1 /* $NetBSD: compress.c,v 1.8 2013/12/01 19:32:15 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 1.72 2013/11/18 17:54:58 christos Exp $") 42 #else 43 __RCSID("$NetBSD: compress.c,v 1.8 2013/12/01 19:32:15 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 #ifndef __MINGW32__ 55 #include <sys/ioctl.h> 56 #endif 57 #ifdef HAVE_SYS_WAIT_H 58 #include <sys/wait.h> 59 #endif 60 #if defined(HAVE_SYS_TIME_H) 61 #include <sys/time.h> 62 #endif 63 #if defined(HAVE_ZLIB_H) && defined(HAVE_LIBZ) 64 #define BUILTIN_DECOMPRESS 65 #include <zlib.h> 66 #endif 67 68 private const struct { 69 const char magic[8]; 70 size_t maglen; 71 const char *argv[3]; 72 int silent; 73 } compr[] = { 74 { "\037\235", 2, { "gzip", "-cdq", NULL }, 1 }, /* compressed */ 75 /* Uncompress can get stuck; so use gzip first if we have it 76 * Idea from Damien Clark, thanks! */ 77 { "\037\235", 2, { "uncompress", "-c", NULL }, 1 }, /* compressed */ 78 { "\037\213", 2, { "gzip", "-cdq", NULL }, 1 }, /* gzipped */ 79 { "\037\236", 2, { "gzip", "-cdq", NULL }, 1 }, /* frozen */ 80 { "\037\240", 2, { "gzip", "-cdq", NULL }, 1 }, /* SCO LZH */ 81 /* the standard pack utilities do not accept standard input */ 82 { "\037\036", 2, { "gzip", "-cdq", NULL }, 0 }, /* packed */ 83 { "PK\3\4", 4, { "gzip", "-cdq", NULL }, 1 }, /* pkzipped, */ 84 /* ...only first file examined */ 85 { "BZh", 3, { "bzip2", "-cd", NULL }, 1 }, /* bzip2-ed */ 86 { "LZIP", 4, { "lzip", "-cdq", NULL }, 1 }, 87 { "\3757zXZ\0",6,{ "xz", "-cd", NULL }, 1 }, /* XZ Utils */ 88 { "LRZI", 4, { "lrzip", "-dqo-", NULL }, 1 }, /* LRZIP */ 89 }; 90 91 #define NODATA ((size_t)~0) 92 93 private ssize_t swrite(int, const void *, size_t); 94 #if HAVE_FORK 95 private size_t ncompr = sizeof(compr) / sizeof(compr[0]); 96 private size_t uncompressbuf(struct magic_set *, int, size_t, 97 const unsigned char *, unsigned char **, size_t); 98 #ifdef BUILTIN_DECOMPRESS 99 private size_t uncompressgzipped(struct magic_set *, const unsigned char *, 100 unsigned char **, size_t); 101 #endif 102 103 protected int 104 file_zmagic(struct magic_set *ms, int fd, const char *name, 105 const unsigned char *buf, size_t nbytes) 106 { 107 unsigned char *newbuf = NULL; 108 size_t i, nsz; 109 int rv = 0; 110 int mime = ms->flags & MAGIC_MIME; 111 112 if ((ms->flags & MAGIC_COMPRESS) == 0) 113 return 0; 114 115 for (i = 0; i < ncompr; i++) { 116 if (nbytes < compr[i].maglen) 117 continue; 118 if (memcmp(buf, compr[i].magic, compr[i].maglen) == 0 && 119 (nsz = uncompressbuf(ms, fd, i, buf, &newbuf, 120 nbytes)) != NODATA) { 121 ms->flags &= ~MAGIC_COMPRESS; 122 rv = -1; 123 if (file_buffer(ms, -1, name, newbuf, nsz) == -1) 124 goto error; 125 126 if (mime == MAGIC_MIME || mime == 0) { 127 if (file_printf(ms, mime ? 128 " compressed-encoding=" : " (") == -1) 129 goto error; 130 if (file_buffer(ms, -1, NULL, buf, nbytes) == -1) 131 goto error; 132 if (!mime && file_printf(ms, ")") == -1) 133 goto error; 134 } 135 136 rv = 1; 137 break; 138 } 139 } 140 error: 141 free(newbuf); 142 ms->flags |= MAGIC_COMPRESS; 143 return rv; 144 } 145 #endif 146 /* 147 * `safe' write for sockets and pipes. 148 */ 149 private ssize_t 150 swrite(int fd, const void *buf, size_t n) 151 { 152 ssize_t rv; 153 size_t rn = n; 154 155 do 156 switch (rv = write(fd, buf, n)) { 157 case -1: 158 if (errno == EINTR) 159 continue; 160 return -1; 161 default: 162 n -= rv; 163 buf = CAST(const char *, buf) + rv; 164 break; 165 } 166 while (n > 0); 167 return rn; 168 } 169 170 171 /* 172 * `safe' read for sockets and pipes. 173 */ 174 protected ssize_t 175 sread(int fd, void *buf, size_t n, int canbepipe __attribute__((__unused__))) 176 { 177 ssize_t rv; 178 #ifdef FIONREAD 179 int t = 0; 180 #endif 181 size_t rn = n; 182 183 if (fd == STDIN_FILENO) 184 goto nocheck; 185 186 #ifdef FIONREAD 187 if (canbepipe && (ioctl(fd, FIONREAD, &t) == -1 || t == 0)) { 188 #ifdef FD_ZERO 189 ssize_t cnt; 190 for (cnt = 0;; cnt++) { 191 fd_set check; 192 struct timeval tout = {0, 100 * 1000}; 193 int selrv; 194 195 FD_ZERO(&check); 196 FD_SET(fd, &check); 197 198 /* 199 * Avoid soft deadlock: do not read if there 200 * is nothing to read from sockets and pipes. 201 */ 202 selrv = select(fd + 1, &check, NULL, NULL, &tout); 203 if (selrv == -1) { 204 if (errno == EINTR || errno == EAGAIN) 205 continue; 206 } else if (selrv == 0 && cnt >= 5) { 207 return 0; 208 } else 209 break; 210 } 211 #endif 212 (void)ioctl(fd, FIONREAD, &t); 213 } 214 215 if (t > 0 && (size_t)t < n) { 216 n = t; 217 rn = n; 218 } 219 #endif 220 221 nocheck: 222 do 223 switch ((rv = read(fd, buf, n))) { 224 case -1: 225 if (errno == EINTR) 226 continue; 227 return -1; 228 case 0: 229 return rn - n; 230 default: 231 n -= rv; 232 buf = ((char *)buf) + rv; 233 break; 234 } 235 while (n > 0); 236 return rn; 237 } 238 239 protected int 240 file_pipe2file(struct magic_set *ms, int fd, const void *startbuf, 241 size_t nbytes) 242 { 243 char buf[4096]; 244 ssize_t r; 245 int tfd; 246 247 (void)strlcpy(buf, "/tmp/file.XXXXXX", sizeof buf); 248 #ifndef HAVE_MKSTEMP 249 { 250 char *ptr = mktemp(buf); 251 tfd = open(ptr, O_RDWR|O_TRUNC|O_EXCL|O_CREAT, 0600); 252 r = errno; 253 (void)unlink(ptr); 254 errno = r; 255 } 256 #else 257 { 258 int te; 259 tfd = mkstemp(buf); 260 te = errno; 261 (void)unlink(buf); 262 errno = te; 263 } 264 #endif 265 if (tfd == -1) { 266 file_error(ms, errno, 267 "cannot create temporary file for pipe copy"); 268 return -1; 269 } 270 271 if (swrite(tfd, startbuf, nbytes) != (ssize_t)nbytes) 272 r = 1; 273 else { 274 while ((r = sread(fd, buf, sizeof(buf), 1)) > 0) 275 if (swrite(tfd, buf, (size_t)r) != r) 276 break; 277 } 278 279 switch (r) { 280 case -1: 281 file_error(ms, errno, "error copying from pipe to temp file"); 282 return -1; 283 case 0: 284 break; 285 default: 286 file_error(ms, errno, "error while writing to temp file"); 287 return -1; 288 } 289 290 /* 291 * We duplicate the file descriptor, because fclose on a 292 * tmpfile will delete the file, but any open descriptors 293 * can still access the phantom inode. 294 */ 295 if ((fd = dup2(tfd, fd)) == -1) { 296 file_error(ms, errno, "could not dup descriptor for temp file"); 297 return -1; 298 } 299 (void)close(tfd); 300 if (lseek(fd, (off_t)0, SEEK_SET) == (off_t)-1) { 301 file_badseek(ms); 302 return -1; 303 } 304 return fd; 305 } 306 #if HAVE_FORK 307 #ifdef BUILTIN_DECOMPRESS 308 309 #define FHCRC (1 << 1) 310 #define FEXTRA (1 << 2) 311 #define FNAME (1 << 3) 312 #define FCOMMENT (1 << 4) 313 314 private size_t 315 uncompressgzipped(struct magic_set *ms, const unsigned char *old, 316 unsigned char **newch, size_t n) 317 { 318 unsigned char flg = old[3]; 319 size_t data_start = 10; 320 z_stream z; 321 int rc; 322 323 if (flg & FEXTRA) { 324 if (data_start+1 >= n) 325 return 0; 326 data_start += 2 + old[data_start] + old[data_start + 1] * 256; 327 } 328 if (flg & FNAME) { 329 while(data_start < n && old[data_start]) 330 data_start++; 331 data_start++; 332 } 333 if(flg & FCOMMENT) { 334 while(data_start < n && old[data_start]) 335 data_start++; 336 data_start++; 337 } 338 if(flg & FHCRC) 339 data_start += 2; 340 341 if (data_start >= n) 342 return 0; 343 if ((*newch = CAST(unsigned char *, malloc(HOWMANY + 1))) == NULL) { 344 return 0; 345 } 346 347 /* XXX: const castaway, via strchr */ 348 z.next_in = (Bytef *)strchr((const char *)old + data_start, 349 old[data_start]); 350 z.avail_in = CAST(uint32_t, (n - data_start)); 351 z.next_out = *newch; 352 z.avail_out = HOWMANY; 353 z.zalloc = Z_NULL; 354 z.zfree = Z_NULL; 355 z.opaque = Z_NULL; 356 357 /* LINTED bug in header macro */ 358 rc = inflateInit2(&z, -15); 359 if (rc != Z_OK) { 360 file_error(ms, 0, "zlib: %s", z.msg); 361 return 0; 362 } 363 364 rc = inflate(&z, Z_SYNC_FLUSH); 365 if (rc != Z_OK && rc != Z_STREAM_END) { 366 file_error(ms, 0, "zlib: %s", z.msg); 367 return 0; 368 } 369 370 n = (size_t)z.total_out; 371 (void)inflateEnd(&z); 372 373 /* let's keep the nul-terminate tradition */ 374 (*newch)[n] = '\0'; 375 376 return n; 377 } 378 #endif 379 380 private size_t 381 uncompressbuf(struct magic_set *ms, int fd, size_t method, 382 const unsigned char *old, unsigned char **newch, size_t n) 383 { 384 int fdin[2], fdout[2]; 385 ssize_t r; 386 pid_t pid; 387 388 #ifdef BUILTIN_DECOMPRESS 389 /* FIXME: This doesn't cope with bzip2 */ 390 if (method == 2) 391 return uncompressgzipped(ms, old, newch, n); 392 #endif 393 (void)fflush(stdout); 394 (void)fflush(stderr); 395 396 if ((fd != -1 && pipe(fdin) == -1) || pipe(fdout) == -1) { 397 file_error(ms, errno, "cannot create pipe"); 398 return NODATA; 399 } 400 switch (pid = fork()) { 401 case 0: /* child */ 402 (void) close(0); 403 if (fd != -1) { 404 if (dup(fd) == -1) 405 _exit(1); 406 (void) lseek(0, (off_t)0, SEEK_SET); 407 } else { 408 if (dup(fdin[0]) == -1) 409 _exit(1); 410 (void) close(fdin[0]); 411 (void) close(fdin[1]); 412 } 413 414 (void) close(1); 415 if (dup(fdout[1]) == -1) 416 _exit(1); 417 (void) close(fdout[0]); 418 (void) close(fdout[1]); 419 #ifndef DEBUG 420 if (compr[method].silent) 421 (void)close(2); 422 #endif 423 424 (void)execvp(compr[method].argv[0], 425 (char *const *)(intptr_t)compr[method].argv); 426 #ifdef DEBUG 427 (void)fprintf(stderr, "exec `%s' failed (%s)\n", 428 compr[method].argv[0], strerror(errno)); 429 #endif 430 exit(1); 431 /*NOTREACHED*/ 432 case -1: 433 file_error(ms, errno, "could not fork"); 434 return NODATA; 435 436 default: /* parent */ 437 (void) close(fdout[1]); 438 if (fd == -1) { 439 (void) close(fdin[0]); 440 /* 441 * fork again, to avoid blocking because both 442 * pipes filled 443 */ 444 switch (fork()) { 445 case 0: /* child */ 446 (void)close(fdout[0]); 447 if (swrite(fdin[1], old, n) != (ssize_t)n) { 448 #ifdef DEBUG 449 (void)fprintf(stderr, 450 "Write failed (%s)\n", 451 strerror(errno)); 452 #endif 453 exit(1); 454 } 455 exit(0); 456 /*NOTREACHED*/ 457 458 case -1: 459 #ifdef DEBUG 460 (void)fprintf(stderr, "Fork failed (%s)\n", 461 strerror(errno)); 462 #endif 463 exit(1); 464 /*NOTREACHED*/ 465 466 default: /* parent */ 467 break; 468 } 469 (void) close(fdin[1]); 470 fdin[1] = -1; 471 } 472 473 if ((*newch = (unsigned char *) malloc(HOWMANY + 1)) == NULL) { 474 #ifdef DEBUG 475 (void)fprintf(stderr, "Malloc failed (%s)\n", 476 strerror(errno)); 477 #endif 478 n = 0; 479 goto err; 480 } 481 if ((r = sread(fdout[0], *newch, HOWMANY, 0)) <= 0) { 482 #ifdef DEBUG 483 (void)fprintf(stderr, "Read failed (%s)\n", 484 strerror(errno)); 485 #endif 486 free(*newch); 487 n = 0; 488 *newch = NULL; 489 goto err; 490 } else { 491 n = r; 492 } 493 /* NUL terminate, as every buffer is handled here. */ 494 (*newch)[n] = '\0'; 495 err: 496 if (fdin[1] != -1) 497 (void) close(fdin[1]); 498 (void) close(fdout[0]); 499 #ifdef WNOHANG 500 while (waitpid(pid, NULL, WNOHANG) != -1) 501 continue; 502 #else 503 (void)wait(NULL); 504 #endif 505 (void) close(fdin[0]); 506 507 return n; 508 } 509 } 510 #endif 511