compress.c revision 191736
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. 15 * 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. 27 */ 28/* 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 34 */ 35#include "file.h" 36 37#ifndef lint 38FILE_RCSID("@(#)$File: compress.c,v 1.61 2009/02/03 20:27:51 christos Exp $") 39#endif 40 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 <sys/ioctl.h> 49#ifdef HAVE_SYS_WAIT_H 50#include <sys/wait.h> 51#endif 52#if defined(HAVE_SYS_TIME_H) 53#include <sys/time.h> 54#endif 55#if defined(HAVE_ZLIB_H) && defined(HAVE_LIBZ) 56#define BUILTIN_DECOMPRESS 57#include <zlib.h> 58#endif 59 60private const struct { 61 const char magic[8]; 62 size_t maglen; 63 const char *argv[3]; 64 int silent; 65} compr[] = { 66 { "\037\235", 2, { "gzip", "-cdq", NULL }, 1 }, /* compressed */ 67 /* Uncompress can get stuck; so use gzip first if we have it 68 * Idea from Damien Clark, thanks! */ 69 { "\037\235", 2, { "uncompress", "-c", NULL }, 1 }, /* compressed */ 70 { "\037\213", 2, { "gzip", "-cdq", NULL }, 1 }, /* gzipped */ 71 { "\037\236", 2, { "gzip", "-cdq", NULL }, 1 }, /* frozen */ 72 { "\037\240", 2, { "gzip", "-cdq", NULL }, 1 }, /* SCO LZH */ 73 /* the standard pack utilities do not accept standard input */ 74 { "\037\036", 2, { "gzip", "-cdq", NULL }, 0 }, /* packed */ 75 { "PK\3\4", 4, { "gzip", "-cdq", NULL }, 1 }, /* pkzipped, */ 76 /* ...only first file examined */ 77 { "BZh", 3, { "bzip2", "-cd", NULL }, 1 }, /* bzip2-ed */ 78 { "LZIP", 4, { "lzip", "-cdq", NULL }, 1 }, 79}; 80 81private size_t ncompr = sizeof(compr) / sizeof(compr[0]); 82 83#define NODATA ((size_t)~0) 84 85 86private ssize_t swrite(int, const void *, size_t); 87private size_t uncompressbuf(struct magic_set *, int, size_t, 88 const unsigned char *, unsigned char **, size_t); 89#ifdef BUILTIN_DECOMPRESS 90private size_t uncompressgzipped(struct magic_set *, const unsigned char *, 91 unsigned char **, size_t); 92#endif 93 94protected int 95file_zmagic(struct magic_set *ms, int fd, const char *name, 96 const unsigned char *buf, size_t nbytes) 97{ 98 unsigned char *newbuf = NULL; 99 size_t i, nsz; 100 int rv = 0; 101 int mime = ms->flags & MAGIC_MIME; 102 103 if ((ms->flags & MAGIC_COMPRESS) == 0) 104 return 0; 105 106 for (i = 0; i < ncompr; i++) { 107 if (nbytes < compr[i].maglen) 108 continue; 109 if (memcmp(buf, compr[i].magic, compr[i].maglen) == 0 && 110 (nsz = uncompressbuf(ms, fd, i, buf, &newbuf, 111 nbytes)) != NODATA) { 112 ms->flags &= ~MAGIC_COMPRESS; 113 rv = -1; 114 if (file_buffer(ms, -1, name, newbuf, nsz) == -1) 115 goto error; 116 117 if (mime == MAGIC_MIME || mime == 0) { 118 if (file_printf(ms, mime ? 119 " compressed-encoding=" : " (") == -1) 120 goto error; 121 } 122 123 if ((mime == 0 || mime & MAGIC_MIME_ENCODING) && 124 file_buffer(ms, -1, NULL, buf, nbytes) == -1) 125 goto error; 126 127 if (!mime && file_printf(ms, ")") == -1) 128 goto error; 129 rv = 1; 130 break; 131 } 132 } 133error: 134 if (newbuf) 135 free(newbuf); 136 ms->flags |= MAGIC_COMPRESS; 137 return rv; 138} 139 140/* 141 * `safe' write for sockets and pipes. 142 */ 143private ssize_t 144swrite(int fd, const void *buf, size_t n) 145{ 146 int rv; 147 size_t rn = n; 148 149 do 150 switch (rv = write(fd, buf, n)) { 151 case -1: 152 if (errno == EINTR) 153 continue; 154 return -1; 155 default: 156 n -= rv; 157 buf = ((const char *)buf) + rv; 158 break; 159 } 160 while (n > 0); 161 return rn; 162} 163 164 165/* 166 * `safe' read for sockets and pipes. 167 */ 168protected ssize_t 169sread(int fd, void *buf, size_t n, int canbepipe) 170{ 171 int rv, cnt; 172#ifdef FIONREAD 173 int t = 0; 174#endif 175 size_t rn = n; 176 177 if (fd == STDIN_FILENO) 178 goto nocheck; 179 180#ifdef FIONREAD 181 if ((canbepipe && (ioctl(fd, FIONREAD, &t) == -1)) || (t == 0)) { 182#ifdef FD_ZERO 183 for (cnt = 0;; cnt++) { 184 fd_set check; 185 struct timeval tout = {0, 100 * 1000}; 186 int selrv; 187 188 FD_ZERO(&check); 189 FD_SET(fd, &check); 190 191 /* 192 * Avoid soft deadlock: do not read if there 193 * is nothing to read from sockets and pipes. 194 */ 195 selrv = select(fd + 1, &check, NULL, NULL, &tout); 196 if (selrv == -1) { 197 if (errno == EINTR || errno == EAGAIN) 198 continue; 199 } else if (selrv == 0 && cnt >= 5) { 200 return 0; 201 } else 202 break; 203 } 204#endif 205 (void)ioctl(fd, FIONREAD, &t); 206 } 207 208 if (t > 0 && (size_t)t < n) { 209 n = t; 210 rn = n; 211 } 212#endif 213 214nocheck: 215 do 216 switch ((rv = read(fd, buf, n))) { 217 case -1: 218 if (errno == EINTR) 219 continue; 220 return -1; 221 case 0: 222 return rn - n; 223 default: 224 n -= rv; 225 buf = ((char *)buf) + rv; 226 break; 227 } 228 while (n > 0); 229 return rn; 230} 231 232protected int 233file_pipe2file(struct magic_set *ms, int fd, const void *startbuf, 234 size_t nbytes) 235{ 236 char buf[4096]; 237 int r, tfd; 238 239 (void)strlcpy(buf, "/tmp/file.XXXXXX", sizeof buf); 240#ifndef HAVE_MKSTEMP 241 { 242 char *ptr = mktemp(buf); 243 tfd = open(ptr, O_RDWR|O_TRUNC|O_EXCL|O_CREAT, 0600); 244 r = errno; 245 (void)unlink(ptr); 246 errno = r; 247 } 248#else 249 tfd = mkstemp(buf); 250 r = errno; 251 (void)unlink(buf); 252 errno = r; 253#endif 254 if (tfd == -1) { 255 file_error(ms, errno, 256 "cannot create temporary file for pipe copy"); 257 return -1; 258 } 259 260 if (swrite(tfd, startbuf, nbytes) != (ssize_t)nbytes) 261 r = 1; 262 else { 263 while ((r = sread(fd, buf, sizeof(buf), 1)) > 0) 264 if (swrite(tfd, buf, (size_t)r) != r) 265 break; 266 } 267 268 switch (r) { 269 case -1: 270 file_error(ms, errno, "error copying from pipe to temp file"); 271 return -1; 272 case 0: 273 break; 274 default: 275 file_error(ms, errno, "error while writing to temp file"); 276 return -1; 277 } 278 279 /* 280 * We duplicate the file descriptor, because fclose on a 281 * tmpfile will delete the file, but any open descriptors 282 * can still access the phantom inode. 283 */ 284 if ((fd = dup2(tfd, fd)) == -1) { 285 file_error(ms, errno, "could not dup descriptor for temp file"); 286 return -1; 287 } 288 (void)close(tfd); 289 if (lseek(fd, (off_t)0, SEEK_SET) == (off_t)-1) { 290 file_badseek(ms); 291 return -1; 292 } 293 return fd; 294} 295 296#ifdef BUILTIN_DECOMPRESS 297 298#define FHCRC (1 << 1) 299#define FEXTRA (1 << 2) 300#define FNAME (1 << 3) 301#define FCOMMENT (1 << 4) 302 303private size_t 304uncompressgzipped(struct magic_set *ms, const unsigned char *old, 305 unsigned char **newch, size_t n) 306{ 307 unsigned char flg = old[3]; 308 size_t data_start = 10; 309 z_stream z; 310 int rc; 311 312 if (flg & FEXTRA) { 313 if (data_start+1 >= n) 314 return 0; 315 data_start += 2 + old[data_start] + old[data_start + 1] * 256; 316 } 317 if (flg & FNAME) { 318 while(data_start < n && old[data_start]) 319 data_start++; 320 data_start++; 321 } 322 if(flg & FCOMMENT) { 323 while(data_start < n && old[data_start]) 324 data_start++; 325 data_start++; 326 } 327 if(flg & FHCRC) 328 data_start += 2; 329 330 if (data_start >= n) 331 return 0; 332 if ((*newch = CAST(unsigned char *, malloc(HOWMANY + 1))) == NULL) { 333 return 0; 334 } 335 336 /* XXX: const castaway, via strchr */ 337 z.next_in = (Bytef *)strchr((const char *)old + data_start, 338 old[data_start]); 339 z.avail_in = n - data_start; 340 z.next_out = *newch; 341 z.avail_out = HOWMANY; 342 z.zalloc = Z_NULL; 343 z.zfree = Z_NULL; 344 z.opaque = Z_NULL; 345 346 rc = inflateInit2(&z, -15); 347 if (rc != Z_OK) { 348 file_error(ms, 0, "zlib: %s", z.msg); 349 return 0; 350 } 351 352 rc = inflate(&z, Z_SYNC_FLUSH); 353 if (rc != Z_OK && rc != Z_STREAM_END) { 354 file_error(ms, 0, "zlib: %s", z.msg); 355 return 0; 356 } 357 358 n = (size_t)z.total_out; 359 (void)inflateEnd(&z); 360 361 /* let's keep the nul-terminate tradition */ 362 (*newch)[n] = '\0'; 363 364 return n; 365} 366#endif 367 368private size_t 369uncompressbuf(struct magic_set *ms, int fd, size_t method, 370 const unsigned char *old, unsigned char **newch, size_t n) 371{ 372 int fdin[2], fdout[2]; 373 int r; 374 375#ifdef BUILTIN_DECOMPRESS 376 /* FIXME: This doesn't cope with bzip2 */ 377 if (method == 2) 378 return uncompressgzipped(ms, old, newch, n); 379#endif 380 (void)fflush(stdout); 381 (void)fflush(stderr); 382 383 if ((fd != -1 && pipe(fdin) == -1) || pipe(fdout) == -1) { 384 file_error(ms, errno, "cannot create pipe"); 385 return NODATA; 386 } 387 switch (fork()) { 388 case 0: /* child */ 389 (void) close(0); 390 if (fd != -1) { 391 (void) dup(fd); 392 (void) lseek(0, (off_t)0, SEEK_SET); 393 } else { 394 (void) dup(fdin[0]); 395 (void) close(fdin[0]); 396 (void) close(fdin[1]); 397 } 398 399 (void) close(1); 400 (void) dup(fdout[1]); 401 (void) close(fdout[0]); 402 (void) close(fdout[1]); 403#ifndef DEBUG 404 if (compr[method].silent) 405 (void)close(2); 406#endif 407 408 (void)execvp(compr[method].argv[0], 409 (char *const *)(intptr_t)compr[method].argv); 410#ifdef DEBUG 411 (void)fprintf(stderr, "exec `%s' failed (%s)\n", 412 compr[method].argv[0], strerror(errno)); 413#endif 414 exit(1); 415 /*NOTREACHED*/ 416 case -1: 417 file_error(ms, errno, "could not fork"); 418 return NODATA; 419 420 default: /* parent */ 421 (void) close(fdout[1]); 422 if (fd == -1) { 423 (void) close(fdin[0]); 424 /* 425 * fork again, to avoid blocking because both 426 * pipes filled 427 */ 428 switch (fork()) { 429 case 0: /* child */ 430 (void)close(fdout[0]); 431 if (swrite(fdin[1], old, n) != (ssize_t)n) { 432#ifdef DEBUG 433 (void)fprintf(stderr, 434 "Write failed (%s)\n", 435 strerror(errno)); 436#endif 437 exit(1); 438 } 439 exit(0); 440 /*NOTREACHED*/ 441 442 case -1: 443#ifdef DEBUG 444 (void)fprintf(stderr, "Fork failed (%s)\n", 445 strerror(errno)); 446#endif 447 exit(1); 448 /*NOTREACHED*/ 449 450 default: /* parent */ 451 break; 452 } 453 (void) close(fdin[1]); 454 fdin[1] = -1; 455 } 456 457 if ((*newch = (unsigned char *) malloc(HOWMANY + 1)) == NULL) { 458#ifdef DEBUG 459 (void)fprintf(stderr, "Malloc failed (%s)\n", 460 strerror(errno)); 461#endif 462 n = 0; 463 goto err; 464 } 465 if ((r = sread(fdout[0], *newch, HOWMANY, 0)) <= 0) { 466#ifdef DEBUG 467 (void)fprintf(stderr, "Read failed (%s)\n", 468 strerror(errno)); 469#endif 470 free(*newch); 471 n = 0; 472 newch[0] = '\0'; 473 goto err; 474 } else { 475 n = r; 476 } 477 /* NUL terminate, as every buffer is handled here. */ 478 (*newch)[n] = '\0'; 479err: 480 if (fdin[1] != -1) 481 (void) close(fdin[1]); 482 (void) close(fdout[0]); 483#ifdef WNOHANG 484 while (waitpid(-1, NULL, WNOHANG) != -1) 485 continue; 486#else 487 (void)wait(NULL); 488#endif 489 return n; 490 } 491} 492