1 /* $NetBSD: compress.c,v 1.4 2011/09/16 21:06:26 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 1.67 2011/09/01 12:12:37 christos Exp $") 42 #else 43 __RCSID("$NetBSD: compress.c,v 1.4 2011/09/16 21:06:26 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 } 131 132 if ((mime == 0 || mime & MAGIC_MIME_ENCODING) && 133 file_buffer(ms, -1, NULL, buf, nbytes) == -1) 134 goto error; 135 136 if (!mime && file_printf(ms, ")") == -1) 137 goto error; 138 rv = 1; 139 break; 140 } 141 } 142 error: 143 if (newbuf) 144 free(newbuf); 145 ms->flags |= MAGIC_COMPRESS; 146 return rv; 147 } 148 #endif 149 /* 150 * `safe' write for sockets and pipes. 151 */ 152 private ssize_t 153 swrite(int fd, const void *buf, size_t n) 154 { 155 ssize_t rv; 156 size_t rn = n; 157 158 do 159 switch (rv = write(fd, buf, n)) { 160 case -1: 161 if (errno == EINTR) 162 continue; 163 return -1; 164 default: 165 n -= rv; 166 buf = CAST(const char *, buf) + rv; 167 break; 168 } 169 while (n > 0); 170 return rn; 171 } 172 173 174 /* 175 * `safe' read for sockets and pipes. 176 */ 177 protected ssize_t 178 sread(int fd, void *buf, size_t n, int canbepipe __attribute__ ((unused))) 179 { 180 ssize_t rv; 181 #ifdef FD_ZERO 182 ssize_t cnt; 183 #endif 184 #ifdef FIONREAD 185 int t = 0; 186 #endif 187 size_t rn = n; 188 189 if (fd == STDIN_FILENO) 190 goto nocheck; 191 192 #ifdef FIONREAD 193 if ((canbepipe && (ioctl(fd, FIONREAD, &t) == -1)) || (t == 0)) { 194 #ifdef FD_ZERO 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 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 #ifdef HAVE_MKSTEMP 252 int te; 253 #endif 254 255 (void)strlcpy(buf, "/tmp/file.XXXXXX", sizeof buf); 256 #ifndef HAVE_MKSTEMP 257 { 258 char *ptr = mktemp(buf); 259 tfd = open(ptr, O_RDWR|O_TRUNC|O_EXCL|O_CREAT, 0600); 260 r = errno; 261 (void)unlink(ptr); 262 errno = r; 263 } 264 #else 265 tfd = mkstemp(buf); 266 te = errno; 267 (void)unlink(buf); 268 errno = te; 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 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 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 ssize_t r; 391 pid_t pid; 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 (pid = fork()) { 406 case 0: /* child */ 407 (void) close(0); 408 if (fd != -1) { 409 (void) dup(fd); 410 (void) lseek(0, (off_t)0, SEEK_SET); 411 } else { 412 (void) dup(fdin[0]); 413 (void) close(fdin[0]); 414 (void) close(fdin[1]); 415 } 416 417 (void) close(1); 418 (void) dup(fdout[1]); 419 (void) close(fdout[0]); 420 (void) close(fdout[1]); 421 #ifndef DEBUG 422 if (compr[method].silent) 423 (void)close(2); 424 #endif 425 426 (void)execvp(compr[method].argv[0], 427 (char *const *)(intptr_t)compr[method].argv); 428 #ifdef DEBUG 429 (void)fprintf(stderr, "exec `%s' failed (%s)\n", 430 compr[method].argv[0], strerror(errno)); 431 #endif 432 exit(1); 433 /*NOTREACHED*/ 434 case -1: 435 file_error(ms, errno, "could not fork"); 436 return NODATA; 437 438 default: /* parent */ 439 (void) close(fdout[1]); 440 if (fd == -1) { 441 (void) close(fdin[0]); 442 /* 443 * fork again, to avoid blocking because both 444 * pipes filled 445 */ 446 switch (fork()) { 447 case 0: /* child */ 448 (void)close(fdout[0]); 449 if (swrite(fdin[1], old, n) != (ssize_t)n) { 450 #ifdef DEBUG 451 (void)fprintf(stderr, 452 "Write failed (%s)\n", 453 strerror(errno)); 454 #endif 455 exit(1); 456 } 457 exit(0); 458 /*NOTREACHED*/ 459 460 case -1: 461 #ifdef DEBUG 462 (void)fprintf(stderr, "Fork failed (%s)\n", 463 strerror(errno)); 464 #endif 465 exit(1); 466 /*NOTREACHED*/ 467 468 default: /* parent */ 469 break; 470 } 471 (void) close(fdin[1]); 472 fdin[1] = -1; 473 } 474 475 if ((*newch = (unsigned char *) malloc(HOWMANY + 1)) == NULL) { 476 #ifdef DEBUG 477 (void)fprintf(stderr, "Malloc failed (%s)\n", 478 strerror(errno)); 479 #endif 480 n = 0; 481 goto err; 482 } 483 if ((r = sread(fdout[0], *newch, HOWMANY, 0)) <= 0) { 484 #ifdef DEBUG 485 (void)fprintf(stderr, "Read failed (%s)\n", 486 strerror(errno)); 487 #endif 488 free(*newch); 489 n = 0; 490 newch[0] = '\0'; 491 goto err; 492 } else { 493 n = r; 494 } 495 /* NUL terminate, as every buffer is handled here. */ 496 (*newch)[n] = '\0'; 497 err: 498 if (fdin[1] != -1) 499 (void) close(fdin[1]); 500 (void) close(fdout[0]); 501 #ifdef WNOHANG 502 while (waitpid(pid, NULL, WNOHANG) != -1) 503 continue; 504 #else 505 (void)wait(NULL); 506 #endif 507 (void) close(fdin[0]); 508 509 return n; 510 } 511 } 512 #endif 513