maint: fix typo in previous patch
[gzip.git] / gzip.c
blob9602271003110e02e7ef279d3bfafb0aebcdd81d
1 /* gzip (GNU zip) -- compress files with zip algorithm and 'compress' interface
3 Copyright (C) 1999, 2001-2002, 2006-2007, 2009-2022 Free Software
4 Foundation, Inc.
5 Copyright (C) 1992-1993 Jean-loup Gailly
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 3, or (at your option)
10 any later version.
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with this program; if not, write to the Free Software Foundation,
19 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
22 * The unzip code was written and put in the public domain by Mark Adler.
23 * Portions of the lzw code are derived from the public domain 'compress'
24 * written by Spencer Thomas, Joe Orost, James Woods, Jim McKie, Steve Davies,
25 * Ken Turkowski, Dave Mack and Peter Jannesen.
27 * See the license_msg below and the file COPYING for the software license.
28 * See the file algorithm.doc for the compression algorithms and file formats.
31 static char const *const license_msg[] = {
32 "Copyright (C) 2022 Free Software Foundation, Inc.",
33 "Copyright (C) 1993 Jean-loup Gailly.",
34 "This is free software. You may redistribute copies of it under the terms of",
35 "the GNU General Public License <https://www.gnu.org/licenses/gpl.html>.",
36 "There is NO WARRANTY, to the extent permitted by law.",
37 0};
39 /* Compress files with zip algorithm and 'compress' interface.
40 * See help() function below for all options.
41 * Outputs:
42 * file.gz: compressed file with same mode, owner, and utimes
43 * or stdout with -c option or if stdin used as input.
44 * If the output file name had to be truncated, the original name is kept
45 * in the compressed file.
46 * On MSDOS, file.tmp -> file.tmz.
48 * Using gz on MSDOS would create too many file name conflicts. For
49 * example, foo.txt -> foo.tgz (.tgz must be reserved as shorthand for
50 * tar.gz). Similarly, foo.dir and foo.doc would both be mapped to foo.dgz.
51 * I also considered 12345678.txt -> 12345txt.gz but this truncates the name
52 * too heavily. There is no ideal solution given the MSDOS 8+3 limitation.
54 * For the meaning of all compilation flags, see comments in Makefile.in.
57 #include <config.h>
58 #include <ctype.h>
59 #include <sys/types.h>
60 #include <signal.h>
61 #include <stdalign.h>
62 #include <stdbool.h>
63 #include <stddef.h>
64 #include <sys/stat.h>
65 #include <errno.h>
67 #include "tailor.h"
68 #include "gzip.h"
69 #include "intprops.h"
70 #include "lzw.h"
71 #include "revision.h"
72 #include "timespec.h"
74 #include "dirname.h"
75 #include "fcntl--.h"
76 #include "filename.h"
77 #include "getopt.h"
78 #include "ignore-value.h"
79 #include "stat-time.h"
80 #include "version.h"
81 #include "xalloc.h"
82 #include "yesno.h"
84 /* configuration */
86 #include <limits.h>
87 #include <unistd.h>
88 #include <stdlib.h>
89 #include <errno.h>
91 #ifndef NO_DIR
92 # define NO_DIR 0
93 #endif
94 #if !NO_DIR
95 # include <dirent.h>
96 # include <savedir.h>
97 #endif
99 #ifndef NO_UTIME
100 # include <utimens.h>
101 #endif
103 #ifndef MAX_PATH_LEN
104 # define MAX_PATH_LEN 1024 /* max pathname length */
105 #endif
107 #ifndef SEEK_END
108 # define SEEK_END 2
109 #endif
111 #ifndef CHAR_BIT
112 # define CHAR_BIT 8
113 #endif
115 #ifdef off_t
116 off_t lseek (int fd, off_t offset, int whence);
117 #endif
119 #ifndef HAVE_WORKING_O_NOFOLLOW
120 # define HAVE_WORKING_O_NOFOLLOW 0
121 #endif
123 /* Separator for file name parts (see shorten_name()) */
124 #ifdef NO_MULTIPLE_DOTS
125 # define PART_SEP "-"
126 #else
127 # define PART_SEP "."
128 #endif
130 /* global buffers */
132 /* With IBM_Z_DFLTCC, DEFLATE COMPRESSION works faster with
133 page-aligned input and output buffers, and requires page-aligned
134 windows; the alignment requirement is 4096. On other platforms
135 alignment doesn't hurt, and alignment up to 4096 is portable so
136 let's do that. */
137 #if defined HAVE_C_ALIGNASOF || defined alignas
138 # define BUFFER_ALIGNED alignas (4096)
139 #else
140 # define BUFFER_ALIGNED /**/
141 #endif
142 DECLARE(uch BUFFER_ALIGNED, inbuf, INBUFSIZ +INBUF_EXTRA);
143 DECLARE(uch BUFFER_ALIGNED, outbuf, OUTBUFSIZ+OUTBUF_EXTRA);
144 DECLARE(ush, d_buf, DIST_BUFSIZE);
145 DECLARE(uch BUFFER_ALIGNED, window, 2L*WSIZE);
146 #ifndef MAXSEG_64K
147 DECLARE(ush, tab_prefix, 1L<<BITS);
148 #else
149 DECLARE(ush, tab_prefix0, 1L<<(BITS-1));
150 DECLARE(ush, tab_prefix1, 1L<<(BITS-1));
151 #endif
153 /* local variables */
155 /* If true, pretend that standard input is a tty. This option
156 is deliberately not documented, and only for testing. */
157 static bool presume_input_tty;
159 /* If true, transfer output data to the output file's storage device
160 when supported. Otherwise, if the system crashes around the time
161 gzip is run, the user might lose both input and output data. See:
162 Pillai TS et al. All file systems are not created equal: on the
163 complexity of crafting crash-consistent applications. OSDI'14. 2014:433-48.
164 https://www.usenix.org/conference/osdi14/technical-sessions/presentation/pillai */
165 static bool synchronous;
167 static int ascii = 0; /* convert end-of-lines to local OS conventions */
168 int to_stdout = 0; /* output to stdout (-c) */
169 static int decompress = 0; /* decompress (-d) */
170 static int force = 0; /* don't ask questions, compress links (-f) */
171 static int keep = 0; /* keep (don't delete) input files */
172 static int no_name = -1; /* don't save or restore the original file name */
173 static int no_time = -1; /* don't save or restore the original file time */
174 static int recursive = 0; /* recurse through directories (-r) */
175 static int list = 0; /* list the file contents (-l) */
176 #ifndef DEBUG
177 static
178 #endif
179 int verbose = 0; /* be verbose (-v) */
180 int quiet = 0; /* be very quiet (-q) */
181 int test = 0; /* test .gz file integrity */
182 static int foreground = 0; /* set if program run in foreground */
183 char *program_name; /* program name */
184 int maxbits = BITS; /* max bits per code for LZW */
185 int method = DEFLATED;/* compression method */
186 int level = 6; /* compression level */
187 int exit_code = OK; /* program exit code */
188 int save_orig_name; /* set if original name must be saved */
189 static int last_member; /* set for .zip and .Z files */
190 static int part_nb; /* number of parts in .gz file */
191 off_t ifile_size; /* input file size, -1 for devices (debug only) */
192 static char *env; /* contents of GZIP env variable */
193 static char const *z_suffix; /* default suffix (can be set with --suffix) */
194 static size_t z_len; /* strlen(z_suffix) */
196 /* The original timestamp (modification time). If the original is
197 unknown, TIME_STAMP.tv_nsec is negative. If the original is
198 greater than struct timespec range, TIME_STAMP is the maximal
199 struct timespec value; this can happen on hosts with 32-bit signed
200 time_t because the gzip format's MTIME is 32-bit unsigned.
201 The original cannot be less than struct timespec range. */
202 struct timespec time_stamp;
204 /* The set of signals that are caught. */
205 static sigset_t caught_signals;
207 /* If nonzero then exit with status WARNING, rather than with the usual
208 signal status, on receipt of a signal with this value. This
209 suppresses a "Broken Pipe" message with some shells. */
210 static int volatile exiting_signal;
212 /* If nonnegative, close this file descriptor and unlink remove_ofname
213 on error. */
214 static int volatile remove_ofname_fd = -1;
215 static char volatile remove_ofname[MAX_PATH_LEN];
217 static bool stdin_was_read;
219 off_t bytes_in; /* number of input bytes */
220 off_t bytes_out; /* number of output bytes */
221 static off_t total_in; /* input bytes for all files */
222 static off_t total_out; /* output bytes for all files */
223 char ifname[MAX_PATH_LEN]; /* input file name */
224 char ofname[MAX_PATH_LEN]; /* output file name */
225 static char dfname[MAX_PATH_LEN]; /* name of dir containing output file */
226 static struct stat istat; /* status for input file */
227 int ifd; /* input file descriptor */
228 int ofd; /* output file descriptor */
229 static int dfd = -1; /* output directory file descriptor */
230 unsigned insize; /* valid bytes in inbuf */
231 unsigned inptr; /* index of next byte to be processed in inbuf */
232 unsigned outcnt; /* bytes in output buffer */
233 int rsync = 0; /* make rsyncable chunks */
235 static int handled_sig[] =
237 /* SIGINT must be first, as 'foreground' depends on it. */
238 SIGINT
240 #ifdef SIGHUP
241 , SIGHUP
242 #endif
243 #if SIGPIPE
244 , SIGPIPE
245 #endif
246 #ifdef SIGTERM
247 , SIGTERM
248 #endif
249 #ifdef SIGXCPU
250 , SIGXCPU
251 #endif
252 #ifdef SIGXFSZ
253 , SIGXFSZ
254 #endif
257 /* For long options that have no equivalent short option, use a
258 non-character as a pseudo short option, starting with CHAR_MAX + 1. */
259 enum
261 PRESUME_INPUT_TTY_OPTION = CHAR_MAX + 1,
262 RSYNCABLE_OPTION,
263 SYNCHRONOUS_OPTION,
265 /* A value greater than all valid long options, used as a flag to
266 distinguish options derived from the GZIP environment variable. */
267 ENV_OPTION
270 static char const shortopts[] = "ab:cdfhH?klLmMnNqrS:tvVZ123456789";
272 static const struct option longopts[] =
274 /* { name has_arg *flag val } */
275 {"ascii", 0, 0, 'a'}, /* ascii text mode */
276 {"to-stdout", 0, 0, 'c'}, /* write output on standard output */
277 {"stdout", 0, 0, 'c'}, /* write output on standard output */
278 {"decompress", 0, 0, 'd'}, /* decompress */
279 {"uncompress", 0, 0, 'd'}, /* decompress */
280 /* {"encrypt", 0, 0, 'e'}, encrypt */
281 {"force", 0, 0, 'f'}, /* force overwrite of output file */
282 {"help", 0, 0, 'h'}, /* give help */
283 /* {"pkzip", 0, 0, 'k'}, force output in pkzip format */
284 {"keep", 0, 0, 'k'}, /* keep (don't delete) input files */
285 {"list", 0, 0, 'l'}, /* list .gz file contents */
286 {"license", 0, 0, 'L'}, /* display software license */
287 {"no-name", 0, 0, 'n'}, /* don't save or restore original name & time */
288 {"name", 0, 0, 'N'}, /* save or restore original name & time */
289 {"-presume-input-tty", no_argument, NULL, PRESUME_INPUT_TTY_OPTION},
290 {"quiet", 0, 0, 'q'}, /* quiet mode */
291 {"silent", 0, 0, 'q'}, /* quiet mode */
292 {"synchronous",0, 0, SYNCHRONOUS_OPTION},
293 {"recursive", 0, 0, 'r'}, /* recurse through directories */
294 {"suffix", 1, 0, 'S'}, /* use given suffix instead of .gz */
295 {"test", 0, 0, 't'}, /* test compressed file integrity */
296 {"verbose", 0, 0, 'v'}, /* verbose mode */
297 {"version", 0, 0, 'V'}, /* display version number */
298 {"fast", 0, 0, '1'}, /* compress faster */
299 {"best", 0, 0, '9'}, /* compress better */
300 {"lzw", 0, 0, 'Z'}, /* make output compatible with old compress */
301 {"bits", 1, 0, 'b'}, /* max number of bits per code (implies -Z) */
302 {"rsyncable", 0, 0, RSYNCABLE_OPTION}, /* make rsync-friendly archive */
303 { 0, 0, 0, 0 }
306 /* local functions */
308 _Noreturn static void try_help (void);
309 static void help (void);
310 static void license (void);
311 static void version (void);
312 static int input_eof (void);
313 static void treat_stdin (void);
314 static void treat_file (char *iname);
315 static int create_outfile (void);
316 static char *get_suffix (char *name);
317 static int open_input_file (char *iname, struct stat *sbuf);
318 static void discard_input_bytes (size_t nbytes, unsigned int flags);
319 static int make_ofname (void);
320 static void shorten_name (char *name);
321 static int get_method (int in);
322 static void do_list (int method);
323 static int check_ofname (void);
324 static void copy_stat (struct stat *ifstat);
325 static void install_signal_handlers (void);
326 static void remove_output_file (bool);
327 static void abort_gzip_signal (int);
328 _Noreturn static void do_exit (int exitcode);
329 static void finish_out (void);
330 int main (int argc, char **argv);
331 static int (*work) (int infile, int outfile) = zip; /* function to call */
333 #if ! NO_DIR
334 static void treat_dir (int fd, char *dir);
335 #endif
337 #define strequ(s1, s2) (strcmp((s1),(s2)) == 0)
339 static void
340 try_help ()
342 fprintf (stderr, "Try `%s --help' for more information.\n",
343 program_name);
344 do_exit (ERROR);
347 /* ======================================================================== */
348 static void
349 help ()
351 static char const* const help_msg[] = {
352 "Compress or uncompress FILEs (by default, compress FILES in-place).",
354 "Mandatory arguments to long options are mandatory for short options too.",
356 #if O_BINARY
357 " -a, --ascii ascii text; convert end-of-line using local conventions",
358 #endif
359 " -c, --stdout write on standard output, keep original files unchanged",
360 " -d, --decompress decompress",
361 /* -e, --encrypt encrypt */
362 " -f, --force force overwrite of output file and compress links",
363 " -h, --help give this help",
364 /* -k, --pkzip force output in pkzip format */
365 " -k, --keep keep (don't delete) input files",
366 " -l, --list list compressed file contents",
367 " -L, --license display software license",
368 #ifdef UNDOCUMENTED
369 " -m do not save or restore the original modification time",
370 " -M, --time save or restore the original modification time",
371 #endif
372 " -n, --no-name do not save or restore the original name and timestamp",
373 " -N, --name save or restore the original name and timestamp",
374 " -q, --quiet suppress all warnings",
375 #if ! NO_DIR
376 " -r, --recursive operate recursively on directories",
377 #endif
378 " --rsyncable make rsync-friendly archive",
379 " -S, --suffix=SUF use suffix SUF on compressed files",
380 " --synchronous synchronous output (safer if system crashes, but slower)",
381 " -t, --test test compressed file integrity",
382 " -v, --verbose verbose mode",
383 " -V, --version display version number",
384 " -1, --fast compress faster",
385 " -9, --best compress better",
387 "With no FILE, or when FILE is -, read standard input.",
389 "Report bugs to <bug-gzip@gnu.org>.",
391 char const *const *p = help_msg;
393 printf ("Usage: %s [OPTION]... [FILE]...\n", program_name);
394 while (*p) printf ("%s\n", *p++);
397 /* ======================================================================== */
398 static void
399 license ()
401 char const *const *p = license_msg;
403 printf ("%s %s\n", program_name, Version);
404 while (*p) printf ("%s\n", *p++);
407 /* ======================================================================== */
408 static void
409 version ()
411 license ();
412 printf ("\n");
413 printf ("Written by Jean-loup Gailly.\n");
416 static void
417 progerror (char const *string)
419 int e = errno;
420 fprintf (stderr, "%s: ", program_name);
421 errno = e;
422 perror(string);
423 exit_code = ERROR;
426 /* ======================================================================== */
427 int main (int argc, char **argv)
429 int file_count; /* number of files to process */
430 size_t proglen; /* length of program_name */
431 char **argv_copy;
432 int env_argc;
433 char **env_argv;
435 EXPAND(argc, argv); /* wild card expansion if necessary */
437 program_name = gzip_base_name (argv[0]);
438 proglen = strlen (program_name);
440 /* Suppress .exe for MSDOS and OS/2: */
441 if (4 < proglen && strequ (program_name + proglen - 4, ".exe"))
442 program_name[proglen - 4] = '\0';
444 /* Add options in GZIP environment variable if there is one */
445 argv_copy = argv;
446 env = add_envopt (&env_argc, &argv_copy, OPTIONS_VAR);
447 env_argv = env ? argv_copy : NULL;
449 #ifndef GNU_STANDARD
450 # define GNU_STANDARD 1
451 #endif
452 #if !GNU_STANDARD
453 /* For compatibility with old compress, use program name as an option.
454 * Unless you compile with -DGNU_STANDARD=0, this program will behave as
455 * gzip even if it is invoked under the name gunzip or zcat.
457 * Systems which do not support links can still use -d or -dc.
458 * Ignore an .exe extension for MSDOS and OS/2.
460 if (strncmp (program_name, "un", 2) == 0 /* ungzip, uncompress */
461 || strncmp (program_name, "gun", 3) == 0) /* gunzip */
462 decompress = 1;
463 else if (strequ (program_name + 1, "cat") /* zcat, pcat, gcat */
464 || strequ (program_name, "gzcat")) /* gzcat */
465 decompress = to_stdout = 1;
466 #endif
468 z_suffix = Z_SUFFIX;
469 z_len = strlen(z_suffix);
471 while (true) {
472 int optc;
473 int longind = -1;
475 if (env_argv)
477 if (env_argv[optind] && strequ (env_argv[optind], "--"))
478 optc = ENV_OPTION + '-';
479 else
481 optc = getopt_long (env_argc, env_argv, shortopts, longopts,
482 &longind);
483 if (0 <= optc)
484 optc += ENV_OPTION;
485 else
487 if (optind != env_argc)
489 fprintf (stderr,
490 ("%s: %s: non-option in "OPTIONS_VAR
491 " environment variable\n"),
492 program_name, env_argv[optind]);
493 try_help ();
496 /* Wait until here before warning, so that GZIP='-q'
497 doesn't warn. */
498 if (env_argc != 1 && !quiet)
499 fprintf (stderr,
500 ("%s: warning: "OPTIONS_VAR" environment variable"
501 " is deprecated; use an alias or script\n"),
502 program_name);
504 /* Start processing ARGC and ARGV instead. */
505 free (env_argv);
506 env_argv = NULL;
507 optind = 1;
508 longind = -1;
513 if (!env_argv)
514 optc = getopt_long (argc, argv, shortopts, longopts, &longind);
515 if (optc < 0)
516 break;
518 switch (optc) {
519 case 'a':
520 ascii = 1; break;
521 case 'b':
522 maxbits = atoi(optarg);
523 for (; *optarg; optarg++)
524 if (! ('0' <= *optarg && *optarg <= '9'))
526 fprintf (stderr, "%s: -b operand is not an integer\n",
527 program_name);
528 try_help ();
530 break;
531 case 'c':
532 to_stdout = 1; break;
533 case 'd':
534 decompress = 1; break;
535 case 'f':
536 force++; break;
537 case 'h': case 'H':
538 help (); finish_out (); break;
539 case 'k':
540 keep = 1; break;
541 case 'l':
542 list = decompress = test = to_stdout = 1; break;
543 case 'L':
544 license (); finish_out (); break;
545 case 'm': /* undocumented, may change later */
546 no_time = 1; break;
547 case 'M': /* undocumented, may change later */
548 no_time = 0; break;
549 case 'n':
550 case 'n' + ENV_OPTION:
551 no_name = no_time = 1; break;
552 case 'N':
553 case 'N' + ENV_OPTION:
554 no_name = no_time = 0; break;
555 case PRESUME_INPUT_TTY_OPTION:
556 presume_input_tty = true; break;
557 case 'q':
558 case 'q' + ENV_OPTION:
559 quiet = 1; verbose = 0; break;
560 case 'r':
561 #if NO_DIR
562 fprintf (stderr, "%s: -r not supported on this system\n",
563 program_name);
564 try_help ();
565 #else
566 recursive = 1;
567 #endif
568 break;
570 case RSYNCABLE_OPTION:
571 case RSYNCABLE_OPTION + ENV_OPTION:
572 rsync = 1;
573 break;
574 case 'S':
575 #ifdef NO_MULTIPLE_DOTS
576 if (*optarg == '.') optarg++;
577 #endif
578 z_len = strlen(optarg);
579 z_suffix = optarg;
580 break;
581 case SYNCHRONOUS_OPTION:
582 synchronous = true;
583 break;
584 case 't':
585 test = decompress = to_stdout = 1;
586 break;
587 case 'v':
588 case 'v' + ENV_OPTION:
589 verbose++; quiet = 0; break;
590 case 'V':
591 version (); finish_out (); break;
592 case 'Z':
593 fprintf(stderr, "%s: -Z not supported in this version\n",
594 program_name);
595 try_help ();
596 break;
597 case '1' + ENV_OPTION: case '2' + ENV_OPTION: case '3' + ENV_OPTION:
598 case '4' + ENV_OPTION: case '5' + ENV_OPTION: case '6' + ENV_OPTION:
599 case '7' + ENV_OPTION: case '8' + ENV_OPTION: case '9' + ENV_OPTION:
600 optc -= ENV_OPTION;
601 FALLTHROUGH;
602 case '1': case '2': case '3': case '4':
603 case '5': case '6': case '7': case '8': case '9':
604 level = optc - '0';
605 break;
607 default:
608 if (ENV_OPTION <= optc && optc != ENV_OPTION + '?')
610 /* Output a diagnostic, since getopt_long didn't. */
611 fprintf (stderr, "%s: ", program_name);
612 if (longind < 0)
613 fprintf (stderr, "-%c: ", optc - ENV_OPTION);
614 else
615 fprintf (stderr, "--%s: ", longopts[longind].name);
616 fprintf (stderr, ("option not valid in "OPTIONS_VAR
617 " environment variable\n"));
619 try_help ();
621 } /* loop on all arguments */
623 /* By default, save name and timestamp on compression but do not
624 * restore them on decompression.
626 if (no_time < 0) no_time = decompress;
627 if (no_name < 0) no_name = decompress;
629 file_count = argc - optind;
631 #if O_BINARY
632 #else
633 if (ascii && !quiet) {
634 fprintf(stderr, "%s: option --ascii ignored on this system\n",
635 program_name);
637 #endif
638 if (z_len == 0 || z_len > MAX_SUFFIX) {
639 fprintf(stderr, "%s: invalid suffix '%s'\n", program_name, z_suffix);
640 do_exit(ERROR);
643 /* Allocate all global buffers (for DYN_ALLOC option) */
644 ALLOC(uch, inbuf, INBUFSIZ +INBUF_EXTRA);
645 ALLOC(uch, outbuf, OUTBUFSIZ+OUTBUF_EXTRA);
646 ALLOC(ush, d_buf, DIST_BUFSIZE);
647 ALLOC(uch, window, 2L*WSIZE);
648 #ifndef MAXSEG_64K
649 ALLOC(ush, tab_prefix, 1L<<BITS);
650 #else
651 ALLOC(ush, tab_prefix0, 1L<<(BITS-1));
652 ALLOC(ush, tab_prefix1, 1L<<(BITS-1));
653 #endif
655 #if SIGPIPE
656 exiting_signal = quiet ? SIGPIPE : 0;
657 #endif
658 install_signal_handlers ();
660 /* And get to work */
661 if (file_count != 0) {
662 if (to_stdout && !test && (!decompress || !ascii)) {
663 SET_BINARY_MODE (STDOUT_FILENO);
665 while (optind < argc) {
666 treat_file(argv[optind++]);
668 } else { /* Standard input */
669 treat_stdin();
671 if (stdin_was_read && close (STDIN_FILENO) != 0)
673 strcpy (ifname, "stdin");
674 read_error ();
676 if (list)
678 /* Output any totals, and check for output errors. */
679 if (!quiet && 1 < file_count)
680 do_list (-1);
681 if (fflush (stdout) != 0)
682 write_error ();
684 if (to_stdout
685 && ((synchronous
686 && fdatasync (STDOUT_FILENO) != 0 && errno != EINVAL)
687 || close (STDOUT_FILENO) != 0)
688 && errno != EBADF)
689 write_error ();
690 do_exit(exit_code);
693 /* Return nonzero when at end of file on input. */
694 static int
695 input_eof ()
697 if (!decompress || last_member)
698 return 1;
700 if (inptr == insize)
702 if (insize != INBUFSIZ || fill_inbuf (1) == EOF)
703 return 1;
705 /* Unget the char that fill_inbuf got. */
706 inptr = 0;
709 return 0;
712 static void
713 get_input_size_and_time (void)
715 ifile_size = -1;
716 time_stamp.tv_nsec = -1;
718 /* Record the input file's size and timestamp only if it is a
719 regular file. Doing this for the timestamp helps to keep gzip's
720 output more reproducible when it is used as part of a
721 pipeline. */
723 if (S_ISREG (istat.st_mode))
725 ifile_size = istat.st_size;
726 if (!no_time || list)
727 time_stamp = get_stat_mtime (&istat);
731 /* ========================================================================
732 * Compress or decompress stdin
734 static void
735 treat_stdin ()
737 if (!force && !list
738 && (presume_input_tty
739 || isatty (decompress ? STDIN_FILENO : STDOUT_FILENO))) {
740 /* Do not send compressed data to the terminal or read it from
741 * the terminal. We get here when user invoked the program
742 * without parameters, so be helpful. According to the GNU standards:
744 * If there is one behavior you think is most useful when the output
745 * is to a terminal, and another that you think is most useful when
746 * the output is a file or a pipe, then it is usually best to make
747 * the default behavior the one that is useful with output to a
748 * terminal, and have an option for the other behavior.
750 * Here we use the --force option to get the other behavior.
752 if (! quiet)
753 fprintf (stderr,
754 ("%s: compressed data not %s a terminal."
755 " Use -f to force %scompression.\n"
756 "For help, type: %s -h\n"),
757 program_name,
758 decompress ? "read from" : "written to",
759 decompress ? "de" : "",
760 program_name);
761 do_exit(ERROR);
764 if (decompress || !ascii) {
765 SET_BINARY_MODE (STDIN_FILENO);
767 if (!test && (!decompress || !ascii)) {
768 SET_BINARY_MODE (STDOUT_FILENO);
770 strcpy(ifname, "stdin");
771 strcpy(ofname, "stdout");
773 /* Get the file's timestamp and size. */
774 if (fstat (STDIN_FILENO, &istat) != 0)
776 progerror ("standard input");
777 do_exit (ERROR);
780 get_input_size_and_time ();
782 clear_bufs(); /* clear input and output buffers */
783 to_stdout = 1;
784 part_nb = 0;
785 ifd = STDIN_FILENO;
786 stdin_was_read = true;
788 if (decompress) {
789 method = get_method(ifd);
790 if (method < 0) {
791 do_exit(exit_code); /* error message already emitted */
795 /* Actually do the compression/decompression. Loop over zipped members.
797 for (;;) {
798 if (work (STDIN_FILENO, STDOUT_FILENO) != OK)
799 return;
801 if (input_eof ())
802 break;
804 method = get_method(ifd);
805 if (method < 0) return; /* error message already emitted */
806 bytes_out = 0; /* required for length check */
809 if (list)
811 do_list (method);
812 return;
815 if (verbose) {
816 if (test) {
817 fprintf(stderr, " OK\n");
819 } else if (!decompress) {
820 display_ratio(bytes_in-(bytes_out-header_bytes), bytes_in, stderr);
821 fprintf(stderr, "\n");
822 #ifdef DISPLAY_STDIN_RATIO
823 } else {
824 display_ratio(bytes_out-(bytes_in-header_bytes), bytes_out,stderr);
825 fprintf(stderr, "\n");
826 #endif
831 static char const dot = '.';
833 /* True if the cached directory for calls to openat etc. is DIR, with
834 length DIRLEN. DIR need not be null-terminated. DIRLEN must be
835 less than MAX_PATH_LEN. */
836 static bool
837 atdir_eq (char const *dir, ptrdiff_t dirlen)
839 if (dirlen == 0)
840 dir = &dot, dirlen = 1;
841 return memcmp (dfname, dir, dirlen) == 0 && !dfname[dirlen];
844 /* Set the directory used for calls to openat etc. to be the directory
845 DIR, with length DIRLEN. DIR need not be null-terminated.
846 DIRLEN must be less than MAX_PATH_LEN. Return a file descriptor for
847 the directory, or -1 if one could not be obtained. */
848 static int
849 atdir_set (char const *dir, ptrdiff_t dirlen)
851 /* Don't bother opening directories on older systems that
852 lack openat and unlinkat. It's not worth the porting hassle. */
853 #if HAVE_OPENAT && HAVE_UNLINKAT
854 enum { try_opening_directories = true };
855 #else
856 enum { try_opening_directories = false };
857 #endif
859 if (try_opening_directories && ! atdir_eq (dir, dirlen))
861 if (0 <= dfd)
862 close (dfd);
863 if (dirlen == 0)
864 dir = &dot, dirlen = 1;
865 memcpy (dfname, dir, dirlen);
866 dfname[dirlen] = '\0';
867 dfd = open (dfname, O_SEARCH | O_DIRECTORY);
870 return dfd;
873 /* ========================================================================
874 * Compress or decompress the given file
876 static void
877 treat_file (char *iname)
879 /* Accept "-" as synonym for stdin */
880 if (strequ(iname, "-")) {
881 int cflag = to_stdout;
882 treat_stdin();
883 to_stdout = cflag;
884 return;
887 /* Check if the input file is present, set ifname and istat: */
888 ifd = open_input_file (iname, &istat);
889 if (ifd < 0)
890 return;
892 /* If the input name is that of a directory, recurse or ignore: */
893 if (S_ISDIR(istat.st_mode)) {
894 #if ! NO_DIR
895 if (recursive) {
896 treat_dir (ifd, iname);
897 /* Warning: ifname is now garbage */
898 return;
900 #endif
901 close (ifd);
902 WARN ((stderr, "%s: %s is a directory -- ignored\n",
903 program_name, ifname));
904 return;
907 if (! to_stdout)
909 if (! S_ISREG (istat.st_mode))
911 WARN ((stderr,
912 "%s: %s is not a directory or a regular file - ignored\n",
913 program_name, ifname));
914 close (ifd);
915 return;
917 if (istat.st_mode & S_ISUID)
919 WARN ((stderr, "%s: %s is set-user-ID on execution - ignored\n",
920 program_name, ifname));
921 close (ifd);
922 return;
924 if (istat.st_mode & S_ISGID)
926 WARN ((stderr, "%s: %s is set-group-ID on execution - ignored\n",
927 program_name, ifname));
928 close (ifd);
929 return;
932 if (! force)
934 if (istat.st_mode & S_ISVTX)
936 WARN ((stderr,
937 "%s: %s has the sticky bit set - file ignored\n",
938 program_name, ifname));
939 close (ifd);
940 return;
942 if (2 <= istat.st_nlink)
944 WARN ((stderr, "%s: %s has %lu other link%s -- file ignored\n",
945 program_name, ifname,
946 (unsigned long int) istat.st_nlink - 1,
947 istat.st_nlink == 2 ? "" : "s"));
948 close (ifd);
949 return;
954 get_input_size_and_time ();
956 /* Generate output file name. For -r and (-t or -l), skip files
957 * without a valid gzip suffix (check done in make_ofname).
959 if (to_stdout && !test) {
960 strcpy(ofname, "stdout");
962 } else if (make_ofname() != OK) {
963 close (ifd);
964 return;
967 clear_bufs(); /* clear input and output buffers */
968 part_nb = 0;
970 if (decompress) {
971 method = get_method(ifd); /* updates ofname if original given */
972 if (method < 0) {
973 close(ifd);
974 return; /* error message already emitted */
978 /* If compressing to a file, check if ofname is not ambiguous
979 * because the operating system truncates names. Otherwise, generate
980 * a new ofname and save the original name in the compressed file.
982 if (to_stdout) {
983 ofd = STDOUT_FILENO;
984 /* Keep remove_ofname_fd negative. */
985 } else {
986 if (create_outfile() != OK) return;
988 if (!decompress && save_orig_name && !verbose && !quiet) {
989 fprintf(stderr, "%s: %s compressed to %s\n",
990 program_name, ifname, ofname);
993 /* Keep the name even if not truncated except with --no-name: */
994 if (!save_orig_name) save_orig_name = !no_name;
996 if (verbose && !list) {
997 fprintf(stderr, "%s:\t", ifname);
1000 /* Actually do the compression/decompression. Loop over zipped members.
1002 for (;;) {
1003 if ((*work)(ifd, ofd) != OK) {
1004 method = -1; /* force cleanup */
1005 break;
1008 if (input_eof ())
1009 break;
1011 method = get_method(ifd);
1012 if (method < 0) break; /* error message already emitted */
1013 bytes_out = 0; /* required for length check */
1016 if (close (ifd) != 0)
1017 read_error ();
1019 if (list)
1021 do_list (method);
1022 return;
1025 if (!to_stdout)
1027 copy_stat (&istat);
1029 if ((synchronous
1030 && ((0 <= dfd && fdatasync (dfd) != 0 && errno != EINVAL)
1031 || (fsync (ofd) != 0 && errno != EINVAL)))
1032 || close (ofd) != 0)
1033 write_error ();
1035 if (!keep)
1037 sigset_t oldset;
1038 int unlink_errno;
1039 char *ifbase = last_component (ifname);
1040 int ufd = atdir_eq (ifname, ifbase - ifname) ? dfd : -1;
1041 int res;
1043 sigprocmask (SIG_BLOCK, &caught_signals, &oldset);
1044 remove_ofname_fd = -1;
1045 res = ufd < 0 ? xunlink (ifname) : unlinkat (ufd, ifbase, 0);
1046 unlink_errno = res == 0 ? 0 : errno;
1047 sigprocmask (SIG_SETMASK, &oldset, NULL);
1049 if (unlink_errno)
1051 WARN ((stderr, "%s: ", program_name));
1052 if (!quiet)
1054 errno = unlink_errno;
1055 perror (ifname);
1061 if (method == -1) {
1062 if (!to_stdout)
1063 remove_output_file (false);
1064 return;
1067 /* Display statistics */
1068 if(verbose) {
1069 if (test) {
1070 fprintf(stderr, " OK");
1071 } else if (decompress) {
1072 display_ratio(bytes_out-(bytes_in-header_bytes), bytes_out,stderr);
1073 } else {
1074 display_ratio(bytes_in-(bytes_out-header_bytes), bytes_in, stderr);
1076 if (!test)
1077 fprintf(stderr, " -- %s %s", keep ? "created" : "replaced with",
1078 ofname);
1079 fprintf(stderr, "\n");
1083 static void
1084 volatile_strcpy (char volatile *dst, char const volatile *src)
1086 while ((*dst++ = *src++))
1087 continue;
1090 /* ========================================================================
1091 * Create the output file. Return OK or ERROR.
1092 * Try several times if necessary to avoid truncating the z_suffix. For
1093 * example, do not create a compressed file of name "1234567890123."
1094 * Sets save_orig_name to true if the file name has been truncated.
1095 * IN assertions: the input file has already been open (ifd is set) and
1096 * ofname has already been updated if there was an original name.
1097 * OUT assertions: ifd and ofd are closed in case of error.
1099 static int
1100 create_outfile ()
1102 int name_shortened = 0;
1103 int flags = (O_WRONLY | O_CREAT | O_EXCL
1104 | (ascii && decompress ? 0 : O_BINARY));
1105 char const *base = ofname;
1106 int atfd = AT_FDCWD;
1108 if (!keep)
1110 char const *b = last_component (ofname);
1111 int f = atdir_set (ofname, b - ofname);
1112 if (0 <= f)
1114 base = b;
1115 atfd = f;
1119 for (;;)
1121 int open_errno;
1122 sigset_t oldset;
1124 volatile_strcpy (remove_ofname, ofname);
1126 sigprocmask (SIG_BLOCK, &caught_signals, &oldset);
1127 remove_ofname_fd = ofd = openat (atfd, base, flags, S_IRUSR | S_IWUSR);
1128 open_errno = errno;
1129 sigprocmask (SIG_SETMASK, &oldset, NULL);
1131 if (0 <= ofd)
1132 break;
1134 switch (open_errno)
1136 #ifdef ENAMETOOLONG
1137 case ENAMETOOLONG:
1138 shorten_name (ofname);
1139 name_shortened = 1;
1140 break;
1141 #endif
1143 case EEXIST:
1144 if (check_ofname () != OK)
1146 close (ifd);
1147 return ERROR;
1149 break;
1151 default:
1152 progerror (ofname);
1153 close (ifd);
1154 return ERROR;
1158 if (name_shortened && decompress)
1160 /* name might be too long if an original name was saved */
1161 WARN ((stderr, "%s: %s: warning, name truncated\n",
1162 program_name, ofname));
1165 return OK;
1168 /* ========================================================================
1169 * Return a pointer to the 'z' suffix of a file name, or NULL. For all
1170 * systems, ".gz", ".z", ".Z", ".taz", ".tgz", "-gz", "-z" and "_z" are
1171 * accepted suffixes, in addition to the value of the --suffix option.
1172 * ".tgz" is a useful convention for tar.z files on systems limited
1173 * to 3 characters extensions. On such systems, ".?z" and ".??z" are
1174 * also accepted suffixes. For Unix, we do not want to accept any
1175 * .??z suffix as indicating a compressed file; some people use .xyz
1176 * to denote volume data.
1178 static char *
1179 get_suffix (char *name)
1181 int nlen, slen;
1182 char suffix[MAX_SUFFIX+3]; /* last chars of name, forced to lower case */
1183 static char const *known_suffixes[] =
1184 {NULL, ".gz", ".z", ".taz", ".tgz", "-gz", "-z", "_z",
1185 #ifdef MAX_EXT_CHARS
1186 "z",
1187 #endif
1188 NULL, NULL};
1189 char const **suf;
1190 bool suffix_of_builtin = false;
1192 /* Normally put Z_SUFFIX at the start of KNOWN_SUFFIXES, but if it
1193 is a suffix of one of them, put it at the end. */
1194 for (suf = known_suffixes + 1; *suf; suf++)
1196 size_t suflen = strlen (*suf);
1197 if (z_len < suflen && strequ (z_suffix, *suf + suflen - z_len))
1199 suffix_of_builtin = true;
1200 break;
1204 char *z_lower = xstrdup(z_suffix);
1205 strlwr(z_lower);
1206 known_suffixes[suffix_of_builtin
1207 ? sizeof known_suffixes / sizeof *known_suffixes - 2
1208 : 0] = z_lower;
1209 suf = known_suffixes + suffix_of_builtin;
1211 nlen = strlen(name);
1212 if (nlen <= MAX_SUFFIX+2) {
1213 strcpy(suffix, name);
1214 } else {
1215 strcpy(suffix, name+nlen-MAX_SUFFIX-2);
1217 strlwr(suffix);
1218 slen = strlen(suffix);
1219 char *match = NULL;
1220 do {
1221 int s = strlen(*suf);
1222 if (slen > s && ! ISSLASH (suffix[slen - s - 1])
1223 && strequ(suffix + slen - s, *suf)) {
1224 match = name+nlen-s;
1225 break;
1227 } while (*++suf != NULL);
1228 free(z_lower);
1230 return match;
1234 /* Open file NAME with the given flags and store its status
1235 into *ST. Return a file descriptor to the newly opened file, or -1
1236 (setting errno) on failure. */
1237 static int
1238 open_and_stat (char *name, int flags, struct stat *st)
1240 int fd;
1241 int atfd = AT_FDCWD;
1242 char const *base = name;
1244 /* Refuse to follow symbolic links unless -c or -f. */
1245 if (!to_stdout && !force)
1247 if (HAVE_WORKING_O_NOFOLLOW)
1248 flags |= O_NOFOLLOW;
1249 else
1251 #ifdef S_ISLNK
1252 if (lstat (name, st) != 0)
1253 return -1;
1254 else if (S_ISLNK (st->st_mode))
1256 errno = ELOOP;
1257 return -1;
1259 #endif
1263 if (!keep)
1265 char const *b = last_component (name);
1266 int f = atdir_set (name, b - name);
1267 if (0 <= f)
1269 base = b;
1270 atfd = f;
1274 fd = openat (atfd, base, flags);
1275 if (0 <= fd && fstat (fd, st) != 0)
1277 int e = errno;
1278 close (fd);
1279 errno = e;
1280 return -1;
1282 return fd;
1286 /* ========================================================================
1287 * Set ifname to the input file name (with a suffix appended if necessary)
1288 * and istat to its stats. For decompression, if no file exists with the
1289 * original name, try adding successively z_suffix, .gz, .z, -z and .Z.
1290 * For MSDOS, we try only z_suffix and z.
1291 * Return an open file descriptor or -1.
1293 static int
1294 open_input_file (char *iname, struct stat *sbuf)
1296 int ilen; /* strlen(ifname) */
1297 int z_suffix_errno = 0;
1298 static char const *suffixes[] = {NULL, ".gz", ".z", "-z", ".Z", NULL};
1299 char const **suf = suffixes;
1300 char const *s;
1301 #ifdef NO_MULTIPLE_DOTS
1302 char *dot; /* pointer to ifname extension, or NULL */
1303 #endif
1304 int fd;
1305 int open_flags = (O_RDONLY | O_NONBLOCK | O_NOCTTY
1306 | (ascii && !decompress ? 0 : O_BINARY));
1308 *suf = z_suffix;
1310 if (sizeof ifname - 1 <= strlen (iname))
1311 goto name_too_long;
1313 strcpy(ifname, iname);
1315 /* If input file exists, return OK. */
1316 fd = open_and_stat (ifname, open_flags, sbuf);
1317 if (0 <= fd)
1318 return fd;
1320 if (!decompress || errno != ENOENT) {
1321 progerror(ifname);
1322 return -1;
1324 /* File.ext doesn't exist. Try adding a suffix. */
1325 s = get_suffix(ifname);
1326 if (s != NULL) {
1327 progerror(ifname); /* ifname already has z suffix and does not exist */
1328 return -1;
1330 #ifdef NO_MULTIPLE_DOTS
1331 dot = strrchr(ifname, '.');
1332 if (dot == NULL) {
1333 strcat(ifname, ".");
1334 dot = strrchr(ifname, '.');
1336 #endif
1337 ilen = strlen(ifname);
1338 if (strequ(z_suffix, ".gz")) suf++;
1340 /* Search for all suffixes */
1341 do {
1342 char const *s0 = s = *suf;
1343 strcpy (ifname, iname);
1344 #ifdef NO_MULTIPLE_DOTS
1345 if (*s == '.') s++;
1346 if (*dot == '\0') strcpy (dot, ".");
1347 #endif
1348 #ifdef MAX_EXT_CHARS
1349 if (MAX_EXT_CHARS < strlen (s) + strlen (dot + 1))
1350 dot[MAX_EXT_CHARS + 1 - strlen (s)] = '\0';
1351 #endif
1352 if (sizeof ifname <= ilen + strlen (s))
1353 goto name_too_long;
1354 strcat(ifname, s);
1355 fd = open_and_stat (ifname, open_flags, sbuf);
1356 if (0 <= fd)
1357 return fd;
1358 if (errno != ENOENT)
1360 progerror (ifname);
1361 return -1;
1363 if (strequ (s0, z_suffix))
1364 z_suffix_errno = errno;
1365 } while (*++suf != NULL);
1367 /* No suffix found, complain using z_suffix: */
1368 strcpy(ifname, iname);
1369 #ifdef NO_MULTIPLE_DOTS
1370 if (*dot == '\0') strcpy(dot, ".");
1371 #endif
1372 #ifdef MAX_EXT_CHARS
1373 if (MAX_EXT_CHARS < z_len + strlen (dot + 1))
1374 dot[MAX_EXT_CHARS + 1 - z_len] = '\0';
1375 #endif
1376 strcat(ifname, z_suffix);
1377 errno = z_suffix_errno;
1378 progerror(ifname);
1379 return -1;
1381 name_too_long:
1382 fprintf (stderr, "%s: %s: file name too long\n", program_name, iname);
1383 exit_code = ERROR;
1384 return -1;
1387 /* ========================================================================
1388 * Generate ofname given ifname. Return OK, or WARNING if file must be skipped.
1389 * Sets save_orig_name to true if the file name has been truncated.
1391 static int
1392 make_ofname ()
1394 char *suff; /* ofname z suffix */
1396 strcpy(ofname, ifname);
1397 /* strip a version number if any and get the gzip suffix if present: */
1398 suff = get_suffix(ofname);
1400 if (decompress) {
1401 if (suff == NULL) {
1402 /* With -t or -l, try all files (even without .gz suffix)
1403 * except with -r (behave as with just -dr).
1405 if (!recursive && test)
1406 return OK;
1408 /* Avoid annoying messages with -r */
1409 if (verbose || (!recursive && !quiet)) {
1410 WARN((stderr,"%s: %s: unknown suffix -- ignored\n",
1411 program_name, ifname));
1413 return WARNING;
1415 /* Make a special case for .tgz and .taz: */
1416 strlwr(suff);
1417 if (strequ(suff, ".tgz") || strequ(suff, ".taz")) {
1418 strcpy(suff, ".tar");
1419 } else {
1420 *suff = '\0'; /* strip the z suffix */
1422 /* ofname might be changed later if infile contains an original name */
1424 } else if (suff && ! force) {
1425 /* Avoid annoying messages with -r (see treat_dir()) */
1426 if (verbose || (!recursive && !quiet)) {
1427 /* Don't use WARN, as it affects exit status. */
1428 fprintf (stderr, "%s: %s already has %s suffix -- unchanged\n",
1429 program_name, ifname, suff);
1431 return WARNING;
1432 } else {
1433 save_orig_name = 0;
1435 #ifdef NO_MULTIPLE_DOTS
1436 suff = strrchr(ofname, '.');
1437 if (suff == NULL) {
1438 if (sizeof ofname <= strlen (ofname) + 1)
1439 goto name_too_long;
1440 strcat(ofname, ".");
1441 # ifdef MAX_EXT_CHARS
1442 if (strequ(z_suffix, "z")) {
1443 if (sizeof ofname <= strlen (ofname) + 2)
1444 goto name_too_long;
1445 strcat(ofname, "gz"); /* enough room */
1446 return OK;
1448 /* On the Atari and some versions of MSDOS,
1449 * ENAMETOOLONG does not work correctly. So we
1450 * must truncate here.
1452 } else if (strlen(suff)-1 + z_len > MAX_SUFFIX) {
1453 suff[MAX_SUFFIX+1-z_len] = '\0';
1454 save_orig_name = 1;
1455 # endif
1457 #endif /* NO_MULTIPLE_DOTS */
1458 if (sizeof ofname <= strlen (ofname) + z_len)
1459 goto name_too_long;
1460 strcat(ofname, z_suffix);
1462 } /* decompress ? */
1463 return OK;
1465 name_too_long:
1466 WARN ((stderr, "%s: %s: file name too long\n", program_name, ifname));
1467 return WARNING;
1470 /* Discard NBYTES input bytes from the input, or up through the next
1471 zero byte if NBYTES == (size_t) -1. If FLAGS say that the header
1472 CRC should be computed, update the CRC accordingly. */
1473 static void
1474 discard_input_bytes (size_t nbytes, unsigned int flags)
1476 while (nbytes != 0)
1478 uch c = get_byte ();
1479 if (flags & HEADER_CRC)
1480 updcrc (&c, 1);
1481 if (nbytes != (size_t) -1)
1482 nbytes--;
1483 else if (! c)
1484 break;
1488 /* ========================================================================
1489 * Check the magic number of the input file and update ofname if an
1490 * original name was given and to_stdout is not set.
1491 * Return the compression method, -1 for error, -2 for warning.
1492 * Set inptr to the offset of the next byte to be processed.
1493 * Updates time_stamp if there is one and neither -m nor -n is used.
1494 * This function may be called repeatedly for an input file consisting
1495 * of several contiguous gzip'ed members.
1496 * 'in' is the input file descriptor.
1497 * IN assertions: there is at least one remaining compressed member.
1498 * If the member is a zip file, it must be the only one.
1500 static int
1501 get_method (int in)
1503 uch flags; /* compression flags */
1504 uch magic[10]; /* magic header */
1505 int imagic0; /* first magic byte or EOF */
1506 int imagic1; /* like magic[1], but can represent EOF */
1507 ulg stamp; /* timestamp */
1509 /* If --force and --stdout, zcat == cat, so do not complain about
1510 * premature end of file: use try_byte instead of get_byte.
1512 if (force && to_stdout) {
1513 imagic0 = try_byte();
1514 magic[0] = imagic0;
1515 imagic1 = try_byte ();
1516 magic[1] = imagic1;
1517 /* If try_byte returned EOF, magic[1] == (char) EOF. */
1518 } else {
1519 magic[0] = get_byte ();
1520 imagic0 = 0;
1521 if (magic[0]) {
1522 magic[1] = get_byte ();
1523 imagic1 = 0; /* avoid lint warning */
1524 } else {
1525 imagic1 = try_byte ();
1526 magic[1] = imagic1;
1529 method = -1; /* unknown yet */
1530 part_nb++; /* number of parts in gzip file */
1531 header_bytes = 0;
1532 last_member = 0;
1533 /* assume multiple members in gzip file except for record oriented I/O */
1535 if (memcmp(magic, GZIP_MAGIC, 2) == 0
1536 || memcmp(magic, OLD_GZIP_MAGIC, 2) == 0) {
1538 method = (int)get_byte();
1539 if (method != DEFLATED) {
1540 fprintf(stderr,
1541 "%s: %s: unknown method %d -- not supported\n",
1542 program_name, ifname, method);
1543 exit_code = ERROR;
1544 return -1;
1546 work = unzip;
1547 flags = (uch)get_byte();
1549 if ((flags & ENCRYPTED) != 0) {
1550 fprintf(stderr,
1551 "%s: %s is encrypted -- not supported\n",
1552 program_name, ifname);
1553 exit_code = ERROR;
1554 return -1;
1556 if ((flags & RESERVED) != 0) {
1557 fprintf(stderr,
1558 "%s: %s has flags 0x%x -- not supported\n",
1559 program_name, ifname, flags);
1560 exit_code = ERROR;
1561 if (force <= 1) return -1;
1563 stamp = (ulg)get_byte();
1564 stamp |= ((ulg)get_byte()) << 8;
1565 stamp |= ((ulg)get_byte()) << 16;
1566 stamp |= ((ulg)get_byte()) << 24;
1567 if (stamp != 0 && !no_time)
1569 if (stamp <= TYPE_MAXIMUM (time_t))
1571 time_stamp.tv_sec = stamp;
1572 time_stamp.tv_nsec = 0;
1574 else
1576 WARN ((stderr,
1577 "%s: %s: MTIME %lu out of range for this platform\n",
1578 program_name, ifname, stamp));
1579 time_stamp.tv_sec = TYPE_MAXIMUM (time_t);
1580 time_stamp.tv_nsec = TIMESPEC_RESOLUTION - 1;
1584 magic[8] = get_byte (); /* Ignore extra flags. */
1585 magic[9] = get_byte (); /* Ignore OS type. */
1587 if (flags & HEADER_CRC)
1589 magic[2] = DEFLATED;
1590 magic[3] = flags;
1591 magic[4] = stamp & 0xff;
1592 magic[5] = (stamp >> 8) & 0xff;
1593 magic[6] = (stamp >> 16) & 0xff;
1594 magic[7] = stamp >> 24;
1595 updcrc (NULL, 0);
1596 updcrc (magic, 10);
1599 if ((flags & EXTRA_FIELD) != 0) {
1600 uch lenbuf[2];
1601 unsigned int len = lenbuf[0] = get_byte ();
1602 len |= (lenbuf[1] = get_byte ()) << 8;
1603 if (verbose) {
1604 fprintf(stderr,"%s: %s: extra field of %u bytes ignored\n",
1605 program_name, ifname, len);
1607 if (flags & HEADER_CRC)
1608 updcrc (lenbuf, 2);
1609 discard_input_bytes (len, flags);
1612 /* Get original file name if it was truncated */
1613 if ((flags & ORIG_NAME) != 0) {
1614 if (no_name || (to_stdout && !list) || part_nb > 1) {
1615 /* Discard the old name */
1616 discard_input_bytes (-1, flags);
1617 } else {
1618 /* Copy the base name. Keep a directory prefix intact. */
1619 char *p = gzip_base_name (ofname);
1620 char *base = p;
1621 for (;;) {
1622 *p = (char) get_byte ();
1623 if (*p++ == '\0') break;
1624 if (p >= ofname+sizeof(ofname)) {
1625 gzip_error ("corrupted input -- file name too large");
1628 if (flags & HEADER_CRC)
1629 updcrc ((uch *) base, p - base);
1630 p = gzip_base_name (base);
1631 memmove (base, p, strlen (p) + 1);
1632 /* If necessary, adapt the name to local OS conventions: */
1633 if (!list) {
1634 MAKE_LEGAL_NAME(base);
1635 if (base) list=0; /* avoid warning about unused variable */
1637 } /* no_name || to_stdout */
1638 } /* ORIG_NAME */
1640 /* Discard file comment if any */
1641 if ((flags & COMMENT) != 0) {
1642 discard_input_bytes (-1, flags);
1645 if (flags & HEADER_CRC)
1647 unsigned int crc16 = updcrc (magic, 0) & 0xffff;
1648 unsigned int header16 = get_byte ();
1649 header16 |= ((unsigned int) get_byte ()) << 8;
1650 if (header16 != crc16)
1652 fprintf (stderr,
1653 "%s: %s: header checksum 0x%04x != computed checksum 0x%04x\n",
1654 program_name, ifname, header16, crc16);
1655 exit_code = ERROR;
1656 if (force <= 1)
1657 return -1;
1661 if (part_nb == 1) {
1662 header_bytes = inptr + 2*4; /* include crc and size */
1665 } else if (memcmp(magic, PKZIP_MAGIC, 2) == 0 && inptr == 2
1666 && memcmp((char*)inbuf, PKZIP_MAGIC, 4) == 0) {
1667 /* To simplify the code, we support a zip file when alone only.
1668 * We are thus guaranteed that the entire local header fits in inbuf.
1670 inptr = 0;
1671 work = unzip;
1672 if (check_zipfile(in) != OK) return -1;
1673 /* check_zipfile may get ofname from the local header */
1674 last_member = 1;
1676 } else if (memcmp(magic, PACK_MAGIC, 2) == 0) {
1677 work = unpack;
1678 method = PACKED;
1680 } else if (memcmp(magic, LZW_MAGIC, 2) == 0) {
1681 work = unlzw;
1682 method = COMPRESSED;
1683 last_member = 1;
1685 } else if (memcmp(magic, LZH_MAGIC, 2) == 0) {
1686 work = unlzh;
1687 method = LZHED;
1688 last_member = 1;
1690 } else if (force && to_stdout && !list) { /* pass input unchanged */
1691 method = STORED;
1692 work = copy;
1693 if (imagic1 != EOF)
1694 inptr--;
1695 last_member = 1;
1696 if (imagic0 != EOF) {
1697 write_buf (STDOUT_FILENO, magic, 1);
1700 if (method >= 0) return method;
1702 if (part_nb == 1) {
1703 fprintf (stderr, "\n%s: %s: not in gzip format\n",
1704 program_name, ifname);
1705 exit_code = ERROR;
1706 return -1;
1707 } else {
1708 if (magic[0] == 0)
1710 int inbyte;
1711 for (inbyte = imagic1; inbyte == 0; inbyte = try_byte ())
1712 continue;
1713 if (inbyte == EOF)
1715 if (verbose)
1716 WARN ((stderr, "\n%s: %s: decompression OK, trailing zero bytes ignored\n",
1717 program_name, ifname));
1718 return -3;
1722 WARN((stderr, "\n%s: %s: decompression OK, trailing garbage ignored\n",
1723 program_name, ifname));
1724 return -2;
1728 /* ========================================================================
1729 * Display the characteristics of the compressed file.
1730 * If the given method is < 0, display the accumulated totals.
1731 * IN assertions: time_stamp, header_bytes and ifile_size are initialized.
1733 static void
1734 do_list (int method)
1736 ulg crc; /* original crc */
1737 static int first_time = 1;
1738 static char const *const methods[MAX_METHODS] = {
1739 "store", /* 0 */
1740 "compr", /* 1 */
1741 "pack ", /* 2 */
1742 "lzh ", /* 3 */
1743 "", "", "", "", /* 4 to 7 reserved */
1744 "defla"}; /* 8 */
1745 int positive_off_t_width = INT_STRLEN_BOUND (off_t) - 1;
1747 if (first_time && method >= 0) {
1748 first_time = 0;
1749 if (verbose) {
1750 printf("method crc date time ");
1752 if (!quiet) {
1753 printf("%*.*s %*.*s ratio uncompressed_name\n",
1754 positive_off_t_width, positive_off_t_width, "compressed",
1755 positive_off_t_width, positive_off_t_width, "uncompressed");
1757 } else if (method < 0) {
1758 if (total_in <= 0 || total_out <= 0) return;
1759 if (verbose) {
1760 printf(" ");
1762 if (verbose || !quiet) {
1763 fprint_off(stdout, total_in, positive_off_t_width);
1764 printf(" ");
1765 fprint_off(stdout, total_out, positive_off_t_width);
1766 printf(" ");
1768 display_ratio(total_out-(total_in-header_bytes), total_out, stdout);
1769 /* header_bytes is not meaningful but used to ensure the same
1770 * ratio if there is a single file.
1772 printf(" (totals)\n");
1773 return;
1775 crc = (ulg)~0; /* unknown */
1777 if (method == DEFLATED && !last_member) {
1778 crc = unzip_crc;
1781 if (verbose)
1783 static char const month_abbr[][4]
1784 = { "Jan", "Feb", "Mar", "Apr", "May", "Jun",
1785 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec" };
1786 struct tm *tm = localtime (&time_stamp.tv_sec);
1787 printf ("%5s %08lx ", methods[method], crc);
1788 if (tm)
1789 printf ("%s%3d %02d:%02d ", month_abbr[tm->tm_mon],
1790 tm->tm_mday, tm->tm_hour, tm->tm_min);
1791 else
1792 printf ("??? ?? ??:?? ");
1794 fprint_off(stdout, bytes_in, positive_off_t_width);
1795 printf(" ");
1796 fprint_off(stdout, bytes_out, positive_off_t_width);
1797 printf(" ");
1798 if (bytes_in == -1L) {
1799 total_in = -1L;
1800 bytes_in = bytes_out = header_bytes = 0;
1801 } else if (total_in >= 0) {
1802 total_in += bytes_in;
1804 if (bytes_out == -1L) {
1805 total_out = -1L;
1806 bytes_in = bytes_out = header_bytes = 0;
1807 } else if (total_out >= 0) {
1808 total_out += bytes_out;
1810 display_ratio(bytes_out-(bytes_in-header_bytes), bytes_out, stdout);
1811 printf(" %s\n", ofname);
1814 /* ========================================================================
1815 * Shorten the given name by one character, or replace a .tar extension
1816 * with .tgz. Truncate the last part of the name which is longer than
1817 * MIN_PART characters: 1234.678.012.gz -> 123.678.012.gz. If the name
1818 * has only parts shorter than MIN_PART truncate the longest part.
1819 * For decompression, just remove the last character of the name.
1821 * IN assertion: for compression, the suffix of the given name is z_suffix.
1823 static void
1824 shorten_name (char *name)
1826 int len; /* length of name without z_suffix */
1827 char *trunc = NULL; /* character to be truncated */
1828 int plen; /* current part length */
1829 int min_part = MIN_PART; /* current minimum part length */
1830 char *p;
1832 len = strlen(name);
1833 if (decompress) {
1834 if (len <= 1)
1835 gzip_error ("name too short");
1836 name[len-1] = '\0';
1837 return;
1839 p = get_suffix(name);
1840 if (! p)
1841 gzip_error ("can't recover suffix\n");
1842 *p = '\0';
1843 save_orig_name = 1;
1845 /* compress 1234567890.tar to 1234567890.tgz */
1846 if (len > 4 && strequ(p-4, ".tar")) {
1847 strcpy(p-4, ".tgz");
1848 return;
1850 /* Try keeping short extensions intact:
1851 * 1234.678.012.gz -> 123.678.012.gz
1853 do {
1854 p = last_component (name);
1855 while (*p) {
1856 plen = strcspn(p, PART_SEP);
1857 p += plen;
1858 if (plen > min_part) trunc = p-1;
1859 if (*p) p++;
1861 } while (trunc == NULL && --min_part != 0);
1863 if (trunc != NULL) {
1864 do {
1865 trunc[0] = trunc[1];
1866 } while (*trunc++);
1867 trunc--;
1868 } else {
1869 trunc = strrchr(name, PART_SEP[0]);
1870 if (!trunc)
1871 gzip_error ("internal error in shorten_name");
1872 if (trunc[1] == '\0') trunc--; /* force truncation */
1874 strcpy(trunc, z_suffix);
1877 /* ========================================================================
1878 * The compressed file already exists, so ask for confirmation.
1879 * Return ERROR if the file must be skipped.
1881 static int
1882 check_ofname ()
1884 /* Ask permission to overwrite the existing file */
1885 if (!force) {
1886 int ok = 0;
1887 fprintf (stderr, "%s: %s already exists;", program_name, ofname);
1888 if (foreground && (presume_input_tty || isatty (STDIN_FILENO))) {
1889 fprintf(stderr, " do you wish to overwrite (y or n)? ");
1890 fflush(stderr);
1891 ok = yesno();
1893 if (!ok) {
1894 fprintf(stderr, "\tnot overwritten\n");
1895 if (exit_code == OK) exit_code = WARNING;
1896 return ERROR;
1899 if (xunlink (ofname)) {
1900 progerror(ofname);
1901 return ERROR;
1903 return OK;
1906 /* Change the owner and group of a file. FD is a file descriptor for
1907 the file and NAME its name. Change it to user UID and to group GID.
1908 If UID or GID is -1, though, do not change the corresponding user
1909 or group. */
1910 #if ! (HAVE_FCHOWN || HAVE_CHOWN)
1911 /* The types uid_t and gid_t do not exist on mingw, so don't assume them. */
1912 # define do_chown(fd, name, uid, gid) ((void) 0)
1913 #else
1914 static void
1915 do_chown (int fd, char const *name, uid_t uid, gid_t gid)
1917 # if HAVE_FCHOWN
1918 ignore_value (fchown (fd, uid, gid));
1919 # else
1920 ignore_value (chown (name, uid, gid));
1921 # endif
1923 #endif
1925 /* ========================================================================
1926 * Copy modes, times, ownership from input file to output file.
1927 * IN assertion: to_stdout is false.
1929 static void
1930 copy_stat (struct stat *ifstat)
1932 mode_t mode = ifstat->st_mode & S_IRWXUGO;
1933 int r;
1935 #ifndef NO_UTIME
1936 bool restoring;
1937 struct timespec timespec[2];
1938 timespec[0] = get_stat_atime (ifstat);
1939 timespec[1] = get_stat_mtime (ifstat);
1940 restoring = (decompress && 0 <= time_stamp.tv_nsec
1941 && ! (timespec[1].tv_sec == time_stamp.tv_sec
1942 && timespec[1].tv_nsec == time_stamp.tv_nsec));
1943 if (restoring)
1944 timespec[1] = time_stamp;
1946 if (fdutimens (ofd, ofname, timespec) == 0)
1948 if (restoring && 1 < verbose) {
1949 fprintf(stderr, "%s: timestamp restored\n", ofname);
1952 else
1954 int e = errno;
1955 WARN ((stderr, "%s: ", program_name));
1956 if (!quiet)
1958 errno = e;
1959 perror (ofname);
1962 #endif
1964 /* Change the group first, then the permissions, then the owner.
1965 That way, the permissions will be correct on systems that allow
1966 users to give away files, without introducing a security hole.
1967 Security depends on permissions not containing the setuid or
1968 setgid bits. */
1970 do_chown (ofd, ofname, -1, ifstat->st_gid);
1972 #if HAVE_FCHMOD
1973 r = fchmod (ofd, mode);
1974 #else
1975 r = chmod (ofname, mode);
1976 #endif
1977 if (r != 0) {
1978 int e = errno;
1979 WARN ((stderr, "%s: ", program_name));
1980 if (!quiet) {
1981 errno = e;
1982 perror(ofname);
1986 do_chown (ofd, ofname, ifstat->st_uid, -1);
1989 #if ! NO_DIR
1991 /* ========================================================================
1992 * Recurse through the given directory.
1994 static void
1995 treat_dir (int fd, char *dir)
1997 DIR *dirp;
1998 char nbuf[MAX_PATH_LEN];
1999 char *entries;
2000 char const *entry;
2001 size_t entrylen;
2003 dirp = fdopendir (fd);
2005 if (dirp == NULL) {
2006 progerror(dir);
2007 close (fd);
2008 return ;
2011 entries = streamsavedir (dirp, SAVEDIR_SORT_NONE);
2012 if (! entries)
2013 progerror (dir);
2014 if (closedir (dirp) != 0)
2015 progerror (dir);
2016 if (! entries)
2017 return;
2019 for (entry = entries; *entry; entry += entrylen + 1) {
2020 size_t len = strlen (dir);
2021 entrylen = strlen (entry);
2022 if (strequ (entry, ".") || strequ (entry, ".."))
2023 continue;
2024 if (len + entrylen < MAX_PATH_LEN - 2) {
2025 strcpy(nbuf,dir);
2026 if (*last_component (nbuf) && !ISSLASH (nbuf[len - 1]))
2027 nbuf[len++] = '/';
2028 strcpy (nbuf + len, entry);
2029 treat_file(nbuf);
2030 } else {
2031 fprintf(stderr,"%s: %s/%s: pathname too long\n",
2032 program_name, dir, entry);
2033 exit_code = ERROR;
2036 free (entries);
2038 #endif /* ! NO_DIR */
2040 /* Make sure signals get handled properly. */
2042 static void
2043 install_signal_handlers ()
2045 int nsigs = sizeof handled_sig / sizeof handled_sig[0];
2046 int i;
2047 struct sigaction act;
2049 sigemptyset (&caught_signals);
2050 for (i = 0; i < nsigs; i++)
2052 sigaction (handled_sig[i], NULL, &act);
2053 if (act.sa_handler != SIG_IGN)
2054 sigaddset (&caught_signals, handled_sig[i]);
2057 act.sa_handler = abort_gzip_signal;
2058 act.sa_mask = caught_signals;
2059 act.sa_flags = 0;
2061 for (i = 0; i < nsigs; i++)
2062 if (sigismember (&caught_signals, handled_sig[i]))
2064 if (i == 0)
2065 foreground = 1;
2066 sigaction (handled_sig[i], &act, NULL);
2070 /* ========================================================================
2071 * Free all dynamically allocated variables and exit with the given code.
2073 static void
2074 do_exit (int exitcode)
2076 static int in_exit = 0;
2078 if (in_exit) exit(exitcode);
2079 in_exit = 1;
2080 free(env);
2081 env = NULL;
2082 FREE(inbuf);
2083 FREE(outbuf);
2084 FREE(d_buf);
2085 FREE(window);
2086 #ifndef MAXSEG_64K
2087 FREE(tab_prefix);
2088 #else
2089 FREE(tab_prefix0);
2090 FREE(tab_prefix1);
2091 #endif
2092 exit(exitcode);
2095 static void
2096 finish_out ()
2098 if (fclose (stdout) != 0)
2099 write_error ();
2100 do_exit (OK);
2103 /* ========================================================================
2104 * Close and unlink the output file.
2106 static void
2107 remove_output_file (bool signals_already_blocked)
2109 int fd;
2110 sigset_t oldset;
2112 if (!signals_already_blocked)
2113 sigprocmask (SIG_BLOCK, &caught_signals, &oldset);
2114 fd = remove_ofname_fd;
2115 if (0 <= fd)
2117 char fname[MAX_PATH_LEN];
2118 remove_ofname_fd = -1;
2119 close (fd);
2120 volatile_strcpy (fname, remove_ofname);
2121 xunlink (fname);
2123 if (!signals_already_blocked)
2124 sigprocmask (SIG_SETMASK, &oldset, NULL);
2127 /* ========================================================================
2128 * Error handler.
2130 void
2131 abort_gzip ()
2133 remove_output_file (false);
2134 do_exit(ERROR);
2137 /* ========================================================================
2138 * Signal handler.
2140 static void
2141 abort_gzip_signal (int sig)
2143 remove_output_file (true);
2144 if (sig == exiting_signal)
2145 _exit (WARNING);
2146 signal (sig, SIG_DFL);
2147 raise (sig);