1 /* $NetBSD: compress.c,v 1.2 2009/05/08 17:28:01 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 1.63 2009/03/23 14:21:51 christos Exp $") 42 #else 43 __RCSID("$NetBSD: compress.c,v 1.2 2009/05/08 17:28:01 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 <sys/ioctl.h> 55 #ifdef HAVE_SYS_WAIT_H 56 #include <sys/wait.h> 57 #endif 58 #if defined(HAVE_SYS_TIME_H) 59 #include <sys/time.h> 60 #endif 61 #if defined(HAVE_ZLIB_H) && defined(HAVE_LIBZ) 62 #define BUILTIN_DECOMPRESS 63 #include <zlib.h> 64 #endif 65 66 private const struct { 67 const char magic[8]; 68 size_t maglen; 69 const char *argv[3]; 70 int silent; 71 } compr[] = { 72 { "\037\235", 2, { "gzip", "-cdq", NULL }, 1 }, /* compressed */ 73 /* Uncompress can get stuck; so use gzip first if we have it 74 * Idea from Damien Clark, thanks! */ 75 { "\037\235", 2, { "uncompress", "-c", NULL }, 1 }, /* compressed */ 76 { "\037\213", 2, { "gzip", "-cdq", NULL }, 1 }, /* gzipped */ 77 { "\037\236", 2, { "gzip", "-cdq", NULL }, 1 }, /* frozen */ 78 { "\037\240", 2, { "gzip", "-cdq", NULL }, 1 }, /* SCO LZH */ 79 /* the standard pack utilities do not accept standard input */ 80 { "\037\036", 2, { "gzip", "-cdq", NULL }, 0 }, /* packed */ 81 { "PK\3\4", 4, { "gzip", "-cdq", NULL }, 1 }, /* pkzipped, */ 82 /* ...only first file examined */ 83 { "BZh", 3, { "bzip2", "-cd", NULL }, 1 }, /* bzip2-ed */ 84 { "LZIP", 4, { "lzip", "-cdq", NULL }, 1 }, 85 { "\3757zXZ\0",6,{ "xz", "-cd", NULL }, 1 }, /* XZ Utils */ 86 }; 87 88 private size_t ncompr = sizeof(compr) / sizeof(compr[0]); 89 90 #define NODATA ((size_t)~0) 91 92 93 private ssize_t swrite(int, const void *, size_t); 94 private size_t uncompressbuf(struct magic_set *, int, size_t, 95 const unsigned char *, unsigned char **, size_t); 96 #ifdef BUILTIN_DECOMPRESS 97 private size_t uncompressgzipped(struct magic_set *, const unsigned char *, 98 unsigned char **, size_t); 99 #endif 100 101 protected int 102 file_zmagic(struct magic_set *ms, int fd, const char *name, 103 const unsigned char *buf, size_t nbytes) 104 { 105 unsigned char *newbuf = NULL; 106 size_t i, nsz; 107 int rv = 0; 108 int mime = ms->flags & MAGIC_MIME; 109 110 if ((ms->flags & MAGIC_COMPRESS) == 0) 111 return 0; 112 113 for (i = 0; i < ncompr; i++) { 114 if (nbytes < compr[i].maglen) 115 continue; 116 if (memcmp(buf, compr[i].magic, compr[i].maglen) == 0 && 117 (nsz = uncompressbuf(ms, fd, i, buf, &newbuf, 118 nbytes)) != NODATA) { 119 ms->flags &= ~MAGIC_COMPRESS; 120 rv = -1; 121 if (file_buffer(ms, -1, name, newbuf, nsz) == -1) 122 goto error; 123 124 if (mime == MAGIC_MIME || mime == 0) { 125 if (file_printf(ms, mime ? 126 " compressed-encoding=" : " (") == -1) 127 goto error; 128 } 129 130 if ((mime == 0 || mime & MAGIC_MIME_ENCODING) && 131 file_buffer(ms, -1, NULL, buf, nbytes) == -1) 132 goto error; 133 134 if (!mime && file_printf(ms, ")") == -1) 135 goto error; 136 rv = 1; 137 break; 138 } 139 } 140 error: 141 if (newbuf) 142 free(newbuf); 143 ms->flags |= MAGIC_COMPRESS; 144 return rv; 145 } 146 147 /* 148 * `safe' write for sockets and pipes. 149 */ 150 private ssize_t 151 swrite(int fd, const void *buf, size_t n) 152 { 153 ssize_t rv; 154 size_t rn = n; 155 156 do 157 switch (rv = write(fd, buf, n)) { 158 case -1: 159 if (errno == EINTR) 160 continue; 161 return -1; 162 default: 163 n -= rv; 164 buf = CAST(const char *, buf) + rv; 165 break; 166 } 167 while (n > 0); 168 return rn; 169 } 170 171 172 /* 173 * `safe' read for sockets and pipes. 174 */ 175 protected ssize_t 176 sread(int fd, void *buf, size_t n, int canbepipe) 177 { 178 ssize_t rv, cnt; 179 #ifdef FIONREAD 180 int t = 0; 181 #endif 182 size_t rn = n; 183 184 if (fd == STDIN_FILENO) 185 goto nocheck; 186 187 #ifdef FIONREAD 188 if ((canbepipe && (ioctl(fd, FIONREAD, &t) == -1)) || (t == 0)) { 189 #ifdef FD_ZERO 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, te; 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 tfd = mkstemp(buf); 258 te = errno; 259 (void)unlink(buf); 260 errno = te; 261 #endif 262 if (tfd == -1) { 263 file_error(ms, errno, 264 "cannot create temporary file for pipe copy"); 265 return -1; 266 } 267 268 if (swrite(tfd, startbuf, nbytes) != (ssize_t)nbytes) 269 r = 1; 270 else { 271 while ((r = sread(fd, buf, sizeof(buf), 1)) > 0) 272 if (swrite(tfd, buf, (size_t)r) != r) 273 break; 274 } 275 276 switch (r) { 277 case -1: 278 file_error(ms, errno, "error copying from pipe to temp file"); 279 return -1; 280 case 0: 281 break; 282 default: 283 file_error(ms, errno, "error while writing to temp file"); 284 return -1; 285 } 286 287 /* 288 * We duplicate the file descriptor, because fclose on a 289 * tmpfile will delete the file, but any open descriptors 290 * can still access the phantom inode. 291 */ 292 if ((fd = dup2(tfd, fd)) == -1) { 293 file_error(ms, errno, "could not dup descriptor for temp file"); 294 return -1; 295 } 296 (void)close(tfd); 297 if (lseek(fd, (off_t)0, SEEK_SET) == (off_t)-1) { 298 file_badseek(ms); 299 return -1; 300 } 301 return fd; 302 } 303 304 #ifdef BUILTIN_DECOMPRESS 305 306 #define FHCRC (1 << 1) 307 #define FEXTRA (1 << 2) 308 #define FNAME (1 << 3) 309 #define FCOMMENT (1 << 4) 310 311 private size_t 312 uncompressgzipped(struct magic_set *ms, const unsigned char *old, 313 unsigned char **newch, size_t n) 314 { 315 unsigned char flg = old[3]; 316 size_t data_start = 10; 317 z_stream z; 318 int rc; 319 320 if (flg & FEXTRA) { 321 if (data_start+1 >= n) 322 return 0; 323 data_start += 2 + old[data_start] + old[data_start + 1] * 256; 324 } 325 if (flg & FNAME) { 326 while(data_start < n && old[data_start]) 327 data_start++; 328 data_start++; 329 } 330 if(flg & FCOMMENT) { 331 while(data_start < n && old[data_start]) 332 data_start++; 333 data_start++; 334 } 335 if(flg & FHCRC) 336 data_start += 2; 337 338 if (data_start >= n) 339 return 0; 340 if ((*newch = CAST(unsigned char *, malloc(HOWMANY + 1))) == NULL) { 341 return 0; 342 } 343 344 /* XXX: const castaway, via strchr */ 345 z.next_in = (Bytef *)strchr((const char *)old + data_start, 346 old[data_start]); 347 z.avail_in = CAST(uint32_t, (n - data_start)); 348 z.next_out = *newch; 349 z.avail_out = HOWMANY; 350 z.zalloc = Z_NULL; 351 z.zfree = Z_NULL; 352 z.opaque = Z_NULL; 353 354 /* LINTED bug in header macro */ 355 rc = inflateInit2(&z, -15); 356 if (rc != Z_OK) { 357 file_error(ms, 0, "zlib: %s", z.msg); 358 return 0; 359 } 360 361 rc = inflate(&z, Z_SYNC_FLUSH); 362 if (rc != Z_OK && rc != Z_STREAM_END) { 363 file_error(ms, 0, "zlib: %s", z.msg); 364 return 0; 365 } 366 367 n = (size_t)z.total_out; 368 (void)inflateEnd(&z); 369 370 /* let's keep the nul-terminate tradition */ 371 (*newch)[n] = '\0'; 372 373 return n; 374 } 375 #endif 376 377 private size_t 378 uncompressbuf(struct magic_set *ms, int fd, size_t method, 379 const unsigned char *old, unsigned char **newch, size_t n) 380 { 381 int fdin[2], fdout[2]; 382 ssize_t r; 383 384 #ifdef BUILTIN_DECOMPRESS 385 /* FIXME: This doesn't cope with bzip2 */ 386 if (method == 2) 387 return uncompressgzipped(ms, old, newch, n); 388 #endif 389 (void)fflush(stdout); 390 (void)fflush(stderr); 391 392 if ((fd != -1 && pipe(fdin) == -1) || pipe(fdout) == -1) { 393 file_error(ms, errno, "cannot create pipe"); 394 return NODATA; 395 } 396 switch (fork()) { 397 case 0: /* child */ 398 (void) close(0); 399 if (fd != -1) { 400 (void) dup(fd); 401 (void) lseek(0, (off_t)0, SEEK_SET); 402 } else { 403 (void) dup(fdin[0]); 404 (void) close(fdin[0]); 405 (void) close(fdin[1]); 406 } 407 408 (void) close(1); 409 (void) dup(fdout[1]); 410 (void) close(fdout[0]); 411 (void) close(fdout[1]); 412 #ifndef DEBUG 413 if (compr[method].silent) 414 (void)close(2); 415 #endif 416 417 (void)execvp(compr[method].argv[0], 418 (char *const *)(intptr_t)compr[method].argv); 419 #ifdef DEBUG 420 (void)fprintf(stderr, "exec `%s' failed (%s)\n", 421 compr[method].argv[0], strerror(errno)); 422 #endif 423 exit(1); 424 /*NOTREACHED*/ 425 case -1: 426 file_error(ms, errno, "could not fork"); 427 return NODATA; 428 429 default: /* parent */ 430 (void) close(fdout[1]); 431 if (fd == -1) { 432 (void) close(fdin[0]); 433 /* 434 * fork again, to avoid blocking because both 435 * pipes filled 436 */ 437 switch (fork()) { 438 case 0: /* child */ 439 (void)close(fdout[0]); 440 if (swrite(fdin[1], old, n) != (ssize_t)n) { 441 #ifdef DEBUG 442 (void)fprintf(stderr, 443 "Write failed (%s)\n", 444 strerror(errno)); 445 #endif 446 exit(1); 447 } 448 exit(0); 449 /*NOTREACHED*/ 450 451 case -1: 452 #ifdef DEBUG 453 (void)fprintf(stderr, "Fork failed (%s)\n", 454 strerror(errno)); 455 #endif 456 exit(1); 457 /*NOTREACHED*/ 458 459 default: /* parent */ 460 break; 461 } 462 (void) close(fdin[1]); 463 fdin[1] = -1; 464 } 465 466 if ((*newch = (unsigned char *) malloc(HOWMANY + 1)) == NULL) { 467 #ifdef DEBUG 468 (void)fprintf(stderr, "Malloc failed (%s)\n", 469 strerror(errno)); 470 #endif 471 n = 0; 472 goto err; 473 } 474 if ((r = sread(fdout[0], *newch, HOWMANY, 0)) <= 0) { 475 #ifdef DEBUG 476 (void)fprintf(stderr, "Read failed (%s)\n", 477 strerror(errno)); 478 #endif 479 free(*newch); 480 n = 0; 481 newch[0] = '\0'; 482 goto err; 483 } else { 484 n = r; 485 } 486 /* NUL terminate, as every buffer is handled here. */ 487 (*newch)[n] = '\0'; 488 err: 489 if (fdin[1] != -1) 490 (void) close(fdin[1]); 491 (void) close(fdout[0]); 492 #ifdef WNOHANG 493 while (waitpid(-1, NULL, WNOHANG) != -1) 494 continue; 495 #else 496 (void)wait(NULL); 497 #endif 498 (void) close(fdin[0]); 499 500 return n; 501 } 502 } 503