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 * apprentice - make one pass through /etc/magic, learning its secrets. 30 */ 31 32 #include "file.h" 33 34 #ifndef lint 35 FILE_RCSID("@(#)$File: apprentice.c,v 1.376 2026/05/09 22:44:32 christos Exp $") 36 #endif /* lint */ 37 38 #include "magic.h" 39 #include "swap.h" 40 #include <stdlib.h> 41 #ifdef HAVE_UNISTD_H 42 #include <unistd.h> 43 #endif 44 #include <stddef.h> 45 #include <string.h> 46 #include <assert.h> 47 #include <ctype.h> 48 #include <fcntl.h> 49 #ifdef QUICK 50 #include <sys/mman.h> 51 #endif 52 #include <dirent.h> 53 #include <limits.h> 54 55 56 #define EATAB {while (*l && isascii(CAST(unsigned char, *l)) && \ 57 isspace(CAST(unsigned char, *l))) ++l;} 58 #define LOWCASE(l) (isupper(CAST(unsigned char, l)) ? \ 59 tolower(CAST(unsigned char, l)) : (l)) 60 /* 61 * Work around a bug in headers on Digital Unix. 62 * At least confirmed for: OSF1 V4.0 878 63 */ 64 #if defined(__osf__) && defined(__DECC) 65 #ifdef MAP_FAILED 66 #undef MAP_FAILED 67 #endif 68 #endif 69 70 #ifndef MAP_FAILED 71 #define MAP_FAILED (void *) -1 72 #endif 73 74 #ifndef MAP_FILE 75 #define MAP_FILE 0 76 #endif 77 78 #define ALLOC_CHUNK CAST(size_t, 10) 79 #define ALLOC_INCR CAST(size_t, 200) 80 81 #define MAP_TYPE_USER 0 82 #define MAP_TYPE_MALLOC 1 83 #define MAP_TYPE_MMAP 2 84 85 struct magic_entry { 86 struct magic *mp; 87 uint32_t cont_count; 88 uint32_t max_count; 89 }; 90 91 struct magic_entry_set { 92 struct magic_entry *me; 93 uint32_t count; 94 uint32_t max; 95 }; 96 97 struct magic_map { 98 void *p; 99 size_t len; 100 int type; 101 struct magic *magic[MAGIC_SETS]; 102 uint32_t nmagic[MAGIC_SETS]; 103 }; 104 105 file_private int file_formats[FILE_NAMES_SIZE]; 106 file_private const size_t file_nformats = FILE_NAMES_SIZE; 107 file_protected const char *file_names[FILE_NAMES_SIZE]; 108 file_protected const size_t file_nnames = FILE_NAMES_SIZE; 109 110 file_private int getvalue(struct magic_set *ms, struct magic *, const char **, int); 111 file_private int hextoint(int); 112 file_private const char *getstr(struct magic_set *, struct magic *, const char *, 113 int); 114 file_private int parse(struct magic_set *, struct magic_entry *, const char *, 115 const char *, size_t, size_t, int); 116 file_private void eatsize(const char **); 117 file_private int apprentice_1(struct magic_set *, const char *, int); 118 file_private ssize_t apprentice_magic_strength_1(const struct magic *); 119 file_private int apprentice_sort(const void *, const void *); 120 file_private void apprentice_list(struct mlist *, int ); 121 file_private struct magic_map *apprentice_load(struct magic_set *, 122 const char *, int); 123 file_private struct mlist *mlist_alloc(void); 124 file_private void mlist_free_all(struct magic_set *); 125 file_private void mlist_free(struct mlist *); 126 file_private void byteswap(struct magic *, uint32_t); 127 file_private void bs1(struct magic *); 128 129 file_private char *mkdbname(struct magic_set *, const char *, int); 130 file_private struct magic_map *apprentice_buf(struct magic_set *, struct magic *, 131 size_t); 132 file_private struct magic_map *apprentice_map(struct magic_set *, const char *); 133 file_private int check_buffer(struct magic_set *, struct magic_map *, const char *); 134 file_private void apprentice_unmap(struct magic_map *); 135 file_private int apprentice_compile(struct magic_set *, struct magic_map *, 136 const char *); 137 file_private int check_format_type(const char *, int, const char **); 138 file_private int check_format(struct magic_set *, struct magic *); 139 file_private int get_op(char); 140 file_private int parse_mime(struct magic_set *, struct magic_entry *, const char *, 141 size_t); 142 file_private int parse_strength(struct magic_set *, struct magic_entry *, 143 const char *, size_t); 144 file_private int parse_apple(struct magic_set *, struct magic_entry *, const char *, 145 size_t); 146 file_private int parse_ext(struct magic_set *, struct magic_entry *, const char *, 147 size_t); 148 149 150 file_private size_t magicsize = sizeof(struct magic); 151 152 file_private const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc"; 153 154 file_private struct { 155 const char *name; 156 size_t len; 157 int (*fun)(struct magic_set *, struct magic_entry *, const char *, 158 size_t); 159 } bang[] = { 160 #define DECLARE_FIELD(name) { # name, sizeof(# name) - 1, parse_ ## name } 161 DECLARE_FIELD(mime), 162 DECLARE_FIELD(apple), 163 DECLARE_FIELD(ext), 164 DECLARE_FIELD(strength), 165 #undef DECLARE_FIELD 166 { NULL, 0, NULL } 167 }; 168 169 #ifdef COMPILE_ONLY 170 171 int main(int, char *[]); 172 173 int 174 main(int argc, char *argv[]) 175 { 176 int ret; 177 struct magic_set *ms; 178 char *progname; 179 180 if ((progname = strrchr(argv[0], '/')) != NULL) 181 progname++; 182 else 183 progname = argv[0]; 184 185 if (argc != 2) { 186 (void)fprintf(stderr, "Usage: %s file\n", progname); 187 return 1; 188 } 189 190 if ((ms = magic_open(MAGIC_CHECK)) == NULL) { 191 (void)fprintf(stderr, "%s: %s\n", progname, strerror(errno)); 192 return 1; 193 } 194 ret = magic_compile(ms, argv[1]) == -1 ? 1 : 0; 195 if (ret == 1) 196 (void)fprintf(stderr, "%s: %s\n", progname, magic_error(ms)); 197 magic_close(ms); 198 return ret; 199 } 200 #endif /* COMPILE_ONLY */ 201 202 struct type_tbl_s { 203 const char name[16]; 204 const size_t len; 205 const int type; 206 const int format; 207 }; 208 209 /* 210 * XXX - the actual Single UNIX Specification says that "long" means "long", 211 * as in the C data type, but we treat it as meaning "4-byte integer". 212 * Given that the OS X version of file 5.04 did the same, I guess that passes 213 * the actual test; having "long" be dependent on how big a "long" is on 214 * the machine running "file" is silly. 215 */ 216 static const struct type_tbl_s type_tbl[] = { 217 # define XX(s) s, (sizeof(s) - 1) 218 # define XX_NULL "", 0 219 { XX("invalid"), FILE_INVALID, FILE_FMT_NONE }, 220 { XX("byte"), FILE_BYTE, FILE_FMT_NUM }, 221 { XX("short"), FILE_SHORT, FILE_FMT_NUM }, 222 { XX("default"), FILE_DEFAULT, FILE_FMT_NONE }, 223 { XX("long"), FILE_LONG, FILE_FMT_NUM }, 224 { XX("string"), FILE_STRING, FILE_FMT_STR }, 225 { XX("date"), FILE_DATE, FILE_FMT_STR }, 226 { XX("beshort"), FILE_BESHORT, FILE_FMT_NUM }, 227 { XX("belong"), FILE_BELONG, FILE_FMT_NUM }, 228 { XX("bedate"), FILE_BEDATE, FILE_FMT_STR }, 229 { XX("leshort"), FILE_LESHORT, FILE_FMT_NUM }, 230 { XX("lelong"), FILE_LELONG, FILE_FMT_NUM }, 231 { XX("ledate"), FILE_LEDATE, FILE_FMT_STR }, 232 { XX("pstring"), FILE_PSTRING, FILE_FMT_STR }, 233 { XX("ldate"), FILE_LDATE, FILE_FMT_STR }, 234 { XX("beldate"), FILE_BELDATE, FILE_FMT_STR }, 235 { XX("leldate"), FILE_LELDATE, FILE_FMT_STR }, 236 { XX("regex"), FILE_REGEX, FILE_FMT_STR }, 237 { XX("bestring16"), FILE_BESTRING16, FILE_FMT_STR }, 238 { XX("lestring16"), FILE_LESTRING16, FILE_FMT_STR }, 239 { XX("search"), FILE_SEARCH, FILE_FMT_STR }, 240 { XX("medate"), FILE_MEDATE, FILE_FMT_STR }, 241 { XX("meldate"), FILE_MELDATE, FILE_FMT_STR }, 242 { XX("melong"), FILE_MELONG, FILE_FMT_NUM }, 243 { XX("quad"), FILE_QUAD, FILE_FMT_QUAD }, 244 { XX("lequad"), FILE_LEQUAD, FILE_FMT_QUAD }, 245 { XX("bequad"), FILE_BEQUAD, FILE_FMT_QUAD }, 246 { XX("qdate"), FILE_QDATE, FILE_FMT_STR }, 247 { XX("leqdate"), FILE_LEQDATE, FILE_FMT_STR }, 248 { XX("beqdate"), FILE_BEQDATE, FILE_FMT_STR }, 249 { XX("qldate"), FILE_QLDATE, FILE_FMT_STR }, 250 { XX("leqldate"), FILE_LEQLDATE, FILE_FMT_STR }, 251 { XX("beqldate"), FILE_BEQLDATE, FILE_FMT_STR }, 252 { XX("float"), FILE_FLOAT, FILE_FMT_FLOAT }, 253 { XX("befloat"), FILE_BEFLOAT, FILE_FMT_FLOAT }, 254 { XX("lefloat"), FILE_LEFLOAT, FILE_FMT_FLOAT }, 255 { XX("double"), FILE_DOUBLE, FILE_FMT_DOUBLE }, 256 { XX("bedouble"), FILE_BEDOUBLE, FILE_FMT_DOUBLE }, 257 { XX("ledouble"), FILE_LEDOUBLE, FILE_FMT_DOUBLE }, 258 { XX("leid3"), FILE_LEID3, FILE_FMT_NUM }, 259 { XX("beid3"), FILE_BEID3, FILE_FMT_NUM }, 260 { XX("indirect"), FILE_INDIRECT, FILE_FMT_NUM }, 261 { XX("qwdate"), FILE_QWDATE, FILE_FMT_STR }, 262 { XX("leqwdate"), FILE_LEQWDATE, FILE_FMT_STR }, 263 { XX("beqwdate"), FILE_BEQWDATE, FILE_FMT_STR }, 264 { XX("name"), FILE_NAME, FILE_FMT_NONE }, 265 { XX("use"), FILE_USE, FILE_FMT_NONE }, 266 { XX("clear"), FILE_CLEAR, FILE_FMT_NONE }, 267 { XX("der"), FILE_DER, FILE_FMT_STR }, 268 { XX("guid"), FILE_GUID, FILE_FMT_STR }, 269 { XX("leguid"), FILE_LEGUID, FILE_FMT_STR }, 270 { XX("beguid"), FILE_BEGUID, FILE_FMT_STR }, 271 { XX("offset"), FILE_OFFSET, FILE_FMT_QUAD }, 272 { XX("bevarint"), FILE_BEVARINT, FILE_FMT_STR }, 273 { XX("levarint"), FILE_LEVARINT, FILE_FMT_STR }, 274 { XX("msdosdate"), FILE_MSDOSDATE, FILE_FMT_STR }, 275 { XX("lemsdosdate"), FILE_LEMSDOSDATE, FILE_FMT_STR }, 276 { XX("bemsdosdate"), FILE_BEMSDOSDATE, FILE_FMT_STR }, 277 { XX("msdostime"), FILE_MSDOSTIME, FILE_FMT_STR }, 278 { XX("lemsdostime"), FILE_LEMSDOSTIME, FILE_FMT_STR }, 279 { XX("bemsdostime"), FILE_BEMSDOSTIME, FILE_FMT_STR }, 280 { XX("octal"), FILE_OCTAL, FILE_FMT_STR }, 281 { XX_NULL, FILE_INVALID, FILE_FMT_NONE }, 282 }; 283 284 /* 285 * These are not types, and cannot be preceded by "u" to make them 286 * unsigned. 287 */ 288 static const struct type_tbl_s special_tbl[] = { 289 { XX("der"), FILE_DER, FILE_FMT_STR }, 290 { XX("name"), FILE_NAME, FILE_FMT_STR }, 291 { XX("use"), FILE_USE, FILE_FMT_STR }, 292 { XX("octal"), FILE_OCTAL, FILE_FMT_STR }, 293 { XX_NULL, FILE_INVALID, FILE_FMT_NONE }, 294 }; 295 # undef XX 296 # undef XX_NULL 297 298 file_private int 299 get_type(const struct type_tbl_s *tbl, const char *l, const char **t) 300 { 301 const struct type_tbl_s *p; 302 303 for (p = tbl; p->len; p++) { 304 if (strncmp(l, p->name, p->len) == 0) { 305 if (t) 306 *t = l + p->len; 307 break; 308 } 309 } 310 return p->type; 311 } 312 313 file_private off_t 314 maxoff_t(void) { 315 if (/*CONSTCOND*/sizeof(off_t) == sizeof(int)) 316 return CAST(off_t, INT_MAX); 317 if (/*CONSTCOND*/sizeof(off_t) == sizeof(long)) 318 return CAST(off_t, LONG_MAX); 319 return 0x7fffffff; 320 } 321 322 file_private int 323 get_standard_integer_type(const char *l, const char **t) 324 { 325 int type; 326 327 if (l[0] == '\0') 328 return FILE_INVALID; 329 if (l[1] == '\0') 330 return FILE_INVALID; 331 else if (isalpha(CAST(unsigned char, l[1]))) { 332 switch (l[1]) { 333 case 'C': 334 /* "dC" and "uC" */ 335 type = FILE_BYTE; 336 break; 337 case 'S': 338 /* "dS" and "uS" */ 339 type = FILE_SHORT; 340 break; 341 case 'I': 342 case 'L': 343 /* 344 * "dI", "dL", "uI", and "uL". 345 * 346 * XXX - the actual Single UNIX Specification says 347 * that "L" means "long", as in the C data type, 348 * but we treat it as meaning "4-byte integer". 349 * Given that the OS X version of file 5.04 did 350 * the same, I guess that passes the actual SUS 351 * validation suite; having "dL" be dependent on 352 * how big a "long" is on the machine running 353 * "file" is silly. 354 */ 355 type = FILE_LONG; 356 break; 357 case 'Q': 358 /* "dQ" and "uQ" */ 359 type = FILE_QUAD; 360 break; 361 default: 362 /* "d{anything else}", "u{anything else}" */ 363 return FILE_INVALID; 364 } 365 l += 2; 366 } else if (isdigit(CAST(unsigned char, l[1]))) { 367 /* 368 * "d{num}" and "u{num}"; we only support {num} values 369 * of 1, 2, 4, and 8 - the Single UNIX Specification 370 * doesn't say anything about whether arbitrary 371 * values should be supported, but both the Solaris 10 372 * and OS X Mountain Lion versions of file passed the 373 * Single UNIX Specification validation suite, and 374 * neither of them support values bigger than 8 or 375 * non-power-of-2 values. 376 */ 377 if (isdigit(CAST(unsigned char, l[2]))) { 378 /* Multi-digit, so > 9 */ 379 return FILE_INVALID; 380 } 381 switch (l[1]) { 382 case '1': 383 type = FILE_BYTE; 384 break; 385 case '2': 386 type = FILE_SHORT; 387 break; 388 case '4': 389 type = FILE_LONG; 390 break; 391 case '8': 392 type = FILE_QUAD; 393 break; 394 default: 395 /* XXX - what about 3, 5, 6, or 7? */ 396 return FILE_INVALID; 397 } 398 l += 2; 399 } else { 400 /* 401 * "d" or "u" by itself. 402 */ 403 type = FILE_LONG; 404 ++l; 405 } 406 if (t) 407 *t = l; 408 return type; 409 } 410 411 file_private void 412 init_file_tables(void) 413 { 414 static int done = 0; 415 const struct type_tbl_s *p; 416 417 if (done) 418 return; 419 done++; 420 421 for (p = type_tbl; p->len; p++) { 422 assert(p->type < FILE_NAMES_SIZE); 423 file_names[p->type] = p->name; 424 file_formats[p->type] = p->format; 425 } 426 assert(p - type_tbl == FILE_NAMES_SIZE); 427 } 428 429 file_private int 430 add_mlist(struct mlist *mlp, struct magic_map *map, size_t idx) 431 { 432 struct mlist *ml; 433 434 mlp->map = NULL; 435 if ((ml = CAST(struct mlist *, malloc(sizeof(*ml)))) == NULL) 436 return -1; 437 438 ml->map = idx == 0 ? map : NULL; 439 ml->magic = map->magic[idx]; 440 ml->nmagic = map->nmagic[idx]; 441 if (ml->nmagic) { 442 ml->magic_rxcomp = CAST(file_regex_t **, 443 calloc(ml->nmagic, sizeof(*ml->magic_rxcomp))); 444 if (ml->magic_rxcomp == NULL) { 445 free(ml); 446 return -1; 447 } 448 } else 449 ml->magic_rxcomp = NULL; 450 mlp->prev->next = ml; 451 ml->prev = mlp->prev; 452 ml->next = mlp; 453 mlp->prev = ml; 454 return 0; 455 } 456 457 /* 458 * Handle one file or directory. 459 */ 460 file_private int 461 apprentice_1(struct magic_set *ms, const char *fn, int action) 462 { 463 struct magic_map *map; 464 #ifndef COMPILE_ONLY 465 size_t i; 466 #endif 467 468 if (magicsize != FILE_MAGICSIZE) { 469 file_error(ms, 0, "magic element size %lu != %lu", 470 CAST(unsigned long, sizeof(*map->magic[0])), 471 CAST(unsigned long, FILE_MAGICSIZE)); 472 return -1; 473 } 474 475 if (action == FILE_COMPILE) { 476 map = apprentice_load(ms, fn, action); 477 if (map == NULL) 478 return -1; 479 return apprentice_compile(ms, map, fn); 480 } 481 482 #ifndef COMPILE_ONLY 483 map = apprentice_map(ms, fn); 484 if (map == NULL) { 485 if (ms->flags & MAGIC_CHECK) 486 file_magwarn(ms, "using regular magic file `%s'", fn); 487 map = apprentice_load(ms, fn, action); 488 if (map == NULL) 489 return -1; 490 } 491 492 for (i = 0; i < MAGIC_SETS; i++) { 493 if (add_mlist(ms->mlist[i], map, i) == -1) { 494 /* failed to add to any list, free explicitly */ 495 if (i == 0) 496 apprentice_unmap(map); 497 else 498 mlist_free_all(ms); 499 file_oomem(ms, sizeof(*ms->mlist[0])); 500 return -1; 501 } 502 } 503 504 if (action == FILE_LIST) { 505 for (i = 0; i < MAGIC_SETS; i++) { 506 printf("Set %" SIZE_T_FORMAT "u:\nBinary patterns:\n", 507 i); 508 apprentice_list(ms->mlist[i], BINTEST); 509 printf("Text patterns:\n"); 510 apprentice_list(ms->mlist[i], TEXTTEST); 511 } 512 } 513 return 0; 514 #else 515 return 0; 516 #endif /* COMPILE_ONLY */ 517 } 518 519 file_protected void 520 file_ms_free(struct magic_set *ms) 521 { 522 size_t i; 523 if (ms == NULL) 524 return; 525 for (i = 0; i < MAGIC_SETS; i++) 526 mlist_free(ms->mlist[i]); 527 free(ms->o.pbuf); 528 free(ms->o.buf); 529 free(ms->c.li); 530 free(ms->fnamebuf); 531 #ifdef USE_C_LOCALE 532 freelocale(ms->c_lc_ctype); 533 #endif 534 free(ms); 535 } 536 537 file_protected struct magic_set * 538 file_ms_alloc(int flags) 539 { 540 struct magic_set *ms; 541 size_t i, len; 542 543 if ((ms = CAST(struct magic_set *, calloc(CAST(size_t, 1u), 544 sizeof(*ms)))) == NULL) 545 return NULL; 546 547 if (magic_setflags(ms, flags) == -1) { 548 errno = EINVAL; 549 goto free; 550 } 551 552 ms->o.buf = ms->o.pbuf = NULL; 553 ms->o.blen = 0; 554 len = (ms->c.len = 10) * sizeof(*ms->c.li); 555 556 if ((ms->c.li = CAST(struct level_info *, malloc(len))) == NULL) 557 goto free; 558 559 ms->event_flags = 0; 560 ms->error = -1; 561 for (i = 0; i < MAGIC_SETS; i++) 562 ms->mlist[i] = NULL; 563 ms->fnamebuf = NULL; 564 ms->file = "unknown"; 565 ms->line = 0; 566 ms->magwarn = 0; 567 ms->indir_max = FILE_INDIR_MAX; 568 ms->name_max = FILE_NAME_MAX; 569 ms->elf_shnum_max = FILE_ELF_SHNUM_MAX; 570 ms->elf_shsize_max = FILE_ELF_SHSIZE_MAX; 571 ms->elf_phnum_max = FILE_ELF_PHNUM_MAX; 572 ms->elf_notes_max = FILE_ELF_NOTES_MAX; 573 ms->regex_max = FILE_REGEX_MAX; 574 ms->bytes_max = FILE_BYTES_MAX; 575 ms->encoding_max = FILE_ENCODING_MAX; 576 ms->magwarn_max = FILE_MAGWARN_MAX; 577 #ifdef USE_C_LOCALE 578 ms->c_lc_ctype = newlocale(LC_CTYPE_MASK, "C", 0); 579 assert(ms->c_lc_ctype != NULL); 580 #endif 581 return ms; 582 free: 583 free(ms); 584 return NULL; 585 } 586 587 file_private void 588 apprentice_unmap(struct magic_map *map) 589 { 590 size_t i; 591 char *p; 592 if (map == NULL) 593 return; 594 595 switch (map->type) { 596 case MAP_TYPE_USER: 597 break; 598 case MAP_TYPE_MALLOC: 599 p = CAST(char *, map->p); 600 for (i = 0; i < MAGIC_SETS; i++) { 601 char *b = RCAST(char *, map->magic[i]); 602 if (p != NULL && b >= p && b <= p + map->len) 603 continue; 604 free(b); 605 } 606 free(p); 607 break; 608 #ifdef QUICK 609 case MAP_TYPE_MMAP: 610 if (map->p && map->p != MAP_FAILED) 611 (void)munmap(map->p, map->len); 612 break; 613 #endif 614 default: 615 fprintf(stderr, "Bad map type %d", map->type); 616 abort(); 617 } 618 free(map); 619 } 620 621 file_private struct mlist * 622 mlist_alloc(void) 623 { 624 struct mlist *mlist; 625 if ((mlist = CAST(struct mlist *, calloc(1, sizeof(*mlist)))) == NULL) { 626 return NULL; 627 } 628 mlist->next = mlist->prev = mlist; 629 return mlist; 630 } 631 632 file_private void 633 mlist_free_all(struct magic_set *ms) 634 { 635 size_t i; 636 637 for (i = 0; i < MAGIC_SETS; i++) { 638 mlist_free(ms->mlist[i]); 639 ms->mlist[i] = NULL; 640 } 641 } 642 643 file_private void 644 mlist_free_one(struct mlist *ml) 645 { 646 size_t i; 647 648 if (ml->map) 649 apprentice_unmap(CAST(struct magic_map *, ml->map)); 650 651 for (i = 0; i < ml->nmagic; ++i) { 652 if (ml->magic_rxcomp[i]) { 653 file_regfree(ml->magic_rxcomp[i]); 654 free(ml->magic_rxcomp[i]); 655 ml->magic_rxcomp[i] = NULL; 656 } 657 } 658 free(ml->magic_rxcomp); 659 ml->magic_rxcomp = NULL; 660 free(ml); 661 } 662 663 file_private void 664 mlist_free(struct mlist *mlist) 665 { 666 struct mlist *ml, *next; 667 668 if (mlist == NULL) 669 return; 670 671 for (ml = mlist->next; ml != mlist;) { 672 next = ml->next; 673 mlist_free_one(ml); 674 ml = next; 675 } 676 mlist_free_one(mlist); 677 } 678 679 #ifndef COMPILE_ONLY 680 /* void **bufs: an array of compiled magic files */ 681 file_protected int 682 buffer_apprentice(struct magic_set *ms, struct magic **bufs, 683 size_t *sizes, size_t nbufs) 684 { 685 size_t i, j; 686 struct magic_map *map; 687 688 if (nbufs == 0) 689 return -1; 690 691 (void)file_reset(ms, 0); 692 693 init_file_tables(); 694 695 for (i = 0; i < MAGIC_SETS; i++) { 696 mlist_free(ms->mlist[i]); 697 if ((ms->mlist[i] = mlist_alloc()) == NULL) { 698 file_oomem(ms, sizeof(*ms->mlist[0])); 699 goto fail; 700 } 701 } 702 703 for (i = 0; i < nbufs; i++) { 704 map = apprentice_buf(ms, bufs[i], sizes[i]); 705 if (map == NULL) 706 goto fail; 707 708 for (j = 0; j < MAGIC_SETS; j++) { 709 if (add_mlist(ms->mlist[j], map, j) == -1) { 710 file_oomem(ms, sizeof(*ms->mlist[0])); 711 goto fail; 712 } 713 } 714 } 715 716 return 0; 717 fail: 718 mlist_free_all(ms); 719 return -1; 720 } 721 #endif 722 723 /* const char *fn: list of magic files and directories */ 724 file_protected int 725 file_apprentice(struct magic_set *ms, const char *fn, int action) 726 { 727 char *p; 728 int fileerr, errs = -1; 729 size_t i, j; 730 731 (void)file_reset(ms, 0); 732 733 if ((fn = magic_getpath(fn, action)) == NULL) 734 return -1; 735 736 init_file_tables(); 737 738 free(ms->fnamebuf); 739 ms->file = NULL; 740 if ((ms->fnamebuf = strdup(fn)) == NULL) { 741 file_oomem(ms, strlen(fn)); 742 return -1; 743 } 744 745 for (i = 0; i < MAGIC_SETS; i++) { 746 mlist_free(ms->mlist[i]); 747 if ((ms->mlist[i] = mlist_alloc()) == NULL) { 748 file_oomem(ms, sizeof(*ms->mlist[0])); 749 for (j = 0; j < i; j++) { 750 mlist_free(ms->mlist[j]); 751 ms->mlist[j] = NULL; 752 } 753 return -1; 754 } 755 } 756 fn = ms->fnamebuf; 757 758 while (fn) { 759 p = CCAST(char *, strchr(fn, PATHSEP)); 760 if (p) 761 *p++ = '\0'; 762 if (*fn == '\0') 763 break; 764 fileerr = apprentice_1(ms, fn, action); 765 errs = MAX(errs, fileerr); 766 fn = p; 767 } 768 769 if (errs == -1) { 770 for (i = 0; i < MAGIC_SETS; i++) { 771 mlist_free(ms->mlist[i]); 772 ms->mlist[i] = NULL; 773 } 774 file_error(ms, 0, "could not find any valid magic files!"); 775 return -1; 776 } 777 778 #if 0 779 /* 780 * Always leave the database loaded 781 */ 782 if (action == FILE_LOAD) 783 return 0; 784 785 for (i = 0; i < MAGIC_SETS; i++) { 786 mlist_free(ms->mlist[i]); 787 ms->mlist[i] = NULL; 788 } 789 #endif 790 791 switch (action) { 792 case FILE_LOAD: 793 case FILE_COMPILE: 794 case FILE_CHECK: 795 case FILE_LIST: 796 return 0; 797 default: 798 file_error(ms, 0, "Invalid action %d", action); 799 return -1; 800 } 801 } 802 803 /* 804 * Compute the real length of a magic expression, for the purposes 805 * of determining how "strong" a magic expression is (approximating 806 * how specific its matches are): 807 * - magic characters count 0 unless escaped. 808 * - [] expressions count 1 809 * - {} expressions count 0 810 * - regular characters or escaped magic characters count 1 811 * - 0 length expressions count as one 812 */ 813 file_private size_t 814 nonmagic(const char *str) 815 { 816 const char *p; 817 size_t rv = 0; 818 819 for (p = str; *p; p++) 820 switch (*p) { 821 case '\\': /* Escaped anything counts 1 */ 822 if (!*++p) 823 p--; 824 rv++; 825 continue; 826 case '?': /* Magic characters count 0 */ 827 case '*': 828 case '.': 829 case '+': 830 case '^': 831 case '$': 832 continue; 833 case '[': /* Bracketed expressions count 1 the ']' */ 834 while (*p && *p != ']') 835 p++; 836 p--; 837 continue; 838 case '{': /* Braced expressions count 0 */ 839 while (*p && *p != '}') 840 p++; 841 if (!*p) 842 p--; 843 continue; 844 default: /* Anything else counts 1 */ 845 rv++; 846 continue; 847 } 848 849 return rv == 0 ? 1 : rv; /* Return at least 1 */ 850 } 851 852 853 file_private size_t 854 typesize(int type) 855 { 856 switch (type) { 857 case FILE_BYTE: 858 return 1; 859 860 case FILE_SHORT: 861 case FILE_LESHORT: 862 case FILE_BESHORT: 863 case FILE_MSDOSDATE: 864 case FILE_BEMSDOSDATE: 865 case FILE_LEMSDOSDATE: 866 case FILE_MSDOSTIME: 867 case FILE_BEMSDOSTIME: 868 case FILE_LEMSDOSTIME: 869 return 2; 870 871 case FILE_LONG: 872 case FILE_LELONG: 873 case FILE_BELONG: 874 case FILE_MELONG: 875 return 4; 876 877 case FILE_DATE: 878 case FILE_LEDATE: 879 case FILE_BEDATE: 880 case FILE_MEDATE: 881 case FILE_LDATE: 882 case FILE_LELDATE: 883 case FILE_BELDATE: 884 case FILE_MELDATE: 885 case FILE_FLOAT: 886 case FILE_BEFLOAT: 887 case FILE_LEFLOAT: 888 case FILE_BEID3: 889 case FILE_LEID3: 890 return 4; 891 892 case FILE_QUAD: 893 case FILE_BEQUAD: 894 case FILE_LEQUAD: 895 case FILE_QDATE: 896 case FILE_LEQDATE: 897 case FILE_BEQDATE: 898 case FILE_QLDATE: 899 case FILE_LEQLDATE: 900 case FILE_BEQLDATE: 901 case FILE_QWDATE: 902 case FILE_LEQWDATE: 903 case FILE_BEQWDATE: 904 case FILE_DOUBLE: 905 case FILE_BEDOUBLE: 906 case FILE_LEDOUBLE: 907 case FILE_OFFSET: 908 case FILE_BEVARINT: 909 case FILE_LEVARINT: 910 return 8; 911 912 case FILE_LEGUID: 913 case FILE_BEGUID: 914 case FILE_GUID: 915 return 16; 916 917 default: 918 return FILE_BADSIZE; 919 } 920 } 921 922 /* 923 * Get weight of this magic entry, for sorting purposes. 924 */ 925 file_private ssize_t 926 apprentice_magic_strength_1(const struct magic *m) 927 { 928 #define MULT 10U 929 size_t ts, v; 930 ssize_t val = 2 * MULT; /* baseline strength */ 931 932 switch (m->type) { 933 case FILE_DEFAULT: /* make sure this sorts last */ 934 if (m->factor_op != FILE_FACTOR_OP_NONE) { 935 file_magwarn1("Unsupported factor_op in default %d", 936 m->factor_op); 937 } 938 return 0; 939 940 case FILE_BYTE: 941 case FILE_SHORT: 942 case FILE_LESHORT: 943 case FILE_BESHORT: 944 case FILE_LONG: 945 case FILE_LELONG: 946 case FILE_BELONG: 947 case FILE_MELONG: 948 case FILE_DATE: 949 case FILE_LEDATE: 950 case FILE_BEDATE: 951 case FILE_MEDATE: 952 case FILE_LDATE: 953 case FILE_LELDATE: 954 case FILE_BELDATE: 955 case FILE_MELDATE: 956 case FILE_FLOAT: 957 case FILE_BEFLOAT: 958 case FILE_LEFLOAT: 959 case FILE_QUAD: 960 case FILE_BEQUAD: 961 case FILE_LEQUAD: 962 case FILE_QDATE: 963 case FILE_LEQDATE: 964 case FILE_BEQDATE: 965 case FILE_QLDATE: 966 case FILE_LEQLDATE: 967 case FILE_BEQLDATE: 968 case FILE_QWDATE: 969 case FILE_LEQWDATE: 970 case FILE_BEQWDATE: 971 case FILE_DOUBLE: 972 case FILE_BEDOUBLE: 973 case FILE_LEDOUBLE: 974 case FILE_BEVARINT: 975 case FILE_LEVARINT: 976 case FILE_GUID: 977 case FILE_LEGUID: 978 case FILE_BEGUID: 979 case FILE_BEID3: 980 case FILE_LEID3: 981 case FILE_OFFSET: 982 case FILE_MSDOSDATE: 983 case FILE_BEMSDOSDATE: 984 case FILE_LEMSDOSDATE: 985 case FILE_MSDOSTIME: 986 case FILE_BEMSDOSTIME: 987 case FILE_LEMSDOSTIME: 988 ts = typesize(m->type); 989 if (ts == FILE_BADSIZE) { 990 (void)fprintf(stderr, "Bad size for type %d\n", 991 m->type); 992 abort(); 993 } 994 val += ts * MULT; 995 break; 996 997 case FILE_PSTRING: 998 case FILE_STRING: 999 case FILE_OCTAL: 1000 val += m->vallen * MULT; 1001 break; 1002 1003 case FILE_BESTRING16: 1004 case FILE_LESTRING16: 1005 val += m->vallen * MULT / 2; 1006 break; 1007 1008 case FILE_SEARCH: 1009 if (m->vallen == 0) 1010 break; 1011 val += m->vallen * MAX(MULT / m->vallen, 1); 1012 break; 1013 1014 case FILE_REGEX: 1015 v = nonmagic(m->value.s); 1016 val += v * MAX(MULT / v, 1); 1017 break; 1018 1019 case FILE_INDIRECT: 1020 case FILE_NAME: 1021 case FILE_USE: 1022 case FILE_CLEAR: 1023 break; 1024 1025 case FILE_DER: 1026 val += MULT; 1027 break; 1028 1029 default: 1030 (void)fprintf(stderr, "Bad type %d\n", m->type); 1031 abort(); 1032 } 1033 1034 switch (m->reln) { 1035 case 'x': /* matches anything penalize */ 1036 case '!': /* matches almost anything penalize */ 1037 val = 0; 1038 break; 1039 1040 case '=': /* Exact match, prefer */ 1041 val += MULT; 1042 break; 1043 1044 case '>': 1045 case '<': /* comparison match reduce strength */ 1046 val -= 2 * MULT; 1047 break; 1048 1049 case '^': 1050 case '&': /* masking bits, we could count them too */ 1051 val -= MULT; 1052 break; 1053 1054 default: 1055 (void)fprintf(stderr, "Bad relation %c\n", m->reln); 1056 abort(); 1057 } 1058 1059 return val; 1060 } 1061 1062 1063 /*ARGSUSED*/ 1064 file_protected size_t 1065 file_magic_strength(const struct magic *m, 1066 size_t nmagic __attribute__((__unused__))) 1067 { 1068 ssize_t val = apprentice_magic_strength_1(m); 1069 1070 #ifdef notyet 1071 if (m->desc[0] == '\0') { 1072 size_t i; 1073 /* 1074 * Magic entries with no description get their continuations 1075 * added 1076 */ 1077 for (i = 1; m[i].cont_level != 0 && i < MIN(nmagic, 3); i++) { 1078 ssize_t v = apprentice_magic_strength_1(&m[i]) >> 1079 (i + 1); 1080 val += v; 1081 if (m[i].desc[0] != '\0') 1082 break; 1083 } 1084 } 1085 #endif 1086 1087 switch (m->factor_op) { 1088 case FILE_FACTOR_OP_NONE: 1089 break; 1090 case FILE_FACTOR_OP_PLUS: 1091 val += m->factor; 1092 break; 1093 case FILE_FACTOR_OP_MINUS: 1094 val -= m->factor; 1095 break; 1096 case FILE_FACTOR_OP_TIMES: 1097 val *= m->factor; 1098 break; 1099 case FILE_FACTOR_OP_DIV: 1100 val /= m->factor; 1101 break; 1102 default: 1103 (void)fprintf(stderr, "Bad factor_op %u\n", m->factor_op); 1104 abort(); 1105 } 1106 1107 if (val <= 0) /* ensure we only return 0 for FILE_DEFAULT */ 1108 val = 1; 1109 1110 #ifndef notyet 1111 /* 1112 * Magic entries with no description get a bonus because they depend 1113 * on subsequent magic entries to print something. 1114 */ 1115 if (m->desc[0] == '\0') 1116 val++; 1117 #endif 1118 1119 return val; 1120 } 1121 1122 /* 1123 * Sort callback for sorting entries by "strength" (basically length) 1124 */ 1125 file_private int 1126 apprentice_sort(const void *a, const void *b) 1127 { 1128 const struct magic_entry *ma = CAST(const struct magic_entry *, a); 1129 const struct magic_entry *mb = CAST(const struct magic_entry *, b); 1130 size_t sa = file_magic_strength(ma->mp, ma->cont_count); 1131 size_t sb = file_magic_strength(mb->mp, mb->cont_count); 1132 if (sa == sb) { 1133 struct magic mpa = *ma->mp; 1134 struct magic mpb = *mb->mp; 1135 mpa.lineno = mpb.lineno = 0; 1136 int x = memcmp(&mpa, &mpb, sizeof(mpa)); 1137 if (x == 0) { 1138 // Don't warn for DER 1139 if (mpa.type == FILE_DER) 1140 return 0; 1141 file_magwarn1("Duplicate magic entry `%s'", 1142 ma->mp->desc); 1143 #ifndef COMPILE_ONLY 1144 file_mdump(ma->mp); 1145 file_mdump(mb->mp); 1146 #endif 1147 return 0; 1148 } 1149 return x > 0 ? -1 : 1; 1150 } 1151 return sa > sb ? -1 : 1; 1152 } 1153 1154 /* 1155 * Shows sorted patterns list in the order which is used for the matching 1156 */ 1157 file_private void 1158 apprentice_list(struct mlist *mlist, int mode) 1159 { 1160 uint32_t magindex, descindex, mimeindex, lineindex; 1161 struct mlist *ml; 1162 for (ml = mlist->next; ml != mlist; ml = ml->next) { 1163 for (magindex = 0; magindex < ml->nmagic; magindex++) { 1164 struct magic *m = &ml->magic[magindex]; 1165 if ((m->flag & mode) != mode) { 1166 /* Skip sub-tests */ 1167 while (magindex + 1 < ml->nmagic && 1168 ml->magic[magindex + 1].cont_level != 0) 1169 ++magindex; 1170 continue; /* Skip to next top-level test*/ 1171 } 1172 1173 /* 1174 * Try to iterate over the tree until we find item with 1175 * description/mimetype. 1176 */ 1177 lineindex = descindex = mimeindex = magindex; 1178 for (; magindex + 1 < ml->nmagic && 1179 ml->magic[magindex + 1].cont_level != 0; 1180 magindex++) { 1181 uint32_t mi = magindex + 1; 1182 if (*ml->magic[descindex].desc == '\0' 1183 && *ml->magic[mi].desc) 1184 descindex = mi; 1185 if (*ml->magic[mimeindex].mimetype == '\0' 1186 && *ml->magic[mi].mimetype) 1187 mimeindex = mi; 1188 } 1189 1190 printf("Strength = %3" SIZE_T_FORMAT "u@%u: %s [%s]\n", 1191 file_magic_strength(m, ml->nmagic - magindex), 1192 ml->magic[lineindex].lineno, 1193 ml->magic[descindex].desc, 1194 ml->magic[mimeindex].mimetype); 1195 } 1196 } 1197 } 1198 1199 file_private void 1200 set_test_type(struct magic *mstart, struct magic *m) 1201 { 1202 switch (m->type) { 1203 case FILE_BYTE: 1204 case FILE_SHORT: 1205 case FILE_LONG: 1206 case FILE_DATE: 1207 case FILE_BESHORT: 1208 case FILE_BELONG: 1209 case FILE_BEDATE: 1210 case FILE_LESHORT: 1211 case FILE_LELONG: 1212 case FILE_LEDATE: 1213 case FILE_LDATE: 1214 case FILE_BELDATE: 1215 case FILE_LELDATE: 1216 case FILE_MEDATE: 1217 case FILE_MELDATE: 1218 case FILE_MELONG: 1219 case FILE_QUAD: 1220 case FILE_LEQUAD: 1221 case FILE_BEQUAD: 1222 case FILE_QDATE: 1223 case FILE_LEQDATE: 1224 case FILE_BEQDATE: 1225 case FILE_QLDATE: 1226 case FILE_LEQLDATE: 1227 case FILE_BEQLDATE: 1228 case FILE_QWDATE: 1229 case FILE_LEQWDATE: 1230 case FILE_BEQWDATE: 1231 case FILE_FLOAT: 1232 case FILE_BEFLOAT: 1233 case FILE_LEFLOAT: 1234 case FILE_DOUBLE: 1235 case FILE_BEDOUBLE: 1236 case FILE_LEDOUBLE: 1237 case FILE_BEVARINT: 1238 case FILE_LEVARINT: 1239 case FILE_DER: 1240 case FILE_GUID: 1241 case FILE_LEGUID: 1242 case FILE_BEGUID: 1243 case FILE_OFFSET: 1244 case FILE_MSDOSDATE: 1245 case FILE_BEMSDOSDATE: 1246 case FILE_LEMSDOSDATE: 1247 case FILE_MSDOSTIME: 1248 case FILE_BEMSDOSTIME: 1249 case FILE_LEMSDOSTIME: 1250 case FILE_OCTAL: 1251 mstart->flag |= BINTEST; 1252 break; 1253 case FILE_STRING: 1254 case FILE_PSTRING: 1255 case FILE_BESTRING16: 1256 case FILE_LESTRING16: 1257 case FILE_REGEX: 1258 case FILE_SEARCH: 1259 /* Check for override */ 1260 if (mstart->str_flags & STRING_BINTEST) 1261 mstart->flag |= BINTEST; 1262 if (mstart->str_flags & STRING_TEXTTEST) 1263 mstart->flag |= TEXTTEST; 1264 1265 if (mstart->flag & (TEXTTEST|BINTEST)) 1266 break; 1267 1268 /* 1269 * XXX: Compatibility: For string types assume binary match 1270 * in reality it is better in the long term to change all 1271 * the magic to be string/b where appropriate 1272 */ 1273 if (m->type != FILE_REGEX && m->type != FILE_SEARCH) { 1274 mstart->flag |= BINTEST; 1275 break; 1276 } 1277 1278 /* binary test if pattern is not text */ 1279 if (file_looks_utf8(m->value.us, CAST(size_t, m->vallen), NULL, 1280 NULL) <= 0) 1281 mstart->flag |= BINTEST; 1282 else 1283 mstart->flag |= TEXTTEST; 1284 break; 1285 case FILE_DEFAULT: 1286 /* can't deduce anything; we shouldn't see this at the 1287 top level anyway */ 1288 break; 1289 case FILE_INVALID: 1290 default: 1291 /* invalid search type, but no need to complain here */ 1292 break; 1293 } 1294 } 1295 1296 file_private int 1297 addentry(struct magic_set *ms, struct magic_entry *me, 1298 struct magic_entry_set *mset) 1299 { 1300 size_t i = me->mp->type == FILE_NAME ? 1 : 0; 1301 if (mset[i].me == NULL || mset[i].count == mset[i].max) { 1302 struct magic_entry *mp; 1303 1304 size_t incr = mset[i].max + ALLOC_INCR; 1305 if ((mp = CAST(struct magic_entry *, 1306 realloc(mset[i].me, sizeof(*mp) * incr))) == 1307 NULL) { 1308 file_oomem(ms, sizeof(*mp) * incr); 1309 return -1; 1310 } 1311 (void)memset(&mp[mset[i].count], 0, sizeof(*mp) * 1312 ALLOC_INCR); 1313 mset[i].me = mp; 1314 mset[i].max = CAST(uint32_t, incr); 1315 assert(mset[i].max == incr); 1316 } 1317 mset[i].me[mset[i].count++] = *me; 1318 memset(me, 0, sizeof(*me)); 1319 return 0; 1320 } 1321 1322 /* 1323 * Load and parse one file. 1324 */ 1325 file_private void 1326 load_1(struct magic_set *ms, int action, const char *fn, int *errs, 1327 struct magic_entry_set *mset) 1328 { 1329 size_t lineno = 0, llen = 0; 1330 char *line = NULL; 1331 ssize_t len; 1332 struct magic_entry me; 1333 1334 FILE *f = fopen(ms->file = fn, "r"); 1335 if (f == NULL) { 1336 if (errno != ENOENT) 1337 file_error(ms, errno, "cannot read magic file `%s'", 1338 fn); 1339 (*errs)++; 1340 return; 1341 } 1342 1343 memset(&me, 0, sizeof(me)); 1344 /* read and parse this file */ 1345 for (ms->line = 1; (len = getline(&line, &llen, f)) != -1; 1346 ms->line++) { 1347 if (ms->magwarn >= ms->magwarn_max) 1348 break; 1349 if (len == 0) /* null line, garbage, etc */ 1350 continue; 1351 if (line[len - 1] == '\n') { 1352 lineno++; 1353 line[len - 1] = '\0'; /* delete newline */ 1354 } 1355 switch (line[0]) { 1356 case '\0': /* empty, do not parse */ 1357 case '#': /* comment, do not parse */ 1358 continue; 1359 case '!': 1360 if (line[1] == ':') { 1361 size_t i; 1362 1363 for (i = 0; bang[i].name != NULL; i++) { 1364 if (CAST(size_t, len - 2) > bang[i].len && 1365 memcmp(bang[i].name, line + 2, 1366 bang[i].len) == 0) 1367 break; 1368 } 1369 if (bang[i].name == NULL) { 1370 file_error(ms, 0, 1371 "Unknown !: entry `%s'", line); 1372 (*errs)++; 1373 continue; 1374 } 1375 if (me.mp == NULL) { 1376 file_error(ms, 0, 1377 "No current entry for :!%s type", 1378 bang[i].name); 1379 (*errs)++; 1380 continue; 1381 } 1382 if ((*bang[i].fun)(ms, &me, 1383 line + bang[i].len + 2, 1384 len - bang[i].len - 2) != 0) { 1385 (*errs)++; 1386 continue; 1387 } 1388 continue; 1389 } 1390 /*FALLTHROUGH*/ 1391 default: 1392 again: 1393 switch (parse(ms, &me, fn, line, llen, lineno, action)) { 1394 case 0: 1395 continue; 1396 case 1: 1397 (void)addentry(ms, &me, mset); 1398 goto again; 1399 default: 1400 (*errs)++; 1401 break; 1402 } 1403 } 1404 } 1405 if (me.mp) 1406 (void)addentry(ms, &me, mset); 1407 free(line); 1408 (void)fclose(f); 1409 } 1410 1411 /* 1412 * parse a file or directory of files 1413 * const char *fn: name of magic file or directory 1414 */ 1415 file_private int 1416 cmpstrp(const void *p1, const void *p2) 1417 { 1418 return strcmp(*RCAST(char *const *, p1), *RCAST(char *const *, p2)); 1419 } 1420 1421 1422 file_private uint32_t 1423 set_text_binary(struct magic_set *ms, struct magic_entry *me, uint32_t nme, 1424 uint32_t starttest) 1425 { 1426 static const char text[] = "text"; 1427 static const char binary[] = "binary"; 1428 static const size_t len = sizeof(text); 1429 1430 uint32_t i = starttest; 1431 1432 do { 1433 set_test_type(me[starttest].mp, me[i].mp); 1434 if ((ms->flags & MAGIC_DEBUG) == 0) 1435 continue; 1436 (void)fprintf(stderr, "%s%s%s: %s\n", 1437 me[i].mp->mimetype, 1438 me[i].mp->mimetype[0] == '\0' ? "" : "; ", 1439 me[i].mp->desc[0] ? me[i].mp->desc : "(no description)", 1440 me[i].mp->flag & BINTEST ? binary : text); 1441 if (me[i].mp->flag & BINTEST) { 1442 char *p = strstr(me[i].mp->desc, text); 1443 if (p && (p == me[i].mp->desc || 1444 isspace(CAST(unsigned char, p[-1]))) && 1445 (p + len - me[i].mp->desc == MAXstring 1446 || (p[len] == '\0' || 1447 isspace(CAST(unsigned char, p[len]))))) 1448 (void)fprintf(stderr, "*** Possible " 1449 "binary test for text type\n"); 1450 } 1451 } while (++i < nme && me[i].mp->cont_level != 0); 1452 return i; 1453 } 1454 1455 file_private void 1456 set_last_default(struct magic_set *ms, struct magic_entry *me, uint32_t nme) 1457 { 1458 uint32_t i; 1459 for (i = 0; i < nme; i++) { 1460 if (me[i].mp->cont_level == 0 && 1461 me[i].mp->type == FILE_DEFAULT) { 1462 while (++i < nme) 1463 if (me[i].mp->cont_level == 0) 1464 break; 1465 if (i != nme) { 1466 /* XXX - Ugh! */ 1467 ms->line = me[i].mp->lineno; 1468 file_magwarn(ms, 1469 "level 0 \"default\" did not sort last"); 1470 } 1471 return; 1472 } 1473 } 1474 } 1475 1476 file_private int 1477 coalesce_entries(struct magic_set *ms, struct magic_entry *me, uint32_t nme, 1478 struct magic **ma, uint32_t *nma) 1479 { 1480 uint32_t i; 1481 size_t mentrycount = 0; 1482 size_t slen; 1483 1484 for (i = 0; i < nme; i++) 1485 mentrycount += me[i].cont_count; 1486 1487 if (mentrycount > UINT32_MAX / sizeof(**ma)) { 1488 file_error(ms, 0, "too many magic entries (%zu)", mentrycount); 1489 return -1; 1490 } 1491 1492 if (mentrycount == 0) { 1493 *ma = NULL; 1494 *nma = 0; 1495 return 0; 1496 } 1497 1498 slen = sizeof(**ma) * mentrycount; 1499 if ((*ma = CAST(struct magic *, malloc(slen))) == NULL) { 1500 file_oomem(ms, slen); 1501 return -1; 1502 } 1503 1504 mentrycount = 0; 1505 for (i = 0; i < nme; i++) { 1506 (void)memcpy(*ma + mentrycount, me[i].mp, 1507 me[i].cont_count * sizeof(**ma)); 1508 mentrycount += me[i].cont_count; 1509 } 1510 *nma = mentrycount; 1511 return 0; 1512 } 1513 1514 file_private void 1515 magic_entry_free(struct magic_entry *me, uint32_t nme) 1516 { 1517 uint32_t i; 1518 if (me == NULL) 1519 return; 1520 for (i = 0; i < nme; i++) 1521 free(me[i].mp); 1522 free(me); 1523 } 1524 1525 file_private struct magic_map * 1526 apprentice_load(struct magic_set *ms, const char *fn, int action) 1527 { 1528 int errs = 0; 1529 uint32_t i, j; 1530 size_t files = 0, maxfiles = 0; 1531 char **filearr = NULL, *mfn; 1532 struct stat st; 1533 struct magic_map *map; 1534 struct magic_entry_set mset[MAGIC_SETS]; 1535 DIR *dir; 1536 struct dirent *d; 1537 1538 memset(mset, 0, sizeof(mset)); 1539 ms->flags |= MAGIC_CHECK; /* Enable checks for parsed files */ 1540 1541 1542 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) 1543 { 1544 file_oomem(ms, sizeof(*map)); 1545 return NULL; 1546 } 1547 map->type = MAP_TYPE_MALLOC; 1548 1549 /* print silly verbose header for USG compat. */ 1550 if (action == FILE_CHECK) 1551 (void)fprintf(stderr, "%s\n", usg_hdr); 1552 1553 /* load directory or file */ 1554 if (stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) { 1555 dir = opendir(fn); 1556 if (!dir) { 1557 errs++; 1558 goto out; 1559 } 1560 while ((d = readdir(dir)) != NULL) { 1561 if (d->d_name[0] == '.') 1562 continue; 1563 if (asprintf(&mfn, "%s/%s", fn, d->d_name) < 0) { 1564 file_oomem(ms, 1565 strlen(fn) + strlen(d->d_name) + 2); 1566 errs++; 1567 closedir(dir); 1568 goto out; 1569 } 1570 if (stat(mfn, &st) == -1 || !S_ISREG(st.st_mode)) { 1571 free(mfn); 1572 continue; 1573 } 1574 if (files >= maxfiles) { 1575 size_t mlen; 1576 char **nfilearr; 1577 maxfiles = (maxfiles + 1) * 2; 1578 mlen = maxfiles * sizeof(*filearr); 1579 if ((nfilearr = CAST(char **, 1580 realloc(filearr, mlen))) == NULL) { 1581 file_oomem(ms, mlen); 1582 free(mfn); 1583 closedir(dir); 1584 errs++; 1585 goto out; 1586 } 1587 filearr = nfilearr; 1588 } 1589 filearr[files++] = mfn; 1590 } 1591 closedir(dir); 1592 if (filearr) { 1593 qsort(filearr, files, sizeof(*filearr), cmpstrp); 1594 for (i = 0; i < files; i++) { 1595 load_1(ms, action, filearr[i], &errs, mset); 1596 free(filearr[i]); 1597 } 1598 free(filearr); 1599 filearr = NULL; 1600 } 1601 } else 1602 load_1(ms, action, fn, &errs, mset); 1603 if (errs) 1604 goto out; 1605 1606 for (j = 0; j < MAGIC_SETS; j++) { 1607 /* Set types of tests */ 1608 for (i = 0; i < mset[j].count; ) { 1609 if (mset[j].me[i].mp->cont_level != 0) { 1610 i++; 1611 continue; 1612 } 1613 i = set_text_binary(ms, mset[j].me, mset[j].count, i); 1614 } 1615 if (mset[j].me) 1616 qsort(mset[j].me, mset[j].count, sizeof(*mset[0].me), 1617 apprentice_sort); 1618 1619 /* 1620 * Make sure that any level 0 "default" line is last 1621 * (if one exists). 1622 */ 1623 set_last_default(ms, mset[j].me, mset[j].count); 1624 1625 /* coalesce per file arrays into a single one, if needed */ 1626 if (mset[j].count == 0) 1627 continue; 1628 1629 if (coalesce_entries(ms, mset[j].me, mset[j].count, 1630 &map->magic[j], &map->nmagic[j]) == -1) { 1631 errs++; 1632 goto out; 1633 } 1634 } 1635 1636 out: 1637 free(filearr); 1638 for (j = 0; j < MAGIC_SETS; j++) 1639 magic_entry_free(mset[j].me, mset[j].count); 1640 1641 if (errs) { 1642 apprentice_unmap(map); 1643 return NULL; 1644 } 1645 return map; 1646 } 1647 1648 /* 1649 * extend the sign bit if the comparison is to be signed 1650 */ 1651 file_protected uint64_t 1652 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v) 1653 { 1654 if (!(m->flag & UNSIGNED)) { 1655 switch(m->type) { 1656 /* 1657 * Do not remove the casts below. They are 1658 * vital. When later compared with the data, 1659 * the sign extension must have happened. 1660 */ 1661 case FILE_BYTE: 1662 v = CAST(signed char, v); 1663 break; 1664 case FILE_SHORT: 1665 case FILE_BESHORT: 1666 case FILE_LESHORT: 1667 v = CAST(short, v); 1668 break; 1669 case FILE_DATE: 1670 case FILE_BEDATE: 1671 case FILE_LEDATE: 1672 case FILE_MEDATE: 1673 case FILE_LDATE: 1674 case FILE_BELDATE: 1675 case FILE_LELDATE: 1676 case FILE_MELDATE: 1677 case FILE_LONG: 1678 case FILE_BELONG: 1679 case FILE_LELONG: 1680 case FILE_MELONG: 1681 case FILE_FLOAT: 1682 case FILE_BEFLOAT: 1683 case FILE_LEFLOAT: 1684 case FILE_MSDOSDATE: 1685 case FILE_BEMSDOSDATE: 1686 case FILE_LEMSDOSDATE: 1687 case FILE_MSDOSTIME: 1688 case FILE_BEMSDOSTIME: 1689 case FILE_LEMSDOSTIME: 1690 v = CAST(int32_t, v); 1691 break; 1692 case FILE_QUAD: 1693 case FILE_BEQUAD: 1694 case FILE_LEQUAD: 1695 case FILE_QDATE: 1696 case FILE_QLDATE: 1697 case FILE_QWDATE: 1698 case FILE_BEQDATE: 1699 case FILE_BEQLDATE: 1700 case FILE_BEQWDATE: 1701 case FILE_LEQDATE: 1702 case FILE_LEQLDATE: 1703 case FILE_LEQWDATE: 1704 case FILE_DOUBLE: 1705 case FILE_BEDOUBLE: 1706 case FILE_LEDOUBLE: 1707 case FILE_OFFSET: 1708 case FILE_BEVARINT: 1709 case FILE_LEVARINT: 1710 v = CAST(int64_t, v); 1711 break; 1712 case FILE_STRING: 1713 case FILE_PSTRING: 1714 case FILE_BESTRING16: 1715 case FILE_LESTRING16: 1716 case FILE_REGEX: 1717 case FILE_SEARCH: 1718 case FILE_DEFAULT: 1719 case FILE_INDIRECT: 1720 case FILE_NAME: 1721 case FILE_USE: 1722 case FILE_CLEAR: 1723 case FILE_DER: 1724 case FILE_GUID: 1725 case FILE_LEGUID: 1726 case FILE_BEGUID: 1727 case FILE_OCTAL: 1728 break; 1729 default: 1730 if (ms->flags & MAGIC_CHECK) 1731 file_magwarn(ms, "cannot happen: m->type=%d\n", 1732 m->type); 1733 return FILE_BADSIZE; 1734 } 1735 } 1736 return v; 1737 } 1738 1739 file_private int 1740 string_modifier_check(struct magic_set *ms, struct magic *m) 1741 { 1742 if ((ms->flags & MAGIC_CHECK) == 0) 1743 return 0; 1744 1745 if ((m->type != FILE_REGEX || (m->str_flags & REGEX_LINE_COUNT) == 0) && 1746 (m->type != FILE_PSTRING && (m->str_flags & PSTRING_LEN) != 0)) { 1747 file_magwarn(ms, 1748 "'/BHhLl' modifiers are only allowed for pascal strings\n"); 1749 return -1; 1750 } 1751 switch (m->type) { 1752 case FILE_BESTRING16: 1753 case FILE_LESTRING16: 1754 if (m->str_flags != 0) { 1755 file_magwarn(ms, 1756 "no modifiers allowed for 16-bit strings\n"); 1757 return -1; 1758 } 1759 break; 1760 case FILE_STRING: 1761 case FILE_PSTRING: 1762 if ((m->str_flags & REGEX_OFFSET_START) != 0) { 1763 file_magwarn(ms, 1764 "'/%c' only allowed on regex and search\n", 1765 CHAR_REGEX_OFFSET_START); 1766 return -1; 1767 } 1768 break; 1769 case FILE_SEARCH: 1770 if (m->str_range == 0) { 1771 file_magwarn(ms, 1772 "missing range; defaulting to %d\n", 1773 STRING_DEFAULT_RANGE); 1774 m->str_range = STRING_DEFAULT_RANGE; 1775 return -1; 1776 } 1777 break; 1778 case FILE_REGEX: 1779 if ((m->str_flags & STRING_COMPACT_WHITESPACE) != 0) { 1780 file_magwarn(ms, "'/%c' not allowed on regex\n", 1781 CHAR_COMPACT_WHITESPACE); 1782 return -1; 1783 } 1784 if ((m->str_flags & STRING_COMPACT_OPTIONAL_WHITESPACE) != 0) { 1785 file_magwarn(ms, "'/%c' not allowed on regex\n", 1786 CHAR_COMPACT_OPTIONAL_WHITESPACE); 1787 return -1; 1788 } 1789 break; 1790 default: 1791 file_magwarn(ms, "coding error: m->type=%d\n", 1792 m->type); 1793 return -1; 1794 } 1795 return 0; 1796 } 1797 1798 file_private int 1799 get_op(char c) 1800 { 1801 switch (c) { 1802 case '&': 1803 return FILE_OPAND; 1804 case '|': 1805 return FILE_OPOR; 1806 case '^': 1807 return FILE_OPXOR; 1808 case '+': 1809 return FILE_OPADD; 1810 case '-': 1811 return FILE_OPMINUS; 1812 case '*': 1813 return FILE_OPMULTIPLY; 1814 case '/': 1815 return FILE_OPDIVIDE; 1816 case '%': 1817 return FILE_OPMODULO; 1818 default: 1819 return -1; 1820 } 1821 } 1822 1823 #ifdef ENABLE_CONDITIONALS 1824 file_private int 1825 get_cond(const char *l, const char **t) 1826 { 1827 static const struct cond_tbl_s { 1828 char name[8]; 1829 size_t len; 1830 int cond; 1831 } cond_tbl[] = { 1832 { "if", 2, COND_IF }, 1833 { "elif", 4, COND_ELIF }, 1834 { "else", 4, COND_ELSE }, 1835 { "", 0, COND_NONE }, 1836 }; 1837 const struct cond_tbl_s *p; 1838 1839 for (p = cond_tbl; p->len; p++) { 1840 if (strncmp(l, p->name, p->len) == 0 && 1841 isspace(CAST(unsigned char, l[p->len]))) { 1842 if (t) 1843 *t = l + p->len; 1844 break; 1845 } 1846 } 1847 return p->cond; 1848 } 1849 1850 file_private int 1851 check_cond(struct magic_set *ms, int cond, uint32_t cont_level) 1852 { 1853 int last_cond; 1854 last_cond = ms->c.li[cont_level].last_cond; 1855 1856 switch (cond) { 1857 case COND_IF: 1858 if (last_cond != COND_NONE && last_cond != COND_ELIF) { 1859 if (ms->flags & MAGIC_CHECK) 1860 file_magwarn(ms, "syntax error: `if'"); 1861 return -1; 1862 } 1863 last_cond = COND_IF; 1864 break; 1865 1866 case COND_ELIF: 1867 if (last_cond != COND_IF && last_cond != COND_ELIF) { 1868 if (ms->flags & MAGIC_CHECK) 1869 file_magwarn(ms, "syntax error: `elif'"); 1870 return -1; 1871 } 1872 last_cond = COND_ELIF; 1873 break; 1874 1875 case COND_ELSE: 1876 if (last_cond != COND_IF && last_cond != COND_ELIF) { 1877 if (ms->flags & MAGIC_CHECK) 1878 file_magwarn(ms, "syntax error: `else'"); 1879 return -1; 1880 } 1881 last_cond = COND_NONE; 1882 break; 1883 1884 case COND_NONE: 1885 last_cond = COND_NONE; 1886 break; 1887 } 1888 1889 ms->c.li[cont_level].last_cond = last_cond; 1890 return 0; 1891 } 1892 #endif /* ENABLE_CONDITIONALS */ 1893 1894 file_private int 1895 parse_indirect_modifier(struct magic_set *ms, struct magic *m, const char **lp) 1896 { 1897 const char *l = *lp; 1898 1899 while (!isspace(CAST(unsigned char, *++l))) 1900 switch (*l) { 1901 case CHAR_INDIRECT_RELATIVE: 1902 m->str_flags |= INDIRECT_RELATIVE; 1903 break; 1904 default: 1905 if (ms->flags & MAGIC_CHECK) 1906 file_magwarn(ms, "indirect modifier `%c' " 1907 "invalid", *l); 1908 *lp = l; 1909 return -1; 1910 } 1911 *lp = l; 1912 return 0; 1913 } 1914 1915 file_private void 1916 parse_op_modifier(struct magic_set *ms, struct magic *m, const char **lp, 1917 int op) 1918 { 1919 const char *l = *lp; 1920 char *t; 1921 uint64_t val; 1922 1923 ++l; 1924 m->mask_op |= op; 1925 val = CAST(uint64_t, strtoull(l, &t, 0)); 1926 l = t; 1927 m->num_mask = file_signextend(ms, m, val); 1928 eatsize(&l); 1929 *lp = l; 1930 } 1931 1932 file_private int 1933 parse_string_modifier(struct magic_set *ms, struct magic *m, const char **lp) 1934 { 1935 const char *l = *lp; 1936 char *t; 1937 int have_range = 0; 1938 1939 while (!isspace(CAST(unsigned char, *++l))) { 1940 switch (*l) { 1941 case '0': case '1': case '2': 1942 case '3': case '4': case '5': 1943 case '6': case '7': case '8': 1944 case '9': 1945 if (have_range && (ms->flags & MAGIC_CHECK)) 1946 file_magwarn(ms, "multiple ranges"); 1947 have_range = 1; 1948 m->str_range = CAST(uint32_t, strtoul(l, &t, 0)); 1949 if (m->str_range == 0) 1950 file_magwarn(ms, "zero range"); 1951 l = t - 1; 1952 break; 1953 case CHAR_COMPACT_WHITESPACE: 1954 m->str_flags |= STRING_COMPACT_WHITESPACE; 1955 break; 1956 case CHAR_COMPACT_OPTIONAL_WHITESPACE: 1957 m->str_flags |= STRING_COMPACT_OPTIONAL_WHITESPACE; 1958 break; 1959 case CHAR_IGNORE_LOWERCASE: 1960 m->str_flags |= STRING_IGNORE_LOWERCASE; 1961 break; 1962 case CHAR_IGNORE_UPPERCASE: 1963 m->str_flags |= STRING_IGNORE_UPPERCASE; 1964 break; 1965 case CHAR_REGEX_OFFSET_START: 1966 m->str_flags |= REGEX_OFFSET_START; 1967 break; 1968 case CHAR_BINTEST: 1969 m->str_flags |= STRING_BINTEST; 1970 break; 1971 case CHAR_TEXTTEST: 1972 m->str_flags |= STRING_TEXTTEST; 1973 break; 1974 case CHAR_TRIM: 1975 m->str_flags |= STRING_TRIM; 1976 break; 1977 case CHAR_FULL_WORD: 1978 m->str_flags |= STRING_FULL_WORD; 1979 break; 1980 case CHAR_PSTRING_1_LE: 1981 #define SET_LENGTH(a) m->str_flags = (m->str_flags & ~PSTRING_LEN) | (a) 1982 if (m->type != FILE_PSTRING) 1983 goto bad; 1984 SET_LENGTH(PSTRING_1_LE); 1985 break; 1986 case CHAR_PSTRING_2_BE: 1987 if (m->type != FILE_PSTRING) 1988 goto bad; 1989 SET_LENGTH(PSTRING_2_BE); 1990 break; 1991 case CHAR_PSTRING_2_LE: 1992 if (m->type != FILE_PSTRING) 1993 goto bad; 1994 SET_LENGTH(PSTRING_2_LE); 1995 break; 1996 case CHAR_PSTRING_4_BE: 1997 if (m->type != FILE_PSTRING) 1998 goto bad; 1999 SET_LENGTH(PSTRING_4_BE); 2000 break; 2001 case CHAR_PSTRING_4_LE: 2002 switch (m->type) { 2003 case FILE_PSTRING: 2004 case FILE_REGEX: 2005 break; 2006 default: 2007 goto bad; 2008 } 2009 SET_LENGTH(PSTRING_4_LE); 2010 break; 2011 case CHAR_PSTRING_LENGTH_INCLUDES_ITSELF: 2012 if (m->type != FILE_PSTRING) 2013 goto bad; 2014 m->str_flags |= PSTRING_LENGTH_INCLUDES_ITSELF; 2015 break; 2016 default: 2017 bad: 2018 if (ms->flags & MAGIC_CHECK) 2019 file_magwarn(ms, "string modifier `%c' " 2020 "invalid", *l); 2021 goto out; 2022 } 2023 /* allow multiple '/' for readability */ 2024 if (l[1] == '/' && !isspace(CAST(unsigned char, l[2]))) 2025 l++; 2026 } 2027 if (string_modifier_check(ms, m) == -1) 2028 goto out; 2029 *lp = l; 2030 return 0; 2031 out: 2032 *lp = l; 2033 return -1; 2034 } 2035 2036 /* 2037 * parse one line from magic file, put into magic[index++] if valid 2038 */ 2039 file_private int 2040 parse(struct magic_set *ms, struct magic_entry *me, const char *file, 2041 const char *line, size_t llen, size_t lineno, int action) 2042 { 2043 #ifdef ENABLE_CONDITIONALS 2044 static uint32_t last_cont_level = 0; 2045 #endif 2046 size_t i; 2047 struct magic *m; 2048 const char *l = line, *el = line + llen; 2049 char *t; 2050 int op; 2051 uint32_t cont_level; 2052 int32_t diff; 2053 2054 cont_level = 0; 2055 2056 /* 2057 * Parse the offset. 2058 */ 2059 while (*l == '>') { 2060 if (l >= el) 2061 return -1; 2062 ++l; /* step over */ 2063 cont_level++; 2064 } 2065 #ifdef ENABLE_CONDITIONALS 2066 if (cont_level == 0 || cont_level > last_cont_level) 2067 if (file_check_mem(ms, cont_level) == -1) 2068 return -1; 2069 last_cont_level = cont_level; 2070 #endif 2071 if (cont_level != 0) { 2072 if (me->mp == NULL) { 2073 file_magerror(ms, "No current entry for continuation"); 2074 return -1; 2075 } 2076 if (me->cont_count == 0) { 2077 file_magerror(ms, "Continuations present with 0 count"); 2078 return -1; 2079 } 2080 m = &me->mp[me->cont_count - 1]; 2081 diff = CAST(int32_t, cont_level) - CAST(int32_t, m->cont_level); 2082 if (diff > 1) 2083 file_magwarn(ms, "New continuation level %u is more " 2084 "than one larger than current level %u", cont_level, 2085 m->cont_level); 2086 if (me->cont_count == me->max_count) { 2087 struct magic *nm; 2088 size_t cnt = me->max_count + ALLOC_CHUNK; 2089 if ((nm = CAST(struct magic *, realloc(me->mp, 2090 sizeof(*nm) * cnt))) == NULL) { 2091 file_oomem(ms, sizeof(*nm) * cnt); 2092 return -1; 2093 } 2094 me->mp = nm; 2095 me->max_count = CAST(uint32_t, cnt); 2096 } 2097 m = &me->mp[me->cont_count++]; 2098 (void)memset(m, 0, sizeof(*m)); 2099 m->cont_level = cont_level; 2100 } else { 2101 static const size_t len = sizeof(*m) * ALLOC_CHUNK; 2102 if (me->mp != NULL) 2103 return 1; 2104 if ((m = CAST(struct magic *, malloc(len))) == NULL) { 2105 file_oomem(ms, len); 2106 return -1; 2107 } 2108 me->mp = m; 2109 me->max_count = ALLOC_CHUNK; 2110 (void)memset(m, 0, sizeof(*m)); 2111 m->factor_op = FILE_FACTOR_OP_NONE; 2112 m->cont_level = 0; 2113 me->cont_count = 1; 2114 } 2115 m->lineno = CAST(uint32_t, lineno); 2116 2117 if (*l == '&') { /* m->cont_level == 0 checked below. */ 2118 ++l; /* step over */ 2119 m->flag |= OFFADD; 2120 } 2121 if (*l == '(') { 2122 ++l; /* step over */ 2123 m->flag |= INDIR; 2124 if (m->flag & OFFADD) 2125 m->flag = (m->flag & ~OFFADD) | INDIROFFADD; 2126 2127 if (*l == '&') { /* m->cont_level == 0 checked below */ 2128 ++l; /* step over */ 2129 m->flag |= OFFADD; 2130 } 2131 } 2132 /* Indirect offsets are not valid at level 0. */ 2133 if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD))) { 2134 if (ms->flags & MAGIC_CHECK) 2135 file_magwarn(ms, "relative offset at level 0"); 2136 return -1; 2137 } 2138 2139 /* get offset, then skip over it */ 2140 if (*l == '-' || *l == '+') { 2141 ++l; /* step over */ 2142 m->flag |= l[-1] == '-' ? OFFNEGATIVE : OFFPOSITIVE; 2143 } 2144 m->offset = CAST(int32_t, strtol(l, &t, 0)); 2145 if (l == t) { 2146 if (ms->flags & MAGIC_CHECK) 2147 file_magwarn(ms, "offset `%s' invalid", l); 2148 return -1; 2149 } 2150 2151 l = t; 2152 2153 if (m->flag & INDIR) { 2154 m->in_type = FILE_LONG; 2155 m->in_offset = 0; 2156 m->in_op = 0; 2157 /* 2158 * read [.,lbs][+-]nnnnn) 2159 */ 2160 if (*l == '.' || *l == ',') { 2161 if (*l == ',') 2162 m->in_op |= FILE_OPSIGNED; 2163 l++; 2164 switch (*l) { 2165 case 'l': 2166 m->in_type = FILE_LELONG; 2167 break; 2168 case 'L': 2169 m->in_type = FILE_BELONG; 2170 break; 2171 case 'm': 2172 m->in_type = FILE_MELONG; 2173 break; 2174 case 'h': 2175 case 's': 2176 m->in_type = FILE_LESHORT; 2177 break; 2178 case 'H': 2179 case 'S': 2180 m->in_type = FILE_BESHORT; 2181 break; 2182 case 'c': 2183 case 'b': 2184 case 'C': 2185 case 'B': 2186 m->in_type = FILE_BYTE; 2187 break; 2188 case 'e': 2189 case 'f': 2190 case 'g': 2191 m->in_type = FILE_LEDOUBLE; 2192 break; 2193 case 'E': 2194 case 'F': 2195 case 'G': 2196 m->in_type = FILE_BEDOUBLE; 2197 break; 2198 case 'i': 2199 m->in_type = FILE_LEID3; 2200 break; 2201 case 'I': 2202 m->in_type = FILE_BEID3; 2203 break; 2204 case 'o': 2205 m->in_type = FILE_OCTAL; 2206 break; 2207 case 'q': 2208 m->in_type = FILE_LEQUAD; 2209 break; 2210 case 'Q': 2211 m->in_type = FILE_BEQUAD; 2212 break; 2213 default: 2214 if (ms->flags & MAGIC_CHECK) 2215 file_magwarn(ms, 2216 "indirect offset type `%c' invalid", 2217 *l); 2218 return -1; 2219 } 2220 l++; 2221 } 2222 2223 if (*l == '~') { 2224 m->in_op |= FILE_OPINVERSE; 2225 l++; 2226 } 2227 if ((op = get_op(*l)) != -1) { 2228 m->in_op |= op; 2229 l++; 2230 } 2231 if (*l == '(') { 2232 m->in_op |= FILE_OPINDIRECT; 2233 l++; 2234 } 2235 if (isdigit(CAST(unsigned char, *l)) || *l == '-') { 2236 m->in_offset = CAST(int32_t, strtol(l, &t, 0)); 2237 if (l == t) { 2238 if (ms->flags & MAGIC_CHECK) 2239 file_magwarn(ms, 2240 "in_offset `%s' invalid", l); 2241 return -1; 2242 } 2243 l = t; 2244 } 2245 if (*l++ != ')' || 2246 ((m->in_op & FILE_OPINDIRECT) && *l++ != ')')) { 2247 if (ms->flags & MAGIC_CHECK) 2248 file_magwarn(ms, 2249 "missing ')' in indirect offset"); 2250 return -1; 2251 } 2252 } 2253 EATAB; 2254 2255 #ifdef ENABLE_CONDITIONALS 2256 m->cond = get_cond(l, &l); 2257 if (check_cond(ms, m->cond, cont_level) == -1) 2258 return -1; 2259 2260 EATAB; 2261 #endif 2262 2263 /* 2264 * Parse the type. 2265 */ 2266 if (*l == 'u') { 2267 /* 2268 * Try it as a keyword type prefixed by "u"; match what 2269 * follows the "u". If that fails, try it as an SUS 2270 * integer type. 2271 */ 2272 m->type = get_type(type_tbl, l + 1, &l); 2273 if (m->type == FILE_INVALID) { 2274 /* 2275 * Not a keyword type; parse it as an SUS type, 2276 * 'u' possibly followed by a number or C/S/L. 2277 */ 2278 m->type = get_standard_integer_type(l, &l); 2279 } 2280 /* It's unsigned. */ 2281 if (m->type != FILE_INVALID) 2282 m->flag |= UNSIGNED; 2283 } else { 2284 /* 2285 * Try it as a keyword type. If that fails, try it as 2286 * an SUS integer type if it begins with "d" or as an 2287 * SUS string type if it begins with "s". In any case, 2288 * it's not unsigned. 2289 */ 2290 m->type = get_type(type_tbl, l, &l); 2291 if (m->type == FILE_INVALID) { 2292 /* 2293 * Not a keyword type; parse it as an SUS type, 2294 * either 'd' possibly followed by a number or 2295 * C/S/L, or just 's'. 2296 */ 2297 if (*l == 'd') 2298 m->type = get_standard_integer_type(l, &l); 2299 else if (*l == 's' 2300 && !isalpha(CAST(unsigned char, l[1]))) { 2301 m->type = FILE_STRING; 2302 ++l; 2303 } 2304 } 2305 } 2306 2307 if (m->type == FILE_INVALID) { 2308 /* Not found - try it as a special keyword. */ 2309 m->type = get_type(special_tbl, l, &l); 2310 } 2311 2312 if (m->type == FILE_INVALID) { 2313 if (ms->flags & MAGIC_CHECK) 2314 file_magwarn(ms, "type `%s' invalid", l); 2315 return -1; 2316 } 2317 2318 if (m->type == FILE_NAME && cont_level != 0) { 2319 if (ms->flags & MAGIC_CHECK) 2320 file_magwarn(ms, "`name%s' entries can only be " 2321 "declared at top level", l); 2322 return -1; 2323 } 2324 2325 /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */ 2326 /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */ 2327 2328 m->mask_op = 0; 2329 if (*l == '~') { 2330 if (!IS_STRING(m->type)) 2331 m->mask_op |= FILE_OPINVERSE; 2332 else if (ms->flags & MAGIC_CHECK) 2333 file_magwarn(ms, "'~' invalid for string types"); 2334 ++l; 2335 } 2336 m->str_range = 0; 2337 m->str_flags = m->type == FILE_PSTRING ? PSTRING_1_LE : 0; 2338 if ((op = get_op(*l)) != -1) { 2339 if (IS_STRING(m->type)) { 2340 int r; 2341 2342 if (op != FILE_OPDIVIDE) { 2343 if (ms->flags & MAGIC_CHECK) 2344 file_magwarn(ms, 2345 "invalid string/indirect op: " 2346 "`%c'", *t); 2347 return -1; 2348 } 2349 2350 if (m->type == FILE_INDIRECT) 2351 r = parse_indirect_modifier(ms, m, &l); 2352 else 2353 r = parse_string_modifier(ms, m, &l); 2354 if (r == -1) 2355 return -1; 2356 } else 2357 parse_op_modifier(ms, m, &l, op); 2358 } 2359 2360 /* 2361 * We used to set mask to all 1's here, instead let's just not do 2362 * anything if mask = 0 (unless you have a better idea) 2363 */ 2364 EATAB; 2365 2366 switch (*l) { 2367 case '>': 2368 case '<': 2369 m->reln = *l; 2370 ++l; 2371 if (*l == '=') { 2372 if (ms->flags & MAGIC_CHECK) { 2373 file_magwarn(ms, "%c= not supported", 2374 m->reln); 2375 return -1; 2376 } 2377 ++l; 2378 } 2379 break; 2380 /* Old-style anding: "0 byte &0x80 dynamically linked" */ 2381 case '&': 2382 case '^': 2383 case '=': 2384 m->reln = *l; 2385 ++l; 2386 if (*l == '=') { 2387 /* HP compat: ignore &= etc. */ 2388 ++l; 2389 } 2390 break; 2391 case '!': 2392 m->reln = *l; 2393 ++l; 2394 break; 2395 default: 2396 m->reln = '='; /* the default relation */ 2397 if (*l == 'x' && ((isascii(CAST(unsigned char, l[1])) && 2398 isspace(CAST(unsigned char, l[1]))) || !l[1])) { 2399 m->reln = *l; 2400 ++l; 2401 } 2402 break; 2403 } 2404 2405 2406 /* 2407 * Grab the value part, except for an 'x' reln. 2408 */ 2409 if (m->reln != 'x') { 2410 if (*l == '\0') { 2411 file_magwarn(ms, "incomplete magic `%s'", line); 2412 return -1; 2413 } 2414 if (getvalue(ms, m, &l, action)) 2415 return -1; 2416 } 2417 2418 /* 2419 * TODO finish this macro and start using it! 2420 * #define offsetcheck {if (offset > ms->bytes_max -1) 2421 * magwarn("offset too big"); } 2422 */ 2423 2424 /* 2425 * Now get last part - the description 2426 */ 2427 EATAB; 2428 if (l[0] == '\b') { 2429 ++l; 2430 m->flag |= NOSPACE; 2431 } else if ((l[0] == '\\') && (l[1] == 'b')) { 2432 ++l; 2433 ++l; 2434 m->flag |= NOSPACE; 2435 } 2436 for (i = 0; i < sizeof(m->desc) && (m->desc[i++] = *l++) != '\0';) 2437 continue; 2438 if (m->desc[0] == '\0') { 2439 // Tuck in the filename for debugging. 2440 strlcpy(m->desc + 1, file, sizeof(m->desc) - 1); 2441 } 2442 if (i == sizeof(m->desc)) { 2443 m->desc[sizeof(m->desc) - 1] = '\0'; 2444 if (ms->flags & MAGIC_CHECK) 2445 file_magwarn(ms, "description `%s' truncated", m->desc); 2446 } 2447 2448 /* 2449 * We only do this check while compiling, or if any of the magic 2450 * files were not compiled. 2451 */ 2452 if (ms->flags & MAGIC_CHECK) { 2453 if (check_format(ms, m) == -1) 2454 return -1; 2455 } 2456 #ifndef COMPILE_ONLY 2457 if (action == FILE_CHECK) { 2458 file_mdump(m); 2459 } 2460 #endif 2461 m->mimetype[0] = '\0'; /* initialise MIME type to none */ 2462 return 0; 2463 } 2464 2465 /* 2466 * parse a STRENGTH annotation line from magic file, put into magic[index - 1] 2467 * if valid 2468 */ 2469 /*ARGSUSED*/ 2470 file_private int 2471 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line, 2472 size_t len __attribute__((__unused__))) 2473 { 2474 const char *l = line; 2475 char *el; 2476 unsigned long factor; 2477 char sbuf[512]; 2478 struct magic *m = &me->mp[0]; 2479 2480 if (m->factor_op != FILE_FACTOR_OP_NONE) { 2481 file_magwarn(ms, 2482 "Current entry already has a strength type: %c %d", 2483 m->factor_op, m->factor); 2484 return -1; 2485 } 2486 if (m->type == FILE_NAME) { 2487 file_magwarn(ms, "%s: Strength setting is not supported in " 2488 "\"name\" magic entries", 2489 file_printable(ms, sbuf, sizeof(sbuf), m->value.s, 2490 sizeof(m->value.s))); 2491 return -1; 2492 } 2493 EATAB; 2494 switch (*l) { 2495 case FILE_FACTOR_OP_NONE: 2496 break; 2497 case FILE_FACTOR_OP_PLUS: 2498 case FILE_FACTOR_OP_MINUS: 2499 case FILE_FACTOR_OP_TIMES: 2500 case FILE_FACTOR_OP_DIV: 2501 m->factor_op = *l++; 2502 break; 2503 default: 2504 file_magwarn(ms, "Unknown factor op `%c'", *l); 2505 return -1; 2506 } 2507 EATAB; 2508 factor = strtoul(l, &el, 0); 2509 if (factor > 255) { 2510 file_magwarn(ms, "Too large factor `%lu'", factor); 2511 goto out; 2512 } 2513 if (*el && !isspace(CAST(unsigned char, *el))) { 2514 file_magwarn(ms, "Bad factor `%s'", l); 2515 goto out; 2516 } 2517 m->factor = CAST(uint8_t, factor); 2518 if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) { 2519 file_magwarn(ms, "Cannot have factor op `%c' and factor %u", 2520 m->factor_op, m->factor); 2521 goto out; 2522 } 2523 return 0; 2524 out: 2525 m->factor_op = FILE_FACTOR_OP_NONE; 2526 m->factor = 0; 2527 return -1; 2528 } 2529 2530 file_private int 2531 goodchar(unsigned char x, const char *extra) 2532 { 2533 return (isascii(x) && isalnum(x)) || strchr(extra, x); 2534 } 2535 2536 file_private int 2537 parse_extra(struct magic_set *ms, struct magic_entry *me, const char *line, 2538 size_t llen, off_t off, size_t len, const char *name, const char *extra, 2539 int nt) 2540 { 2541 size_t i; 2542 const char *l = line; 2543 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1]; 2544 char *buf = CAST(char *, CAST(void *, m)) + off; 2545 2546 if (buf[0] != '\0') { 2547 len = nt ? strlen(buf) : len; 2548 file_magwarn(ms, "Current entry already has a %s type " 2549 "`%.*s', new type `%s'", name, CAST(int, len), buf, l); 2550 return -1; 2551 } 2552 2553 if (*m->desc == '\0') { 2554 file_magwarn(ms, "Current entry does not yet have a " 2555 "description for adding a %s type", name); 2556 return -1; 2557 } 2558 2559 EATAB; 2560 for (i = 0; *l && i < llen && i < len && goodchar(*l, extra); 2561 buf[i++] = *l++) 2562 continue; 2563 2564 if (i == len && *l) { 2565 if (nt) 2566 buf[len - 1] = '\0'; 2567 if (ms->flags & MAGIC_CHECK) 2568 file_magwarn(ms, "%s type `%s' truncated %" 2569 SIZE_T_FORMAT "u", name, line, i); 2570 } else { 2571 if (!isspace(CAST(unsigned char, *l)) && !goodchar(*l, extra)) 2572 file_magwarn(ms, "%s type `%s' has bad char '%c'", 2573 name, line, *l); 2574 if (nt) 2575 buf[i] = '\0'; 2576 } 2577 2578 if (i > 0) 2579 return 0; 2580 2581 file_magerror(ms, "Bad magic entry '%s'", line); 2582 return -1; 2583 } 2584 2585 /* 2586 * Parse an Apple CREATOR/TYPE annotation from magic file and put it into 2587 * magic[index - 1] 2588 */ 2589 file_private int 2590 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line, 2591 size_t len) 2592 { 2593 return parse_extra(ms, me, line, len, 2594 CAST(off_t, offsetof(struct magic, apple)), 2595 sizeof(me->mp[me->cont_count - 1].apple), "APPLE", "!+-./?", 0); 2596 } 2597 2598 /* 2599 * Parse a comma-separated list of extensions 2600 */ 2601 file_private int 2602 parse_ext(struct magic_set *ms, struct magic_entry *me, const char *line, 2603 size_t len) 2604 { 2605 return parse_extra(ms, me, line, len, 2606 CAST(off_t, offsetof(struct magic, ext)), 2607 sizeof(me->mp[me->cont_count - 1].ext), "EXTENSION", ",!+-/@?_$&~.", 2608 0); 2609 /* & for b&w */ 2610 /* ~ for journal~ */ 2611 } 2612 2613 /* 2614 * parse a MIME annotation line from magic file, put into magic[index - 1] 2615 * if valid 2616 */ 2617 file_private int 2618 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line, 2619 size_t len) 2620 { 2621 return parse_extra(ms, me, line, len, 2622 CAST(off_t, offsetof(struct magic, mimetype)), 2623 sizeof(me->mp[me->cont_count - 1].mimetype), "MIME", "+-/.$?:{};=", 2624 1); 2625 } 2626 2627 file_private int 2628 check_format_type(const char *ptr, int type, const char **estr) 2629 { 2630 int quad = 0, h; 2631 size_t len, cnt; 2632 if (*ptr == '\0') { 2633 /* Missing format string; bad */ 2634 *estr = "missing format spec"; 2635 return -1; 2636 } 2637 2638 switch (file_formats[type]) { 2639 case FILE_FMT_QUAD: 2640 quad = 1; 2641 /*FALLTHROUGH*/ 2642 case FILE_FMT_NUM: 2643 if (quad == 0) { 2644 switch (type) { 2645 case FILE_BYTE: 2646 h = 2; 2647 break; 2648 case FILE_SHORT: 2649 case FILE_BESHORT: 2650 case FILE_LESHORT: 2651 h = 1; 2652 break; 2653 case FILE_LONG: 2654 case FILE_BELONG: 2655 case FILE_LELONG: 2656 case FILE_MELONG: 2657 case FILE_LEID3: 2658 case FILE_BEID3: 2659 case FILE_INDIRECT: 2660 h = 0; 2661 break; 2662 default: 2663 fprintf(stderr, "Bad number format %d", type); 2664 abort(); 2665 } 2666 } else 2667 h = 0; 2668 while (*ptr && strchr("+-.#", *ptr) != NULL) 2669 ptr++; 2670 #define CHECKLEN() do { \ 2671 for (len = cnt = 0; isdigit(CAST(unsigned char, *ptr)); ptr++, cnt++) \ 2672 len = len * 10 + (*ptr - '0'); \ 2673 if (cnt > 5 || len > 1024) \ 2674 goto toolong; \ 2675 } while (/*CONSTCOND*/0) 2676 2677 CHECKLEN(); 2678 if (*ptr == '.') 2679 ptr++; 2680 CHECKLEN(); 2681 if (quad) { 2682 if (*ptr++ != 'l') 2683 goto invalid; 2684 if (*ptr++ != 'l') 2685 goto invalid; 2686 } 2687 2688 switch (*ptr++) { 2689 #ifdef STRICT_FORMAT /* "long" formats are int formats for us */ 2690 /* so don't accept the 'l' modifier */ 2691 case 'l': 2692 switch (*ptr++) { 2693 case 'i': 2694 case 'd': 2695 case 'u': 2696 case 'o': 2697 case 'x': 2698 case 'X': 2699 if (h == 0) 2700 return 0; 2701 /*FALLTHROUGH*/ 2702 default: 2703 goto invalid; 2704 } 2705 2706 /* 2707 * Don't accept h and hh modifiers. They make writing 2708 * magic entries more complicated, for very little benefit 2709 */ 2710 case 'h': 2711 if (h-- <= 0) 2712 goto invalid; 2713 switch (*ptr++) { 2714 case 'h': 2715 if (h-- <= 0) 2716 goto invalid; 2717 switch (*ptr++) { 2718 case 'i': 2719 case 'd': 2720 case 'u': 2721 case 'o': 2722 case 'x': 2723 case 'X': 2724 return 0; 2725 default: 2726 goto invalid; 2727 } 2728 case 'i': 2729 case 'd': 2730 case 'u': 2731 case 'o': 2732 case 'x': 2733 case 'X': 2734 if (h == 0) 2735 return 0; 2736 /*FALLTHROUGH*/ 2737 default: 2738 goto invalid; 2739 } 2740 #endif 2741 case 'c': 2742 if (h == 2) 2743 return 0; 2744 goto invalid; 2745 case 'i': 2746 case 'd': 2747 case 'u': 2748 case 'o': 2749 case 'x': 2750 case 'X': 2751 #ifdef STRICT_FORMAT 2752 if (h == 0) 2753 return 0; 2754 /*FALLTHROUGH*/ 2755 #else 2756 return 0; 2757 #endif 2758 default: 2759 goto invalid; 2760 } 2761 2762 case FILE_FMT_FLOAT: 2763 case FILE_FMT_DOUBLE: 2764 if (*ptr == '-') 2765 ptr++; 2766 if (*ptr == '.') 2767 ptr++; 2768 CHECKLEN(); 2769 if (*ptr == '.') 2770 ptr++; 2771 CHECKLEN(); 2772 switch (*ptr++) { 2773 case 'e': 2774 case 'E': 2775 case 'f': 2776 case 'F': 2777 case 'g': 2778 case 'G': 2779 return 0; 2780 2781 default: 2782 goto invalid; 2783 } 2784 2785 2786 case FILE_FMT_STR: 2787 if (*ptr == '-') 2788 ptr++; 2789 while (isdigit(CAST(unsigned char, *ptr))) 2790 ptr++; 2791 if (*ptr == '.') { 2792 ptr++; 2793 while (isdigit(CAST(unsigned char , *ptr))) 2794 ptr++; 2795 } 2796 2797 switch (*ptr++) { 2798 case 's': 2799 return 0; 2800 default: 2801 goto invalid; 2802 } 2803 2804 default: 2805 /* internal error */ 2806 fprintf(stderr, "Bad file format %d", type); 2807 abort(); 2808 } 2809 invalid: 2810 *estr = "not valid"; 2811 return -1; 2812 toolong: 2813 *estr = "too long"; 2814 return -1; 2815 } 2816 2817 /* 2818 * Check that the optional printf format in description matches 2819 * the type of the magic. 2820 */ 2821 file_private int 2822 check_format(struct magic_set *ms, struct magic *m) 2823 { 2824 char *ptr; 2825 const char *estr; 2826 2827 for (ptr = m->desc; *ptr; ptr++) 2828 if (*ptr == '%') 2829 break; 2830 if (*ptr == '\0') { 2831 /* No format string; ok */ 2832 return 1; 2833 } 2834 2835 assert(file_nformats == file_nnames); 2836 2837 if (m->type >= file_nformats) { 2838 file_magwarn(ms, "Internal error inconsistency between " 2839 "m->type and format strings"); 2840 return -1; 2841 } 2842 if (file_formats[m->type] == FILE_FMT_NONE) { 2843 file_magwarn(ms, "No format string for `%s' with description " 2844 "`%s'", m->desc, file_names[m->type]); 2845 return -1; 2846 } 2847 2848 ptr++; 2849 if (check_format_type(ptr, m->type, &estr) == -1) { 2850 /* 2851 * TODO: this error message is unhelpful if the format 2852 * string is not one character long 2853 */ 2854 file_magwarn(ms, "Printf format is %s for type " 2855 "`%s' in description `%s'", estr, 2856 file_names[m->type], m->desc); 2857 return -1; 2858 } 2859 2860 for (; *ptr; ptr++) { 2861 if (*ptr == '%') { 2862 file_magwarn(ms, 2863 "Too many format strings (should have at most one) " 2864 "for `%s' with description `%s'", 2865 file_names[m->type], m->desc); 2866 return -1; 2867 } 2868 } 2869 return 0; 2870 } 2871 2872 /* 2873 * Read a numeric value from a pointer, into the value union of a magic 2874 * pointer, according to the magic type. Update the string pointer to point 2875 * just after the number read. Return 0 for success, non-zero for failure. 2876 */ 2877 file_no_overflow file_private int 2878 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action) 2879 { 2880 char *ep; 2881 uint64_t ull; 2882 int y; 2883 2884 switch (m->type) { 2885 case FILE_BESTRING16: 2886 case FILE_LESTRING16: 2887 case FILE_STRING: 2888 case FILE_PSTRING: 2889 case FILE_REGEX: 2890 case FILE_SEARCH: 2891 case FILE_NAME: 2892 case FILE_USE: 2893 case FILE_DER: 2894 case FILE_OCTAL: 2895 *p = getstr(ms, m, *p, action == FILE_COMPILE); 2896 if (*p == NULL) { 2897 if (ms->flags & MAGIC_CHECK) 2898 file_magwarn(ms, "cannot get string from `%s'", 2899 m->value.s); 2900 return -1; 2901 } 2902 if (m->type == FILE_REGEX) { 2903 file_regex_t rx; 2904 int rc = file_regcomp(ms, &rx, m->value.s, 2905 REG_EXTENDED | REGEX_ICASE(m)); 2906 if (rc == 0) { 2907 file_regfree(&rx); 2908 } 2909 return rc ? -1 : 0; 2910 } 2911 return 0; 2912 default: 2913 if (m->reln == 'x') 2914 return 0; 2915 break; 2916 } 2917 2918 switch (m->type) { 2919 case FILE_FLOAT: 2920 case FILE_BEFLOAT: 2921 case FILE_LEFLOAT: 2922 errno = 0; 2923 #ifdef HAVE_STRTOF 2924 m->value.f = strtof(*p, &ep); 2925 #else 2926 m->value.f = (float)strtod(*p, &ep); 2927 #endif 2928 if (errno == 0) 2929 *p = ep; 2930 return 0; 2931 case FILE_DOUBLE: 2932 case FILE_BEDOUBLE: 2933 case FILE_LEDOUBLE: 2934 errno = 0; 2935 m->value.d = strtod(*p, &ep); 2936 if (errno == 0) 2937 *p = ep; 2938 return 0; 2939 case FILE_BEGUID: 2940 case FILE_LEGUID: 2941 case FILE_GUID: 2942 if (file_parse_guid(*p, m->value.guid) == -1) { 2943 file_magwarn(ms, "Error parsing guid `%s'", *p); 2944 return -1; 2945 } 2946 *p += FILE_GUID_SIZE - 1; 2947 return 0; 2948 default: 2949 errno = 0; 2950 ull = CAST(uint64_t, strtoull(*p, &ep, 0)); 2951 m->value.q = file_signextend(ms, m, ull); 2952 if (*p == ep) { 2953 file_magwarn(ms, "Unparsable number `%s'", *p); 2954 return -1; 2955 } else { 2956 size_t ts = typesize(m->type); 2957 uint64_t x; 2958 const char *q; 2959 2960 if (ts == FILE_BADSIZE) { 2961 file_magwarn(ms, 2962 "Expected numeric type got `%s'", 2963 type_tbl[m->type].name); 2964 return -1; 2965 } 2966 for (q = *p; isspace(CAST(unsigned char, *q)); q++) 2967 continue; 2968 if (*q == '-' && ull != UINT64_MAX) 2969 ull = -CAST(int64_t, ull); 2970 switch (ts) { 2971 case 1: 2972 x = CAST(uint64_t, ull & ~0xffULL); 2973 y = (x & ~0xffULL) != ~0xffULL; 2974 break; 2975 case 2: 2976 x = CAST(uint64_t, ull & ~0xffffULL); 2977 y = (x & ~0xffffULL) != ~0xffffULL; 2978 break; 2979 case 4: 2980 x = CAST(uint64_t, ull & ~0xffffffffULL); 2981 y = (x & ~0xffffffffULL) != ~0xffffffffULL; 2982 break; 2983 case 8: 2984 x = 0; 2985 y = 0; 2986 break; 2987 default: 2988 fprintf(stderr, "Bad width %zu", ts); 2989 abort(); 2990 } 2991 if (x && y) { 2992 file_magwarn(ms, "Overflow for numeric" 2993 " type `%s' value %#" PRIx64, 2994 type_tbl[m->type].name, ull); 2995 return -1; 2996 } 2997 } 2998 if (errno == 0) { 2999 *p = ep; 3000 eatsize(p); 3001 } 3002 return 0; 3003 } 3004 } 3005 3006 /* 3007 * Convert a string containing C character escapes. Stop at an unescaped 3008 * space or tab. 3009 * Copy the converted version to "m->value.s", and the length in m->vallen. 3010 * Return updated scan pointer as function result. Warn if set. 3011 */ 3012 file_private const char * 3013 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn) 3014 { 3015 const char *origs = s; 3016 char *p = m->value.s; 3017 size_t plen = sizeof(m->value.s); 3018 char *origp = p; 3019 char *pmax = p + plen - 1; 3020 int c; 3021 int val; 3022 size_t bracket_nesting = 0; 3023 3024 while ((c = *s++) != '\0') { 3025 if (isspace(CAST(unsigned char, c))) 3026 break; 3027 if (p >= pmax) { 3028 file_error(ms, 0, "string too long: `%s'", origs); 3029 return NULL; 3030 } 3031 if (c != '\\') { 3032 if (c == '[') { 3033 bracket_nesting++; 3034 } 3035 if (c == ']' && bracket_nesting > 0) { 3036 bracket_nesting--; 3037 } 3038 *p++ = CAST(char, c); 3039 continue; 3040 } 3041 switch(c = *s++) { 3042 3043 case '\0': 3044 if (warn) 3045 file_magwarn(ms, "incomplete escape"); 3046 s--; 3047 goto out; 3048 case '.': 3049 if (m->type == FILE_REGEX && 3050 bracket_nesting == 0 && warn) { 3051 file_magwarn(ms, "escaped dot ('.') found, " 3052 "use \\\\. instead"); 3053 } 3054 warn = 0; /* already did */ 3055 /*FALLTHROUGH*/ 3056 case '\t': 3057 if (warn) { 3058 file_magwarn(ms, 3059 "escaped tab found, use \\\\t instead"); 3060 warn = 0; /* already did */ 3061 } 3062 /*FALLTHROUGH*/ 3063 default: 3064 if (warn) { 3065 if (isprint(CAST(unsigned char, c))) { 3066 /* Allow escaping of 3067 * ``relations'' */ 3068 if (strchr("<>&^=!", c) == NULL 3069 && (m->type != FILE_REGEX || 3070 strchr("[]().*?^$|{}", c) 3071 == NULL)) { 3072 file_magwarn(ms, "no " 3073 "need to escape " 3074 "`%c'", c); 3075 } 3076 } else { 3077 file_magwarn(ms, 3078 "unknown escape sequence: " 3079 "\\%03o", c); 3080 } 3081 } 3082 /*FALLTHROUGH*/ 3083 /* space, perhaps force people to use \040? */ 3084 case ' ': 3085 #if 0 3086 /* 3087 * Other things people escape, but shouldn't need to, 3088 * so we disallow them 3089 */ 3090 case '\'': 3091 case '"': 3092 case '?': 3093 #endif 3094 /* Relations */ 3095 case '>': 3096 case '<': 3097 case '&': 3098 case '^': 3099 case '=': 3100 case '!': 3101 /* and backslash itself */ 3102 case '\\': 3103 *p++ = CAST(char, c); 3104 break; 3105 3106 case 'a': 3107 *p++ = '\a'; 3108 break; 3109 3110 case 'b': 3111 *p++ = '\b'; 3112 break; 3113 3114 case 'f': 3115 *p++ = '\f'; 3116 break; 3117 3118 case 'n': 3119 *p++ = '\n'; 3120 break; 3121 3122 case 'r': 3123 *p++ = '\r'; 3124 break; 3125 3126 case 't': 3127 *p++ = '\t'; 3128 break; 3129 3130 case 'v': 3131 *p++ = '\v'; 3132 break; 3133 3134 /* \ and up to 3 octal digits */ 3135 case '0': 3136 case '1': 3137 case '2': 3138 case '3': 3139 case '4': 3140 case '5': 3141 case '6': 3142 case '7': 3143 val = c - '0'; 3144 c = *s++; /* try for 2 */ 3145 if (c >= '0' && c <= '7') { 3146 val = (val << 3) | (c - '0'); 3147 c = *s++; /* try for 3 */ 3148 if (c >= '0' && c <= '7') 3149 val = (val << 3) | (c-'0'); 3150 else 3151 --s; 3152 } 3153 else 3154 --s; 3155 *p++ = CAST(char, val); 3156 break; 3157 3158 /* \x and up to 2 hex digits */ 3159 case 'x': 3160 val = 'x'; /* Default if no digits */ 3161 c = hextoint(*s++); /* Get next char */ 3162 if (c >= 0) { 3163 val = c; 3164 c = hextoint(*s++); 3165 if (c >= 0) 3166 val = (val << 4) + c; 3167 else 3168 --s; 3169 } else 3170 --s; 3171 *p++ = CAST(char, val); 3172 break; 3173 } 3174 } 3175 --s; 3176 out: 3177 *p = '\0'; 3178 if (CAST(size_t, (p - origp)) > sizeof(m->value.s)) 3179 return NULL; 3180 m->vallen = CAST(unsigned char, (p - origp)); 3181 if (m->type == FILE_PSTRING) { 3182 size_t l = file_pstring_length_size(ms, m); 3183 if (l == FILE_BADSIZE) 3184 return NULL; 3185 if (m->vallen + l > sizeof(m->value.s)) 3186 return NULL; 3187 m->vallen += CAST(unsigned char, l); 3188 } 3189 return s; 3190 } 3191 3192 3193 /* Single hex char to int; -1 if not a hex char. */ 3194 file_private int 3195 hextoint(int c) 3196 { 3197 if (!isascii(CAST(unsigned char, c))) 3198 return -1; 3199 if (isdigit(CAST(unsigned char, c))) 3200 return c - '0'; 3201 if ((c >= 'a') && (c <= 'f')) 3202 return c + 10 - 'a'; 3203 if (( c>= 'A') && (c <= 'F')) 3204 return c + 10 - 'A'; 3205 return -1; 3206 } 3207 3208 3209 /* 3210 * Print a string containing C character escapes. 3211 */ 3212 file_protected void 3213 file_showstr(FILE *fp, const char *s, size_t len) 3214 { 3215 char c; 3216 3217 for (;;) { 3218 if (len == FILE_BADSIZE) { 3219 c = *s++; 3220 if (c == '\0') 3221 break; 3222 } 3223 else { 3224 if (len-- == 0) 3225 break; 3226 c = *s++; 3227 } 3228 if (c >= 040 && c <= 0176) /* TODO isprint && !iscntrl */ 3229 (void) fputc(c, fp); 3230 else { 3231 (void) fputc('\\', fp); 3232 switch (c) { 3233 case '\a': 3234 (void) fputc('a', fp); 3235 break; 3236 3237 case '\b': 3238 (void) fputc('b', fp); 3239 break; 3240 3241 case '\f': 3242 (void) fputc('f', fp); 3243 break; 3244 3245 case '\n': 3246 (void) fputc('n', fp); 3247 break; 3248 3249 case '\r': 3250 (void) fputc('r', fp); 3251 break; 3252 3253 case '\t': 3254 (void) fputc('t', fp); 3255 break; 3256 3257 case '\v': 3258 (void) fputc('v', fp); 3259 break; 3260 3261 default: 3262 (void) fprintf(fp, "%.3o", c & 0377); 3263 break; 3264 } 3265 } 3266 } 3267 } 3268 3269 /* 3270 * eatsize(): Eat the size spec from a number [eg. 10UL] 3271 */ 3272 file_private void 3273 eatsize(const char **p) 3274 { 3275 const char *l = *p; 3276 3277 if (LOWCASE(*l) == 'u') 3278 l++; 3279 3280 switch (LOWCASE(*l)) { 3281 case 'l': /* long */ 3282 case 's': /* short */ 3283 case 'h': /* short */ 3284 case 'b': /* char/byte */ 3285 case 'c': /* char/byte */ 3286 l++; 3287 /*FALLTHROUGH*/ 3288 default: 3289 break; 3290 } 3291 3292 *p = l; 3293 } 3294 3295 /* 3296 * handle a buffer containing a compiled file. 3297 */ 3298 file_private struct magic_map * 3299 apprentice_buf(struct magic_set *ms, struct magic *buf, size_t len) 3300 { 3301 struct magic_map *map; 3302 3303 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) { 3304 file_oomem(ms, sizeof(*map)); 3305 return NULL; 3306 } 3307 map->len = len; 3308 map->p = buf; 3309 map->type = MAP_TYPE_USER; 3310 if (check_buffer(ms, map, "buffer") != 0) { 3311 apprentice_unmap(map); 3312 return NULL; 3313 } 3314 return map; 3315 } 3316 3317 /* 3318 * handle a compiled file. 3319 */ 3320 3321 file_private struct magic_map * 3322 apprentice_map(struct magic_set *ms, const char *fn) 3323 { 3324 int fd; 3325 struct stat st; 3326 char *dbname = NULL; 3327 struct magic_map *map; 3328 struct magic_map *rv = NULL; 3329 3330 fd = -1; 3331 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) { 3332 file_oomem(ms, sizeof(*map)); 3333 goto error; 3334 } 3335 map->type = MAP_TYPE_USER; /* unspecified */ 3336 3337 dbname = mkdbname(ms, fn, 0); 3338 if (dbname == NULL) 3339 goto error; 3340 3341 if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1) 3342 goto error; 3343 3344 if (fstat(fd, &st) == -1) { 3345 file_error(ms, errno, "cannot stat `%s'", dbname); 3346 goto error; 3347 } 3348 if (st.st_size < 8 || st.st_size > maxoff_t()) { 3349 file_error(ms, 0, "file `%s' is too %s", dbname, 3350 st.st_size < 8 ? "small" : "large"); 3351 goto error; 3352 } 3353 3354 map->len = CAST(size_t, st.st_size); 3355 #ifdef QUICK 3356 map->type = MAP_TYPE_MMAP; 3357 if ((map->p = mmap(0, CAST(size_t, st.st_size), PROT_READ|PROT_WRITE, 3358 MAP_PRIVATE|MAP_FILE, fd, CAST(off_t, 0))) == MAP_FAILED) { 3359 file_error(ms, errno, "cannot map `%s'", dbname); 3360 goto error; 3361 } 3362 #else 3363 map->type = MAP_TYPE_MALLOC; 3364 if ((map->p = CAST(void *, malloc(map->len))) == NULL) { 3365 file_oomem(ms, map->len); 3366 goto error; 3367 } 3368 if (read(fd, map->p, map->len) != (ssize_t)map->len) { 3369 file_badread(ms); 3370 goto error; 3371 } 3372 #endif 3373 (void)close(fd); 3374 fd = -1; 3375 3376 if (check_buffer(ms, map, dbname) != 0) { 3377 goto error; 3378 } 3379 #ifdef QUICK 3380 if (mprotect(map->p, CAST(size_t, st.st_size), PROT_READ) == -1) { 3381 file_error(ms, errno, "cannot mprotect `%s'", dbname); 3382 goto error; 3383 } 3384 #endif 3385 3386 free(dbname); 3387 return map; 3388 3389 error: 3390 if (fd != -1) 3391 (void)close(fd); 3392 apprentice_unmap(map); 3393 free(dbname); 3394 return rv; 3395 } 3396 3397 file_private int 3398 check_buffer(struct magic_set *ms, struct magic_map *map, const char *dbname) 3399 { 3400 uint32_t *ptr; 3401 uint32_t entries, nentries; 3402 uint32_t version; 3403 int i, needsbyteswap; 3404 3405 entries = CAST(uint32_t, map->len / sizeof(struct magic)); 3406 if (entries < MAGIC_SETS) { 3407 file_error(ms, 0, "Too few magic entries %u in `%s'", 3408 entries, dbname); 3409 return -1; 3410 } 3411 if ((entries * sizeof(struct magic)) != map->len) { 3412 file_error(ms, 0, "Size of `%s' %" SIZE_T_FORMAT "u is not " 3413 "a multiple of %" SIZE_T_FORMAT "u", 3414 dbname, map->len, sizeof(struct magic)); 3415 return -1; 3416 } 3417 3418 ptr = CAST(uint32_t *, map->p); 3419 if (*ptr != MAGICNO) { 3420 if (file_swap4(*ptr) != MAGICNO) { 3421 file_error(ms, 0, "bad magic in `%s'", dbname); 3422 return -1; 3423 } 3424 needsbyteswap = 1; 3425 } else 3426 needsbyteswap = 0; 3427 if (needsbyteswap) 3428 version = file_swap4(ptr[1]); 3429 else 3430 version = ptr[1]; 3431 if (version != VERSIONNO) { 3432 file_error(ms, 0, "File %s supports only version %d magic " 3433 "files. `%s' is version %d", VERSION, 3434 VERSIONNO, dbname, version); 3435 return -1; 3436 } 3437 map->magic[0] = CAST(struct magic *, map->p) + 1; 3438 nentries = 0; 3439 for (i = 0; i < MAGIC_SETS; i++) { 3440 if (needsbyteswap) 3441 map->nmagic[i] = file_swap4(ptr[i + 2]); 3442 else 3443 map->nmagic[i] = ptr[i + 2]; 3444 if (map->nmagic[i] > entries) { 3445 file_error(ms, 0, "nmagic[%u] too large in `%s'", i, 3446 dbname); 3447 return -1; 3448 } 3449 if (i != MAGIC_SETS - 1) 3450 map->magic[i + 1] = map->magic[i] + map->nmagic[i]; 3451 nentries += map->nmagic[i]; 3452 } 3453 if (entries != nentries + 1) { 3454 file_error(ms, 0, "Inconsistent entries in `%s' %u != %u", 3455 dbname, entries, nentries + 1); 3456 return -1; 3457 } 3458 if (needsbyteswap) 3459 for (i = 0; i < MAGIC_SETS; i++) 3460 byteswap(map->magic[i], map->nmagic[i]); 3461 return 0; 3462 } 3463 3464 /* 3465 * handle an mmaped file. 3466 */ 3467 file_private int 3468 apprentice_compile(struct magic_set *ms, struct magic_map *map, const char *fn) 3469 { 3470 static const size_t nm = sizeof(*map->nmagic) * MAGIC_SETS; 3471 static const size_t m = sizeof(**map->magic); 3472 int fd = -1; 3473 size_t len; 3474 char *dbname; 3475 int rv = -1; 3476 uint32_t i; 3477 union { 3478 struct magic m; 3479 uint32_t h[2 + MAGIC_SETS]; 3480 } hdr; 3481 3482 dbname = mkdbname(ms, fn, 1); 3483 3484 if (dbname == NULL) 3485 goto out; 3486 3487 if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) 3488 { 3489 file_error(ms, errno, "cannot open `%s'", dbname); 3490 goto out; 3491 } 3492 memset(&hdr, 0, sizeof(hdr)); 3493 hdr.h[0] = MAGICNO; 3494 hdr.h[1] = VERSIONNO; 3495 memcpy(hdr.h + 2, map->nmagic, nm); 3496 3497 if (write(fd, &hdr, sizeof(hdr)) != CAST(ssize_t, sizeof(hdr))) { 3498 file_error(ms, errno, "error writing `%s'", dbname); 3499 goto out2; 3500 } 3501 3502 for (i = 0; i < MAGIC_SETS; i++) { 3503 len = m * map->nmagic[i]; 3504 if (write(fd, map->magic[i], len) != CAST(ssize_t, len)) { 3505 file_error(ms, errno, "error writing `%s'", dbname); 3506 goto out2; 3507 } 3508 } 3509 3510 rv = 0; 3511 out2: 3512 if (fd != -1) 3513 (void)close(fd); 3514 out: 3515 apprentice_unmap(map); 3516 free(dbname); 3517 return rv; 3518 } 3519 3520 file_private const char ext[] = ".mgc"; 3521 /* 3522 * make a dbname 3523 */ 3524 file_private char * 3525 mkdbname(struct magic_set *ms, const char *fn, int strip) 3526 { 3527 const char *p, *q; 3528 char *buf; 3529 3530 if (strip) { 3531 if ((p = strrchr(fn, '/')) != NULL) 3532 fn = ++p; 3533 } 3534 3535 for (q = fn; *q; q++) 3536 continue; 3537 /* Look for .mgc */ 3538 for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--) 3539 if (*p != *q) 3540 break; 3541 3542 /* Did not find .mgc, restore q */ 3543 if (p >= ext) 3544 for (q = fn; *q; q++) 3545 continue; 3546 3547 q++; 3548 /* Compatibility with old code that looked in .mime */ 3549 if (ms->flags & MAGIC_MIME) { 3550 if (asprintf(&buf, "%.*s.mime%s", CAST(int, q - fn), fn, ext) 3551 < 0) 3552 return NULL; 3553 if (access(buf, R_OK) != -1) { 3554 ms->flags &= MAGIC_MIME_TYPE; 3555 return buf; 3556 } 3557 free(buf); 3558 } 3559 if (asprintf(&buf, "%.*s%s", CAST(int, q - fn), fn, ext) < 0) 3560 return NULL; 3561 3562 /* Compatibility with old code that looked in .mime */ 3563 if (strstr(fn, ".mime") != NULL) 3564 ms->flags &= MAGIC_MIME_TYPE; 3565 return buf; 3566 } 3567 3568 /* 3569 * Byteswap an mmap'ed file if needed 3570 */ 3571 file_private void 3572 byteswap(struct magic *magic, uint32_t nmagic) 3573 { 3574 uint32_t i; 3575 for (i = 0; i < nmagic; i++) 3576 bs1(&magic[i]); 3577 } 3578 3579 file_protected uintmax_t 3580 file_varint2uintmax_t(const unsigned char *us, int t, size_t *l) 3581 { 3582 uintmax_t x = 0; 3583 const unsigned char *c; 3584 if (t == FILE_LEVARINT) { 3585 for (c = us; *c; c++) { 3586 if ((*c & 0x80) == 0) 3587 break; 3588 } 3589 if (l) 3590 *l = c - us + 1; 3591 for (; c >= us; c--) { 3592 x |= *c & 0x7f; 3593 x <<= 7; 3594 } 3595 } else { 3596 for (c = us; *c; c++) { 3597 x |= *c & 0x7f; 3598 if ((*c & 0x80) == 0) 3599 break; 3600 x <<= 7; 3601 } 3602 if (l) 3603 *l = c - us + 1; 3604 } 3605 return x; 3606 } 3607 3608 3609 /* 3610 * byteswap a single magic entry 3611 */ 3612 file_private void 3613 bs1(struct magic *m) 3614 { 3615 m->flag = file_swap2(m->flag); 3616 m->offset = file_swap4(CAST(uint32_t, m->offset)); 3617 m->in_offset = file_swap4(CAST(uint32_t, m->in_offset)); 3618 m->lineno = file_swap4(CAST(uint32_t, m->lineno)); 3619 if (IS_STRING(m->type)) { 3620 m->str_range = file_swap4(m->str_range); 3621 m->str_flags = file_swap4(m->str_flags); 3622 } else { 3623 m->value.q = file_swap8(m->value.q); 3624 m->num_mask = file_swap8(m->num_mask); 3625 } 3626 } 3627 3628 file_protected size_t 3629 file_pstring_length_size(struct magic_set *ms, const struct magic *m) 3630 { 3631 switch (m->str_flags & PSTRING_LEN) { 3632 case PSTRING_1_LE: 3633 return 1; 3634 case PSTRING_2_LE: 3635 case PSTRING_2_BE: 3636 return 2; 3637 case PSTRING_4_LE: 3638 case PSTRING_4_BE: 3639 return 4; 3640 default: 3641 file_error(ms, 0, "corrupt magic file " 3642 "(bad pascal string length %d)", 3643 m->str_flags & PSTRING_LEN); 3644 return FILE_BADSIZE; 3645 } 3646 } 3647 3648 file_protected size_t 3649 file_pstring_get_length(struct magic_set *ms, const struct magic *m, 3650 const char *ss) 3651 { 3652 size_t len = 0; 3653 const unsigned char *s = RCAST(const unsigned char *, ss); 3654 unsigned int s3, s2, s1, s0; 3655 3656 switch (m->str_flags & PSTRING_LEN) { 3657 case PSTRING_1_LE: 3658 len = *s; 3659 break; 3660 case PSTRING_2_LE: 3661 s0 = s[0]; 3662 s1 = s[1]; 3663 len = (s1 << 8) | s0; 3664 break; 3665 case PSTRING_2_BE: 3666 s0 = s[0]; 3667 s1 = s[1]; 3668 len = (s0 << 8) | s1; 3669 break; 3670 case PSTRING_4_LE: 3671 s0 = s[0]; 3672 s1 = s[1]; 3673 s2 = s[2]; 3674 s3 = s[3]; 3675 len = (s3 << 24) | (s2 << 16) | (s1 << 8) | s0; 3676 break; 3677 case PSTRING_4_BE: 3678 s0 = s[0]; 3679 s1 = s[1]; 3680 s2 = s[2]; 3681 s3 = s[3]; 3682 len = (s0 << 24) | (s1 << 16) | (s2 << 8) | s3; 3683 break; 3684 default: 3685 file_error(ms, 0, "corrupt magic file " 3686 "(bad pascal string length %d)", 3687 m->str_flags & PSTRING_LEN); 3688 return FILE_BADSIZE; 3689 } 3690 3691 if (m->str_flags & PSTRING_LENGTH_INCLUDES_ITSELF) { 3692 size_t l = file_pstring_length_size(ms, m); 3693 if (l == FILE_BADSIZE) 3694 return l; 3695 len -= l; 3696 } 3697 3698 return len; 3699 } 3700 3701 file_protected int 3702 file_magicfind(struct magic_set *ms, const char *name, struct mlist *v) 3703 { 3704 uint32_t i, j; 3705 struct mlist *mlist, *ml; 3706 3707 mlist = ms->mlist[1]; 3708 3709 for (ml = mlist->next; ml != mlist; ml = ml->next) { 3710 struct magic *ma = ml->magic; 3711 for (i = 0; i < ml->nmagic; i++) { 3712 if (ma[i].type != FILE_NAME) 3713 continue; 3714 if (strcmp(ma[i].value.s, name) == 0) { 3715 v->magic = &ma[i]; 3716 v->magic_rxcomp = &(ml->magic_rxcomp[i]); 3717 for (j = i + 1; j < ml->nmagic; j++) 3718 if (ma[j].cont_level == 0) 3719 break; 3720 v->nmagic = j - i; 3721 return 0; 3722 } 3723 } 3724 } 3725 return -1; 3726 } 3727