1 /* $NetBSD: compress.c,v 1.10 2015/01/02 21:15:32 christos Exp $ */
2
3 /*
4 * Copyright (c) Ian F. Darwin 1986-1995.
5 * Software written by Ian F. Darwin and others;
6 * maintained 1995-present by Christos Zoulas and others.
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 immediately at the beginning of the file, without modification,
13 * this list of conditions, and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 *
18 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
19 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
22 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28 * SUCH DAMAGE.
29 */
30 /*
31 * compress routines:
32 * zmagic() - returns 0 if not recognized, uncompresses and prints
33 * information if recognized
34 * uncompress(method, old, n, newch) - uncompress old into new,
35 * using method, return sizeof new
36 */
37 #include "file.h"
38
39 #ifndef lint
40 #if 0
41 FILE_RCSID("@(#)$File: compress.c,v 1.77 2014/12/12 16:33:01 christos Exp $")
42 #else
43 __RCSID("$NetBSD: compress.c,v 1.10 2015/01/02 21:15:32 christos Exp $");
44 #endif
45 #endif
46
47 #include "magic.h"
48 #include <stdlib.h>
49 #ifdef HAVE_UNISTD_H
50 #include <unistd.h>
51 #endif
52 #include <string.h>
53 #include <errno.h>
54 #include <signal.h>
55 #if !defined(__MINGW32__) && !defined(WIN32)
56 #include <sys/ioctl.h>
57 #endif
58 #ifdef HAVE_SYS_WAIT_H
59 #include <sys/wait.h>
60 #endif
61 #if defined(HAVE_SYS_TIME_H)
62 #include <sys/time.h>
63 #endif
64 #if defined(HAVE_ZLIB_H) && defined(HAVE_LIBZ)
65 #define BUILTIN_DECOMPRESS
66 #include <zlib.h>
67 #endif
68
69 private const struct {
70 const char magic[8];
71 size_t maglen;
72 const char *argv[3];
73 int silent;
74 } compr[] = {
75 { "\037\235", 2, { "gzip", "-cdq", NULL }, 1 }, /* compressed */
76 /* Uncompress can get stuck; so use gzip first if we have it
77 * Idea from Damien Clark, thanks! */
78 { "\037\235", 2, { "uncompress", "-c", NULL }, 1 }, /* compressed */
79 { "\037\213", 2, { "gzip", "-cdq", NULL }, 1 }, /* gzipped */
80 { "\037\236", 2, { "gzip", "-cdq", NULL }, 1 }, /* frozen */
81 { "\037\240", 2, { "gzip", "-cdq", NULL }, 1 }, /* SCO LZH */
82 /* the standard pack utilities do not accept standard input */
83 { "\037\036", 2, { "gzip", "-cdq", NULL }, 0 }, /* packed */
84 { "PK\3\4", 4, { "gzip", "-cdq", NULL }, 1 }, /* pkzipped, */
85 /* ...only first file examined */
86 { "BZh", 3, { "bzip2", "-cd", NULL }, 1 }, /* bzip2-ed */
87 { "LZIP", 4, { "lzip", "-cdq", NULL }, 1 },
88 { "\3757zXZ\0",6,{ "xz", "-cd", NULL }, 1 }, /* XZ Utils */
89 { "LRZI", 4, { "lrzip", "-dqo-", NULL }, 1 }, /* LRZIP */
90 { "\004\"M\030", 4, { "lz4", "-cd", NULL }, 1 }, /* LZ4 */
91 };
92
93 #define NODATA ((size_t)~0)
94
95 private ssize_t swrite(int, const void *, size_t);
96 #if HAVE_FORK
97 private size_t ncompr = sizeof(compr) / sizeof(compr[0]);
98 private size_t uncompressbuf(struct magic_set *, int, size_t,
99 const unsigned char *, unsigned char **, size_t);
100 #ifdef BUILTIN_DECOMPRESS
101 private size_t uncompressgzipped(struct magic_set *, const unsigned char *,
102 unsigned char **, size_t);
103 #endif
104
105 protected int
file_zmagic(struct magic_set * ms,int fd,const char * name,const unsigned char * buf,size_t nbytes)106 file_zmagic(struct magic_set *ms, int fd, const char *name,
107 const unsigned char *buf, size_t nbytes)
108 {
109 unsigned char *newbuf = NULL;
110 size_t i, nsz;
111 int rv = 0;
112 int mime = ms->flags & MAGIC_MIME;
113 sig_t osigpipe;
114
115 if ((ms->flags & MAGIC_COMPRESS) == 0)
116 return 0;
117
118 osigpipe = signal(SIGPIPE, SIG_IGN);
119 for (i = 0; i < ncompr; i++) {
120 if (nbytes < compr[i].maglen)
121 continue;
122 if (memcmp(buf, compr[i].magic, compr[i].maglen) == 0 &&
123 (nsz = uncompressbuf(ms, fd, i, buf, &newbuf,
124 nbytes)) != NODATA) {
125 ms->flags &= ~MAGIC_COMPRESS;
126 rv = -1;
127 if (file_buffer(ms, -1, name, newbuf, nsz) == -1)
128 goto error;
129
130 if (mime == MAGIC_MIME || mime == 0) {
131 if (file_printf(ms, mime ?
132 " compressed-encoding=" : " (") == -1)
133 goto error;
134 if (file_buffer(ms, -1, NULL, buf, nbytes) == -1)
135 goto error;
136 if (!mime && file_printf(ms, ")") == -1)
137 goto error;
138 }
139
140 rv = 1;
141 break;
142 }
143 }
144 error:
145 (void)signal(SIGPIPE, osigpipe);
146 free(newbuf);
147 ms->flags |= MAGIC_COMPRESS;
148 return rv;
149 }
150 #endif
151 /*
152 * `safe' write for sockets and pipes.
153 */
154 private ssize_t
swrite(int fd,const void * buf,size_t n)155 swrite(int fd, const void *buf, size_t n)
156 {
157 ssize_t rv;
158 size_t rn = n;
159
160 do
161 switch (rv = write(fd, buf, n)) {
162 case -1:
163 if (errno == EINTR)
164 continue;
165 return -1;
166 default:
167 n -= rv;
168 buf = CAST(const char *, buf) + rv;
169 break;
170 }
171 while (n > 0);
172 return rn;
173 }
174
175
176 /*
177 * `safe' read for sockets and pipes.
178 */
179 protected ssize_t
sread(int fd,void * buf,size_t n,int canbepipe)180 sread(int fd, void *buf, size_t n, int canbepipe __attribute__((__unused__)))
181 {
182 ssize_t rv;
183 #ifdef FIONREAD
184 int t = 0;
185 #endif
186 size_t rn = n;
187
188 if (fd == STDIN_FILENO)
189 goto nocheck;
190
191 #ifdef FIONREAD
192 if (canbepipe && (ioctl(fd, FIONREAD, &t) == -1 || t == 0)) {
193 #ifdef FD_ZERO
194 ssize_t cnt;
195 for (cnt = 0;; cnt++) {
196 fd_set check;
197 struct timeval tout = {0, 100 * 1000};
198 int selrv;
199
200 FD_ZERO(&check);
201 FD_SET(fd, &check);
202
203 /*
204 * Avoid soft deadlock: do not read if there
205 * is nothing to read from sockets and pipes.
206 */
207 selrv = select(fd + 1, &check, NULL, NULL, &tout);
208 if (selrv == -1) {
209 if (errno == EINTR || errno == EAGAIN)
210 continue;
211 } else if (selrv == 0 && cnt >= 5) {
212 return 0;
213 } else
214 break;
215 }
216 #endif
217 (void)ioctl(fd, FIONREAD, &t);
218 }
219
220 if (t > 0 && (size_t)t < n) {
221 n = t;
222 rn = n;
223 }
224 #endif
225
226 nocheck:
227 do
228 switch ((rv = read(fd, buf, n))) {
229 case -1:
230 if (errno == EINTR)
231 continue;
232 return -1;
233 case 0:
234 return rn - n;
235 default:
236 n -= rv;
237 buf = ((char *)buf) + rv;
238 break;
239 }
240 while (n > 0);
241 return rn;
242 }
243
244 protected int
file_pipe2file(struct magic_set * ms,int fd,const void * startbuf,size_t nbytes)245 file_pipe2file(struct magic_set *ms, int fd, const void *startbuf,
246 size_t nbytes)
247 {
248 char buf[4096];
249 ssize_t r;
250 int tfd;
251
252 (void)strlcpy(buf, "/tmp/file.XXXXXX", sizeof buf);
253 #ifndef HAVE_MKSTEMP
254 {
255 char *ptr = mktemp(buf);
256 tfd = open(ptr, O_RDWR|O_TRUNC|O_EXCL|O_CREAT, 0600);
257 r = errno;
258 (void)unlink(ptr);
259 errno = r;
260 }
261 #else
262 {
263 int te;
264 tfd = mkstemp(buf);
265 te = errno;
266 (void)unlink(buf);
267 errno = te;
268 }
269 #endif
270 if (tfd == -1) {
271 file_error(ms, errno,
272 "cannot create temporary file for pipe copy");
273 return -1;
274 }
275
276 if (swrite(tfd, startbuf, nbytes) != (ssize_t)nbytes)
277 r = 1;
278 else {
279 while ((r = sread(fd, buf, sizeof(buf), 1)) > 0)
280 if (swrite(tfd, buf, (size_t)r) != r)
281 break;
282 }
283
284 switch (r) {
285 case -1:
286 file_error(ms, errno, "error copying from pipe to temp file");
287 return -1;
288 case 0:
289 break;
290 default:
291 file_error(ms, errno, "error while writing to temp file");
292 return -1;
293 }
294
295 /*
296 * We duplicate the file descriptor, because fclose on a
297 * tmpfile will delete the file, but any open descriptors
298 * can still access the phantom inode.
299 */
300 if ((fd = dup2(tfd, fd)) == -1) {
301 file_error(ms, errno, "could not dup descriptor for temp file");
302 return -1;
303 }
304 (void)close(tfd);
305 if (lseek(fd, (off_t)0, SEEK_SET) == (off_t)-1) {
306 file_badseek(ms);
307 return -1;
308 }
309 return fd;
310 }
311 #if HAVE_FORK
312 #ifdef BUILTIN_DECOMPRESS
313
314 #define FHCRC (1 << 1)
315 #define FEXTRA (1 << 2)
316 #define FNAME (1 << 3)
317 #define FCOMMENT (1 << 4)
318
319 private size_t
uncompressgzipped(struct magic_set * ms,const unsigned char * old,unsigned char ** newch,size_t n)320 uncompressgzipped(struct magic_set *ms, const unsigned char *old,
321 unsigned char **newch, size_t n)
322 {
323 unsigned char flg = old[3];
324 size_t data_start = 10;
325 z_stream z;
326 int rc;
327
328 if (flg & FEXTRA) {
329 if (data_start+1 >= n)
330 return 0;
331 data_start += 2 + old[data_start] + old[data_start + 1] * 256;
332 }
333 if (flg & FNAME) {
334 while(data_start < n && old[data_start])
335 data_start++;
336 data_start++;
337 }
338 if(flg & FCOMMENT) {
339 while(data_start < n && old[data_start])
340 data_start++;
341 data_start++;
342 }
343 if(flg & FHCRC)
344 data_start += 2;
345
346 if (data_start >= n)
347 return 0;
348 if ((*newch = CAST(unsigned char *, malloc(HOWMANY + 1))) == NULL) {
349 return 0;
350 }
351
352 /* XXX: const castaway, via strchr */
353 z.next_in = (Bytef *)strchr((const char *)old + data_start,
354 old[data_start]);
355 z.avail_in = CAST(uint32_t, (n - data_start));
356 z.next_out = *newch;
357 z.avail_out = HOWMANY;
358 z.zalloc = Z_NULL;
359 z.zfree = Z_NULL;
360 z.opaque = Z_NULL;
361
362 /* LINTED bug in header macro */
363 rc = inflateInit2(&z, -15);
364 if (rc != Z_OK) {
365 file_error(ms, 0, "zlib: %s", z.msg);
366 return 0;
367 }
368
369 rc = inflate(&z, Z_SYNC_FLUSH);
370 if (rc != Z_OK && rc != Z_STREAM_END) {
371 file_error(ms, 0, "zlib: %s", z.msg);
372 return 0;
373 }
374
375 n = (size_t)z.total_out;
376 (void)inflateEnd(&z);
377
378 /* let's keep the nul-terminate tradition */
379 (*newch)[n] = '\0';
380
381 return n;
382 }
383 #endif
384
385 private size_t
uncompressbuf(struct magic_set * ms,int fd,size_t method,const unsigned char * old,unsigned char ** newch,size_t n)386 uncompressbuf(struct magic_set *ms, int fd, size_t method,
387 const unsigned char *old, unsigned char **newch, size_t n)
388 {
389 int fdin[2], fdout[2];
390 int status;
391 ssize_t r;
392
393 #ifdef BUILTIN_DECOMPRESS
394 /* FIXME: This doesn't cope with bzip2 */
395 if (method == 2)
396 return uncompressgzipped(ms, old, newch, n);
397 #endif
398 (void)fflush(stdout);
399 (void)fflush(stderr);
400
401 if ((fd != -1 && pipe(fdin) == -1) || pipe(fdout) == -1) {
402 file_error(ms, errno, "cannot create pipe");
403 return NODATA;
404 }
405 switch (fork()) {
406 case 0: /* child */
407 (void) close(0);
408 if (fd != -1) {
409 if (dup(fd) == -1)
410 _exit(1);
411 (void) lseek(0, (off_t)0, SEEK_SET);
412 } else {
413 if (dup(fdin[0]) == -1)
414 _exit(1);
415 (void) close(fdin[0]);
416 (void) close(fdin[1]);
417 }
418
419 (void) close(1);
420 if (dup(fdout[1]) == -1)
421 _exit(1);
422 (void) close(fdout[0]);
423 (void) close(fdout[1]);
424 #ifndef DEBUG
425 if (compr[method].silent)
426 (void)close(2);
427 #endif
428
429 (void)execvp(compr[method].argv[0],
430 (char *const *)(intptr_t)compr[method].argv);
431 #ifdef DEBUG
432 (void)fprintf(stderr, "exec `%s' failed (%s)\n",
433 compr[method].argv[0], strerror(errno));
434 #endif
435 exit(1);
436 /*NOTREACHED*/
437 case -1:
438 file_error(ms, errno, "could not fork");
439 return NODATA;
440
441 default: /* parent */
442 (void) close(fdout[1]);
443 if (fd == -1) {
444 (void) close(fdin[0]);
445 /*
446 * fork again, to avoid blocking because both
447 * pipes filled
448 */
449 switch (fork()) {
450 case 0: /* child */
451 (void)close(fdout[0]);
452 if (swrite(fdin[1], old, n) != (ssize_t)n) {
453 #ifdef DEBUG
454 (void)fprintf(stderr,
455 "Write failed (%s)\n",
456 strerror(errno));
457 #endif
458 exit(1);
459 }
460 exit(0);
461 /*NOTREACHED*/
462
463 case -1:
464 #ifdef DEBUG
465 (void)fprintf(stderr, "Fork failed (%s)\n",
466 strerror(errno));
467 #endif
468 exit(1);
469 /*NOTREACHED*/
470
471 default: /* parent */
472 if (wait(&status) == -1) {
473 #ifdef DEBUG
474 (void)fprintf(stderr,
475 "Wait failed (%s)\n",
476 strerror(errno));
477 #endif
478 exit(1);
479 }
480 exit(WIFEXITED(status) ?
481 WEXITSTATUS(status) : 1);
482 /*NOTREACHED*/
483 }
484 (void) close(fdin[1]);
485 fdin[1] = -1;
486 }
487
488 if ((*newch = (unsigned char *) malloc(HOWMANY + 1)) == NULL) {
489 #ifdef DEBUG
490 (void)fprintf(stderr, "Malloc failed (%s)\n",
491 strerror(errno));
492 #endif
493 n = NODATA;
494 goto err;
495 }
496 if ((r = sread(fdout[0], *newch, HOWMANY, 0)) <= 0) {
497 #ifdef DEBUG
498 (void)fprintf(stderr, "Read failed (%s)\n",
499 strerror(errno));
500 #endif
501 free(*newch);
502 n = NODATA;
503 *newch = NULL;
504 goto err;
505 } else {
506 n = r;
507 }
508 /* NUL terminate, as every buffer is handled here. */
509 (*newch)[n] = '\0';
510 err:
511 if (fdin[1] != -1)
512 (void) close(fdin[1]);
513 (void) close(fdout[0]);
514 if (wait(&status) == -1) {
515 #ifdef DEBUG
516 (void)fprintf(stderr, "Wait failed (%s)\n",
517 strerror(errno));
518 #endif
519 n = NODATA;
520 } else if (!WIFEXITED(status)) {
521 #ifdef DEBUG
522 (void)fprintf(stderr, "Child not exited (0x%x)\n",
523 status);
524 #endif
525 } else if (WEXITSTATUS(status) != 0) {
526 #ifdef DEBUG
527 (void)fprintf(stderr, "Child exited (0x%d)\n",
528 WEXITSTATUS(status));
529 #endif
530 }
531
532 (void) close(fdin[0]);
533
534 return n;
535 }
536 }
537 #endif
538