep = in_tot;
1076 return (out_tot);
1077 }
1078
1079 #ifndef SMALL
1080 /*
1081 * set the owner, mode, flags & utimes using the given file descriptor.
1082 * file is only used in possible warning messages.
1083 */
1084 static void
1085 copymodes(int fd, const struct stat *sbp, const char *file)
1086 {
1087 struct timeval times[2];
1088 struct stat sb;
1089
1090 /*
1091 * If we have no info on the input, give this file some
1092 * default values and return..
1093 */
1094 if (sbp == NULL) {
1095 mode_t mask = umask(022);
1096
1097 (void)fchmod(fd, DEFFILEMODE & ~mask);
1098 (void)umask(mask);
1099 return;
1100 }
1101 sb = *sbp;
1102
1103 /* if the chown fails, remove set-id bits as-per compress(1) */
1104 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) {
1105 if (errno != EPERM)
1106 maybe_warn("couldn't fchown: %s", file);
1107 sb.st_mode &= ~(S_ISUID|S_ISGID);
1108 }
1109
1110 /* we only allow set-id and the 9 normal permission bits */
1111 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO;
1112 if (fchmod(fd, sb.st_mode) < 0)
1113 maybe_warn("couldn't fchmod: %s", file);
1114
1115 /* only try flags if they exist already */
1116 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0)
1117 maybe_warn("couldn't fchflags: %s", file);
1118
1119 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec);
1120 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec);
1121 if (futimes(fd, times) < 0)
1122 maybe_warn("couldn't utimes: %s", file);
1123 }
1124 #endif
1125
1126 /* what sort of file is this? */
1127 static enum filetype
1128 file_gettype(u_char *buf)
1129 {
1130
1131 if (buf[0] == GZIP_MAGIC0 &&
1132 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1))
1133 return FT_GZIP;
1134 else
1135 #ifndef NO_BZIP2_SUPPORT
1136 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 &&
1137 buf[3] >= '0' && buf[3] <= '9')
1138 return FT_BZIP2;
1139 else
1140 #endif
1141 #ifndef NO_COMPRESS_SUPPORT
1142 if (memcmp(buf, Z_MAGIC, 2) == 0)
1143 return FT_Z;
1144 else
1145 #endif
1146 #ifndef NO_PACK_SUPPORT
1147 if (memcmp(buf, PACK_MAGIC, 2) == 0)
1148 return FT_PACK;
1149 else
1150 #endif
1151 #ifndef NO_XZ_SUPPORT
1152 if (memcmp(buf, XZ_MAGIC, 4) == 0) /* XXX: We only have 4 bytes */
1153 return FT_XZ;
1154 else
1155 #endif
1156 #ifndef NO_LZ_SUPPORT
1157 if (memcmp(buf, LZ_MAGIC, 4) == 0)
1158 return FT_LZ;
1159 else
1160 #endif
1161 return FT_UNKNOWN;
1162 }
1163
1164 #ifndef SMALL
1165 /* check the outfile is OK. */
1166 static int
1167 check_outfile(const char *outfile)
1168 {
1169 struct stat sb;
1170 int ok = 1;
1171
1172 if (lflag == 0 && stat(outfile, &sb) == 0) {
1173 if (fflag)
1174 unlink(outfile);
1175 else if (isatty(STDIN_FILENO)) {
1176 char ans[10] = { 'n', '\0' }; /* default */
1177
1178 fprintf(stderr, "%s already exists -- do you wish to "
1179 "overwrite (y or n)? " , outfile);
1180 (void)fgets(ans, sizeof(ans) - 1, stdin);
1181 if (ans[0] != 'y' && ans[0] != 'Y') {
1182 fprintf(stderr, "\tnot overwriting\n");
1183 ok = 0;
1184 } else
1185 unlink(outfile);
1186 } else {
1187 maybe_warnx("%s already exists -- skipping", outfile);
1188 ok = 0;
1189 }
1190 }
1191 return ok;
1192 }
1193
1194 static void
1195 unlink_input(const char *file, const struct stat *sb)
1196 {
1197 struct stat nsb;
1198
1199 if (kflag)
1200 return;
1201 if (stat(file, &nsb) != 0)
1202 /* Must be gone already */
1203 return;
1204 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino)
1205 /* Definitely a different file */
1206 return;
1207 unlink(file);
1208 }
1209
1210 static void
1211 got_siginfo(int signo)
1212 {
1213
1214 print_info = 1;
1215 }
1216
1217 static void
1218 setup_signals(void)
1219 {
1220
1221 signal(SIGINFO, got_siginfo);
1222 }
1223
1224 static void
1225 infile_newdata(size_t newdata)
1226 {
1227
1228 infile_current += newdata;
1229 }
1230 #endif
1231
1232 static void
1233 infile_set(const char *newinfile, off_t total)
1234 {
1235
1236 if (newinfile)
1237 infile = newinfile;
1238 #ifndef SMALL
1239 infile_total = total;
1240 #endif
1241 }
1242
1243 static void
1244 infile_clear(void)
1245 {
1246
1247 infile = NULL;
1248 #ifndef SMALL
1249 infile_total = infile_current = 0;
1250 #endif
1251 }
1252
1253 static const suffixes_t *
1254 check_suffix(char *file, int xlate)
1255 {
1256 const suffixes_t *s;
1257 int len = strlen(file);
1258 char *sp;
1259
1260 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) {
1261 /* if it doesn't fit in "a.suf", don't bother */
1262 if (s->ziplen >= len)
1263 continue;
1264 sp = file + len - s->ziplen;
1265 if (strcmp(s->zipped, sp) != 0)
1266 continue;
1267 if (xlate)
1268 strcpy(sp, s->normal);
1269 return s;
1270 }
1271 return NULL;
1272 }
1273
1274 /*
1275 * compress the given file: create a corresponding .gz file and remove the
1276 * original.
1277 */
1278 static off_t
1279 file_compress(char *file, char *outfile, size_t outsize)
1280 {
1281 int in;
1282 int out;
1283 off_t size, in_size;
1284 #ifndef SMALL
1285 struct stat isb, osb;
1286 const suffixes_t *suff;
1287 #endif
1288
1289 in = open(file, O_RDONLY);
1290 if (in == -1) {
1291 maybe_warn("can't open %s", file);
1292 return -1;
1293 }
1294
1295 #ifndef SMALL
1296 if (fstat(in, &isb) != 0) {
1297 close(in);
1298 maybe_warn("can't stat %s", file);
1299 return -1;
1300 }
1301 infile_set(file, isb.st_size);
1302 #endif
1303
1304 if (cflag == 0) {
1305 #ifndef SMALL
1306 if (isb.st_nlink > 1 && fflag == 0) {
1307 maybe_warnx("%s has %d other link%s -- "
1308 "skipping", file, isb.st_nlink - 1,
1309 isb.st_nlink == 1 ? "" : "s");
1310 close(in);
1311 return -1;
1312 }
1313
1314 if (fflag == 0 && (suff = check_suffix(file, 0))
1315 && suff->zipped[0] != 0) {
1316 maybe_warnx("%s already has %s suffix -- unchanged",
1317 file, suff->zipped);
1318 close(in);
1319 return -1;
1320 }
1321 #endif
1322
1323 /* Add (usually) .gz to filename */
1324 if ((size_t)snprintf(outfile, outsize, "%s%s",
1325 file, suffixes[0].zipped) >= outsize)
1326 memcpy(outfile + outsize - suffixes[0].ziplen - 1,
1327 suffixes[0].zipped, suffixes[0].ziplen + 1);
1328
1329 #ifndef SMALL
1330 if (check_outfile(outfile) == 0) {
1331 close(in);
1332 return -1;
1333 }
1334 #endif
1335 }
1336
1337 if (cflag == 0) {
1338 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600);
1339 if (out == -1) {
1340 maybe_warn("could not create output: %s", outfile);
1341 fclose(stdin);
1342 return -1;
1343 }
1344 } else
1345 out = STDOUT_FILENO;
1346
1347 in_size = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime);
1348
1349 (void)close(in);
1350
1351 /*
1352 * If there was an error, in_size will be -1.
1353 * If we compressed to stdout, just return the size.
1354 * Otherwise stat the file and check it is the correct size.
1355 * We only blow away the file if we can stat the output and it
1356 * has the expected size.
1357 */
1358 if (cflag != 0)
1359 return in_size == -1 ? -1 : size;
1360
1361 #ifndef SMALL
1362 if (fstat(out, &osb) != 0) {
1363 maybe_warn("couldn't stat: %s", outfile);
1364 goto bad_outfile;
1365 }
1366
1367 if (osb.st_size != size) {
1368 maybe_warnx("output file: %s wrong size (%" PRIdOFF
1369 " != %" PRIdOFF "), deleting",
1370 outfile, osb.st_size, size);
1371 goto bad_outfile;
1372 }
1373
1374 copymodes(out, &isb, outfile);
1375 #endif
1376 if (close(out) == -1)
1377 maybe_warn("couldn't close output");
1378
1379 /* output is good, ok to delete input */
1380 unlink_input(file, &isb);
1381 return size;
1382
1383 #ifndef SMALL
1384 bad_outfile:
1385 if (close(out) == -1)
1386 maybe_warn("couldn't close output");
1387
1388 maybe_warnx("leaving original %s", file);
1389 unlink(outfile);
1390 return size;
1391 #endif
1392 }
1393
1394 /* uncompress the given file and remove the original */
1395 static off_t
1396 file_uncompress(char *file, char *outfile, size_t outsize)
1397 {
1398 struct stat isb, osb;
1399 off_t size;
1400 ssize_t rbytes;
1401 unsigned char header1[4];
1402 enum filetype method;
1403 int fd, ofd, zfd = -1;
1404 size_t in_size;
1405 #ifndef SMALL
1406 ssize_t rv;
1407 time_t timestamp = 0;
1408 char name[PATH_MAX + 1];
1409 #endif
1410
1411 /* gather the old name info */
1412
1413 fd = open(file, O_RDONLY);
1414 if (fd < 0) {
1415 maybe_warn("can't open %s", file);
1416 goto lose;
1417 }
1418 if (fstat(fd, &isb) != 0) {
1419 close(fd);
1420 maybe_warn("can't stat %s", file);
1421 goto lose;
1422 }
1423 if (S_ISREG(isb.st_mode))
1424 in_size = isb.st_size;
1425 else
1426 in_size = 0;
1427 infile_set(file, in_size);
1428
1429 strlcpy(outfile, file, outsize);
1430 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) {
1431 maybe_warnx("%s: unknown suffix -- ignored", file);
1432 goto lose;
1433 }
1434
1435 rbytes = read(fd, header1, sizeof header1);
1436 if (rbytes != sizeof header1) {
1437 /* we don't want to fail here. */
1438 #ifndef SMALL
1439 if (fflag)
1440 goto lose;
1441 #endif
1442 if (rbytes == -1)
1443 maybe_warn("can't read %s", file);
1444 else
1445 goto unexpected_EOF;
1446 goto lose;
1447 }
1448 infile_newdata(rbytes);
1449
1450 method = file_gettype(header1);
1451 #ifndef SMALL
1452 if (fflag == 0 && method == FT_UNKNOWN) {
1453 maybe_warnx("%s: not in gzip format", file);
1454 goto lose;
1455 }
1456
1457 #endif
1458
1459 #ifndef SMALL
1460 if (method == FT_GZIP && Nflag) {
1461 unsigned char ts[4]; /* timestamp */
1462
1463 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP);
1464 if (rv >= 0 && rv < (ssize_t)(sizeof ts))
1465 goto unexpected_EOF;
1466 if (rv == -1) {
1467 if (!fflag)
1468 maybe_warn("can't read %s", file);
1469 goto lose;
1470 }
1471 infile_newdata(rv);
1472 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0];
1473
1474 if (header1[3] & ORIG_NAME) {
1475 rbytes = pread(fd, name, sizeof(name) - 1, GZIP_ORIGNAME);
1476 if (rbytes < 0) {
1477 maybe_warn("can't read %s", file);
1478 goto lose;
1479 }
1480 if (name[0] != '\0') {
1481 char *dp, *nf;
1482
1483 /* Make sure that name is NUL-terminated */
1484 name[rbytes] = '\0';
1485
1486 /* strip saved directory name */
1487 nf = strrchr(name, '/');
1488 if (nf == NULL)
1489 nf = name;
1490 else
1491 nf++;
1492
1493 /* preserve original directory name */
1494 dp = strrchr(file, '/');
1495 if (dp == NULL)
1496 dp = file;
1497 else
1498 dp++;
1499 snprintf(outfile, outsize, "%.*s%.*s",
1500 (int) (dp - file),
1501 file, (int) rbytes, nf);
1502 }
1503 }
1504 }
1505 #endif
1506 lseek(fd, 0, SEEK_SET);
1507
1508 if (cflag == 0 || lflag) {
1509 #ifndef SMALL
1510 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) {
1511 maybe_warnx("%s has %d other links -- skipping",
1512 file, isb.st_nlink - 1);
1513 goto lose;
1514 }
1515 if (nflag == 0 && timestamp)
1516 isb.st_mtime = timestamp;
1517 if (check_outfile(outfile) == 0)
1518 goto lose;
1519 #endif
1520 }
1521
1522 if (cflag)
1523 zfd = STDOUT_FILENO;
1524 else if (lflag)
1525 zfd = -1;
1526 else {
1527 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600);
1528 if (zfd == STDOUT_FILENO) {
1529 /* We won't close STDOUT_FILENO later... */
1530 zfd = dup(zfd);
1531 close(STDOUT_FILENO);
1532 }
1533 if (zfd == -1) {
1534 maybe_warn("can't open %s", outfile);
1535 goto lose;
1536 }
1537 }
1538
1539 switch (method) {
1540 #ifndef NO_BZIP2_SUPPORT
1541 case FT_BZIP2:
1542 /* XXX */
1543 if (lflag) {
1544 maybe_warnx("no -l with bzip2 files");
1545 goto lose;
1546 }
1547
1548 size = unbzip2(fd, zfd, NULL, 0, NULL);
1549 break;
1550 #endif
1551
1552 #ifndef NO_COMPRESS_SUPPORT
1553 case FT_Z: {
1554 FILE *in, *out;
1555
1556 /* XXX */
1557 if (lflag) {
1558 maybe_warnx("no -l with Lempel-Ziv files");
1559 goto lose;
1560 }
1561
1562 if ((in = zdopen(fd)) == NULL) {
1563 maybe_warn("zdopen for read: %s", file);
1564 goto lose;
1565 }
1566
1567 out = fdopen(dup(zfd), "w");
1568 if (out == NULL) {
1569 maybe_warn("fdopen for write: %s", outfile);
1570 fclose(in);
1571 goto lose;
1572 }
1573
1574 size = zuncompress(in, out, NULL, 0, NULL);
1575 /* need to fclose() if ferror() is true... */
1576 if (ferror(in) | fclose(in)) {
1577 maybe_warn("failed infile fclose");
1578 unlink(outfile);
1579 (void)fclose(out);
1580 }
1581 if (fclose(out) != 0) {
1582 maybe_warn("failed outfile fclose");
1583 unlink(outfile);
1584 goto lose;
1585 }
1586 break;
1587 }
1588 #endif
1589
1590 #ifndef NO_PACK_SUPPORT
1591 case FT_PACK:
1592 if (lflag) {
1593 maybe_warnx("no -l with packed files");
1594 goto lose;
1595 }
1596
1597 size = unpack(fd, zfd, NULL, 0, NULL);
1598 break;
1599 #endif
1600
1601 #ifndef NO_XZ_SUPPORT
1602 case FT_XZ:
1603 if (lflag) {
1604 size = unxz_len(fd);
1605 print_list_out(in_size, size, file);
1606 return -1;
1607 }
1608 size = unxz(fd, zfd, NULL, 0, NULL);
1609 break;
1610 #endif
1611
1612 #ifndef NO_LZ_SUPPORT
1613 case FT_LZ:
1614 if (lflag) {
1615 maybe_warnx("no -l with lzip files");
1616 goto lose;
1617 }
1618 size = unlz(fd, zfd, NULL, 0, NULL);
1619 break;
1620 #endif
1621 #ifndef SMALL
1622 case FT_UNKNOWN:
1623 if (lflag) {
1624 maybe_warnx("no -l for unknown filetypes");
1625 goto lose;
1626 }
1627 size = cat_fd(NULL, 0, NULL, fd);
1628 break;
1629 #endif
1630 default:
1631 if (lflag) {
1632 print_list(fd, in_size, outfile, isb.st_mtime);
1633 close(fd);
1634 return -1; /* XXX */
1635 }
1636
1637 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file);
1638 break;
1639 }
1640
1641 if (close(fd) != 0)
1642 maybe_warn("couldn't close input");
1643 if (zfd != STDOUT_FILENO && close(zfd) != 0)
1644 maybe_warn("couldn't close output");
1645
1646 if (size == -1) {
1647 if (cflag == 0)
1648 unlink(outfile);
1649 maybe_warnx("%s: uncompress failed", file);
1650 return -1;
1651 }
1652
1653 /* if testing, or we uncompressed to stdout, this is all we need */
1654 #ifndef SMALL
1655 if (tflag)
1656 return size;
1657 #endif
1658 /* if we are uncompressing to stdin, don't remove the file. */
1659 if (cflag)
1660 return size;
1661
1662 /*
1663 * if we create a file...
1664 */
1665 /*
1666 * if we can't stat the file don't remove the file.
1667 */
1668
1669 ofd = open(outfile, O_RDWR, 0);
1670 if (ofd == -1) {
1671 maybe_warn("couldn't open (leaving original): %s",
1672 outfile);
1673 return -1;
1674 }
1675 if (fstat(ofd, &osb) != 0) {
1676 maybe_warn("couldn't stat (leaving original): %s",
1677 outfile);
1678 close(ofd);
1679 return -1;
1680 }
1681 if (osb.st_size != size) {
1682 maybe_warnx("stat gave different size: %" PRIdOFF
1683 " != %" PRIdOFF " (leaving original)",
1684 size, osb.st_size);
1685 close(ofd);
1686 unlink(outfile);
1687 return -1;
1688 }
1689 unlink_input(file, &isb);
1690 #ifndef SMALL
1691 copymodes(ofd, &isb, outfile);
1692 #endif
1693 close(ofd);
1694 return size;
1695
1696 unexpected_EOF:
1697 maybe_warnx("%s: unexpected end of file", file);
1698 lose:
1699 if (fd != -1)
1700 close(fd);
1701 if (zfd != -1 && zfd != STDOUT_FILENO)
1702 close(fd);
1703 return -1;
1704 }
1705
1706 #ifndef SMALL
1707 static void
1708 check_siginfo(void)
1709 {
1710 if (print_info == 0)
1711 return;
1712 if (infile) {
1713 if (infile_total) {
1714 int pcent = (int)((100.0 * infile_current) / infile_total);
1715
1716 fprintf(stderr, "%s: done %llu/%llu bytes %d%%\n",
1717 infile, (unsigned long long)infile_current,
1718 (unsigned long long)infile_total, pcent);
1719 } else
1720 fprintf(stderr, "%s: done %llu bytes\n",
1721 infile, (unsigned long long)infile_current);
1722 }
1723 print_info = 0;
1724 }
1725
1726 static off_t
1727 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd)
1728 {
1729 char buf[BUFLEN];
1730 off_t in_tot;
1731 ssize_t w;
1732
1733 in_tot = count;
1734 w = write_retry(STDOUT_FILENO, prepend, count);
1735 if (w == -1 || (size_t)w != count) {
1736 maybe_warn("write to stdout");
1737 return -1;
1738 }
1739 for (;;) {
1740 ssize_t rv;
1741
1742 rv = read(fd, buf, sizeof buf);
1743 if (rv == 0)
1744 break;
1745 if (rv < 0) {
1746 maybe_warn("read from fd %d", fd);
1747 break;
1748 }
1749 infile_newdata(rv);
1750
1751 if (write_retry(STDOUT_FILENO, buf, rv) != rv) {
1752 maybe_warn("write to stdout");
1753 break;
1754 }
1755 in_tot += rv;
1756 }
1757
1758 if (gsizep)
1759 *gsizep = in_tot;
1760 return (in_tot);
1761 }
1762 #endif
1763
1764 static void
1765 handle_stdin(void)
1766 {
1767 struct stat isb;
1768 unsigned char header1[4];
1769 size_t in_size;
1770 off_t usize, gsize;
1771 enum filetype method;
1772 ssize_t bytes_read;
1773 #ifndef NO_COMPRESS_SUPPORT
1774 FILE *in;
1775 #endif
1776
1777 #ifndef SMALL
1778 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) {
1779 maybe_warnx("standard input is a terminal -- ignoring");
1780 goto out;
1781 }
1782 #endif
1783
1784 if (fstat(STDIN_FILENO, &isb) < 0) {
1785 maybe_warn("fstat");
1786 goto out;
1787 }
1788 if (S_ISREG(isb.st_mode))
1789 in_size = isb.st_size;
1790 else
1791 in_size = 0;
1792 infile_set("(stdin)", in_size);
1793
1794 if (lflag) {
1795 print_list(STDIN_FILENO, in_size, infile, isb.st_mtime);
1796 goto out;
1797 }
1798
1799 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1);
1800 if (bytes_read == -1) {
1801 maybe_warn("can't read stdin");
1802 goto out;
1803 } else if (bytes_read != sizeof(header1)) {
1804 maybe_warnx("(stdin): unexpected end of file");
1805 goto out;
1806 }
1807
1808 method = file_gettype(header1);
1809 switch (method) {
1810 default:
1811 #ifndef SMALL
1812 if (fflag == 0) {
1813 maybe_warnx("unknown compression format");
1814 goto out;
1815 }
1816 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO);
1817 break;
1818 #endif
1819 case FT_GZIP:
1820 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO,
1821 (char *)header1, sizeof header1, &gsize, "(stdin)");
1822 break;
1823 #ifndef NO_BZIP2_SUPPORT
1824 case FT_BZIP2:
1825 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO,
1826 (char *)header1, sizeof header1, &gsize);
1827 break;
1828 #endif
1829 #ifndef NO_COMPRESS_SUPPORT
1830 case FT_Z:
1831 if ((in = zdopen(STDIN_FILENO)) == NULL) {
1832 maybe_warnx("zopen of stdin");
1833 goto out;
1834 }
1835
1836 usize = zuncompress(in, stdout, (char *)header1,
1837 sizeof header1, &gsize);
1838 fclose(in);
1839 break;
1840 #endif
1841 #ifndef NO_PACK_SUPPORT
1842 case FT_PACK:
1843 usize = unpack(STDIN_FILENO, STDOUT_FILENO,
1844 (char *)header1, sizeof header1, &gsize);
1845 break;
1846 #endif
1847 #ifndef NO_XZ_SUPPORT
1848 case FT_XZ:
1849 usize = unxz(STDIN_FILENO, STDOUT_FILENO,
1850 (char *)header1, sizeof header1, &gsize);
1851 break;
1852 #endif
1853 #ifndef NO_LZ_SUPPORT
1854 case FT_LZ:
1855 usize = unlz(STDIN_FILENO, STDOUT_FILENO,
1856 (char *)header1, sizeof header1, &gsize);
1857 break;
1858 #endif
1859 }
1860
1861 #ifndef SMALL
1862 if (vflag && !tflag && usize != -1 && gsize != -1)
1863 print_verbage(NULL, NULL, usize, gsize);
1864 if (vflag && tflag)
1865 print_test("(stdin)", usize != -1);
1866 #else
1867 (void)&usize;
1868 #endif
1869
1870 out:
1871 infile_clear();
1872 }
1873
1874 static void
1875 handle_stdout(void)
1876 {
1877 off_t gsize;
1878 #ifndef SMALL
1879 off_t usize;
1880 struct stat sb;
1881 time_t systime;
1882 uint32_t mtime;
1883 int ret;
1884
1885 infile_set("(stdout)", 0);
1886
1887 if (fflag == 0 && isatty(STDOUT_FILENO)) {
1888 maybe_warnx("standard output is a terminal -- ignoring");
1889 return;
1890 }
1891
1892 /* If stdin is a file use its mtime, otherwise use current time */
1893 ret = fstat(STDIN_FILENO, &sb);
1894 if (ret < 0) {
1895 maybe_warn("Can't stat stdin");
1896 return;
1897 }
1898
1899 if (S_ISREG(sb.st_mode)) {
1900 infile_set("(stdout)", sb.st_size);
1901 mtime = (uint32_t)sb.st_mtime;
1902 } else {
1903 systime = time(NULL);
1904 if (systime == -1) {
1905 maybe_warn("time");
1906 return;
1907 }
1908 mtime = (uint32_t)systime;
1909 }
1910
1911 usize =
1912 #endif
1913 gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime);
1914 #ifndef SMALL
1915 if (vflag && !tflag && usize != -1 && gsize != -1)
1916 print_verbage(NULL, NULL, usize, gsize);
1917 #endif
1918 }
1919
1920 /* do what is asked for, for the path name */
1921 static void
1922 handle_pathname(char *path)
1923 {
1924 char *opath = path, *s = NULL;
1925 ssize_t len;
1926 int slen;
1927 struct stat sb;
1928
1929 /* check for stdout/stdin */
1930 if (path[0] == '-' && path[1] == '\0') {
1931 if (dflag)
1932 handle_stdin();
1933 else
1934 handle_stdout();
1935 return;
1936 }
1937
1938 retry:
1939 if (stat(path, &sb) != 0) {
1940 /* lets try <path>.gz if we're decompressing */
1941 if (dflag && s == NULL && errno == ENOENT) {
1942 len = strlen(path);
1943 slen = suffixes[0].ziplen;
1944 s = malloc(len + slen + 1);
1945 if (s == NULL)
1946 maybe_err("malloc");
1947 memcpy(s, path, len);
1948 memcpy(s + len, suffixes[0].zipped, slen + 1);
1949 path = s;
1950 goto retry;
1951 }
1952 maybe_warn("can't stat: %s", opath);
1953 goto out;
1954 }
1955
1956 if (S_ISDIR(sb.st_mode)) {
1957 #ifndef SMALL
1958 if (rflag)
1959 handle_dir(path);
1960 else
1961 #endif
1962 maybe_warnx("%s is a directory", path);
1963 goto out;
1964 }
1965
1966 if (S_ISREG(sb.st_mode))
1967 handle_file(path, &sb);
1968 else
1969 maybe_warnx("%s is not a regular file", path);
1970
1971 out:
1972 if (s)
1973 free(s);
1974 }
1975
1976 /* compress/decompress a file */
1977 static void
1978 handle_file(char *file, struct stat *sbp)
1979 {
1980 off_t usize, gsize;
1981 char outfile[PATH_MAX];
1982
1983 infile_set(file, sbp->st_size);
1984 if (dflag) {
1985 usize = file_uncompress(file, outfile, sizeof(outfile));
1986 #ifndef SMALL
1987 if (vflag && tflag)
1988 print_test(file, usize != -1);
1989 #endif
1990 if (usize == -1)
1991 return;
1992 gsize = sbp->st_size;
1993 } else {
1994 gsize = file_compress(file, outfile, sizeof(outfile));
1995 if (gsize == -1)
1996 return;
1997 usize = sbp->st_size;
1998 }
1999 infile_clear();
2000
2001 #ifndef SMALL
2002 if (vflag && !tflag)
2003 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize);
2004 #endif
2005 }
2006
2007 #ifndef SMALL
2008 /* this is used with -r to recursively descend directories */
2009 static void
2010 handle_dir(char *dir)
2011 {
2012 char *path_argv[2];
2013 FTS *fts;
2014 FTSENT *entry;
2015
2016 path_argv[0] = dir;
2017 path_argv[1] = 0;
2018 fts = fts_open(path_argv, FTS_PHYSICAL, NULL);
2019 if (fts == NULL) {
2020 warn("couldn't fts_open %s", dir);
2021 return;
2022 }
2023
2024 while ((entry = fts_read(fts))) {
2025 switch(entry->fts_info) {
2026 case FTS_D:
2027 case FTS_DP:
2028 continue;
2029
2030 case FTS_DNR:
2031 case FTS_ERR:
2032 case FTS_NS:
2033 maybe_warn("%s", entry->fts_path);
2034 continue;
2035 case FTS_F:
2036 handle_file(entry->fts_name, entry->fts_statp);
2037 }
2038 }
2039 (void)fts_close(fts);
2040 }
2041 #endif
2042
2043 /* print a ratio - size reduction as a fraction of uncompressed size */
2044 static void
2045 print_ratio(off_t in, off_t out, FILE *where)
2046 {
2047 int percent10; /* 10 * percent */
2048 off_t diff;
2049 char buff[8];
2050 int len;
2051
2052 diff = in - out/2;
2053 if (in == 0 && out == 0)
2054 percent10 = 0;
2055 else if (diff < 0)
2056 /*
2057 * Output is more than double size of input! print -99.9%
2058 * Quite possibly we've failed to get the original size.
2059 */
2060 percent10 = -999;
2061 else {
2062 /*
2063 * We only need 12 bits of result from the final division,
2064 * so reduce the values until a 32bit division will suffice.
2065 */
2066 while (in > 0x100000) {
2067 diff >>= 1;
2068 in >>= 1;
2069 }
2070 if (in != 0)
2071 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000;
2072 else
2073 percent10 = 0;
2074 }
2075
2076 len = snprintf(buff, sizeof buff, "%2.2d.", percent10);
2077 /* Move the '.' to before the last digit */
2078 buff[len - 1] = buff[len - 2];
2079 buff[len - 2] = '.';
2080 fprintf(where, "%5s%%", buff);
2081 }
2082
2083 #ifndef SMALL
2084 /* print compression statistics, and the new name (if there is one!) */
2085 static void
2086 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize)
2087 {
2088 if (file)
2089 fprintf(stderr, "%s:%s ", file,
2090 strlen(file) < 7 ? "\t\t" : "\t");
2091 print_ratio(usize, gsize, stderr);
2092 if (nfile)
2093 fprintf(stderr, " -- replaced with %s", nfile);
2094 fprintf(stderr, "\n");
2095 fflush(stderr);
2096 }
2097
2098 /* print test results */
2099 static void
2100 print_test(const char *file, int ok)
2101 {
2102
2103 if (exit_value == 0 && ok == 0)
2104 exit_value = 1;
2105 fprintf(stderr, "%s:%s %s\n", file,
2106 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK");
2107 fflush(stderr);
2108 }
2109 #endif
2110
2111 /* print a file's info ala --list */
2112 /* eg:
2113 compressed uncompressed ratio uncompressed_name
2114 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar
2115 */
2116 static void
2117 print_list(int fd, off_t out, const char *outfile, time_t ts)
2118 {
2119 static int first = 1;
2120 #ifndef SMALL
2121 static off_t in_tot, out_tot;
2122 uint32_t crc = 0;
2123 #endif
2124 off_t in = 0, rv;
2125
2126 if (first) {
2127 #ifndef SMALL
2128 if (vflag)
2129 printf("method crc date time ");
2130 #endif
2131 if (qflag == 0)
2132 printf(" compressed uncompressed "
2133 "ratio uncompressed_name\n");
2134 }
2135 first = 0;
2136
2137 /* print totals? */
2138 #ifndef SMALL
2139 if (fd == -1) {
2140 in = in_tot;
2141 out = out_tot;
2142 } else
2143 #endif
2144 {
2145 /* read the last 4 bytes - this is the uncompressed size */
2146 rv = lseek(fd, (off_t)(-8), SEEK_END);
2147 if (rv != -1) {
2148 unsigned char buf[8];
2149 uint32_t usize;
2150
2151 rv = read(fd, (char *)buf, sizeof(buf));
2152 if (rv == -1)
2153 maybe_warn("read of uncompressed size");
2154 else if (rv != sizeof(buf))
2155 maybe_warnx("read of uncompressed size");
2156
2157 else {
2158 usize = buf[4];
2159 usize |= (unsigned int)buf[5] << 8;
2160 usize |= (unsigned int)buf[6] << 16;
2161 usize |= (unsigned int)buf[7] << 24;
2162 in = (off_t)usize;
2163 #ifndef SMALL
2164 crc = buf[0];
2165 crc |= (unsigned int)buf[1] << 8;
2166 crc |= (unsigned int)buf[2] << 16;
2167 crc |= (unsigned int)buf[3] << 24;
2168 #endif
2169 }
2170 }
2171 }
2172
2173 #ifndef SMALL
2174 if (vflag && fd == -1)
2175 printf(" ");
2176 else if (vflag) {
2177 char *date = ctime(&ts);
2178
2179 /* skip the day, 1/100th second, and year */
2180 date += 4;
2181 date[12] = 0;
2182 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date);
2183 }
2184 in_tot += in;
2185 out_tot += out;
2186 #endif
2187 print_list_out(out, in, outfile);
2188 }
2189
2190 static void
2191 print_list_out(off_t out, off_t in, const char *outfile)
2192 {
2193 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in);
2194 print_ratio(in, out, stdout);
2195 printf(" %s\n", outfile);
2196 }
2197
2198 /* display the usage of NetBSD gzip */
2199 static void
2200 usage(void)
2201 {
2202
2203 fprintf(stderr, "%s\n", gzip_version);
2204 fprintf(stderr,
2205 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n"
2206 #ifndef SMALL
2207 " -1 --fast fastest (worst) compression\n"
2208 " -2 .. -8 set compression level\n"
2209 " -9 --best best (slowest) compression\n"
2210 " -c --stdout write to stdout, keep original files\n"
2211 " --to-stdout\n"
2212 " -d --decompress uncompress files\n"
2213 " --uncompress\n"
2214 " -f --force force overwriting & compress links\n"
2215 " -h --help display this help\n"
2216 " -k --keep don't delete input files during operation\n"
2217 " -l --list list compressed file contents\n"
2218 " -N --name save or restore original file name and time stamp\n"
2219 " -n --no-name don't save original file name or time stamp\n"
2220 " -q --quiet output no warnings\n"
2221 " -r --recursive recursively compress files in directories\n"
2222 " -S .suf use suffix .suf instead of .gz\n"
2223 " --suffix .suf\n"
2224 " -t --test test compressed file\n"
2225 " -V --version display program version\n"
2226 " -v --verbose print extra statistics\n",
2227 #else
2228 ,
2229 #endif
2230 getprogname());
2231 exit(0);
2232 }
2233
2234 /* display the version of NetBSD gzip */
2235 static void
2236 display_version(void)
2237 {
2238
2239 fprintf(stderr, "%s\n", gzip_version);
2240 exit(0);
2241 }
2242
2243 #ifndef NO_BZIP2_SUPPORT
2244 #include "unbzip2.c"
2245 #endif
2246 #ifndef NO_COMPRESS_SUPPORT
2247 #include "zuncompress.c"
2248 #endif
2249 #ifndef NO_PACK_SUPPORT
2250 #include "unpack.c"
2251 #endif
2252 #ifndef NO_XZ_SUPPORT
2253 #include "unxz.c"
2254 #endif
2255 #ifndef NO_LZ_SUPPORT
2256 #include "unlz.c"
2257 #endif
2258
2259 static ssize_t
2260 read_retry(int fd, void *buf, size_t sz)
2261 {
2262 char *cp = buf;
2263 size_t left = MIN(sz, (size_t) SSIZE_MAX);
2264
2265 while (left > 0) {
2266 ssize_t ret;
2267
2268 ret = read(fd, cp, left);
2269 if (ret == -1) {
2270 return ret;
2271 } else if (ret == 0) {
2272 break; /* EOF */
2273 }
2274 cp += ret;
2275 left -= ret;
2276 }
2277
2278 return sz - left;
2279 }
2280
2281 static ssize_t
2282 write_retry(int fd, const void *buf, size_t sz)
2283 {
2284 const char *cp = buf;
2285 size_t left = MIN(sz, (size_t) SSIZE_MAX);
2286
2287 while (left > 0) {
2288 ssize_t ret;
2289
2290 ret = write(fd, cp, left);
2291 if (ret == -1) {
2292 return ret;
2293 } else if (ret == 0) {
2294 abort(); /* Can't happen */
2295 }
2296 cp += ret;
2297 left -= ret;
2298 }
2299
2300 return sz - left;
2301 }
2302