ldconfig(8): clean up manual page
[dragonfly.git] / contrib / file / src / compress.c
blob2f789cd2bc44825a3a7e1a4954814c46cc02a406
1 /*
2 * Copyright (c) Ian F. Darwin 1986-1995.
3 * Software written by Ian F. Darwin and others;
4 * maintained 1995-present by Christos Zoulas and others.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 * notice immediately at the beginning of the file, without modification,
11 * this list of conditions, and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
20 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26 * SUCH DAMAGE.
29 * compress routines:
30 * zmagic() - returns 0 if not recognized, uncompresses and prints
31 * information if recognized
32 * uncompress(method, old, n, newch) - uncompress old into new,
33 * using method, return sizeof new
35 #include "file.h"
37 #ifndef lint
38 FILE_RCSID("@(#)$File: compress.c,v 1.105 2017/05/25 00:13:03 christos Exp $")
39 #endif
41 #include "magic.h"
42 #include <stdlib.h>
43 #ifdef HAVE_UNISTD_H
44 #include <unistd.h>
45 #endif
46 #include <string.h>
47 #include <errno.h>
48 #include <ctype.h>
49 #include <stdarg.h>
50 #ifdef HAVE_SIGNAL_H
51 #include <signal.h>
52 # ifndef HAVE_SIG_T
53 typedef void (*sig_t)(int);
54 # endif /* HAVE_SIG_T */
55 #endif
56 #if !defined(__MINGW32__) && !defined(WIN32)
57 #include <sys/ioctl.h>
58 #endif
59 #ifdef HAVE_SYS_WAIT_H
60 #include <sys/wait.h>
61 #endif
62 #if defined(HAVE_SYS_TIME_H)
63 #include <sys/time.h>
64 #endif
65 #if defined(HAVE_ZLIB_H) && defined(ZLIBSUPPORT)
66 #define BUILTIN_DECOMPRESS
67 #include <zlib.h>
68 #endif
69 #ifdef DEBUG
70 int tty = -1;
71 #define DPRINTF(...) do { \
72 if (tty == -1) \
73 tty = open("/dev/tty", O_RDWR); \
74 if (tty == -1) \
75 abort(); \
76 dprintf(tty, __VA_ARGS__); \
77 } while (/*CONSTCOND*/0)
78 #else
79 #define DPRINTF(...)
80 #endif
82 #ifdef ZLIBSUPPORT
84 * The following python code is not really used because ZLIBSUPPORT is only
85 * defined if we have a built-in zlib, and the built-in zlib handles that.
86 * That is not true for android where we have zlib.h and not -lz.
88 static const char zlibcode[] =
89 "import sys, zlib; sys.stdout.write(zlib.decompress(sys.stdin.read()))";
91 static const char *zlib_args[] = { "python", "-c", zlibcode, NULL };
93 static int
94 zlibcmp(const unsigned char *buf)
96 unsigned short x = 1;
97 unsigned char *s = CAST(unsigned char *, CAST(void *, &x));
99 if ((buf[0] & 0xf) != 8 || (buf[0] & 0x80) != 0)
100 return 0;
101 if (s[0] != 1) /* endianness test */
102 x = buf[0] | (buf[1] << 8);
103 else
104 x = buf[1] | (buf[0] << 8);
105 if (x % 31)
106 return 0;
107 return 1;
109 #endif
111 #define gzip_flags "-cd"
112 #define lrzip_flags "-do"
113 #define lzip_flags gzip_flags
115 static const char *gzip_args[] = {
116 "gzip", gzip_flags, NULL
118 static const char *uncompress_args[] = {
119 "uncompress", "-c", NULL
121 static const char *bzip2_args[] = {
122 "bzip2", "-cd", NULL
124 static const char *lzip_args[] = {
125 "lzip", lzip_flags, NULL
127 static const char *xz_args[] = {
128 "xz", "-cd", NULL
130 static const char *lrzip_args[] = {
131 "lrzip", lrzip_flags, NULL
133 static const char *lz4_args[] = {
134 "lz4", "-cd", NULL
136 static const char *zstd_args[] = {
137 "zstd", "-cd", NULL
140 private const struct {
141 const void *magic;
142 size_t maglen;
143 const char **argv;
144 } compr[] = {
145 { "\037\235", 2, gzip_args }, /* compressed */
146 /* Uncompress can get stuck; so use gzip first if we have it
147 * Idea from Damien Clark, thanks! */
148 { "\037\235", 2, uncompress_args }, /* compressed */
149 { "\037\213", 2, gzip_args }, /* gzipped */
150 { "\037\236", 2, gzip_args }, /* frozen */
151 { "\037\240", 2, gzip_args }, /* SCO LZH */
152 /* the standard pack utilities do not accept standard input */
153 { "\037\036", 2, gzip_args }, /* packed */
154 { "PK\3\4", 4, gzip_args }, /* pkzipped, */
155 /* ...only first file examined */
156 { "BZh", 3, bzip2_args }, /* bzip2-ed */
157 { "LZIP", 4, lzip_args }, /* lzip-ed */
158 { "\3757zXZ\0", 6, xz_args }, /* XZ Utils */
159 { "LRZI", 4, lrzip_args }, /* LRZIP */
160 { "\004\"M\030",4, lz4_args }, /* LZ4 */
161 { "\x28\xB5\x2F\xFD", 4, zstd_args }, /* zstd */
162 #ifdef ZLIBSUPPORT
163 { RCAST(const void *, zlibcmp), 0, zlib_args }, /* zlib */
164 #endif
167 #define OKDATA 0
168 #define NODATA 1
169 #define ERRDATA 2
171 private ssize_t swrite(int, const void *, size_t);
172 #if HAVE_FORK
173 private size_t ncompr = sizeof(compr) / sizeof(compr[0]);
174 private int uncompressbuf(int, size_t, size_t, const unsigned char *,
175 unsigned char **, size_t *);
176 #ifdef BUILTIN_DECOMPRESS
177 private int uncompresszlib(const unsigned char *, unsigned char **, size_t,
178 size_t *, int);
179 private int uncompressgzipped(const unsigned char *, unsigned char **, size_t,
180 size_t *);
181 #endif
182 static int makeerror(unsigned char **, size_t *, const char *, ...)
183 __attribute__((__format__(__printf__, 3, 4)));
184 private const char *methodname(size_t);
186 protected int
187 file_zmagic(struct magic_set *ms, int fd, const char *name,
188 const unsigned char *buf, size_t nbytes)
190 unsigned char *newbuf = NULL;
191 size_t i, nsz;
192 char *rbuf;
193 file_pushbuf_t *pb;
194 int urv, prv, rv = 0;
195 int mime = ms->flags & MAGIC_MIME;
196 #ifdef HAVE_SIGNAL_H
197 sig_t osigpipe;
198 #endif
200 if ((ms->flags & MAGIC_COMPRESS) == 0)
201 return 0;
203 #ifdef HAVE_SIGNAL_H
204 osigpipe = signal(SIGPIPE, SIG_IGN);
205 #endif
206 for (i = 0; i < ncompr; i++) {
207 int zm;
208 if (nbytes < compr[i].maglen)
209 continue;
210 #ifdef ZLIBSUPPORT
211 if (compr[i].maglen == 0)
212 zm = (RCAST(int (*)(const unsigned char *),
213 CCAST(void *, compr[i].magic)))(buf);
214 else
215 #endif
216 zm = memcmp(buf, compr[i].magic, compr[i].maglen) == 0;
218 if (!zm)
219 continue;
220 nsz = nbytes;
221 urv = uncompressbuf(fd, ms->bytes_max, i, buf, &newbuf, &nsz);
222 DPRINTF("uncompressbuf = %d, %s, %zu\n", urv, (char *)newbuf,
223 nsz);
224 switch (urv) {
225 case OKDATA:
226 case ERRDATA:
228 ms->flags &= ~MAGIC_COMPRESS;
229 if (urv == ERRDATA)
230 prv = file_printf(ms, "%s ERROR: %s",
231 methodname(i), newbuf);
232 else
233 prv = file_buffer(ms, -1, name, newbuf, nsz);
234 if (prv == -1)
235 goto error;
236 rv = 1;
237 if ((ms->flags & MAGIC_COMPRESS_TRANSP) != 0)
238 goto out;
239 if (mime != MAGIC_MIME && mime != 0)
240 goto out;
241 if ((file_printf(ms,
242 mime ? " compressed-encoding=" : " (")) == -1)
243 goto error;
244 if ((pb = file_push_buffer(ms)) == NULL)
245 goto error;
247 * XXX: If file_buffer fails here, we overwrite
248 * the compressed text. FIXME.
250 if (file_buffer(ms, -1, NULL, buf, nbytes) == -1)
251 goto error;
252 if ((rbuf = file_pop_buffer(ms, pb)) != NULL) {
253 if (file_printf(ms, "%s", rbuf) == -1) {
254 free(rbuf);
255 goto error;
257 free(rbuf);
259 if (!mime && file_printf(ms, ")") == -1)
260 goto error;
261 /*FALLTHROUGH*/
262 case NODATA:
263 break;
264 default:
265 abort();
266 /*NOTREACHED*/
267 error:
268 rv = -1;
269 break;
272 out:
273 DPRINTF("rv = %d\n", rv);
275 #ifdef HAVE_SIGNAL_H
276 (void)signal(SIGPIPE, osigpipe);
277 #endif
278 free(newbuf);
279 ms->flags |= MAGIC_COMPRESS;
280 DPRINTF("Zmagic returns %d\n", rv);
281 return rv;
283 #endif
285 * `safe' write for sockets and pipes.
287 private ssize_t
288 swrite(int fd, const void *buf, size_t n)
290 ssize_t rv;
291 size_t rn = n;
294 switch (rv = write(fd, buf, n)) {
295 case -1:
296 if (errno == EINTR)
297 continue;
298 return -1;
299 default:
300 n -= rv;
301 buf = CAST(const char *, buf) + rv;
302 break;
304 while (n > 0);
305 return rn;
310 * `safe' read for sockets and pipes.
312 protected ssize_t
313 sread(int fd, void *buf, size_t n, int canbepipe __attribute__((__unused__)))
315 ssize_t rv;
316 #ifdef FIONREAD
317 int t = 0;
318 #endif
319 size_t rn = n;
321 if (fd == STDIN_FILENO)
322 goto nocheck;
324 #ifdef FIONREAD
325 if (canbepipe && (ioctl(fd, FIONREAD, &t) == -1 || t == 0)) {
326 #ifdef FD_ZERO
327 ssize_t cnt;
328 for (cnt = 0;; cnt++) {
329 fd_set check;
330 struct timeval tout = {0, 100 * 1000};
331 int selrv;
333 FD_ZERO(&check);
334 FD_SET(fd, &check);
337 * Avoid soft deadlock: do not read if there
338 * is nothing to read from sockets and pipes.
340 selrv = select(fd + 1, &check, NULL, NULL, &tout);
341 if (selrv == -1) {
342 if (errno == EINTR || errno == EAGAIN)
343 continue;
344 } else if (selrv == 0 && cnt >= 5) {
345 return 0;
346 } else
347 break;
349 #endif
350 (void)ioctl(fd, FIONREAD, &t);
353 if (t > 0 && (size_t)t < n) {
354 n = t;
355 rn = n;
357 #endif
359 nocheck:
361 switch ((rv = read(fd, buf, n))) {
362 case -1:
363 if (errno == EINTR)
364 continue;
365 return -1;
366 case 0:
367 return rn - n;
368 default:
369 n -= rv;
370 buf = CAST(char *, CCAST(void *, buf)) + rv;
371 break;
373 while (n > 0);
374 return rn;
377 protected int
378 file_pipe2file(struct magic_set *ms, int fd, const void *startbuf,
379 size_t nbytes)
381 char buf[4096];
382 ssize_t r;
383 int tfd;
385 (void)strlcpy(buf, "/tmp/file.XXXXXX", sizeof buf);
386 #ifndef HAVE_MKSTEMP
388 char *ptr = mktemp(buf);
389 tfd = open(ptr, O_RDWR|O_TRUNC|O_EXCL|O_CREAT, 0600);
390 r = errno;
391 (void)unlink(ptr);
392 errno = r;
394 #else
396 int te;
397 tfd = mkstemp(buf);
398 te = errno;
399 (void)unlink(buf);
400 errno = te;
402 #endif
403 if (tfd == -1) {
404 file_error(ms, errno,
405 "cannot create temporary file for pipe copy");
406 return -1;
409 if (swrite(tfd, startbuf, nbytes) != (ssize_t)nbytes)
410 r = 1;
411 else {
412 while ((r = sread(fd, buf, sizeof(buf), 1)) > 0)
413 if (swrite(tfd, buf, (size_t)r) != r)
414 break;
417 switch (r) {
418 case -1:
419 file_error(ms, errno, "error copying from pipe to temp file");
420 return -1;
421 case 0:
422 break;
423 default:
424 file_error(ms, errno, "error while writing to temp file");
425 return -1;
429 * We duplicate the file descriptor, because fclose on a
430 * tmpfile will delete the file, but any open descriptors
431 * can still access the phantom inode.
433 if ((fd = dup2(tfd, fd)) == -1) {
434 file_error(ms, errno, "could not dup descriptor for temp file");
435 return -1;
437 (void)close(tfd);
438 if (lseek(fd, (off_t)0, SEEK_SET) == (off_t)-1) {
439 file_badseek(ms);
440 return -1;
442 return fd;
444 #if HAVE_FORK
445 #ifdef BUILTIN_DECOMPRESS
447 #define FHCRC (1 << 1)
448 #define FEXTRA (1 << 2)
449 #define FNAME (1 << 3)
450 #define FCOMMENT (1 << 4)
453 private int
454 uncompressgzipped(const unsigned char *old, unsigned char **newch,
455 size_t bytes_max, size_t *n)
457 unsigned char flg = old[3];
458 size_t data_start = 10;
460 if (flg & FEXTRA) {
461 if (data_start + 1 >= *n)
462 goto err;
463 data_start += 2 + old[data_start] + old[data_start + 1] * 256;
465 if (flg & FNAME) {
466 while(data_start < *n && old[data_start])
467 data_start++;
468 data_start++;
470 if (flg & FCOMMENT) {
471 while(data_start < *n && old[data_start])
472 data_start++;
473 data_start++;
475 if (flg & FHCRC)
476 data_start += 2;
478 if (data_start >= *n)
479 goto err;
481 *n -= data_start;
482 old += data_start;
483 return uncompresszlib(old, newch, bytes_max, n, 0);
484 err:
485 return makeerror(newch, n, "File too short");
488 private int
489 uncompresszlib(const unsigned char *old, unsigned char **newch,
490 size_t bytes_max, size_t *n, int zlib)
492 int rc;
493 z_stream z;
495 if ((*newch = CAST(unsigned char *, malloc(bytes_max + 1))) == NULL)
496 return makeerror(newch, n, "No buffer, %s", strerror(errno));
498 z.next_in = CCAST(Bytef *, old);
499 z.avail_in = CAST(uint32_t, *n);
500 z.next_out = *newch;
501 z.avail_out = CAST(unsigned int, bytes_max);
502 z.zalloc = Z_NULL;
503 z.zfree = Z_NULL;
504 z.opaque = Z_NULL;
506 /* LINTED bug in header macro */
507 rc = zlib ? inflateInit(&z) : inflateInit2(&z, -15);
508 if (rc != Z_OK)
509 goto err;
511 rc = inflate(&z, Z_SYNC_FLUSH);
512 if (rc != Z_OK && rc != Z_STREAM_END)
513 goto err;
515 *n = (size_t)z.total_out;
516 rc = inflateEnd(&z);
517 if (rc != Z_OK)
518 goto err;
520 /* let's keep the nul-terminate tradition */
521 (*newch)[*n] = '\0';
523 return OKDATA;
524 err:
525 strlcpy((char *)*newch, z.msg ? z.msg : zError(rc), bytes_max);
526 *n = strlen((char *)*newch);
527 return ERRDATA;
529 #endif
531 static int
532 makeerror(unsigned char **buf, size_t *len, const char *fmt, ...)
534 char *msg;
535 va_list ap;
536 int rv;
538 va_start(ap, fmt);
539 rv = vasprintf(&msg, fmt, ap);
540 va_end(ap);
541 if (rv < 0) {
542 *buf = NULL;
543 *len = 0;
544 return NODATA;
546 *buf = (unsigned char *)msg;
547 *len = strlen(msg);
548 return ERRDATA;
551 static void
552 closefd(int *fd, size_t i)
554 if (fd[i] == -1)
555 return;
556 (void) close(fd[i]);
557 fd[i] = -1;
560 static void
561 closep(int *fd)
563 size_t i;
564 for (i = 0; i < 2; i++)
565 closefd(fd, i);
568 static void
569 copydesc(int i, int *fd)
571 int j = fd[i == STDIN_FILENO ? 0 : 1];
572 if (j == i)
573 return;
574 if (dup2(j, i) == -1) {
575 DPRINTF("dup(%d, %d) failed (%s)\n", j, i, strerror(errno));
576 exit(1);
578 closep(fd);
581 static void
582 writechild(int fdp[3][2], const void *old, size_t n)
584 int status;
586 closefd(fdp[STDIN_FILENO], 0);
588 * fork again, to avoid blocking because both
589 * pipes filled
591 switch (fork()) {
592 case 0: /* child */
593 closefd(fdp[STDOUT_FILENO], 0);
594 if (swrite(fdp[STDIN_FILENO][1], old, n) != (ssize_t)n) {
595 DPRINTF("Write failed (%s)\n", strerror(errno));
596 exit(1);
598 exit(0);
599 /*NOTREACHED*/
601 case -1:
602 DPRINTF("Fork failed (%s)\n", strerror(errno));
603 exit(1);
604 /*NOTREACHED*/
606 default: /* parent */
607 if (wait(&status) == -1) {
608 DPRINTF("Wait failed (%s)\n", strerror(errno));
609 exit(1);
611 DPRINTF("Grandchild wait return %#x\n", status);
613 closefd(fdp[STDIN_FILENO], 1);
616 static ssize_t
617 filter_error(unsigned char *ubuf, ssize_t n)
619 char *p;
620 char *buf;
622 ubuf[n] = '\0';
623 buf = (char *)ubuf;
624 while (isspace((unsigned char)*buf))
625 buf++;
626 DPRINTF("Filter error[[[%s]]]\n", buf);
627 if ((p = strchr((char *)buf, '\n')) != NULL)
628 *p = '\0';
629 if ((p = strchr((char *)buf, ';')) != NULL)
630 *p = '\0';
631 if ((p = strrchr((char *)buf, ':')) != NULL) {
632 ++p;
633 while (isspace((unsigned char)*p))
634 p++;
635 n = strlen(p);
636 memmove(ubuf, p, CAST(size_t, n + 1));
638 DPRINTF("Filter error after[[[%s]]]\n", (char *)ubuf);
639 if (islower(*ubuf))
640 *ubuf = toupper(*ubuf);
641 return n;
644 private const char *
645 methodname(size_t method)
647 #ifdef BUILTIN_DECOMPRESS
648 /* FIXME: This doesn't cope with bzip2 */
649 if (method == 2 || compr[method].maglen == 0)
650 return "zlib";
651 #endif
652 return compr[method].argv[0];
655 private int
656 uncompressbuf(int fd, size_t bytes_max, size_t method, const unsigned char *old,
657 unsigned char **newch, size_t* n)
659 int fdp[3][2];
660 int status, rv;
661 size_t i;
662 ssize_t r;
664 #ifdef BUILTIN_DECOMPRESS
665 /* FIXME: This doesn't cope with bzip2 */
666 if (method == 2)
667 return uncompressgzipped(old, newch, bytes_max, n);
668 if (compr[method].maglen == 0)
669 return uncompresszlib(old, newch, bytes_max, n, 1);
670 #endif
671 (void)fflush(stdout);
672 (void)fflush(stderr);
674 for (i = 0; i < __arraycount(fdp); i++)
675 fdp[i][0] = fdp[i][1] = -1;
677 if ((fd == -1 && pipe(fdp[STDIN_FILENO]) == -1) ||
678 pipe(fdp[STDOUT_FILENO]) == -1 || pipe(fdp[STDERR_FILENO]) == -1) {
679 closep(fdp[STDIN_FILENO]);
680 closep(fdp[STDOUT_FILENO]);
681 return makeerror(newch, n, "Cannot create pipe, %s",
682 strerror(errno));
684 switch (fork()) {
685 case 0: /* child */
686 if (fd != -1) {
687 fdp[STDIN_FILENO][0] = fd;
688 (void) lseek(fd, (off_t)0, SEEK_SET);
691 for (i = 0; i < __arraycount(fdp); i++)
692 copydesc(CAST(int, i), fdp[i]);
694 (void)execvp(compr[method].argv[0],
695 (char *const *)(intptr_t)compr[method].argv);
696 dprintf(STDERR_FILENO, "exec `%s' failed, %s",
697 compr[method].argv[0], strerror(errno));
698 exit(1);
699 /*NOTREACHED*/
700 case -1:
701 return makeerror(newch, n, "Cannot fork, %s",
702 strerror(errno));
704 default: /* parent */
705 for (i = 1; i < __arraycount(fdp); i++)
706 closefd(fdp[i], 1);
708 /* Write the buffer data to the child, if we don't have fd */
709 if (fd == -1)
710 writechild(fdp, old, *n);
712 *newch = CAST(unsigned char *, malloc(bytes_max + 1));
713 if (*newch == NULL) {
714 rv = makeerror(newch, n, "No buffer, %s",
715 strerror(errno));
716 goto err;
718 rv = OKDATA;
719 if ((r = sread(fdp[STDOUT_FILENO][0], *newch, bytes_max, 0)) > 0)
720 break;
721 DPRINTF("Read stdout failed %d (%s)\n", fdp[STDOUT_FILENO][0],
722 r != -1 ? strerror(errno) : "no data");
724 rv = ERRDATA;
725 if (r == 0 &&
726 (r = sread(fdp[STDERR_FILENO][0], *newch, bytes_max, 0)) > 0)
728 r = filter_error(*newch, r);
729 break;
731 free(*newch);
732 if (r == 0)
733 rv = makeerror(newch, n, "Read failed, %s",
734 strerror(errno));
735 else
736 rv = makeerror(newch, n, "No data");
737 goto err;
740 *n = r;
741 /* NUL terminate, as every buffer is handled here. */
742 (*newch)[*n] = '\0';
743 err:
744 closefd(fdp[STDIN_FILENO], 1);
745 closefd(fdp[STDOUT_FILENO], 0);
746 closefd(fdp[STDERR_FILENO], 0);
747 if (wait(&status) == -1) {
748 free(*newch);
749 rv = makeerror(newch, n, "Wait failed, %s", strerror(errno));
750 DPRINTF("Child wait return %#x\n", status);
751 } else if (!WIFEXITED(status)) {
752 DPRINTF("Child not exited (%#x)\n", status);
753 } else if (WEXITSTATUS(status) != 0) {
754 DPRINTF("Child exited (%#x)\n", WEXITSTATUS(status));
757 closefd(fdp[STDIN_FILENO], 0);
758 DPRINTF("Returning %p n=%zu rv=%d\n", *newch, *n, rv);
760 return rv;
762 #endif