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 1.173 2011/12/08 12:38:24 rrt Exp $") 36 #endif /* lint */ 37 38 #include "magic.h" 39 #include <stdlib.h> 40 #ifdef HAVE_UNISTD_H 41 #include <unistd.h> 42 #endif 43 #include <string.h> 44 #include <assert.h> 45 #include <ctype.h> 46 #include <fcntl.h> 47 #ifdef QUICK 48 #include <sys/mman.h> 49 #endif 50 #include <dirent.h> 51 52 #define EATAB {while (isascii((unsigned char) *l) && \ 53 isspace((unsigned char) *l)) ++l;} 54 #define LOWCASE(l) (isupper((unsigned char) (l)) ? \ 55 tolower((unsigned char) (l)) : (l)) 56 /* 57 * Work around a bug in headers on Digital Unix. 58 * At least confirmed for: OSF1 V4.0 878 59 */ 60 #if defined(__osf__) && defined(__DECC) 61 #ifdef MAP_FAILED 62 #undef MAP_FAILED 63 #endif 64 #endif 65 66 #ifndef MAP_FAILED 67 #define MAP_FAILED (void *) -1 68 #endif 69 70 #ifndef MAP_FILE 71 #define MAP_FILE 0 72 #endif 73 74 struct magic_entry { 75 struct magic *mp; 76 uint32_t cont_count; 77 uint32_t max_count; 78 }; 79 80 int file_formats[FILE_NAMES_SIZE]; 81 const size_t file_nformats = FILE_NAMES_SIZE; 82 const char *file_names[FILE_NAMES_SIZE]; 83 const size_t file_nnames = FILE_NAMES_SIZE; 84 85 private int getvalue(struct magic_set *ms, struct magic *, const char **, int); 86 private int hextoint(int); 87 private const char *getstr(struct magic_set *, struct magic *, const char *, 88 int); 89 private int parse(struct magic_set *, struct magic_entry **, uint32_t *, 90 const char *, size_t, int); 91 private void eatsize(const char **); 92 private int apprentice_1(struct magic_set *, const char *, int, struct mlist *); 93 private size_t apprentice_magic_strength(const struct magic *); 94 private int apprentice_sort(const void *, const void *); 95 private void apprentice_list(struct mlist *, int ); 96 private int apprentice_load(struct magic_set *, struct magic **, uint32_t *, 97 const char *, int); 98 private void byteswap(struct magic *, uint32_t); 99 private void bs1(struct magic *); 100 private uint16_t swap2(uint16_t); 101 private uint32_t swap4(uint32_t); 102 private uint64_t swap8(uint64_t); 103 private char *mkdbname(struct magic_set *, const char *, int); 104 private int apprentice_map(struct magic_set *, struct magic **, uint32_t *, 105 const char *); 106 private int apprentice_compile(struct magic_set *, struct magic **, uint32_t *, 107 const char *); 108 private int check_format_type(const char *, int); 109 private int check_format(struct magic_set *, struct magic *); 110 private int get_op(char); 111 private int parse_mime(struct magic_set *, struct magic_entry *, const char *); 112 private int parse_strength(struct magic_set *, struct magic_entry *, const char *); 113 private int parse_apple(struct magic_set *, struct magic_entry *, const char *); 114 115 116 private size_t maxmagic = 0; 117 private size_t magicsize = sizeof(struct magic); 118 119 private const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc"; 120 121 private struct { 122 const char *name; 123 size_t len; 124 int (*fun)(struct magic_set *, struct magic_entry *, const char *); 125 } bang[] = { 126 #define DECLARE_FIELD(name) { # name, sizeof(# name) - 1, parse_ ## name } 127 DECLARE_FIELD(mime), 128 DECLARE_FIELD(apple), 129 DECLARE_FIELD(strength), 130 #undef DECLARE_FIELD 131 { NULL, 0, NULL } 132 }; 133 134 #ifdef COMPILE_ONLY 135 136 int main(int, char *[]); 137 138 int 139 main(int argc, char *argv[]) 140 { 141 int ret; 142 struct magic_set *ms; 143 char *progname; 144 145 if ((progname = strrchr(argv[0], '/')) != NULL) 146 progname++; 147 else 148 progname = argv[0]; 149 150 if (argc != 2) { 151 (void)fprintf(stderr, "Usage: %s file\n", progname); 152 return 1; 153 } 154 155 if ((ms = magic_open(MAGIC_CHECK)) == NULL) { 156 (void)fprintf(stderr, "%s: %s\n", progname, strerror(errno)); 157 return 1; 158 } 159 ret = magic_compile(ms, argv[1]) == -1 ? 1 : 0; 160 if (ret == 1) 161 (void)fprintf(stderr, "%s: %s\n", progname, magic_error(ms)); 162 magic_close(ms); 163 return ret; 164 } 165 #endif /* COMPILE_ONLY */ 166 167 static const struct type_tbl_s { 168 const char name[16]; 169 const size_t len; 170 const int type; 171 const int format; 172 } type_tbl[] = { 173 # define XX(s) s, (sizeof(s) - 1) 174 # define XX_NULL "", 0 175 { XX("byte"), FILE_BYTE, FILE_FMT_NUM }, 176 { XX("short"), FILE_SHORT, FILE_FMT_NUM }, 177 { XX("default"), FILE_DEFAULT, FILE_FMT_STR }, 178 { XX("long"), FILE_LONG, FILE_FMT_NUM }, 179 { XX("string"), FILE_STRING, FILE_FMT_STR }, 180 { XX("date"), FILE_DATE, FILE_FMT_STR }, 181 { XX("beshort"), FILE_BESHORT, FILE_FMT_NUM }, 182 { XX("belong"), FILE_BELONG, FILE_FMT_NUM }, 183 { XX("bedate"), FILE_BEDATE, FILE_FMT_STR }, 184 { XX("leshort"), FILE_LESHORT, FILE_FMT_NUM }, 185 { XX("lelong"), FILE_LELONG, FILE_FMT_NUM }, 186 { XX("ledate"), FILE_LEDATE, FILE_FMT_STR }, 187 { XX("pstring"), FILE_PSTRING, FILE_FMT_STR }, 188 { XX("ldate"), FILE_LDATE, FILE_FMT_STR }, 189 { XX("beldate"), FILE_BELDATE, FILE_FMT_STR }, 190 { XX("leldate"), FILE_LELDATE, FILE_FMT_STR }, 191 { XX("regex"), FILE_REGEX, FILE_FMT_STR }, 192 { XX("bestring16"), FILE_BESTRING16, FILE_FMT_STR }, 193 { XX("lestring16"), FILE_LESTRING16, FILE_FMT_STR }, 194 { XX("search"), FILE_SEARCH, FILE_FMT_STR }, 195 { XX("medate"), FILE_MEDATE, FILE_FMT_STR }, 196 { XX("meldate"), FILE_MELDATE, FILE_FMT_STR }, 197 { XX("melong"), FILE_MELONG, FILE_FMT_NUM }, 198 { XX("quad"), FILE_QUAD, FILE_FMT_QUAD }, 199 { XX("lequad"), FILE_LEQUAD, FILE_FMT_QUAD }, 200 { XX("bequad"), FILE_BEQUAD, FILE_FMT_QUAD }, 201 { XX("qdate"), FILE_QDATE, FILE_FMT_STR }, 202 { XX("leqdate"), FILE_LEQDATE, FILE_FMT_STR }, 203 { XX("beqdate"), FILE_BEQDATE, FILE_FMT_STR }, 204 { XX("qldate"), FILE_QLDATE, FILE_FMT_STR }, 205 { XX("leqldate"), FILE_LEQLDATE, FILE_FMT_STR }, 206 { XX("beqldate"), FILE_BEQLDATE, FILE_FMT_STR }, 207 { XX("float"), FILE_FLOAT, FILE_FMT_FLOAT }, 208 { XX("befloat"), FILE_BEFLOAT, FILE_FMT_FLOAT }, 209 { XX("lefloat"), FILE_LEFLOAT, FILE_FMT_FLOAT }, 210 { XX("double"), FILE_DOUBLE, FILE_FMT_DOUBLE }, 211 { XX("bedouble"), FILE_BEDOUBLE, FILE_FMT_DOUBLE }, 212 { XX("ledouble"), FILE_LEDOUBLE, FILE_FMT_DOUBLE }, 213 { XX("leid3"), FILE_LEID3, FILE_FMT_NUM }, 214 { XX("beid3"), FILE_BEID3, FILE_FMT_NUM }, 215 { XX("indirect"), FILE_INDIRECT, FILE_FMT_NONE }, 216 { XX_NULL, FILE_INVALID, FILE_FMT_NONE }, 217 # undef XX 218 # undef XX_NULL 219 }; 220 221 private int 222 get_type(const char *l, const char **t) 223 { 224 const struct type_tbl_s *p; 225 226 for (p = type_tbl; p->len; p++) { 227 if (strncmp(l, p->name, p->len) == 0) { 228 if (t) 229 *t = l + p->len; 230 break; 231 } 232 } 233 return p->type; 234 } 235 236 private void 237 init_file_tables(void) 238 { 239 static int done = 0; 240 const struct type_tbl_s *p; 241 242 if (done) 243 return; 244 done++; 245 246 for (p = type_tbl; p->len; p++) { 247 assert(p->type < FILE_NAMES_SIZE); 248 file_names[p->type] = p->name; 249 file_formats[p->type] = p->format; 250 } 251 } 252 253 /* 254 * Handle one file or directory. 255 */ 256 private int 257 apprentice_1(struct magic_set *ms, const char *fn, int action, 258 struct mlist *mlist) 259 { 260 struct magic *magic = NULL; 261 uint32_t nmagic = 0; 262 struct mlist *ml; 263 int rv = -1; 264 int mapped; 265 266 if (magicsize != FILE_MAGICSIZE) { 267 file_error(ms, 0, "magic element size %lu != %lu", 268 (unsigned long)sizeof(*magic), 269 (unsigned long)FILE_MAGICSIZE); 270 return -1; 271 } 272 273 if (action == FILE_COMPILE) { 274 rv = apprentice_load(ms, &magic, &nmagic, fn, action); 275 if (rv != 0) 276 return -1; 277 rv = apprentice_compile(ms, &magic, &nmagic, fn); 278 free(magic); 279 return rv; 280 } 281 282 #ifndef COMPILE_ONLY 283 if ((rv = apprentice_map(ms, &magic, &nmagic, fn)) == -1) { 284 if (ms->flags & MAGIC_CHECK) 285 file_magwarn(ms, "using regular magic file `%s'", fn); 286 rv = apprentice_load(ms, &magic, &nmagic, fn, action); 287 if (rv != 0) 288 return -1; 289 } 290 291 mapped = rv; 292 293 if (magic == NULL) { 294 file_delmagic(magic, mapped, nmagic); 295 return -1; 296 } 297 298 if ((ml = CAST(struct mlist *, malloc(sizeof(*ml)))) == NULL) { 299 file_delmagic(magic, mapped, nmagic); 300 file_oomem(ms, sizeof(*ml)); 301 return -1; 302 } 303 304 ml->magic = magic; 305 ml->nmagic = nmagic; 306 ml->mapped = mapped; 307 308 mlist->prev->next = ml; 309 ml->prev = mlist->prev; 310 ml->next = mlist; 311 mlist->prev = ml; 312 313 if (action == FILE_LIST) { 314 printf("Binary patterns:\n"); 315 apprentice_list(mlist, BINTEST); 316 printf("Text patterns:\n"); 317 apprentice_list(mlist, TEXTTEST); 318 } 319 320 return 0; 321 #endif /* COMPILE_ONLY */ 322 } 323 324 protected void 325 file_delmagic(struct magic *p, int type, size_t entries) 326 { 327 if (p == NULL) 328 return; 329 switch (type) { 330 case 2: 331 #ifdef QUICK 332 p--; 333 (void)munmap((void *)p, sizeof(*p) * (entries + 1)); 334 break; 335 #else 336 (void)&entries; 337 abort(); 338 /*NOTREACHED*/ 339 #endif 340 case 1: 341 p--; 342 /*FALLTHROUGH*/ 343 case 0: 344 free(p); 345 break; 346 default: 347 abort(); 348 } 349 } 350 351 /* const char *fn: list of magic files and directories */ 352 protected struct mlist * 353 file_apprentice(struct magic_set *ms, const char *fn, int action) 354 { 355 char *p, *mfn; 356 int file_err, errs = -1; 357 struct mlist *mlist; 358 359 if ((fn = magic_getpath(fn, action)) == NULL) 360 return NULL; 361 362 init_file_tables(); 363 364 if ((mfn = strdup(fn)) == NULL) { 365 file_oomem(ms, strlen(fn)); 366 return NULL; 367 } 368 fn = mfn; 369 370 if ((mlist = CAST(struct mlist *, malloc(sizeof(*mlist)))) == NULL) { 371 free(mfn); 372 file_oomem(ms, sizeof(*mlist)); 373 return NULL; 374 } 375 mlist->next = mlist->prev = mlist; 376 377 while (fn) { 378 p = strchr(fn, PATHSEP); 379 if (p) 380 *p++ = '\0'; 381 if (*fn == '\0') 382 break; 383 file_err = apprentice_1(ms, fn, action, mlist); 384 errs = MAX(errs, file_err); 385 fn = p; 386 } 387 if (errs == -1) { 388 free(mfn); 389 free(mlist); 390 mlist = NULL; 391 file_error(ms, 0, "could not find any magic files!"); 392 return NULL; 393 } 394 free(mfn); 395 return mlist; 396 } 397 398 /* 399 * Get weight of this magic entry, for sorting purposes. 400 */ 401 private size_t 402 apprentice_magic_strength(const struct magic *m) 403 { 404 #define MULT 10 405 size_t val = 2 * MULT; /* baseline strength */ 406 407 switch (m->type) { 408 case FILE_DEFAULT: /* make sure this sorts last */ 409 if (m->factor_op != FILE_FACTOR_OP_NONE) 410 abort(); 411 return 0; 412 413 case FILE_BYTE: 414 val += 1 * MULT; 415 break; 416 417 case FILE_SHORT: 418 case FILE_LESHORT: 419 case FILE_BESHORT: 420 val += 2 * MULT; 421 break; 422 423 case FILE_LONG: 424 case FILE_LELONG: 425 case FILE_BELONG: 426 case FILE_MELONG: 427 val += 4 * MULT; 428 break; 429 430 case FILE_PSTRING: 431 case FILE_STRING: 432 val += m->vallen * MULT; 433 break; 434 435 case FILE_BESTRING16: 436 case FILE_LESTRING16: 437 val += m->vallen * MULT / 2; 438 break; 439 440 case FILE_SEARCH: 441 case FILE_REGEX: 442 val += m->vallen * MAX(MULT / m->vallen, 1); 443 break; 444 445 case FILE_DATE: 446 case FILE_LEDATE: 447 case FILE_BEDATE: 448 case FILE_MEDATE: 449 case FILE_LDATE: 450 case FILE_LELDATE: 451 case FILE_BELDATE: 452 case FILE_MELDATE: 453 case FILE_FLOAT: 454 case FILE_BEFLOAT: 455 case FILE_LEFLOAT: 456 val += 4 * MULT; 457 break; 458 459 case FILE_QUAD: 460 case FILE_BEQUAD: 461 case FILE_LEQUAD: 462 case FILE_QDATE: 463 case FILE_LEQDATE: 464 case FILE_BEQDATE: 465 case FILE_QLDATE: 466 case FILE_LEQLDATE: 467 case FILE_BEQLDATE: 468 case FILE_DOUBLE: 469 case FILE_BEDOUBLE: 470 case FILE_LEDOUBLE: 471 val += 8 * MULT; 472 break; 473 474 default: 475 val = 0; 476 (void)fprintf(stderr, "Bad type %d\n", m->type); 477 abort(); 478 } 479 480 switch (m->reln) { 481 case 'x': /* matches anything penalize */ 482 case '!': /* matches almost anything penalize */ 483 val = 0; 484 break; 485 486 case '=': /* Exact match, prefer */ 487 val += MULT; 488 break; 489 490 case '>': 491 case '<': /* comparison match reduce strength */ 492 val -= 2 * MULT; 493 break; 494 495 case '^': 496 case '&': /* masking bits, we could count them too */ 497 val -= MULT; 498 break; 499 500 default: 501 (void)fprintf(stderr, "Bad relation %c\n", m->reln); 502 abort(); 503 } 504 505 if (val == 0) /* ensure we only return 0 for FILE_DEFAULT */ 506 val = 1; 507 508 switch (m->factor_op) { 509 case FILE_FACTOR_OP_NONE: 510 break; 511 case FILE_FACTOR_OP_PLUS: 512 val += m->factor; 513 break; 514 case FILE_FACTOR_OP_MINUS: 515 val -= m->factor; 516 break; 517 case FILE_FACTOR_OP_TIMES: 518 val *= m->factor; 519 break; 520 case FILE_FACTOR_OP_DIV: 521 val /= m->factor; 522 break; 523 default: 524 abort(); 525 } 526 527 /* 528 * Magic entries with no description get a bonus because they depend 529 * on subsequent magic entries to print something. 530 */ 531 if (m->desc[0] == '\0') 532 val++; 533 return val; 534 } 535 536 /* 537 * Sort callback for sorting entries by "strength" (basically length) 538 */ 539 private int 540 apprentice_sort(const void *a, const void *b) 541 { 542 const struct magic_entry *ma = CAST(const struct magic_entry *, a); 543 const struct magic_entry *mb = CAST(const struct magic_entry *, b); 544 size_t sa = apprentice_magic_strength(ma->mp); 545 size_t sb = apprentice_magic_strength(mb->mp); 546 if (sa == sb) 547 return 0; 548 else if (sa > sb) 549 return -1; 550 else 551 return 1; 552 } 553 554 /* 555 * Shows sorted patterns list in the order which is used for the matching 556 */ 557 private void 558 apprentice_list(struct mlist *mlist, int mode) 559 { 560 uint32_t magindex = 0; 561 struct mlist *ml; 562 for (ml = mlist->next; ml != mlist; ml = ml->next) { 563 for (magindex = 0; magindex < ml->nmagic; magindex++) { 564 struct magic *m = &ml->magic[magindex]; 565 if ((m->flag & mode) != mode) { 566 /* Skip sub-tests */ 567 while (magindex + 1 < ml->nmagic && 568 ml->magic[magindex + 1].cont_level != 0) 569 ++magindex; 570 continue; /* Skip to next top-level test*/ 571 } 572 573 /* 574 * Try to iterate over the tree until we find item with 575 * description/mimetype. 576 */ 577 while (magindex + 1 < ml->nmagic && 578 ml->magic[magindex + 1].cont_level != 0 && 579 *ml->magic[magindex].desc == '\0' && 580 *ml->magic[magindex].mimetype == '\0') 581 magindex++; 582 583 printf("Strength = %3" SIZE_T_FORMAT "u : %s [%s]\n", 584 apprentice_magic_strength(m), 585 ml->magic[magindex].desc, 586 ml->magic[magindex].mimetype); 587 } 588 } 589 } 590 591 private void 592 set_test_type(struct magic *mstart, struct magic *m) 593 { 594 switch (m->type) { 595 case FILE_BYTE: 596 case FILE_SHORT: 597 case FILE_LONG: 598 case FILE_DATE: 599 case FILE_BESHORT: 600 case FILE_BELONG: 601 case FILE_BEDATE: 602 case FILE_LESHORT: 603 case FILE_LELONG: 604 case FILE_LEDATE: 605 case FILE_LDATE: 606 case FILE_BELDATE: 607 case FILE_LELDATE: 608 case FILE_MEDATE: 609 case FILE_MELDATE: 610 case FILE_MELONG: 611 case FILE_QUAD: 612 case FILE_LEQUAD: 613 case FILE_BEQUAD: 614 case FILE_QDATE: 615 case FILE_LEQDATE: 616 case FILE_BEQDATE: 617 case FILE_QLDATE: 618 case FILE_LEQLDATE: 619 case FILE_BEQLDATE: 620 case FILE_FLOAT: 621 case FILE_BEFLOAT: 622 case FILE_LEFLOAT: 623 case FILE_DOUBLE: 624 case FILE_BEDOUBLE: 625 case FILE_LEDOUBLE: 626 mstart->flag |= BINTEST; 627 break; 628 case FILE_STRING: 629 case FILE_PSTRING: 630 case FILE_BESTRING16: 631 case FILE_LESTRING16: 632 /* Allow text overrides */ 633 if (mstart->str_flags & STRING_TEXTTEST) 634 mstart->flag |= TEXTTEST; 635 else 636 mstart->flag |= BINTEST; 637 break; 638 case FILE_REGEX: 639 case FILE_SEARCH: 640 #ifndef COMPILE_ONLY 641 if (mstart->str_flags & STRING_BINTEST) 642 mstart->flag |= BINTEST; 643 if (mstart->str_flags & STRING_TEXTTEST) 644 mstart->flag |= TEXTTEST; 645 646 if (mstart->flag & (TEXTTEST|BINTEST)) 647 break; 648 649 /* binary test if pattern is not text */ 650 if (file_looks_utf8(m->value.us, (size_t)m->vallen, NULL, 651 NULL) <= 0) 652 mstart->flag |= BINTEST; 653 else 654 mstart->flag |= TEXTTEST; 655 #endif 656 break; 657 case FILE_DEFAULT: 658 /* can't deduce anything; we shouldn't see this at the 659 top level anyway */ 660 break; 661 case FILE_INVALID: 662 default: 663 /* invalid search type, but no need to complain here */ 664 break; 665 } 666 } 667 668 /* 669 * Load and parse one file. 670 */ 671 private void 672 load_1(struct magic_set *ms, int action, const char *fn, int *errs, 673 struct magic_entry **marray, uint32_t *marraycount) 674 { 675 size_t lineno = 0, llen = 0; 676 char *line = NULL; 677 ssize_t len; 678 679 FILE *f = fopen(ms->file = fn, "r"); 680 if (f == NULL) { 681 if (errno != ENOENT) 682 file_error(ms, errno, "cannot read magic file `%s'", 683 fn); 684 (*errs)++; 685 return; 686 } 687 688 /* read and parse this file */ 689 for (ms->line = 1; (len = getline(&line, &llen, f)) != -1; 690 ms->line++) { 691 if (len == 0) /* null line, garbage, etc */ 692 continue; 693 if (line[len - 1] == '\n') { 694 lineno++; 695 line[len - 1] = '\0'; /* delete newline */ 696 } 697 switch (line[0]) { 698 case '\0': /* empty, do not parse */ 699 case '#': /* comment, do not parse */ 700 continue; 701 case '!': 702 if (line[1] == ':') { 703 size_t i; 704 705 for (i = 0; bang[i].name != NULL; i++) { 706 if ((size_t)(len - 2) > bang[i].len && 707 memcmp(bang[i].name, line + 2, 708 bang[i].len) == 0) 709 break; 710 } 711 if (bang[i].name == NULL) { 712 file_error(ms, 0, 713 "Unknown !: entry `%s'", line); 714 (*errs)++; 715 continue; 716 } 717 if (*marraycount == 0) { 718 file_error(ms, 0, 719 "No current entry for :!%s type", 720 bang[i].name); 721 (*errs)++; 722 continue; 723 } 724 if ((*bang[i].fun)(ms, 725 &(*marray)[*marraycount - 1], 726 line + bang[i].len + 2) != 0) { 727 (*errs)++; 728 continue; 729 } 730 continue; 731 } 732 /*FALLTHROUGH*/ 733 default: 734 if (parse(ms, marray, marraycount, line, lineno, 735 action) != 0) 736 (*errs)++; 737 break; 738 } 739 } 740 free(line); 741 (void)fclose(f); 742 } 743 744 /* 745 * parse a file or directory of files 746 * const char *fn: name of magic file or directory 747 */ 748 private int 749 cmpstrp(const void *p1, const void *p2) 750 { 751 return strcmp(*(char *const *)p1, *(char *const *)p2); 752 } 753 754 private int 755 apprentice_load(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp, 756 const char *fn, int action) 757 { 758 int errs = 0; 759 struct magic_entry *marray; 760 uint32_t marraycount, i, mentrycount = 0, starttest; 761 size_t slen, files = 0, maxfiles = 0; 762 char **filearr = NULL, *mfn; 763 struct stat st; 764 DIR *dir; 765 struct dirent *d; 766 767 ms->flags |= MAGIC_CHECK; /* Enable checks for parsed files */ 768 769 maxmagic = MAXMAGIS; 770 if ((marray = CAST(struct magic_entry *, calloc(maxmagic, 771 sizeof(*marray)))) == NULL) { 772 file_oomem(ms, maxmagic * sizeof(*marray)); 773 return -1; 774 } 775 marraycount = 0; 776 777 /* print silly verbose header for USG compat. */ 778 if (action == FILE_CHECK) 779 (void)fprintf(stderr, "%s\n", usg_hdr); 780 781 /* load directory or file */ 782 if (stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) { 783 dir = opendir(fn); 784 if (!dir) { 785 errs++; 786 goto out; 787 } 788 while ((d = readdir(dir)) != NULL) { 789 if (asprintf(&mfn, "%s/%s", fn, d->d_name) < 0) { 790 file_oomem(ms, 791 strlen(fn) + strlen(d->d_name) + 2); 792 errs++; 793 closedir(dir); 794 goto out; 795 } 796 if (stat(mfn, &st) == -1 || !S_ISREG(st.st_mode)) { 797 free(mfn); 798 continue; 799 } 800 if (files >= maxfiles) { 801 size_t mlen; 802 maxfiles = (maxfiles + 1) * 2; 803 mlen = maxfiles * sizeof(*filearr); 804 if ((filearr = CAST(char **, 805 realloc(filearr, mlen))) == NULL) { 806 file_oomem(ms, mlen); 807 free(mfn); 808 closedir(dir); 809 errs++; 810 goto out; 811 } 812 } 813 filearr[files++] = mfn; 814 } 815 closedir(dir); 816 qsort(filearr, files, sizeof(*filearr), cmpstrp); 817 for (i = 0; i < files; i++) { 818 load_1(ms, action, filearr[i], &errs, &marray, 819 &marraycount); 820 free(filearr[i]); 821 } 822 free(filearr); 823 } else 824 load_1(ms, action, fn, &errs, &marray, &marraycount); 825 if (errs) 826 goto out; 827 828 /* Set types of tests */ 829 for (i = 0; i < marraycount; ) { 830 if (marray[i].mp->cont_level != 0) { 831 i++; 832 continue; 833 } 834 835 starttest = i; 836 do { 837 static const char text[] = "text"; 838 static const char binary[] = "binary"; 839 static const size_t len = sizeof(text); 840 set_test_type(marray[starttest].mp, marray[i].mp); 841 if ((ms->flags & MAGIC_DEBUG) == 0) 842 continue; 843 (void)fprintf(stderr, "%s%s%s: %s\n", 844 marray[i].mp->mimetype, 845 marray[i].mp->mimetype[0] == '\0' ? "" : "; ", 846 marray[i].mp->desc[0] ? marray[i].mp->desc : 847 "(no description)", 848 marray[i].mp->flag & BINTEST ? binary : text); 849 if (marray[i].mp->flag & BINTEST) { 850 char *p = strstr(marray[i].mp->desc, text); 851 if (p && (p == marray[i].mp->desc || 852 isspace((unsigned char)p[-1])) && 853 (p + len - marray[i].mp->desc == 854 MAXstring || (p[len] == '\0' || 855 isspace((unsigned char)p[len])))) 856 (void)fprintf(stderr, "*** Possible " 857 "binary test for text type\n"); 858 } 859 } while (++i < marraycount && marray[i].mp->cont_level != 0); 860 } 861 862 qsort(marray, marraycount, sizeof(*marray), apprentice_sort); 863 864 /* 865 * Make sure that any level 0 "default" line is last (if one exists). 866 */ 867 for (i = 0; i < marraycount; i++) { 868 if (marray[i].mp->cont_level == 0 && 869 marray[i].mp->type == FILE_DEFAULT) { 870 while (++i < marraycount) 871 if (marray[i].mp->cont_level == 0) 872 break; 873 if (i != marraycount) { 874 /* XXX - Ugh! */ 875 ms->line = marray[i].mp->lineno; 876 file_magwarn(ms, 877 "level 0 \"default\" did not sort last"); 878 } 879 break; 880 } 881 } 882 883 for (i = 0; i < marraycount; i++) 884 mentrycount += marray[i].cont_count; 885 886 slen = sizeof(**magicp) * mentrycount; 887 if ((*magicp = CAST(struct magic *, malloc(slen))) == NULL) { 888 file_oomem(ms, slen); 889 errs++; 890 goto out; 891 } 892 893 mentrycount = 0; 894 for (i = 0; i < marraycount; i++) { 895 (void)memcpy(*magicp + mentrycount, marray[i].mp, 896 marray[i].cont_count * sizeof(**magicp)); 897 mentrycount += marray[i].cont_count; 898 } 899 out: 900 for (i = 0; i < marraycount; i++) 901 free(marray[i].mp); 902 free(marray); 903 if (errs) { 904 *magicp = NULL; 905 *nmagicp = 0; 906 return errs; 907 } else { 908 *nmagicp = mentrycount; 909 return 0; 910 } 911 912 } 913 914 /* 915 * extend the sign bit if the comparison is to be signed 916 */ 917 protected uint64_t 918 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v) 919 { 920 if (!(m->flag & UNSIGNED)) { 921 switch(m->type) { 922 /* 923 * Do not remove the casts below. They are 924 * vital. When later compared with the data, 925 * the sign extension must have happened. 926 */ 927 case FILE_BYTE: 928 v = (char) v; 929 break; 930 case FILE_SHORT: 931 case FILE_BESHORT: 932 case FILE_LESHORT: 933 v = (short) v; 934 break; 935 case FILE_DATE: 936 case FILE_BEDATE: 937 case FILE_LEDATE: 938 case FILE_MEDATE: 939 case FILE_LDATE: 940 case FILE_BELDATE: 941 case FILE_LELDATE: 942 case FILE_MELDATE: 943 case FILE_LONG: 944 case FILE_BELONG: 945 case FILE_LELONG: 946 case FILE_MELONG: 947 case FILE_FLOAT: 948 case FILE_BEFLOAT: 949 case FILE_LEFLOAT: 950 v = (int32_t) v; 951 break; 952 case FILE_QUAD: 953 case FILE_BEQUAD: 954 case FILE_LEQUAD: 955 case FILE_QDATE: 956 case FILE_QLDATE: 957 case FILE_BEQDATE: 958 case FILE_BEQLDATE: 959 case FILE_LEQDATE: 960 case FILE_LEQLDATE: 961 case FILE_DOUBLE: 962 case FILE_BEDOUBLE: 963 case FILE_LEDOUBLE: 964 v = (int64_t) v; 965 break; 966 case FILE_STRING: 967 case FILE_PSTRING: 968 case FILE_BESTRING16: 969 case FILE_LESTRING16: 970 case FILE_REGEX: 971 case FILE_SEARCH: 972 case FILE_DEFAULT: 973 case FILE_INDIRECT: 974 break; 975 default: 976 if (ms->flags & MAGIC_CHECK) 977 file_magwarn(ms, "cannot happen: m->type=%d\n", 978 m->type); 979 return ~0U; 980 } 981 } 982 return v; 983 } 984 985 private int 986 string_modifier_check(struct magic_set *ms, struct magic *m) 987 { 988 if ((ms->flags & MAGIC_CHECK) == 0) 989 return 0; 990 991 if (m->type != FILE_PSTRING && (m->str_flags & PSTRING_LEN) != 0) { 992 file_magwarn(ms, 993 "'/BHhLl' modifiers are only allowed for pascal strings\n"); 994 return -1; 995 } 996 switch (m->type) { 997 case FILE_BESTRING16: 998 case FILE_LESTRING16: 999 if (m->str_flags != 0) { 1000 file_magwarn(ms, 1001 "no modifiers allowed for 16-bit strings\n"); 1002 return -1; 1003 } 1004 break; 1005 case FILE_STRING: 1006 case FILE_PSTRING: 1007 if ((m->str_flags & REGEX_OFFSET_START) != 0) { 1008 file_magwarn(ms, 1009 "'/%c' only allowed on regex and search\n", 1010 CHAR_REGEX_OFFSET_START); 1011 return -1; 1012 } 1013 break; 1014 case FILE_SEARCH: 1015 if (m->str_range == 0) { 1016 file_magwarn(ms, 1017 "missing range; defaulting to %d\n", 1018 STRING_DEFAULT_RANGE); 1019 m->str_range = STRING_DEFAULT_RANGE; 1020 return -1; 1021 } 1022 break; 1023 case FILE_REGEX: 1024 if ((m->str_flags & STRING_COMPACT_WHITESPACE) != 0) { 1025 file_magwarn(ms, "'/%c' not allowed on regex\n", 1026 CHAR_COMPACT_WHITESPACE); 1027 return -1; 1028 } 1029 if ((m->str_flags & STRING_COMPACT_OPTIONAL_WHITESPACE) != 0) { 1030 file_magwarn(ms, "'/%c' not allowed on regex\n", 1031 CHAR_COMPACT_OPTIONAL_WHITESPACE); 1032 return -1; 1033 } 1034 break; 1035 default: 1036 file_magwarn(ms, "coding error: m->type=%d\n", 1037 m->type); 1038 return -1; 1039 } 1040 return 0; 1041 } 1042 1043 private int 1044 get_op(char c) 1045 { 1046 switch (c) { 1047 case '&': 1048 return FILE_OPAND; 1049 case '|': 1050 return FILE_OPOR; 1051 case '^': 1052 return FILE_OPXOR; 1053 case '+': 1054 return FILE_OPADD; 1055 case '-': 1056 return FILE_OPMINUS; 1057 case '*': 1058 return FILE_OPMULTIPLY; 1059 case '/': 1060 return FILE_OPDIVIDE; 1061 case '%': 1062 return FILE_OPMODULO; 1063 default: 1064 return -1; 1065 } 1066 } 1067 1068 #ifdef ENABLE_CONDITIONALS 1069 private int 1070 get_cond(const char *l, const char **t) 1071 { 1072 static const struct cond_tbl_s { 1073 char name[8]; 1074 size_t len; 1075 int cond; 1076 } cond_tbl[] = { 1077 { "if", 2, COND_IF }, 1078 { "elif", 4, COND_ELIF }, 1079 { "else", 4, COND_ELSE }, 1080 { "", 0, COND_NONE }, 1081 }; 1082 const struct cond_tbl_s *p; 1083 1084 for (p = cond_tbl; p->len; p++) { 1085 if (strncmp(l, p->name, p->len) == 0 && 1086 isspace((unsigned char)l[p->len])) { 1087 if (t) 1088 *t = l + p->len; 1089 break; 1090 } 1091 } 1092 return p->cond; 1093 } 1094 1095 private int 1096 check_cond(struct magic_set *ms, int cond, uint32_t cont_level) 1097 { 1098 int last_cond; 1099 last_cond = ms->c.li[cont_level].last_cond; 1100 1101 switch (cond) { 1102 case COND_IF: 1103 if (last_cond != COND_NONE && last_cond != COND_ELIF) { 1104 if (ms->flags & MAGIC_CHECK) 1105 file_magwarn(ms, "syntax error: `if'"); 1106 return -1; 1107 } 1108 last_cond = COND_IF; 1109 break; 1110 1111 case COND_ELIF: 1112 if (last_cond != COND_IF && last_cond != COND_ELIF) { 1113 if (ms->flags & MAGIC_CHECK) 1114 file_magwarn(ms, "syntax error: `elif'"); 1115 return -1; 1116 } 1117 last_cond = COND_ELIF; 1118 break; 1119 1120 case COND_ELSE: 1121 if (last_cond != COND_IF && last_cond != COND_ELIF) { 1122 if (ms->flags & MAGIC_CHECK) 1123 file_magwarn(ms, "syntax error: `else'"); 1124 return -1; 1125 } 1126 last_cond = COND_NONE; 1127 break; 1128 1129 case COND_NONE: 1130 last_cond = COND_NONE; 1131 break; 1132 } 1133 1134 ms->c.li[cont_level].last_cond = last_cond; 1135 return 0; 1136 } 1137 #endif /* ENABLE_CONDITIONALS */ 1138 1139 /* 1140 * parse one line from magic file, put into magic[index++] if valid 1141 */ 1142 private int 1143 parse(struct magic_set *ms, struct magic_entry **mentryp, uint32_t *nmentryp, 1144 const char *line, size_t lineno, int action) 1145 { 1146 #ifdef ENABLE_CONDITIONALS 1147 static uint32_t last_cont_level = 0; 1148 #endif 1149 size_t i; 1150 struct magic_entry *me; 1151 struct magic *m; 1152 const char *l = line; 1153 char *t; 1154 int op; 1155 uint32_t cont_level; 1156 1157 cont_level = 0; 1158 1159 while (*l == '>') { 1160 ++l; /* step over */ 1161 cont_level++; 1162 } 1163 #ifdef ENABLE_CONDITIONALS 1164 if (cont_level == 0 || cont_level > last_cont_level) 1165 if (file_check_mem(ms, cont_level) == -1) 1166 return -1; 1167 last_cont_level = cont_level; 1168 #endif 1169 1170 #define ALLOC_CHUNK (size_t)10 1171 #define ALLOC_INCR (size_t)200 1172 1173 if (cont_level != 0) { 1174 if (*nmentryp == 0) { 1175 file_error(ms, 0, "No current entry for continuation"); 1176 return -1; 1177 } 1178 me = &(*mentryp)[*nmentryp - 1]; 1179 if (me->cont_count == me->max_count) { 1180 struct magic *nm; 1181 size_t cnt = me->max_count + ALLOC_CHUNK; 1182 if ((nm = CAST(struct magic *, realloc(me->mp, 1183 sizeof(*nm) * cnt))) == NULL) { 1184 file_oomem(ms, sizeof(*nm) * cnt); 1185 return -1; 1186 } 1187 me->mp = m = nm; 1188 me->max_count = CAST(uint32_t, cnt); 1189 } 1190 m = &me->mp[me->cont_count++]; 1191 (void)memset(m, 0, sizeof(*m)); 1192 m->cont_level = cont_level; 1193 } else { 1194 if (*nmentryp == maxmagic) { 1195 struct magic_entry *mp; 1196 1197 maxmagic += ALLOC_INCR; 1198 if ((mp = CAST(struct magic_entry *, 1199 realloc(*mentryp, sizeof(*mp) * maxmagic))) == 1200 NULL) { 1201 file_oomem(ms, sizeof(*mp) * maxmagic); 1202 return -1; 1203 } 1204 (void)memset(&mp[*nmentryp], 0, sizeof(*mp) * 1205 ALLOC_INCR); 1206 *mentryp = mp; 1207 } 1208 me = &(*mentryp)[*nmentryp]; 1209 if (me->mp == NULL) { 1210 size_t len = sizeof(*m) * ALLOC_CHUNK; 1211 if ((m = CAST(struct magic *, malloc(len))) == NULL) { 1212 file_oomem(ms, len); 1213 return -1; 1214 } 1215 me->mp = m; 1216 me->max_count = ALLOC_CHUNK; 1217 } else 1218 m = me->mp; 1219 (void)memset(m, 0, sizeof(*m)); 1220 m->factor_op = FILE_FACTOR_OP_NONE; 1221 m->cont_level = 0; 1222 me->cont_count = 1; 1223 } 1224 m->lineno = CAST(uint32_t, lineno); 1225 1226 if (*l == '&') { /* m->cont_level == 0 checked below. */ 1227 ++l; /* step over */ 1228 m->flag |= OFFADD; 1229 } 1230 if (*l == '(') { 1231 ++l; /* step over */ 1232 m->flag |= INDIR; 1233 if (m->flag & OFFADD) 1234 m->flag = (m->flag & ~OFFADD) | INDIROFFADD; 1235 1236 if (*l == '&') { /* m->cont_level == 0 checked below */ 1237 ++l; /* step over */ 1238 m->flag |= OFFADD; 1239 } 1240 } 1241 /* Indirect offsets are not valid at level 0. */ 1242 if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD))) 1243 if (ms->flags & MAGIC_CHECK) 1244 file_magwarn(ms, "relative offset at level 0"); 1245 1246 /* get offset, then skip over it */ 1247 m->offset = (uint32_t)strtoul(l, &t, 0); 1248 if (l == t) 1249 if (ms->flags & MAGIC_CHECK) 1250 file_magwarn(ms, "offset `%s' invalid", l); 1251 l = t; 1252 1253 if (m->flag & INDIR) { 1254 m->in_type = FILE_LONG; 1255 m->in_offset = 0; 1256 /* 1257 * read [.lbs][+-]nnnnn) 1258 */ 1259 if (*l == '.') { 1260 l++; 1261 switch (*l) { 1262 case 'l': 1263 m->in_type = FILE_LELONG; 1264 break; 1265 case 'L': 1266 m->in_type = FILE_BELONG; 1267 break; 1268 case 'm': 1269 m->in_type = FILE_MELONG; 1270 break; 1271 case 'h': 1272 case 's': 1273 m->in_type = FILE_LESHORT; 1274 break; 1275 case 'H': 1276 case 'S': 1277 m->in_type = FILE_BESHORT; 1278 break; 1279 case 'c': 1280 case 'b': 1281 case 'C': 1282 case 'B': 1283 m->in_type = FILE_BYTE; 1284 break; 1285 case 'e': 1286 case 'f': 1287 case 'g': 1288 m->in_type = FILE_LEDOUBLE; 1289 break; 1290 case 'E': 1291 case 'F': 1292 case 'G': 1293 m->in_type = FILE_BEDOUBLE; 1294 break; 1295 case 'i': 1296 m->in_type = FILE_LEID3; 1297 break; 1298 case 'I': 1299 m->in_type = FILE_BEID3; 1300 break; 1301 default: 1302 if (ms->flags & MAGIC_CHECK) 1303 file_magwarn(ms, 1304 "indirect offset type `%c' invalid", 1305 *l); 1306 break; 1307 } 1308 l++; 1309 } 1310 1311 m->in_op = 0; 1312 if (*l == '~') { 1313 m->in_op |= FILE_OPINVERSE; 1314 l++; 1315 } 1316 if ((op = get_op(*l)) != -1) { 1317 m->in_op |= op; 1318 l++; 1319 } 1320 if (*l == '(') { 1321 m->in_op |= FILE_OPINDIRECT; 1322 l++; 1323 } 1324 if (isdigit((unsigned char)*l) || *l == '-') { 1325 m->in_offset = (int32_t)strtol(l, &t, 0); 1326 if (l == t) 1327 if (ms->flags & MAGIC_CHECK) 1328 file_magwarn(ms, 1329 "in_offset `%s' invalid", l); 1330 l = t; 1331 } 1332 if (*l++ != ')' || 1333 ((m->in_op & FILE_OPINDIRECT) && *l++ != ')')) 1334 if (ms->flags & MAGIC_CHECK) 1335 file_magwarn(ms, 1336 "missing ')' in indirect offset"); 1337 } 1338 EATAB; 1339 1340 #ifdef ENABLE_CONDITIONALS 1341 m->cond = get_cond(l, &l); 1342 if (check_cond(ms, m->cond, cont_level) == -1) 1343 return -1; 1344 1345 EATAB; 1346 #endif 1347 1348 if (*l == 'u') { 1349 ++l; 1350 m->flag |= UNSIGNED; 1351 } 1352 1353 m->type = get_type(l, &l); 1354 if (m->type == FILE_INVALID) { 1355 if (ms->flags & MAGIC_CHECK) 1356 file_magwarn(ms, "type `%s' invalid", l); 1357 return -1; 1358 } 1359 1360 /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */ 1361 /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */ 1362 1363 m->mask_op = 0; 1364 if (*l == '~') { 1365 if (!IS_STRING(m->type)) 1366 m->mask_op |= FILE_OPINVERSE; 1367 else if (ms->flags & MAGIC_CHECK) 1368 file_magwarn(ms, "'~' invalid for string types"); 1369 ++l; 1370 } 1371 m->str_range = 0; 1372 m->str_flags = m->type == FILE_PSTRING ? PSTRING_1_LE : 0; 1373 if ((op = get_op(*l)) != -1) { 1374 if (!IS_STRING(m->type)) { 1375 uint64_t val; 1376 ++l; 1377 m->mask_op |= op; 1378 val = (uint64_t)strtoull(l, &t, 0); 1379 l = t; 1380 m->num_mask = file_signextend(ms, m, val); 1381 eatsize(&l); 1382 } 1383 else if (op == FILE_OPDIVIDE) { 1384 int have_range = 0; 1385 while (!isspace((unsigned char)*++l)) { 1386 switch (*l) { 1387 case '0': case '1': case '2': 1388 case '3': case '4': case '5': 1389 case '6': case '7': case '8': 1390 case '9': 1391 if (have_range && 1392 (ms->flags & MAGIC_CHECK)) 1393 file_magwarn(ms, 1394 "multiple ranges"); 1395 have_range = 1; 1396 m->str_range = CAST(uint32_t, 1397 strtoul(l, &t, 0)); 1398 if (m->str_range == 0) 1399 file_magwarn(ms, 1400 "zero range"); 1401 l = t - 1; 1402 break; 1403 case CHAR_COMPACT_WHITESPACE: 1404 m->str_flags |= 1405 STRING_COMPACT_WHITESPACE; 1406 break; 1407 case CHAR_COMPACT_OPTIONAL_WHITESPACE: 1408 m->str_flags |= 1409 STRING_COMPACT_OPTIONAL_WHITESPACE; 1410 break; 1411 case CHAR_IGNORE_LOWERCASE: 1412 m->str_flags |= STRING_IGNORE_LOWERCASE; 1413 break; 1414 case CHAR_IGNORE_UPPERCASE: 1415 m->str_flags |= STRING_IGNORE_UPPERCASE; 1416 break; 1417 case CHAR_REGEX_OFFSET_START: 1418 m->str_flags |= REGEX_OFFSET_START; 1419 break; 1420 case CHAR_BINTEST: 1421 m->str_flags |= STRING_BINTEST; 1422 break; 1423 case CHAR_TEXTTEST: 1424 m->str_flags |= STRING_TEXTTEST; 1425 break; 1426 case CHAR_PSTRING_1_LE: 1427 if (m->type != FILE_PSTRING) 1428 goto bad; 1429 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_1_LE; 1430 break; 1431 case CHAR_PSTRING_2_BE: 1432 if (m->type != FILE_PSTRING) 1433 goto bad; 1434 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_2_BE; 1435 break; 1436 case CHAR_PSTRING_2_LE: 1437 if (m->type != FILE_PSTRING) 1438 goto bad; 1439 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_2_LE; 1440 break; 1441 case CHAR_PSTRING_4_BE: 1442 if (m->type != FILE_PSTRING) 1443 goto bad; 1444 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_4_BE; 1445 break; 1446 case CHAR_PSTRING_4_LE: 1447 if (m->type != FILE_PSTRING) 1448 goto bad; 1449 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_4_LE; 1450 break; 1451 case CHAR_PSTRING_LENGTH_INCLUDES_ITSELF: 1452 if (m->type != FILE_PSTRING) 1453 goto bad; 1454 m->str_flags |= PSTRING_LENGTH_INCLUDES_ITSELF; 1455 break; 1456 default: 1457 bad: 1458 if (ms->flags & MAGIC_CHECK) 1459 file_magwarn(ms, 1460 "string extension `%c' " 1461 "invalid", *l); 1462 return -1; 1463 } 1464 /* allow multiple '/' for readability */ 1465 if (l[1] == '/' && 1466 !isspace((unsigned char)l[2])) 1467 l++; 1468 } 1469 if (string_modifier_check(ms, m) == -1) 1470 return -1; 1471 } 1472 else { 1473 if (ms->flags & MAGIC_CHECK) 1474 file_magwarn(ms, "invalid string op: %c", *t); 1475 return -1; 1476 } 1477 } 1478 /* 1479 * We used to set mask to all 1's here, instead let's just not do 1480 * anything if mask = 0 (unless you have a better idea) 1481 */ 1482 EATAB; 1483 1484 switch (*l) { 1485 case '>': 1486 case '<': 1487 m->reln = *l; 1488 ++l; 1489 if (*l == '=') { 1490 if (ms->flags & MAGIC_CHECK) { 1491 file_magwarn(ms, "%c= not supported", 1492 m->reln); 1493 return -1; 1494 } 1495 ++l; 1496 } 1497 break; 1498 /* Old-style anding: "0 byte &0x80 dynamically linked" */ 1499 case '&': 1500 case '^': 1501 case '=': 1502 m->reln = *l; 1503 ++l; 1504 if (*l == '=') { 1505 /* HP compat: ignore &= etc. */ 1506 ++l; 1507 } 1508 break; 1509 case '!': 1510 m->reln = *l; 1511 ++l; 1512 break; 1513 default: 1514 m->reln = '='; /* the default relation */ 1515 if (*l == 'x' && ((isascii((unsigned char)l[1]) && 1516 isspace((unsigned char)l[1])) || !l[1])) { 1517 m->reln = *l; 1518 ++l; 1519 } 1520 break; 1521 } 1522 /* 1523 * Grab the value part, except for an 'x' reln. 1524 */ 1525 if (m->reln != 'x' && getvalue(ms, m, &l, action)) 1526 return -1; 1527 1528 /* 1529 * TODO finish this macro and start using it! 1530 * #define offsetcheck {if (offset > HOWMANY-1) 1531 * magwarn("offset too big"); } 1532 */ 1533 1534 /* 1535 * Now get last part - the description 1536 */ 1537 EATAB; 1538 if (l[0] == '\b') { 1539 ++l; 1540 m->flag |= NOSPACE; 1541 } else if ((l[0] == '\\') && (l[1] == 'b')) { 1542 ++l; 1543 ++l; 1544 m->flag |= NOSPACE; 1545 } 1546 for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); ) 1547 continue; 1548 if (i == sizeof(m->desc)) { 1549 m->desc[sizeof(m->desc) - 1] = '\0'; 1550 if (ms->flags & MAGIC_CHECK) 1551 file_magwarn(ms, "description `%s' truncated", m->desc); 1552 } 1553 1554 /* 1555 * We only do this check while compiling, or if any of the magic 1556 * files were not compiled. 1557 */ 1558 if (ms->flags & MAGIC_CHECK) { 1559 if (check_format(ms, m) == -1) 1560 return -1; 1561 } 1562 #ifndef COMPILE_ONLY 1563 if (action == FILE_CHECK) { 1564 file_mdump(m); 1565 } 1566 #endif 1567 m->mimetype[0] = '\0'; /* initialise MIME type to none */ 1568 if (m->cont_level == 0) 1569 ++(*nmentryp); /* make room for next */ 1570 return 0; 1571 } 1572 1573 /* 1574 * parse a STRENGTH annotation line from magic file, put into magic[index - 1] 1575 * if valid 1576 */ 1577 private int 1578 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line) 1579 { 1580 const char *l = line; 1581 char *el; 1582 unsigned long factor; 1583 struct magic *m = &me->mp[0]; 1584 1585 if (m->factor_op != FILE_FACTOR_OP_NONE) { 1586 file_magwarn(ms, 1587 "Current entry already has a strength type: %c %d", 1588 m->factor_op, m->factor); 1589 return -1; 1590 } 1591 EATAB; 1592 switch (*l) { 1593 case FILE_FACTOR_OP_NONE: 1594 case FILE_FACTOR_OP_PLUS: 1595 case FILE_FACTOR_OP_MINUS: 1596 case FILE_FACTOR_OP_TIMES: 1597 case FILE_FACTOR_OP_DIV: 1598 m->factor_op = *l++; 1599 break; 1600 default: 1601 file_magwarn(ms, "Unknown factor op `%c'", *l); 1602 return -1; 1603 } 1604 EATAB; 1605 factor = strtoul(l, &el, 0); 1606 if (factor > 255) { 1607 file_magwarn(ms, "Too large factor `%lu'", factor); 1608 goto out; 1609 } 1610 if (*el && !isspace((unsigned char)*el)) { 1611 file_magwarn(ms, "Bad factor `%s'", l); 1612 goto out; 1613 } 1614 m->factor = (uint8_t)factor; 1615 if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) { 1616 file_magwarn(ms, "Cannot have factor op `%c' and factor %u", 1617 m->factor_op, m->factor); 1618 goto out; 1619 } 1620 return 0; 1621 out: 1622 m->factor_op = FILE_FACTOR_OP_NONE; 1623 m->factor = 0; 1624 return -1; 1625 } 1626 1627 /* 1628 * Parse an Apple CREATOR/TYPE annotation from magic file and put it into 1629 * magic[index - 1] 1630 */ 1631 private int 1632 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line) 1633 { 1634 size_t i; 1635 const char *l = line; 1636 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1]; 1637 1638 if (m->apple[0] != '\0') { 1639 file_magwarn(ms, "Current entry already has a APPLE type " 1640 "`%.8s', new type `%s'", m->mimetype, l); 1641 return -1; 1642 } 1643 1644 EATAB; 1645 for (i = 0; *l && ((isascii((unsigned char)*l) && 1646 isalnum((unsigned char)*l)) || strchr("-+/.", *l)) && 1647 i < sizeof(m->apple); m->apple[i++] = *l++) 1648 continue; 1649 if (i == sizeof(m->apple) && *l) { 1650 /* We don't need to NUL terminate here, printing handles it */ 1651 if (ms->flags & MAGIC_CHECK) 1652 file_magwarn(ms, "APPLE type `%s' truncated %" 1653 SIZE_T_FORMAT "u", line, i); 1654 } 1655 1656 if (i > 0) 1657 return 0; 1658 else 1659 return -1; 1660 } 1661 1662 /* 1663 * parse a MIME annotation line from magic file, put into magic[index - 1] 1664 * if valid 1665 */ 1666 private int 1667 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line) 1668 { 1669 size_t i; 1670 const char *l = line; 1671 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1]; 1672 1673 if (m->mimetype[0] != '\0') { 1674 file_magwarn(ms, "Current entry already has a MIME type `%s'," 1675 " new type `%s'", m->mimetype, l); 1676 return -1; 1677 } 1678 1679 EATAB; 1680 for (i = 0; *l && ((isascii((unsigned char)*l) && 1681 isalnum((unsigned char)*l)) || strchr("-+/.", *l)) && 1682 i < sizeof(m->mimetype); m->mimetype[i++] = *l++) 1683 continue; 1684 if (i == sizeof(m->mimetype)) { 1685 m->mimetype[sizeof(m->mimetype) - 1] = '\0'; 1686 if (ms->flags & MAGIC_CHECK) 1687 file_magwarn(ms, "MIME type `%s' truncated %" 1688 SIZE_T_FORMAT "u", m->mimetype, i); 1689 } else 1690 m->mimetype[i] = '\0'; 1691 1692 if (i > 0) 1693 return 0; 1694 else 1695 return -1; 1696 } 1697 1698 private int 1699 check_format_type(const char *ptr, int type) 1700 { 1701 int quad = 0; 1702 if (*ptr == '\0') { 1703 /* Missing format string; bad */ 1704 return -1; 1705 } 1706 1707 switch (type) { 1708 case FILE_FMT_QUAD: 1709 quad = 1; 1710 /*FALLTHROUGH*/ 1711 case FILE_FMT_NUM: 1712 if (*ptr == '-') 1713 ptr++; 1714 if (*ptr == '.') 1715 ptr++; 1716 while (isdigit((unsigned char)*ptr)) ptr++; 1717 if (*ptr == '.') 1718 ptr++; 1719 while (isdigit((unsigned char)*ptr)) ptr++; 1720 if (quad) { 1721 if (*ptr++ != 'l') 1722 return -1; 1723 if (*ptr++ != 'l') 1724 return -1; 1725 } 1726 1727 switch (*ptr++) { 1728 case 'l': 1729 switch (*ptr++) { 1730 case 'i': 1731 case 'd': 1732 case 'u': 1733 case 'x': 1734 case 'X': 1735 return 0; 1736 default: 1737 return -1; 1738 } 1739 1740 case 'h': 1741 switch (*ptr++) { 1742 case 'h': 1743 switch (*ptr++) { 1744 case 'i': 1745 case 'd': 1746 case 'u': 1747 case 'x': 1748 case 'X': 1749 return 0; 1750 default: 1751 return -1; 1752 } 1753 case 'd': 1754 return 0; 1755 default: 1756 return -1; 1757 } 1758 1759 case 'i': 1760 case 'c': 1761 case 'd': 1762 case 'u': 1763 case 'x': 1764 case 'X': 1765 return 0; 1766 1767 default: 1768 return -1; 1769 } 1770 1771 case FILE_FMT_FLOAT: 1772 case FILE_FMT_DOUBLE: 1773 if (*ptr == '-') 1774 ptr++; 1775 if (*ptr == '.') 1776 ptr++; 1777 while (isdigit((unsigned char)*ptr)) ptr++; 1778 if (*ptr == '.') 1779 ptr++; 1780 while (isdigit((unsigned char)*ptr)) ptr++; 1781 1782 switch (*ptr++) { 1783 case 'e': 1784 case 'E': 1785 case 'f': 1786 case 'F': 1787 case 'g': 1788 case 'G': 1789 return 0; 1790 1791 default: 1792 return -1; 1793 } 1794 1795 1796 case FILE_FMT_STR: 1797 if (*ptr == '-') 1798 ptr++; 1799 while (isdigit((unsigned char )*ptr)) 1800 ptr++; 1801 if (*ptr == '.') { 1802 ptr++; 1803 while (isdigit((unsigned char )*ptr)) 1804 ptr++; 1805 } 1806 1807 switch (*ptr++) { 1808 case 's': 1809 return 0; 1810 default: 1811 return -1; 1812 } 1813 1814 default: 1815 /* internal error */ 1816 abort(); 1817 } 1818 /*NOTREACHED*/ 1819 return -1; 1820 } 1821 1822 /* 1823 * Check that the optional printf format in description matches 1824 * the type of the magic. 1825 */ 1826 private int 1827 check_format(struct magic_set *ms, struct magic *m) 1828 { 1829 char *ptr; 1830 1831 for (ptr = m->desc; *ptr; ptr++) 1832 if (*ptr == '%') 1833 break; 1834 if (*ptr == '\0') { 1835 /* No format string; ok */ 1836 return 1; 1837 } 1838 1839 assert(file_nformats == file_nnames); 1840 1841 if (m->type >= file_nformats) { 1842 file_magwarn(ms, "Internal error inconsistency between " 1843 "m->type and format strings"); 1844 return -1; 1845 } 1846 if (file_formats[m->type] == FILE_FMT_NONE) { 1847 file_magwarn(ms, "No format string for `%s' with description " 1848 "`%s'", m->desc, file_names[m->type]); 1849 return -1; 1850 } 1851 1852 ptr++; 1853 if (check_format_type(ptr, file_formats[m->type]) == -1) { 1854 /* 1855 * TODO: this error message is unhelpful if the format 1856 * string is not one character long 1857 */ 1858 file_magwarn(ms, "Printf format `%c' is not valid for type " 1859 "`%s' in description `%s'", *ptr ? *ptr : '?', 1860 file_names[m->type], m->desc); 1861 return -1; 1862 } 1863 1864 for (; *ptr; ptr++) { 1865 if (*ptr == '%') { 1866 file_magwarn(ms, 1867 "Too many format strings (should have at most one) " 1868 "for `%s' with description `%s'", 1869 file_names[m->type], m->desc); 1870 return -1; 1871 } 1872 } 1873 return 0; 1874 } 1875 1876 /* 1877 * Read a numeric value from a pointer, into the value union of a magic 1878 * pointer, according to the magic type. Update the string pointer to point 1879 * just after the number read. Return 0 for success, non-zero for failure. 1880 */ 1881 private int 1882 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action) 1883 { 1884 switch (m->type) { 1885 case FILE_BESTRING16: 1886 case FILE_LESTRING16: 1887 case FILE_STRING: 1888 case FILE_PSTRING: 1889 case FILE_REGEX: 1890 case FILE_SEARCH: 1891 *p = getstr(ms, m, *p, action == FILE_COMPILE); 1892 if (*p == NULL) { 1893 if (ms->flags & MAGIC_CHECK) 1894 file_magwarn(ms, "cannot get string from `%s'", 1895 m->value.s); 1896 return -1; 1897 } 1898 return 0; 1899 case FILE_FLOAT: 1900 case FILE_BEFLOAT: 1901 case FILE_LEFLOAT: 1902 if (m->reln != 'x') { 1903 char *ep; 1904 #if defined(HAVE_STRTOF) && !defined(COMPILE_ONLY) 1905 m->value.f = strtof(*p, &ep); 1906 #else 1907 m->value.f = (float)strtod(*p, &ep); 1908 #endif 1909 *p = ep; 1910 } 1911 return 0; 1912 case FILE_DOUBLE: 1913 case FILE_BEDOUBLE: 1914 case FILE_LEDOUBLE: 1915 if (m->reln != 'x') { 1916 char *ep; 1917 m->value.d = strtod(*p, &ep); 1918 *p = ep; 1919 } 1920 return 0; 1921 default: 1922 if (m->reln != 'x') { 1923 char *ep; 1924 m->value.q = file_signextend(ms, m, 1925 (uint64_t)strtoull(*p, &ep, 0)); 1926 *p = ep; 1927 eatsize(p); 1928 } 1929 return 0; 1930 } 1931 } 1932 1933 /* 1934 * Convert a string containing C character escapes. Stop at an unescaped 1935 * space or tab. 1936 * Copy the converted version to "m->value.s", and the length in m->vallen. 1937 * Return updated scan pointer as function result. Warn if set. 1938 */ 1939 private const char * 1940 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn) 1941 { 1942 const char *origs = s; 1943 char *p = m->value.s; 1944 size_t plen = sizeof(m->value.s); 1945 char *origp = p; 1946 char *pmax = p + plen - 1; 1947 int c; 1948 int val; 1949 1950 while ((c = *s++) != '\0') { 1951 if (isspace((unsigned char) c)) 1952 break; 1953 if (p >= pmax) { 1954 file_error(ms, 0, "string too long: `%s'", origs); 1955 return NULL; 1956 } 1957 if (c == '\\') { 1958 switch(c = *s++) { 1959 1960 case '\0': 1961 if (warn) 1962 file_magwarn(ms, "incomplete escape"); 1963 goto out; 1964 1965 case '\t': 1966 if (warn) { 1967 file_magwarn(ms, 1968 "escaped tab found, use \\t instead"); 1969 warn = 0; /* already did */ 1970 } 1971 /*FALLTHROUGH*/ 1972 default: 1973 if (warn) { 1974 if (isprint((unsigned char)c)) { 1975 /* Allow escaping of 1976 * ``relations'' */ 1977 if (strchr("<>&^=!", c) == NULL 1978 && (m->type != FILE_REGEX || 1979 strchr("[]().*?^$|{}", c) 1980 == NULL)) { 1981 file_magwarn(ms, "no " 1982 "need to escape " 1983 "`%c'", c); 1984 } 1985 } else { 1986 file_magwarn(ms, 1987 "unknown escape sequence: " 1988 "\\%03o", c); 1989 } 1990 } 1991 /*FALLTHROUGH*/ 1992 /* space, perhaps force people to use \040? */ 1993 case ' ': 1994 #if 0 1995 /* 1996 * Other things people escape, but shouldn't need to, 1997 * so we disallow them 1998 */ 1999 case '\'': 2000 case '"': 2001 case '?': 2002 #endif 2003 /* Relations */ 2004 case '>': 2005 case '<': 2006 case '&': 2007 case '^': 2008 case '=': 2009 case '!': 2010 /* and baskslash itself */ 2011 case '\\': 2012 *p++ = (char) c; 2013 break; 2014 2015 case 'a': 2016 *p++ = '\a'; 2017 break; 2018 2019 case 'b': 2020 *p++ = '\b'; 2021 break; 2022 2023 case 'f': 2024 *p++ = '\f'; 2025 break; 2026 2027 case 'n': 2028 *p++ = '\n'; 2029 break; 2030 2031 case 'r': 2032 *p++ = '\r'; 2033 break; 2034 2035 case 't': 2036 *p++ = '\t'; 2037 break; 2038 2039 case 'v': 2040 *p++ = '\v'; 2041 break; 2042 2043 /* \ and up to 3 octal digits */ 2044 case '0': 2045 case '1': 2046 case '2': 2047 case '3': 2048 case '4': 2049 case '5': 2050 case '6': 2051 case '7': 2052 val = c - '0'; 2053 c = *s++; /* try for 2 */ 2054 if (c >= '0' && c <= '7') { 2055 val = (val << 3) | (c - '0'); 2056 c = *s++; /* try for 3 */ 2057 if (c >= '0' && c <= '7') 2058 val = (val << 3) | (c-'0'); 2059 else 2060 --s; 2061 } 2062 else 2063 --s; 2064 *p++ = (char)val; 2065 break; 2066 2067 /* \x and up to 2 hex digits */ 2068 case 'x': 2069 val = 'x'; /* Default if no digits */ 2070 c = hextoint(*s++); /* Get next char */ 2071 if (c >= 0) { 2072 val = c; 2073 c = hextoint(*s++); 2074 if (c >= 0) 2075 val = (val << 4) + c; 2076 else 2077 --s; 2078 } else 2079 --s; 2080 *p++ = (char)val; 2081 break; 2082 } 2083 } else 2084 *p++ = (char)c; 2085 } 2086 out: 2087 *p = '\0'; 2088 m->vallen = CAST(unsigned char, (p - origp)); 2089 if (m->type == FILE_PSTRING) 2090 m->vallen += (unsigned char)file_pstring_length_size(m); 2091 return s; 2092 } 2093 2094 2095 /* Single hex char to int; -1 if not a hex char. */ 2096 private int 2097 hextoint(int c) 2098 { 2099 if (!isascii((unsigned char) c)) 2100 return -1; 2101 if (isdigit((unsigned char) c)) 2102 return c - '0'; 2103 if ((c >= 'a') && (c <= 'f')) 2104 return c + 10 - 'a'; 2105 if (( c>= 'A') && (c <= 'F')) 2106 return c + 10 - 'A'; 2107 return -1; 2108 } 2109 2110 2111 /* 2112 * Print a string containing C character escapes. 2113 */ 2114 protected void 2115 file_showstr(FILE *fp, const char *s, size_t len) 2116 { 2117 char c; 2118 2119 for (;;) { 2120 if (len == ~0U) { 2121 c = *s++; 2122 if (c == '\0') 2123 break; 2124 } 2125 else { 2126 if (len-- == 0) 2127 break; 2128 c = *s++; 2129 } 2130 if (c >= 040 && c <= 0176) /* TODO isprint && !iscntrl */ 2131 (void) fputc(c, fp); 2132 else { 2133 (void) fputc('\\', fp); 2134 switch (c) { 2135 case '\a': 2136 (void) fputc('a', fp); 2137 break; 2138 2139 case '\b': 2140 (void) fputc('b', fp); 2141 break; 2142 2143 case '\f': 2144 (void) fputc('f', fp); 2145 break; 2146 2147 case '\n': 2148 (void) fputc('n', fp); 2149 break; 2150 2151 case '\r': 2152 (void) fputc('r', fp); 2153 break; 2154 2155 case '\t': 2156 (void) fputc('t', fp); 2157 break; 2158 2159 case '\v': 2160 (void) fputc('v', fp); 2161 break; 2162 2163 default: 2164 (void) fprintf(fp, "%.3o", c & 0377); 2165 break; 2166 } 2167 } 2168 } 2169 } 2170 2171 /* 2172 * eatsize(): Eat the size spec from a number [eg. 10UL] 2173 */ 2174 private void 2175 eatsize(const char **p) 2176 { 2177 const char *l = *p; 2178 2179 if (LOWCASE(*l) == 'u') 2180 l++; 2181 2182 switch (LOWCASE(*l)) { 2183 case 'l': /* long */ 2184 case 's': /* short */ 2185 case 'h': /* short */ 2186 case 'b': /* char/byte */ 2187 case 'c': /* char/byte */ 2188 l++; 2189 /*FALLTHROUGH*/ 2190 default: 2191 break; 2192 } 2193 2194 *p = l; 2195 } 2196 2197 /* 2198 * handle a compiled file. 2199 */ 2200 private int 2201 apprentice_map(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp, 2202 const char *fn) 2203 { 2204 int fd; 2205 struct stat st; 2206 uint32_t *ptr; 2207 uint32_t version; 2208 int needsbyteswap; 2209 char *dbname = NULL; 2210 void *mm = NULL; 2211 2212 dbname = mkdbname(ms, fn, 0); 2213 if (dbname == NULL) 2214 goto error2; 2215 2216 if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1) 2217 goto error2; 2218 2219 if (fstat(fd, &st) == -1) { 2220 file_error(ms, errno, "cannot stat `%s'", dbname); 2221 goto error1; 2222 } 2223 if (st.st_size < 8) { 2224 file_error(ms, 0, "file `%s' is too small", dbname); 2225 goto error1; 2226 } 2227 2228 #ifdef QUICK 2229 if ((mm = mmap(0, (size_t)st.st_size, PROT_READ|PROT_WRITE, 2230 MAP_PRIVATE|MAP_FILE, fd, (off_t)0)) == MAP_FAILED) { 2231 file_error(ms, errno, "cannot map `%s'", dbname); 2232 goto error1; 2233 } 2234 #define RET 2 2235 #else 2236 if ((mm = CAST(void *, malloc((size_t)st.st_size))) == NULL) { 2237 file_oomem(ms, (size_t)st.st_size); 2238 goto error1; 2239 } 2240 if (read(fd, mm, (size_t)st.st_size) != (ssize_t)st.st_size) { 2241 file_badread(ms); 2242 goto error1; 2243 } 2244 #define RET 1 2245 #endif 2246 *magicp = CAST(struct magic *, mm); 2247 (void)close(fd); 2248 fd = -1; 2249 ptr = (uint32_t *)(void *)*magicp; 2250 if (*ptr != MAGICNO) { 2251 if (swap4(*ptr) != MAGICNO) { 2252 file_error(ms, 0, "bad magic in `%s'", dbname); 2253 goto error1; 2254 } 2255 needsbyteswap = 1; 2256 } else 2257 needsbyteswap = 0; 2258 if (needsbyteswap) 2259 version = swap4(ptr[1]); 2260 else 2261 version = ptr[1]; 2262 if (version != VERSIONNO) { 2263 file_error(ms, 0, "File %s supports only version %d magic " 2264 "files. `%s' is version %d", VERSION, 2265 VERSIONNO, dbname, version); 2266 goto error1; 2267 } 2268 *nmagicp = (uint32_t)(st.st_size / sizeof(struct magic)); 2269 if (*nmagicp > 0) 2270 (*nmagicp)--; 2271 (*magicp)++; 2272 if (needsbyteswap) 2273 byteswap(*magicp, *nmagicp); 2274 free(dbname); 2275 return RET; 2276 2277 error1: 2278 if (fd != -1) 2279 (void)close(fd); 2280 if (mm) { 2281 #ifdef QUICK 2282 (void)munmap((void *)mm, (size_t)st.st_size); 2283 #else 2284 free(mm); 2285 #endif 2286 } else { 2287 *magicp = NULL; 2288 *nmagicp = 0; 2289 } 2290 error2: 2291 free(dbname); 2292 return -1; 2293 } 2294 2295 private const uint32_t ar[] = { 2296 MAGICNO, VERSIONNO 2297 }; 2298 /* 2299 * handle an mmaped file. 2300 */ 2301 private int 2302 apprentice_compile(struct magic_set *ms, struct magic **magicp, 2303 uint32_t *nmagicp, const char *fn) 2304 { 2305 int fd = -1; 2306 char *dbname; 2307 int rv = -1; 2308 2309 dbname = mkdbname(ms, fn, 1); 2310 2311 if (dbname == NULL) 2312 goto out; 2313 2314 if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) { 2315 file_error(ms, errno, "cannot open `%s'", dbname); 2316 goto out; 2317 } 2318 2319 if (write(fd, ar, sizeof(ar)) != (ssize_t)sizeof(ar)) { 2320 file_error(ms, errno, "error writing `%s'", dbname); 2321 goto out; 2322 } 2323 2324 if (lseek(fd, (off_t)sizeof(struct magic), SEEK_SET) 2325 != sizeof(struct magic)) { 2326 file_error(ms, errno, "error seeking `%s'", dbname); 2327 goto out; 2328 } 2329 2330 if (write(fd, *magicp, (sizeof(struct magic) * *nmagicp)) 2331 != (ssize_t)(sizeof(struct magic) * *nmagicp)) { 2332 file_error(ms, errno, "error writing `%s'", dbname); 2333 goto out; 2334 } 2335 2336 if (fd != -1) 2337 (void)close(fd); 2338 rv = 0; 2339 out: 2340 free(dbname); 2341 return rv; 2342 } 2343 2344 private const char ext[] = ".mgc"; 2345 /* 2346 * make a dbname 2347 */ 2348 private char * 2349 mkdbname(struct magic_set *ms, const char *fn, int strip) 2350 { 2351 const char *p, *q; 2352 char *buf; 2353 2354 if (strip) { 2355 if ((p = strrchr(fn, '/')) != NULL) 2356 fn = ++p; 2357 } 2358 2359 for (q = fn; *q; q++) 2360 continue; 2361 /* Look for .mgc */ 2362 for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--) 2363 if (*p != *q) 2364 break; 2365 2366 /* Did not find .mgc, restore q */ 2367 if (p >= ext) 2368 while (*q) 2369 q++; 2370 2371 q++; 2372 /* Compatibility with old code that looked in .mime */ 2373 if (ms->flags & MAGIC_MIME) { 2374 asprintf(&buf, "%.*s.mime%s", (int)(q - fn), fn, ext); 2375 if (access(buf, R_OK) != -1) { 2376 ms->flags &= MAGIC_MIME_TYPE; 2377 return buf; 2378 } 2379 free(buf); 2380 } 2381 asprintf(&buf, "%.*s%s", (int)(q - fn), fn, ext); 2382 2383 /* Compatibility with old code that looked in .mime */ 2384 if (strstr(p, ".mime") != NULL) 2385 ms->flags &= MAGIC_MIME_TYPE; 2386 return buf; 2387 } 2388 2389 /* 2390 * Byteswap an mmap'ed file if needed 2391 */ 2392 private void 2393 byteswap(struct magic *magic, uint32_t nmagic) 2394 { 2395 uint32_t i; 2396 for (i = 0; i < nmagic; i++) 2397 bs1(&magic[i]); 2398 } 2399 2400 /* 2401 * swap a short 2402 */ 2403 private uint16_t 2404 swap2(uint16_t sv) 2405 { 2406 uint16_t rv; 2407 uint8_t *s = (uint8_t *)(void *)&sv; 2408 uint8_t *d = (uint8_t *)(void *)&rv; 2409 d[0] = s[1]; 2410 d[1] = s[0]; 2411 return rv; 2412 } 2413 2414 /* 2415 * swap an int 2416 */ 2417 private uint32_t 2418 swap4(uint32_t sv) 2419 { 2420 uint32_t rv; 2421 uint8_t *s = (uint8_t *)(void *)&sv; 2422 uint8_t *d = (uint8_t *)(void *)&rv; 2423 d[0] = s[3]; 2424 d[1] = s[2]; 2425 d[2] = s[1]; 2426 d[3] = s[0]; 2427 return rv; 2428 } 2429 2430 /* 2431 * swap a quad 2432 */ 2433 private uint64_t 2434 swap8(uint64_t sv) 2435 { 2436 uint64_t rv; 2437 uint8_t *s = (uint8_t *)(void *)&sv; 2438 uint8_t *d = (uint8_t *)(void *)&rv; 2439 #if 0 2440 d[0] = s[3]; 2441 d[1] = s[2]; 2442 d[2] = s[1]; 2443 d[3] = s[0]; 2444 d[4] = s[7]; 2445 d[5] = s[6]; 2446 d[6] = s[5]; 2447 d[7] = s[4]; 2448 #else 2449 d[0] = s[7]; 2450 d[1] = s[6]; 2451 d[2] = s[5]; 2452 d[3] = s[4]; 2453 d[4] = s[3]; 2454 d[5] = s[2]; 2455 d[6] = s[1]; 2456 d[7] = s[0]; 2457 #endif 2458 return rv; 2459 } 2460 2461 /* 2462 * byteswap a single magic entry 2463 */ 2464 private void 2465 bs1(struct magic *m) 2466 { 2467 m->cont_level = swap2(m->cont_level); 2468 m->offset = swap4((uint32_t)m->offset); 2469 m->in_offset = swap4((uint32_t)m->in_offset); 2470 m->lineno = swap4((uint32_t)m->lineno); 2471 if (IS_STRING(m->type)) { 2472 m->str_range = swap4(m->str_range); 2473 m->str_flags = swap4(m->str_flags); 2474 } 2475 else { 2476 m->value.q = swap8(m->value.q); 2477 m->num_mask = swap8(m->num_mask); 2478 } 2479 } 2480 2481 protected size_t 2482 file_pstring_length_size(const struct magic *m) 2483 { 2484 switch (m->str_flags & PSTRING_LEN) { 2485 case PSTRING_1_LE: 2486 return 1; 2487 case PSTRING_2_LE: 2488 case PSTRING_2_BE: 2489 return 2; 2490 case PSTRING_4_LE: 2491 case PSTRING_4_BE: 2492 return 4; 2493 default: 2494 abort(); /* Impossible */ 2495 return 1; 2496 } 2497 } 2498 protected size_t 2499 file_pstring_get_length(const struct magic *m, const char *s) 2500 { 2501 size_t len = 0; 2502 2503 switch (m->str_flags & PSTRING_LEN) { 2504 case PSTRING_1_LE: 2505 len = *s; 2506 break; 2507 case PSTRING_2_LE: 2508 len = (s[1] << 8) | s[0]; 2509 break; 2510 case PSTRING_2_BE: 2511 len = (s[0] << 8) | s[1]; 2512 break; 2513 case PSTRING_4_LE: 2514 len = (s[3] << 24) | (s[2] << 16) | (s[1] << 8) | s[0]; 2515 break; 2516 case PSTRING_4_BE: 2517 len = (s[0] << 24) | (s[1] << 16) | (s[2] << 8) | s[3]; 2518 break; 2519 default: 2520 abort(); /* Impossible */ 2521 } 2522 2523 if (m->str_flags & PSTRING_LENGTH_INCLUDES_ITSELF) 2524 len -= file_pstring_length_size(m); 2525 2526 return len; 2527 } 2528