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 1.165 2011/01/16 19:30:36 rrt Exp $") 36 #endif /* lint */ 37 38 #include "magic.h" 39 #include "patchlevel.h" 40 #include <stdlib.h> 41 #ifdef HAVE_UNISTD_H 42 #include <unistd.h> 43 #endif 44 #include <string.h> 45 #include <assert.h> 46 #include <ctype.h> 47 #include <fcntl.h> 48 #ifdef QUICK 49 #include <sys/mman.h> 50 #endif 51 #include <dirent.h> 52 53 #define EATAB {while (isascii((unsigned char) *l) && \ 54 isspace((unsigned char) *l)) ++l;} 55 #define LOWCASE(l) (isupper((unsigned char) (l)) ? \ 56 tolower((unsigned char) (l)) : (l)) 57 /* 58 * Work around a bug in headers on Digital Unix. 59 * At least confirmed for: OSF1 V4.0 878 60 */ 61 #if defined(__osf__) && defined(__DECC) 62 #ifdef MAP_FAILED 63 #undef MAP_FAILED 64 #endif 65 #endif 66 67 #ifndef MAP_FAILED 68 #define MAP_FAILED (void *) -1 69 #endif 70 71 #ifndef MAP_FILE 72 #define MAP_FILE 0 73 #endif 74 75 #ifndef MAXPATHLEN 76 #define MAXPATHLEN 1024 77 #endif 78 79 struct magic_entry { 80 struct magic *mp; 81 uint32_t cont_count; 82 uint32_t max_count; 83 }; 84 85 int file_formats[FILE_NAMES_SIZE]; 86 const size_t file_nformats = FILE_NAMES_SIZE; 87 const char *file_names[FILE_NAMES_SIZE]; 88 const size_t file_nnames = FILE_NAMES_SIZE; 89 90 private int getvalue(struct magic_set *ms, struct magic *, const char **, int); 91 private int hextoint(int); 92 private const char *getstr(struct magic_set *, struct magic *, const char *, 93 int); 94 private int parse(struct magic_set *, struct magic_entry **, uint32_t *, 95 const char *, size_t, int); 96 private void eatsize(const char **); 97 private int apprentice_1(struct magic_set *, const char *, int, struct mlist *); 98 private size_t apprentice_magic_strength(const struct magic *); 99 private int apprentice_sort(const void *, const void *); 100 private void apprentice_list(struct mlist *, int ); 101 private int apprentice_load(struct magic_set *, struct magic **, uint32_t *, 102 const char *, int); 103 private void byteswap(struct magic *, uint32_t); 104 private void bs1(struct magic *); 105 private uint16_t swap2(uint16_t); 106 private uint32_t swap4(uint32_t); 107 private uint64_t swap8(uint64_t); 108 private char *mkdbname(struct magic_set *, const char *, int); 109 private int apprentice_map(struct magic_set *, struct magic **, uint32_t *, 110 const char *); 111 private int apprentice_compile(struct magic_set *, struct magic **, uint32_t *, 112 const char *); 113 private int check_format_type(const char *, int); 114 private int check_format(struct magic_set *, struct magic *); 115 private int get_op(char); 116 private int parse_mime(struct magic_set *, struct magic_entry *, const char *); 117 private int parse_strength(struct magic_set *, struct magic_entry *, const char *); 118 private int parse_apple(struct magic_set *, struct magic_entry *, const char *); 119 120 121 private size_t maxmagic = 0; 122 private size_t magicsize = sizeof(struct magic); 123 124 private const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc"; 125 126 private struct { 127 const char *name; 128 size_t len; 129 int (*fun)(struct magic_set *, struct magic_entry *, const char *); 130 } bang[] = { 131 #define DECLARE_FIELD(name) { # name, sizeof(# name) - 1, parse_ ## name } 132 DECLARE_FIELD(mime), 133 DECLARE_FIELD(apple), 134 DECLARE_FIELD(strength), 135 #undef DECLARE_FIELD 136 { NULL, 0, NULL } 137 }; 138 139 #ifdef COMPILE_ONLY 140 141 int main(int, char *[]); 142 143 int 144 main(int argc, char *argv[]) 145 { 146 int ret; 147 struct magic_set *ms; 148 char *progname; 149 150 if ((progname = strrchr(argv[0], '/')) != NULL) 151 progname++; 152 else 153 progname = argv[0]; 154 155 if (argc != 2) { 156 (void)fprintf(stderr, "Usage: %s file\n", progname); 157 return 1; 158 } 159 160 if ((ms = magic_open(MAGIC_CHECK)) == NULL) { 161 (void)fprintf(stderr, "%s: %s\n", progname, strerror(errno)); 162 return 1; 163 } 164 ret = magic_compile(ms, argv[1]) == -1 ? 1 : 0; 165 if (ret == 1) 166 (void)fprintf(stderr, "%s: %s\n", progname, magic_error(ms)); 167 magic_close(ms); 168 return ret; 169 } 170 #endif /* COMPILE_ONLY */ 171 172 static const struct type_tbl_s { 173 const char name[16]; 174 const size_t len; 175 const int type; 176 const int format; 177 } type_tbl[] = { 178 # define XX(s) s, (sizeof(s) - 1) 179 # define XX_NULL "", 0 180 { XX("byte"), FILE_BYTE, FILE_FMT_NUM }, 181 { XX("short"), FILE_SHORT, FILE_FMT_NUM }, 182 { XX("default"), FILE_DEFAULT, FILE_FMT_STR }, 183 { XX("long"), FILE_LONG, FILE_FMT_NUM }, 184 { XX("string"), FILE_STRING, FILE_FMT_STR }, 185 { XX("date"), FILE_DATE, FILE_FMT_STR }, 186 { XX("beshort"), FILE_BESHORT, FILE_FMT_NUM }, 187 { XX("belong"), FILE_BELONG, FILE_FMT_NUM }, 188 { XX("bedate"), FILE_BEDATE, FILE_FMT_STR }, 189 { XX("leshort"), FILE_LESHORT, FILE_FMT_NUM }, 190 { XX("lelong"), FILE_LELONG, FILE_FMT_NUM }, 191 { XX("ledate"), FILE_LEDATE, FILE_FMT_STR }, 192 { XX("pstring"), FILE_PSTRING, FILE_FMT_STR }, 193 { XX("ldate"), FILE_LDATE, FILE_FMT_STR }, 194 { XX("beldate"), FILE_BELDATE, FILE_FMT_STR }, 195 { XX("leldate"), FILE_LELDATE, FILE_FMT_STR }, 196 { XX("regex"), FILE_REGEX, FILE_FMT_STR }, 197 { XX("bestring16"), FILE_BESTRING16, FILE_FMT_STR }, 198 { XX("lestring16"), FILE_LESTRING16, FILE_FMT_STR }, 199 { XX("search"), FILE_SEARCH, FILE_FMT_STR }, 200 { XX("medate"), FILE_MEDATE, FILE_FMT_STR }, 201 { XX("meldate"), FILE_MELDATE, FILE_FMT_STR }, 202 { XX("melong"), FILE_MELONG, FILE_FMT_NUM }, 203 { XX("quad"), FILE_QUAD, FILE_FMT_QUAD }, 204 { XX("lequad"), FILE_LEQUAD, FILE_FMT_QUAD }, 205 { XX("bequad"), FILE_BEQUAD, FILE_FMT_QUAD }, 206 { XX("qdate"), FILE_QDATE, FILE_FMT_STR }, 207 { XX("leqdate"), FILE_LEQDATE, FILE_FMT_STR }, 208 { XX("beqdate"), FILE_BEQDATE, FILE_FMT_STR }, 209 { XX("qldate"), FILE_QLDATE, FILE_FMT_STR }, 210 { XX("leqldate"), FILE_LEQLDATE, FILE_FMT_STR }, 211 { XX("beqldate"), FILE_BEQLDATE, FILE_FMT_STR }, 212 { XX("float"), FILE_FLOAT, FILE_FMT_FLOAT }, 213 { XX("befloat"), FILE_BEFLOAT, FILE_FMT_FLOAT }, 214 { XX("lefloat"), FILE_LEFLOAT, FILE_FMT_FLOAT }, 215 { XX("double"), FILE_DOUBLE, FILE_FMT_DOUBLE }, 216 { XX("bedouble"), FILE_BEDOUBLE, FILE_FMT_DOUBLE }, 217 { XX("ledouble"), FILE_LEDOUBLE, FILE_FMT_DOUBLE }, 218 { XX("leid3"), FILE_LEID3, FILE_FMT_NUM }, 219 { XX("beid3"), FILE_BEID3, FILE_FMT_NUM }, 220 { XX("indirect"), FILE_INDIRECT, FILE_FMT_NONE }, 221 { XX_NULL, FILE_INVALID, FILE_FMT_NONE }, 222 # undef XX 223 # undef XX_NULL 224 }; 225 226 private int 227 get_type(const char *l, const char **t) 228 { 229 const struct type_tbl_s *p; 230 231 for (p = type_tbl; p->len; p++) { 232 if (strncmp(l, p->name, p->len) == 0) { 233 if (t) 234 *t = l + p->len; 235 break; 236 } 237 } 238 return p->type; 239 } 240 241 private void 242 init_file_tables(void) 243 { 244 static int done = 0; 245 const struct type_tbl_s *p; 246 247 if (done) 248 return; 249 done++; 250 251 for (p = type_tbl; p->len; p++) { 252 assert(p->type < FILE_NAMES_SIZE); 253 file_names[p->type] = p->name; 254 file_formats[p->type] = p->format; 255 } 256 } 257 258 /* 259 * Handle one file or directory. 260 */ 261 private int 262 apprentice_1(struct magic_set *ms, const char *fn, int action, 263 struct mlist *mlist) 264 { 265 struct magic *magic = NULL; 266 uint32_t nmagic = 0; 267 struct mlist *ml; 268 int rv = -1; 269 int mapped; 270 271 if (magicsize != FILE_MAGICSIZE) { 272 file_error(ms, 0, "magic element size %lu != %lu", 273 (unsigned long)sizeof(*magic), 274 (unsigned long)FILE_MAGICSIZE); 275 return -1; 276 } 277 278 if (action == FILE_COMPILE) { 279 rv = apprentice_load(ms, &magic, &nmagic, fn, action); 280 if (rv != 0) 281 return -1; 282 rv = apprentice_compile(ms, &magic, &nmagic, fn); 283 free(magic); 284 return rv; 285 } 286 287 #ifndef COMPILE_ONLY 288 if ((rv = apprentice_map(ms, &magic, &nmagic, fn)) == -1) { 289 if (ms->flags & MAGIC_CHECK) 290 file_magwarn(ms, "using regular magic file `%s'", fn); 291 rv = apprentice_load(ms, &magic, &nmagic, fn, action); 292 if (rv != 0) 293 return -1; 294 } 295 296 mapped = rv; 297 298 if (magic == NULL) { 299 file_delmagic(magic, mapped, nmagic); 300 return -1; 301 } 302 303 if ((ml = CAST(struct mlist *, malloc(sizeof(*ml)))) == NULL) { 304 file_delmagic(magic, mapped, nmagic); 305 file_oomem(ms, sizeof(*ml)); 306 return -1; 307 } 308 309 ml->magic = magic; 310 ml->nmagic = nmagic; 311 ml->mapped = mapped; 312 313 mlist->prev->next = ml; 314 ml->prev = mlist->prev; 315 ml->next = mlist; 316 mlist->prev = ml; 317 318 if (action == FILE_LIST) { 319 printf("Binary patterns:\n"); 320 apprentice_list(mlist, BINTEST); 321 printf("Text patterns:\n"); 322 apprentice_list(mlist, TEXTTEST); 323 } 324 325 return 0; 326 #endif /* COMPILE_ONLY */ 327 } 328 329 protected void 330 file_delmagic(struct magic *p, int type, size_t entries) 331 { 332 if (p == NULL) 333 return; 334 switch (type) { 335 case 2: 336 #ifdef QUICK 337 p--; 338 (void)munmap((void *)p, sizeof(*p) * (entries + 1)); 339 break; 340 #else 341 (void)&entries; 342 abort(); 343 /*NOTREACHED*/ 344 #endif 345 case 1: 346 p--; 347 /*FALLTHROUGH*/ 348 case 0: 349 free(p); 350 break; 351 default: 352 abort(); 353 } 354 } 355 356 /* const char *fn: list of magic files and directories */ 357 protected struct mlist * 358 file_apprentice(struct magic_set *ms, const char *fn, int action) 359 { 360 char *p, *mfn; 361 int file_err, errs = -1; 362 struct mlist *mlist; 363 364 if ((fn = magic_getpath(fn, action)) == NULL) 365 return NULL; 366 367 init_file_tables(); 368 369 if ((mfn = strdup(fn)) == NULL) { 370 file_oomem(ms, strlen(fn)); 371 return NULL; 372 } 373 fn = mfn; 374 375 if ((mlist = CAST(struct mlist *, malloc(sizeof(*mlist)))) == NULL) { 376 free(mfn); 377 file_oomem(ms, sizeof(*mlist)); 378 return NULL; 379 } 380 mlist->next = mlist->prev = mlist; 381 382 while (fn) { 383 p = strchr(fn, PATHSEP); 384 if (p) 385 *p++ = '\0'; 386 if (*fn == '\0') 387 break; 388 file_err = apprentice_1(ms, fn, action, mlist); 389 errs = MAX(errs, file_err); 390 fn = p; 391 } 392 if (errs == -1) { 393 free(mfn); 394 free(mlist); 395 mlist = NULL; 396 file_error(ms, 0, "could not find any magic files!"); 397 return NULL; 398 } 399 free(mfn); 400 return mlist; 401 } 402 403 /* 404 * Get weight of this magic entry, for sorting purposes. 405 */ 406 private size_t 407 apprentice_magic_strength(const struct magic *m) 408 { 409 #define MULT 10 410 size_t val = 2 * MULT; /* baseline strength */ 411 412 switch (m->type) { 413 case FILE_DEFAULT: /* make sure this sorts last */ 414 if (m->factor_op != FILE_FACTOR_OP_NONE) 415 abort(); 416 return 0; 417 418 case FILE_BYTE: 419 val += 1 * MULT; 420 break; 421 422 case FILE_SHORT: 423 case FILE_LESHORT: 424 case FILE_BESHORT: 425 val += 2 * MULT; 426 break; 427 428 case FILE_LONG: 429 case FILE_LELONG: 430 case FILE_BELONG: 431 case FILE_MELONG: 432 val += 4 * MULT; 433 break; 434 435 case FILE_PSTRING: 436 case FILE_STRING: 437 val += m->vallen * MULT; 438 break; 439 440 case FILE_BESTRING16: 441 case FILE_LESTRING16: 442 val += m->vallen * MULT / 2; 443 break; 444 445 case FILE_SEARCH: 446 case FILE_REGEX: 447 val += m->vallen * MAX(MULT / m->vallen, 1); 448 break; 449 450 case FILE_DATE: 451 case FILE_LEDATE: 452 case FILE_BEDATE: 453 case FILE_MEDATE: 454 case FILE_LDATE: 455 case FILE_LELDATE: 456 case FILE_BELDATE: 457 case FILE_MELDATE: 458 case FILE_FLOAT: 459 case FILE_BEFLOAT: 460 case FILE_LEFLOAT: 461 val += 4 * MULT; 462 break; 463 464 case FILE_QUAD: 465 case FILE_BEQUAD: 466 case FILE_LEQUAD: 467 case FILE_QDATE: 468 case FILE_LEQDATE: 469 case FILE_BEQDATE: 470 case FILE_QLDATE: 471 case FILE_LEQLDATE: 472 case FILE_BEQLDATE: 473 case FILE_DOUBLE: 474 case FILE_BEDOUBLE: 475 case FILE_LEDOUBLE: 476 val += 8 * MULT; 477 break; 478 479 default: 480 val = 0; 481 (void)fprintf(stderr, "Bad type %d\n", m->type); 482 abort(); 483 } 484 485 switch (m->reln) { 486 case 'x': /* matches anything penalize */ 487 case '!': /* matches almost anything penalize */ 488 val = 0; 489 break; 490 491 case '=': /* Exact match, prefer */ 492 val += MULT; 493 break; 494 495 case '>': 496 case '<': /* comparison match reduce strength */ 497 val -= 2 * MULT; 498 break; 499 500 case '^': 501 case '&': /* masking bits, we could count them too */ 502 val -= MULT; 503 break; 504 505 default: 506 (void)fprintf(stderr, "Bad relation %c\n", m->reln); 507 abort(); 508 } 509 510 if (val == 0) /* ensure we only return 0 for FILE_DEFAULT */ 511 val = 1; 512 513 switch (m->factor_op) { 514 case FILE_FACTOR_OP_NONE: 515 break; 516 case FILE_FACTOR_OP_PLUS: 517 val += m->factor; 518 break; 519 case FILE_FACTOR_OP_MINUS: 520 val -= m->factor; 521 break; 522 case FILE_FACTOR_OP_TIMES: 523 val *= m->factor; 524 break; 525 case FILE_FACTOR_OP_DIV: 526 val /= m->factor; 527 break; 528 default: 529 abort(); 530 } 531 532 /* 533 * Magic entries with no description get a bonus because they depend 534 * on subsequent magic entries to print something. 535 */ 536 if (m->desc[0] == '\0') 537 val++; 538 return val; 539 } 540 541 /* 542 * Sort callback for sorting entries by "strength" (basically length) 543 */ 544 private int 545 apprentice_sort(const void *a, const void *b) 546 { 547 const struct magic_entry *ma = CAST(const struct magic_entry *, a); 548 const struct magic_entry *mb = CAST(const struct magic_entry *, b); 549 size_t sa = apprentice_magic_strength(ma->mp); 550 size_t sb = apprentice_magic_strength(mb->mp); 551 if (sa == sb) 552 return 0; 553 else if (sa > sb) 554 return -1; 555 else 556 return 1; 557 } 558 559 /* 560 * Shows sorted patterns list in the order which is used for the matching 561 */ 562 private void 563 apprentice_list(struct mlist *mlist, int mode) 564 { 565 uint32_t magindex = 0; 566 struct mlist *ml; 567 for (ml = mlist->next; ml != mlist; ml = ml->next) { 568 for (magindex = 0; magindex < ml->nmagic; magindex++) { 569 struct magic *m = &ml->magic[magindex]; 570 if ((m->flag & mode) != mode) { 571 /* Skip sub-tests */ 572 while (magindex + 1 < ml->nmagic && 573 ml->magic[magindex + 1].cont_level != 0) 574 ++magindex; 575 continue; /* Skip to next top-level test*/ 576 } 577 578 /* 579 * Try to iterate over the tree until we find item with 580 * description/mimetype. 581 */ 582 while (magindex + 1 < ml->nmagic && 583 ml->magic[magindex + 1].cont_level != 0 && 584 *ml->magic[magindex].desc == '\0' && 585 *ml->magic[magindex].mimetype == '\0') 586 magindex++; 587 588 printf("Strength = %3" SIZE_T_FORMAT "u : %s [%s]\n", 589 apprentice_magic_strength(m), 590 ml->magic[magindex].desc, 591 ml->magic[magindex].mimetype); 592 } 593 } 594 } 595 596 private void 597 set_test_type(struct magic *mstart, struct magic *m) 598 { 599 switch (m->type) { 600 case FILE_BYTE: 601 case FILE_SHORT: 602 case FILE_LONG: 603 case FILE_DATE: 604 case FILE_BESHORT: 605 case FILE_BELONG: 606 case FILE_BEDATE: 607 case FILE_LESHORT: 608 case FILE_LELONG: 609 case FILE_LEDATE: 610 case FILE_LDATE: 611 case FILE_BELDATE: 612 case FILE_LELDATE: 613 case FILE_MEDATE: 614 case FILE_MELDATE: 615 case FILE_MELONG: 616 case FILE_QUAD: 617 case FILE_LEQUAD: 618 case FILE_BEQUAD: 619 case FILE_QDATE: 620 case FILE_LEQDATE: 621 case FILE_BEQDATE: 622 case FILE_QLDATE: 623 case FILE_LEQLDATE: 624 case FILE_BEQLDATE: 625 case FILE_FLOAT: 626 case FILE_BEFLOAT: 627 case FILE_LEFLOAT: 628 case FILE_DOUBLE: 629 case FILE_BEDOUBLE: 630 case FILE_LEDOUBLE: 631 case FILE_STRING: 632 case FILE_PSTRING: 633 case FILE_BESTRING16: 634 case FILE_LESTRING16: 635 /* Allow text overrides */ 636 if (mstart->str_flags & STRING_TEXTTEST) 637 mstart->flag |= TEXTTEST; 638 else 639 mstart->flag |= BINTEST; 640 break; 641 case FILE_REGEX: 642 case FILE_SEARCH: 643 #ifndef COMPILE_ONLY 644 if (mstart->str_flags & STRING_BINTEST) 645 mstart->flag |= BINTEST; 646 if (mstart->str_flags & STRING_TEXTTEST) 647 mstart->flag |= TEXTTEST; 648 649 if (mstart->flag & (TEXTTEST|BINTEST)) 650 break; 651 652 /* binary test if pattern is not text */ 653 if (file_looks_utf8(m->value.us, (size_t)m->vallen, NULL, 654 NULL) <= 0) 655 mstart->flag |= BINTEST; 656 else 657 mstart->flag |= TEXTTEST; 658 #endif 659 break; 660 case FILE_DEFAULT: 661 /* can't deduce anything; we shouldn't see this at the 662 top level anyway */ 663 break; 664 case FILE_INVALID: 665 default: 666 /* invalid search type, but no need to complain here */ 667 break; 668 } 669 } 670 671 /* 672 * Load and parse one file. 673 */ 674 private void 675 load_1(struct magic_set *ms, int action, const char *fn, int *errs, 676 struct magic_entry **marray, uint32_t *marraycount) 677 { 678 char line[BUFSIZ]; 679 size_t lineno = 0; 680 FILE *f = fopen(ms->file = fn, "r"); 681 if (f == NULL) { 682 if (errno != ENOENT) 683 file_error(ms, errno, "cannot read magic file `%s'", 684 fn); 685 (*errs)++; 686 } else { 687 /* read and parse this file */ 688 for (ms->line = 1; 689 fgets(line, CAST(int, sizeof(line)), f) != NULL; 690 ms->line++) { 691 size_t len; 692 len = strlen(line); 693 if (len == 0) /* null line, garbage, etc */ 694 continue; 695 if (line[len - 1] == '\n') { 696 lineno++; 697 line[len - 1] = '\0'; /* delete newline */ 698 } 699 if (line[0] == '\0') /* empty, do not parse */ 700 continue; 701 if (line[0] == '#') /* comment, do not parse */ 702 continue; 703 if (line[0] == '!' && line[1] == ':') { 704 size_t i; 705 706 for (i = 0; bang[i].name != NULL; i++) { 707 if (len - 2 > bang[i].len && 708 memcmp(bang[i].name, line + 2, 709 bang[i].len) == 0) 710 break; 711 } 712 if (bang[i].name == NULL) { 713 file_error(ms, 0, 714 "Unknown !: entry `%s'", line); 715 (*errs)++; 716 continue; 717 } 718 if (*marraycount == 0) { 719 file_error(ms, 0, 720 "No current entry for :!%s type", 721 bang[i].name); 722 (*errs)++; 723 continue; 724 } 725 if ((*bang[i].fun)(ms, 726 &(*marray)[*marraycount - 1], 727 line + bang[i].len + 2) != 0) { 728 (*errs)++; 729 continue; 730 } 731 continue; 732 } 733 if (parse(ms, marray, marraycount, line, lineno, 734 action) != 0) 735 (*errs)++; 736 } 737 738 (void)fclose(f); 739 } 740 } 741 742 /* 743 * parse a file or directory of files 744 * const char *fn: name of magic file or directory 745 */ 746 private int 747 cmpstrp(const void *p1, const void *p2) 748 { 749 return strcmp(*(char *const *)p1, *(char *const *)p2); 750 } 751 752 private int 753 apprentice_load(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp, 754 const char *fn, int action) 755 { 756 int errs = 0; 757 struct magic_entry *marray; 758 uint32_t marraycount, i, mentrycount = 0, starttest; 759 size_t slen, files = 0, maxfiles = 0; 760 char subfn[MAXPATHLEN], **filearr = NULL, *mfn; 761 struct stat st; 762 DIR *dir; 763 struct dirent *d; 764 765 ms->flags |= MAGIC_CHECK; /* Enable checks for parsed files */ 766 767 maxmagic = MAXMAGIS; 768 if ((marray = CAST(struct magic_entry *, calloc(maxmagic, 769 sizeof(*marray)))) == NULL) { 770 file_oomem(ms, maxmagic * sizeof(*marray)); 771 return -1; 772 } 773 marraycount = 0; 774 775 /* print silly verbose header for USG compat. */ 776 if (action == FILE_CHECK) 777 (void)fprintf(stderr, "%s\n", usg_hdr); 778 779 /* load directory or file */ 780 if (stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) { 781 dir = opendir(fn); 782 if (!dir) { 783 errs++; 784 goto out; 785 } 786 while ((d = readdir(dir)) != NULL) { 787 (void)snprintf(subfn, sizeof(subfn), "%s/%s", 788 fn, d->d_name); 789 if (stat(subfn, &st) == -1 || !S_ISREG(st.st_mode)) 790 continue; 791 if ((mfn = strdup(subfn)) == NULL) { 792 file_oomem(ms, strlen(subfn)); 793 errs++; 794 goto out; 795 } 796 if (files >= maxfiles) { 797 size_t mlen; 798 maxfiles = (maxfiles + 1) * 2; 799 mlen = maxfiles * sizeof(*filearr); 800 if ((filearr = CAST(char **, 801 realloc(filearr, mlen))) == NULL) { 802 file_oomem(ms, mlen); 803 errs++; 804 goto out; 805 } 806 } 807 filearr[files++] = mfn; 808 } 809 closedir(dir); 810 qsort(filearr, files, sizeof(*filearr), cmpstrp); 811 for (i = 0; i < files; i++) { 812 load_1(ms, action, filearr[i], &errs, &marray, 813 &marraycount); 814 free(filearr[i]); 815 } 816 free(filearr); 817 } else 818 load_1(ms, action, fn, &errs, &marray, &marraycount); 819 if (errs) 820 goto out; 821 822 /* Set types of tests */ 823 for (i = 0; i < marraycount; ) { 824 if (marray[i].mp->cont_level != 0) { 825 i++; 826 continue; 827 } 828 829 starttest = i; 830 do { 831 static const char text[] = "text"; 832 static const char binary[] = "binary"; 833 static const size_t len = sizeof(text); 834 set_test_type(marray[starttest].mp, marray[i].mp); 835 if ((ms->flags & MAGIC_DEBUG) == 0) 836 continue; 837 (void)fprintf(stderr, "%s%s%s: %s\n", 838 marray[i].mp->mimetype, 839 marray[i].mp->mimetype[0] == '\0' ? "" : "; ", 840 marray[i].mp->desc[0] ? marray[i].mp->desc : 841 "(no description)", 842 marray[i].mp->flag & BINTEST ? binary : text); 843 if (marray[i].mp->flag & BINTEST) { 844 char *p = strstr(marray[i].mp->desc, text); 845 if (p && (p == marray[i].mp->desc || 846 isspace((unsigned char)p[-1])) && 847 (p + len - marray[i].mp->desc == 848 MAXstring || (p[len] == '\0' || 849 isspace((unsigned char)p[len])))) 850 (void)fprintf(stderr, "*** Possible " 851 "binary test for text type\n"); 852 } 853 } while (++i < marraycount && marray[i].mp->cont_level != 0); 854 } 855 856 qsort(marray, marraycount, sizeof(*marray), apprentice_sort); 857 858 /* 859 * Make sure that any level 0 "default" line is last (if one exists). 860 */ 861 for (i = 0; i < marraycount; i++) { 862 if (marray[i].mp->cont_level == 0 && 863 marray[i].mp->type == FILE_DEFAULT) { 864 while (++i < marraycount) 865 if (marray[i].mp->cont_level == 0) 866 break; 867 if (i != marraycount) { 868 /* XXX - Ugh! */ 869 ms->line = marray[i].mp->lineno; 870 file_magwarn(ms, 871 "level 0 \"default\" did not sort last"); 872 } 873 break; 874 } 875 } 876 877 for (i = 0; i < marraycount; i++) 878 mentrycount += marray[i].cont_count; 879 880 slen = sizeof(**magicp) * mentrycount; 881 if ((*magicp = CAST(struct magic *, malloc(slen))) == NULL) { 882 file_oomem(ms, slen); 883 errs++; 884 goto out; 885 } 886 887 mentrycount = 0; 888 for (i = 0; i < marraycount; i++) { 889 (void)memcpy(*magicp + mentrycount, marray[i].mp, 890 marray[i].cont_count * sizeof(**magicp)); 891 mentrycount += marray[i].cont_count; 892 } 893 out: 894 for (i = 0; i < marraycount; i++) 895 free(marray[i].mp); 896 free(marray); 897 if (errs) { 898 *magicp = NULL; 899 *nmagicp = 0; 900 return errs; 901 } else { 902 *nmagicp = mentrycount; 903 return 0; 904 } 905 906 } 907 908 /* 909 * extend the sign bit if the comparison is to be signed 910 */ 911 protected uint64_t 912 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v) 913 { 914 if (!(m->flag & UNSIGNED)) { 915 switch(m->type) { 916 /* 917 * Do not remove the casts below. They are 918 * vital. When later compared with the data, 919 * the sign extension must have happened. 920 */ 921 case FILE_BYTE: 922 v = (char) v; 923 break; 924 case FILE_SHORT: 925 case FILE_BESHORT: 926 case FILE_LESHORT: 927 v = (short) v; 928 break; 929 case FILE_DATE: 930 case FILE_BEDATE: 931 case FILE_LEDATE: 932 case FILE_MEDATE: 933 case FILE_LDATE: 934 case FILE_BELDATE: 935 case FILE_LELDATE: 936 case FILE_MELDATE: 937 case FILE_LONG: 938 case FILE_BELONG: 939 case FILE_LELONG: 940 case FILE_MELONG: 941 case FILE_FLOAT: 942 case FILE_BEFLOAT: 943 case FILE_LEFLOAT: 944 v = (int32_t) v; 945 break; 946 case FILE_QUAD: 947 case FILE_BEQUAD: 948 case FILE_LEQUAD: 949 case FILE_QDATE: 950 case FILE_QLDATE: 951 case FILE_BEQDATE: 952 case FILE_BEQLDATE: 953 case FILE_LEQDATE: 954 case FILE_LEQLDATE: 955 case FILE_DOUBLE: 956 case FILE_BEDOUBLE: 957 case FILE_LEDOUBLE: 958 v = (int64_t) v; 959 break; 960 case FILE_STRING: 961 case FILE_PSTRING: 962 case FILE_BESTRING16: 963 case FILE_LESTRING16: 964 case FILE_REGEX: 965 case FILE_SEARCH: 966 case FILE_DEFAULT: 967 case FILE_INDIRECT: 968 break; 969 default: 970 if (ms->flags & MAGIC_CHECK) 971 file_magwarn(ms, "cannot happen: m->type=%d\n", 972 m->type); 973 return ~0U; 974 } 975 } 976 return v; 977 } 978 979 private int 980 string_modifier_check(struct magic_set *ms, struct magic *m) 981 { 982 if ((ms->flags & MAGIC_CHECK) == 0) 983 return 0; 984 985 if (m->type != FILE_PSTRING && (m->str_flags & PSTRING_LEN) != 0) { 986 file_magwarn(ms, 987 "'/BHhLl' modifiers are only allowed for pascal strings\n"); 988 return -1; 989 } 990 switch (m->type) { 991 case FILE_BESTRING16: 992 case FILE_LESTRING16: 993 if (m->str_flags != 0) { 994 file_magwarn(ms, 995 "no modifiers allowed for 16-bit strings\n"); 996 return -1; 997 } 998 break; 999 case FILE_STRING: 1000 case FILE_PSTRING: 1001 if ((m->str_flags & REGEX_OFFSET_START) != 0) { 1002 file_magwarn(ms, 1003 "'/%c' only allowed on regex and search\n", 1004 CHAR_REGEX_OFFSET_START); 1005 return -1; 1006 } 1007 break; 1008 case FILE_SEARCH: 1009 if (m->str_range == 0) { 1010 file_magwarn(ms, 1011 "missing range; defaulting to %d\n", 1012 STRING_DEFAULT_RANGE); 1013 m->str_range = STRING_DEFAULT_RANGE; 1014 return -1; 1015 } 1016 break; 1017 case FILE_REGEX: 1018 if ((m->str_flags & STRING_COMPACT_WHITESPACE) != 0) { 1019 file_magwarn(ms, "'/%c' not allowed on regex\n", 1020 CHAR_COMPACT_WHITESPACE); 1021 return -1; 1022 } 1023 if ((m->str_flags & STRING_COMPACT_OPTIONAL_WHITESPACE) != 0) { 1024 file_magwarn(ms, "'/%c' not allowed on regex\n", 1025 CHAR_COMPACT_OPTIONAL_WHITESPACE); 1026 return -1; 1027 } 1028 break; 1029 default: 1030 file_magwarn(ms, "coding error: m->type=%d\n", 1031 m->type); 1032 return -1; 1033 } 1034 return 0; 1035 } 1036 1037 private int 1038 get_op(char c) 1039 { 1040 switch (c) { 1041 case '&': 1042 return FILE_OPAND; 1043 case '|': 1044 return FILE_OPOR; 1045 case '^': 1046 return FILE_OPXOR; 1047 case '+': 1048 return FILE_OPADD; 1049 case '-': 1050 return FILE_OPMINUS; 1051 case '*': 1052 return FILE_OPMULTIPLY; 1053 case '/': 1054 return FILE_OPDIVIDE; 1055 case '%': 1056 return FILE_OPMODULO; 1057 default: 1058 return -1; 1059 } 1060 } 1061 1062 #ifdef ENABLE_CONDITIONALS 1063 private int 1064 get_cond(const char *l, const char **t) 1065 { 1066 static const struct cond_tbl_s { 1067 char name[8]; 1068 size_t len; 1069 int cond; 1070 } cond_tbl[] = { 1071 { "if", 2, COND_IF }, 1072 { "elif", 4, COND_ELIF }, 1073 { "else", 4, COND_ELSE }, 1074 { "", 0, COND_NONE }, 1075 }; 1076 const struct cond_tbl_s *p; 1077 1078 for (p = cond_tbl; p->len; p++) { 1079 if (strncmp(l, p->name, p->len) == 0 && 1080 isspace((unsigned char)l[p->len])) { 1081 if (t) 1082 *t = l + p->len; 1083 break; 1084 } 1085 } 1086 return p->cond; 1087 } 1088 1089 private int 1090 check_cond(struct magic_set *ms, int cond, uint32_t cont_level) 1091 { 1092 int last_cond; 1093 last_cond = ms->c.li[cont_level].last_cond; 1094 1095 switch (cond) { 1096 case COND_IF: 1097 if (last_cond != COND_NONE && last_cond != COND_ELIF) { 1098 if (ms->flags & MAGIC_CHECK) 1099 file_magwarn(ms, "syntax error: `if'"); 1100 return -1; 1101 } 1102 last_cond = COND_IF; 1103 break; 1104 1105 case COND_ELIF: 1106 if (last_cond != COND_IF && last_cond != COND_ELIF) { 1107 if (ms->flags & MAGIC_CHECK) 1108 file_magwarn(ms, "syntax error: `elif'"); 1109 return -1; 1110 } 1111 last_cond = COND_ELIF; 1112 break; 1113 1114 case COND_ELSE: 1115 if (last_cond != COND_IF && last_cond != COND_ELIF) { 1116 if (ms->flags & MAGIC_CHECK) 1117 file_magwarn(ms, "syntax error: `else'"); 1118 return -1; 1119 } 1120 last_cond = COND_NONE; 1121 break; 1122 1123 case COND_NONE: 1124 last_cond = COND_NONE; 1125 break; 1126 } 1127 1128 ms->c.li[cont_level].last_cond = last_cond; 1129 return 0; 1130 } 1131 #endif /* ENABLE_CONDITIONALS */ 1132 1133 /* 1134 * parse one line from magic file, put into magic[index++] if valid 1135 */ 1136 private int 1137 parse(struct magic_set *ms, struct magic_entry **mentryp, uint32_t *nmentryp, 1138 const char *line, size_t lineno, int action) 1139 { 1140 #ifdef ENABLE_CONDITIONALS 1141 static uint32_t last_cont_level = 0; 1142 #endif 1143 size_t i; 1144 struct magic_entry *me; 1145 struct magic *m; 1146 const char *l = line; 1147 char *t; 1148 int op; 1149 uint32_t cont_level; 1150 1151 cont_level = 0; 1152 1153 while (*l == '>') { 1154 ++l; /* step over */ 1155 cont_level++; 1156 } 1157 #ifdef ENABLE_CONDITIONALS 1158 if (cont_level == 0 || cont_level > last_cont_level) 1159 if (file_check_mem(ms, cont_level) == -1) 1160 return -1; 1161 last_cont_level = cont_level; 1162 #endif 1163 1164 #define ALLOC_CHUNK (size_t)10 1165 #define ALLOC_INCR (size_t)200 1166 1167 if (cont_level != 0) { 1168 if (*nmentryp == 0) { 1169 file_error(ms, 0, "No current entry for continuation"); 1170 return -1; 1171 } 1172 me = &(*mentryp)[*nmentryp - 1]; 1173 if (me->cont_count == me->max_count) { 1174 struct magic *nm; 1175 size_t cnt = me->max_count + ALLOC_CHUNK; 1176 if ((nm = CAST(struct magic *, realloc(me->mp, 1177 sizeof(*nm) * cnt))) == NULL) { 1178 file_oomem(ms, sizeof(*nm) * cnt); 1179 return -1; 1180 } 1181 me->mp = m = nm; 1182 me->max_count = CAST(uint32_t, cnt); 1183 } 1184 m = &me->mp[me->cont_count++]; 1185 (void)memset(m, 0, sizeof(*m)); 1186 m->cont_level = cont_level; 1187 } else { 1188 if (*nmentryp == maxmagic) { 1189 struct magic_entry *mp; 1190 1191 maxmagic += ALLOC_INCR; 1192 if ((mp = CAST(struct magic_entry *, 1193 realloc(*mentryp, sizeof(*mp) * maxmagic))) == 1194 NULL) { 1195 file_oomem(ms, sizeof(*mp) * maxmagic); 1196 return -1; 1197 } 1198 (void)memset(&mp[*nmentryp], 0, sizeof(*mp) * 1199 ALLOC_INCR); 1200 *mentryp = mp; 1201 } 1202 me = &(*mentryp)[*nmentryp]; 1203 if (me->mp == NULL) { 1204 size_t len = sizeof(*m) * ALLOC_CHUNK; 1205 if ((m = CAST(struct magic *, malloc(len))) == NULL) { 1206 file_oomem(ms, len); 1207 return -1; 1208 } 1209 me->mp = m; 1210 me->max_count = ALLOC_CHUNK; 1211 } else 1212 m = me->mp; 1213 (void)memset(m, 0, sizeof(*m)); 1214 m->factor_op = FILE_FACTOR_OP_NONE; 1215 m->cont_level = 0; 1216 me->cont_count = 1; 1217 } 1218 m->lineno = CAST(uint32_t, lineno); 1219 1220 if (*l == '&') { /* m->cont_level == 0 checked below. */ 1221 ++l; /* step over */ 1222 m->flag |= OFFADD; 1223 } 1224 if (*l == '(') { 1225 ++l; /* step over */ 1226 m->flag |= INDIR; 1227 if (m->flag & OFFADD) 1228 m->flag = (m->flag & ~OFFADD) | INDIROFFADD; 1229 1230 if (*l == '&') { /* m->cont_level == 0 checked below */ 1231 ++l; /* step over */ 1232 m->flag |= OFFADD; 1233 } 1234 } 1235 /* Indirect offsets are not valid at level 0. */ 1236 if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD))) 1237 if (ms->flags & MAGIC_CHECK) 1238 file_magwarn(ms, "relative offset at level 0"); 1239 1240 /* get offset, then skip over it */ 1241 m->offset = (uint32_t)strtoul(l, &t, 0); 1242 if (l == t) 1243 if (ms->flags & MAGIC_CHECK) 1244 file_magwarn(ms, "offset `%s' invalid", l); 1245 l = t; 1246 1247 if (m->flag & INDIR) { 1248 m->in_type = FILE_LONG; 1249 m->in_offset = 0; 1250 /* 1251 * read [.lbs][+-]nnnnn) 1252 */ 1253 if (*l == '.') { 1254 l++; 1255 switch (*l) { 1256 case 'l': 1257 m->in_type = FILE_LELONG; 1258 break; 1259 case 'L': 1260 m->in_type = FILE_BELONG; 1261 break; 1262 case 'm': 1263 m->in_type = FILE_MELONG; 1264 break; 1265 case 'h': 1266 case 's': 1267 m->in_type = FILE_LESHORT; 1268 break; 1269 case 'H': 1270 case 'S': 1271 m->in_type = FILE_BESHORT; 1272 break; 1273 case 'c': 1274 case 'b': 1275 case 'C': 1276 case 'B': 1277 m->in_type = FILE_BYTE; 1278 break; 1279 case 'e': 1280 case 'f': 1281 case 'g': 1282 m->in_type = FILE_LEDOUBLE; 1283 break; 1284 case 'E': 1285 case 'F': 1286 case 'G': 1287 m->in_type = FILE_BEDOUBLE; 1288 break; 1289 case 'i': 1290 m->in_type = FILE_LEID3; 1291 break; 1292 case 'I': 1293 m->in_type = FILE_BEID3; 1294 break; 1295 default: 1296 if (ms->flags & MAGIC_CHECK) 1297 file_magwarn(ms, 1298 "indirect offset type `%c' invalid", 1299 *l); 1300 break; 1301 } 1302 l++; 1303 } 1304 1305 m->in_op = 0; 1306 if (*l == '~') { 1307 m->in_op |= FILE_OPINVERSE; 1308 l++; 1309 } 1310 if ((op = get_op(*l)) != -1) { 1311 m->in_op |= op; 1312 l++; 1313 } 1314 if (*l == '(') { 1315 m->in_op |= FILE_OPINDIRECT; 1316 l++; 1317 } 1318 if (isdigit((unsigned char)*l) || *l == '-') { 1319 m->in_offset = (int32_t)strtol(l, &t, 0); 1320 if (l == t) 1321 if (ms->flags & MAGIC_CHECK) 1322 file_magwarn(ms, 1323 "in_offset `%s' invalid", l); 1324 l = t; 1325 } 1326 if (*l++ != ')' || 1327 ((m->in_op & FILE_OPINDIRECT) && *l++ != ')')) 1328 if (ms->flags & MAGIC_CHECK) 1329 file_magwarn(ms, 1330 "missing ')' in indirect offset"); 1331 } 1332 EATAB; 1333 1334 #ifdef ENABLE_CONDITIONALS 1335 m->cond = get_cond(l, &l); 1336 if (check_cond(ms, m->cond, cont_level) == -1) 1337 return -1; 1338 1339 EATAB; 1340 #endif 1341 1342 if (*l == 'u') { 1343 ++l; 1344 m->flag |= UNSIGNED; 1345 } 1346 1347 m->type = get_type(l, &l); 1348 if (m->type == FILE_INVALID) { 1349 if (ms->flags & MAGIC_CHECK) 1350 file_magwarn(ms, "type `%s' invalid", l); 1351 return -1; 1352 } 1353 1354 /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */ 1355 /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */ 1356 1357 m->mask_op = 0; 1358 if (*l == '~') { 1359 if (!IS_STRING(m->type)) 1360 m->mask_op |= FILE_OPINVERSE; 1361 else if (ms->flags & MAGIC_CHECK) 1362 file_magwarn(ms, "'~' invalid for string types"); 1363 ++l; 1364 } 1365 m->str_range = 0; 1366 m->str_flags = m->type == FILE_PSTRING ? PSTRING_1_LE : 0; 1367 if ((op = get_op(*l)) != -1) { 1368 if (!IS_STRING(m->type)) { 1369 uint64_t val; 1370 ++l; 1371 m->mask_op |= op; 1372 val = (uint64_t)strtoull(l, &t, 0); 1373 l = t; 1374 m->num_mask = file_signextend(ms, m, val); 1375 eatsize(&l); 1376 } 1377 else if (op == FILE_OPDIVIDE) { 1378 int have_range = 0; 1379 while (!isspace((unsigned char)*++l)) { 1380 switch (*l) { 1381 case '0': case '1': case '2': 1382 case '3': case '4': case '5': 1383 case '6': case '7': case '8': 1384 case '9': 1385 if (have_range && 1386 (ms->flags & MAGIC_CHECK)) 1387 file_magwarn(ms, 1388 "multiple ranges"); 1389 have_range = 1; 1390 m->str_range = CAST(uint32_t, 1391 strtoul(l, &t, 0)); 1392 if (m->str_range == 0) 1393 file_magwarn(ms, 1394 "zero range"); 1395 l = t - 1; 1396 break; 1397 case CHAR_COMPACT_WHITESPACE: 1398 m->str_flags |= 1399 STRING_COMPACT_WHITESPACE; 1400 break; 1401 case CHAR_COMPACT_OPTIONAL_WHITESPACE: 1402 m->str_flags |= 1403 STRING_COMPACT_OPTIONAL_WHITESPACE; 1404 break; 1405 case CHAR_IGNORE_LOWERCASE: 1406 m->str_flags |= STRING_IGNORE_LOWERCASE; 1407 break; 1408 case CHAR_IGNORE_UPPERCASE: 1409 m->str_flags |= STRING_IGNORE_UPPERCASE; 1410 break; 1411 case CHAR_REGEX_OFFSET_START: 1412 m->str_flags |= REGEX_OFFSET_START; 1413 break; 1414 case CHAR_BINTEST: 1415 m->str_flags |= STRING_BINTEST; 1416 break; 1417 case CHAR_TEXTTEST: 1418 m->str_flags |= STRING_TEXTTEST; 1419 break; 1420 case CHAR_PSTRING_1_LE: 1421 if (m->type != FILE_PSTRING) 1422 goto bad; 1423 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_1_LE; 1424 break; 1425 case CHAR_PSTRING_2_BE: 1426 if (m->type != FILE_PSTRING) 1427 goto bad; 1428 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_2_BE; 1429 break; 1430 case CHAR_PSTRING_2_LE: 1431 if (m->type != FILE_PSTRING) 1432 goto bad; 1433 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_2_LE; 1434 break; 1435 case CHAR_PSTRING_4_BE: 1436 if (m->type != FILE_PSTRING) 1437 goto bad; 1438 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_4_BE; 1439 break; 1440 case CHAR_PSTRING_4_LE: 1441 if (m->type != FILE_PSTRING) 1442 goto bad; 1443 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_4_LE; 1444 break; 1445 case CHAR_PSTRING_LENGTH_INCLUDES_ITSELF: 1446 if (m->type != FILE_PSTRING) 1447 goto bad; 1448 m->str_flags |= PSTRING_LENGTH_INCLUDES_ITSELF; 1449 break; 1450 bad: 1451 default: 1452 if (ms->flags & MAGIC_CHECK) 1453 file_magwarn(ms, 1454 "string extension `%c' " 1455 "invalid", *l); 1456 return -1; 1457 } 1458 /* allow multiple '/' for readability */ 1459 if (l[1] == '/' && 1460 !isspace((unsigned char)l[2])) 1461 l++; 1462 } 1463 if (string_modifier_check(ms, m) == -1) 1464 return -1; 1465 } 1466 else { 1467 if (ms->flags & MAGIC_CHECK) 1468 file_magwarn(ms, "invalid string op: %c", *t); 1469 return -1; 1470 } 1471 } 1472 /* 1473 * We used to set mask to all 1's here, instead let's just not do 1474 * anything if mask = 0 (unless you have a better idea) 1475 */ 1476 EATAB; 1477 1478 switch (*l) { 1479 case '>': 1480 case '<': 1481 m->reln = *l; 1482 ++l; 1483 if (*l == '=') { 1484 if (ms->flags & MAGIC_CHECK) { 1485 file_magwarn(ms, "%c= not supported", 1486 m->reln); 1487 return -1; 1488 } 1489 ++l; 1490 } 1491 break; 1492 /* Old-style anding: "0 byte &0x80 dynamically linked" */ 1493 case '&': 1494 case '^': 1495 case '=': 1496 m->reln = *l; 1497 ++l; 1498 if (*l == '=') { 1499 /* HP compat: ignore &= etc. */ 1500 ++l; 1501 } 1502 break; 1503 case '!': 1504 m->reln = *l; 1505 ++l; 1506 break; 1507 default: 1508 m->reln = '='; /* the default relation */ 1509 if (*l == 'x' && ((isascii((unsigned char)l[1]) && 1510 isspace((unsigned char)l[1])) || !l[1])) { 1511 m->reln = *l; 1512 ++l; 1513 } 1514 break; 1515 } 1516 /* 1517 * Grab the value part, except for an 'x' reln. 1518 */ 1519 if (m->reln != 'x' && getvalue(ms, m, &l, action)) 1520 return -1; 1521 1522 /* 1523 * TODO finish this macro and start using it! 1524 * #define offsetcheck {if (offset > HOWMANY-1) 1525 * magwarn("offset too big"); } 1526 */ 1527 1528 /* 1529 * Now get last part - the description 1530 */ 1531 EATAB; 1532 if (l[0] == '\b') { 1533 ++l; 1534 m->flag |= NOSPACE; 1535 } else if ((l[0] == '\\') && (l[1] == 'b')) { 1536 ++l; 1537 ++l; 1538 m->flag |= NOSPACE; 1539 } 1540 for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); ) 1541 continue; 1542 if (i == sizeof(m->desc)) { 1543 m->desc[sizeof(m->desc) - 1] = '\0'; 1544 if (ms->flags & MAGIC_CHECK) 1545 file_magwarn(ms, "description `%s' truncated", m->desc); 1546 } 1547 1548 /* 1549 * We only do this check while compiling, or if any of the magic 1550 * files were not compiled. 1551 */ 1552 if (ms->flags & MAGIC_CHECK) { 1553 if (check_format(ms, m) == -1) 1554 return -1; 1555 } 1556 #ifndef COMPILE_ONLY 1557 if (action == FILE_CHECK) { 1558 file_mdump(m); 1559 } 1560 #endif 1561 m->mimetype[0] = '\0'; /* initialise MIME type to none */ 1562 if (m->cont_level == 0) 1563 ++(*nmentryp); /* make room for next */ 1564 return 0; 1565 } 1566 1567 /* 1568 * parse a STRENGTH annotation line from magic file, put into magic[index - 1] 1569 * if valid 1570 */ 1571 private int 1572 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line) 1573 { 1574 const char *l = line; 1575 char *el; 1576 unsigned long factor; 1577 struct magic *m = &me->mp[0]; 1578 1579 if (m->factor_op != FILE_FACTOR_OP_NONE) { 1580 file_magwarn(ms, 1581 "Current entry already has a strength type: %c %d", 1582 m->factor_op, m->factor); 1583 return -1; 1584 } 1585 EATAB; 1586 switch (*l) { 1587 case FILE_FACTOR_OP_NONE: 1588 case FILE_FACTOR_OP_PLUS: 1589 case FILE_FACTOR_OP_MINUS: 1590 case FILE_FACTOR_OP_TIMES: 1591 case FILE_FACTOR_OP_DIV: 1592 m->factor_op = *l++; 1593 break; 1594 default: 1595 file_magwarn(ms, "Unknown factor op `%c'", *l); 1596 return -1; 1597 } 1598 EATAB; 1599 factor = strtoul(l, &el, 0); 1600 if (factor > 255) { 1601 file_magwarn(ms, "Too large factor `%lu'", factor); 1602 goto out; 1603 } 1604 if (*el && !isspace((unsigned char)*el)) { 1605 file_magwarn(ms, "Bad factor `%s'", l); 1606 goto out; 1607 } 1608 m->factor = (uint8_t)factor; 1609 if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) { 1610 file_magwarn(ms, "Cannot have factor op `%c' and factor %u", 1611 m->factor_op, m->factor); 1612 goto out; 1613 } 1614 return 0; 1615 out: 1616 m->factor_op = FILE_FACTOR_OP_NONE; 1617 m->factor = 0; 1618 return -1; 1619 } 1620 1621 /* 1622 * Parse an Apple CREATOR/TYPE annotation from magic file and put it into 1623 * magic[index - 1] 1624 */ 1625 private int 1626 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line) 1627 { 1628 size_t i; 1629 const char *l = line; 1630 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1]; 1631 1632 if (m->apple[0] != '\0') { 1633 file_magwarn(ms, "Current entry already has a APPLE type " 1634 "`%.8s', new type `%s'", m->mimetype, l); 1635 return -1; 1636 } 1637 1638 EATAB; 1639 for (i = 0; *l && ((isascii((unsigned char)*l) && 1640 isalnum((unsigned char)*l)) || strchr("-+/.", *l)) && 1641 i < sizeof(m->apple); m->apple[i++] = *l++) 1642 continue; 1643 if (i == sizeof(m->apple) && *l) { 1644 /* We don't need to NUL terminate here, printing handles it */ 1645 if (ms->flags & MAGIC_CHECK) 1646 file_magwarn(ms, "APPLE type `%s' truncated %" 1647 SIZE_T_FORMAT "u", line, i); 1648 } 1649 1650 if (i > 0) 1651 return 0; 1652 else 1653 return -1; 1654 } 1655 1656 /* 1657 * parse a MIME annotation line from magic file, put into magic[index - 1] 1658 * if valid 1659 */ 1660 private int 1661 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line) 1662 { 1663 size_t i; 1664 const char *l = line; 1665 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1]; 1666 1667 if (m->mimetype[0] != '\0') { 1668 file_magwarn(ms, "Current entry already has a MIME type `%s'," 1669 " new type `%s'", m->mimetype, l); 1670 return -1; 1671 } 1672 1673 EATAB; 1674 for (i = 0; *l && ((isascii((unsigned char)*l) && 1675 isalnum((unsigned char)*l)) || strchr("-+/.", *l)) && 1676 i < sizeof(m->mimetype); m->mimetype[i++] = *l++) 1677 continue; 1678 if (i == sizeof(m->mimetype)) { 1679 m->mimetype[sizeof(m->mimetype) - 1] = '\0'; 1680 if (ms->flags & MAGIC_CHECK) 1681 file_magwarn(ms, "MIME type `%s' truncated %" 1682 SIZE_T_FORMAT "u", m->mimetype, i); 1683 } else 1684 m->mimetype[i] = '\0'; 1685 1686 if (i > 0) 1687 return 0; 1688 else 1689 return -1; 1690 } 1691 1692 private int 1693 check_format_type(const char *ptr, int type) 1694 { 1695 int quad = 0; 1696 if (*ptr == '\0') { 1697 /* Missing format string; bad */ 1698 return -1; 1699 } 1700 1701 switch (type) { 1702 case FILE_FMT_QUAD: 1703 quad = 1; 1704 /*FALLTHROUGH*/ 1705 case FILE_FMT_NUM: 1706 if (*ptr == '-') 1707 ptr++; 1708 if (*ptr == '.') 1709 ptr++; 1710 while (isdigit((unsigned char)*ptr)) ptr++; 1711 if (*ptr == '.') 1712 ptr++; 1713 while (isdigit((unsigned char)*ptr)) ptr++; 1714 if (quad) { 1715 if (*ptr++ != 'l') 1716 return -1; 1717 if (*ptr++ != 'l') 1718 return -1; 1719 } 1720 1721 switch (*ptr++) { 1722 case 'l': 1723 switch (*ptr++) { 1724 case 'i': 1725 case 'd': 1726 case 'u': 1727 case 'x': 1728 case 'X': 1729 return 0; 1730 default: 1731 return -1; 1732 } 1733 1734 case 'h': 1735 switch (*ptr++) { 1736 case 'h': 1737 switch (*ptr++) { 1738 case 'i': 1739 case 'd': 1740 case 'u': 1741 case 'x': 1742 case 'X': 1743 return 0; 1744 default: 1745 return -1; 1746 } 1747 case 'd': 1748 return 0; 1749 default: 1750 return -1; 1751 } 1752 1753 case 'i': 1754 case 'c': 1755 case 'd': 1756 case 'u': 1757 case 'x': 1758 case 'X': 1759 return 0; 1760 1761 default: 1762 return -1; 1763 } 1764 1765 case FILE_FMT_FLOAT: 1766 case FILE_FMT_DOUBLE: 1767 if (*ptr == '-') 1768 ptr++; 1769 if (*ptr == '.') 1770 ptr++; 1771 while (isdigit((unsigned char)*ptr)) ptr++; 1772 if (*ptr == '.') 1773 ptr++; 1774 while (isdigit((unsigned char)*ptr)) ptr++; 1775 1776 switch (*ptr++) { 1777 case 'e': 1778 case 'E': 1779 case 'f': 1780 case 'F': 1781 case 'g': 1782 case 'G': 1783 return 0; 1784 1785 default: 1786 return -1; 1787 } 1788 1789 1790 case FILE_FMT_STR: 1791 if (*ptr == '-') 1792 ptr++; 1793 while (isdigit((unsigned char )*ptr)) 1794 ptr++; 1795 if (*ptr == '.') { 1796 ptr++; 1797 while (isdigit((unsigned char )*ptr)) 1798 ptr++; 1799 } 1800 1801 switch (*ptr++) { 1802 case 's': 1803 return 0; 1804 default: 1805 return -1; 1806 } 1807 1808 default: 1809 /* internal error */ 1810 abort(); 1811 } 1812 /*NOTREACHED*/ 1813 return -1; 1814 } 1815 1816 /* 1817 * Check that the optional printf format in description matches 1818 * the type of the magic. 1819 */ 1820 private int 1821 check_format(struct magic_set *ms, struct magic *m) 1822 { 1823 char *ptr; 1824 1825 for (ptr = m->desc; *ptr; ptr++) 1826 if (*ptr == '%') 1827 break; 1828 if (*ptr == '\0') { 1829 /* No format string; ok */ 1830 return 1; 1831 } 1832 1833 assert(file_nformats == file_nnames); 1834 1835 if (m->type >= file_nformats) { 1836 file_magwarn(ms, "Internal error inconsistency between " 1837 "m->type and format strings"); 1838 return -1; 1839 } 1840 if (file_formats[m->type] == FILE_FMT_NONE) { 1841 file_magwarn(ms, "No format string for `%s' with description " 1842 "`%s'", m->desc, file_names[m->type]); 1843 return -1; 1844 } 1845 1846 ptr++; 1847 if (check_format_type(ptr, file_formats[m->type]) == -1) { 1848 /* 1849 * TODO: this error message is unhelpful if the format 1850 * string is not one character long 1851 */ 1852 file_magwarn(ms, "Printf format `%c' is not valid for type " 1853 "`%s' in description `%s'", *ptr ? *ptr : '?', 1854 file_names[m->type], m->desc); 1855 return -1; 1856 } 1857 1858 for (; *ptr; ptr++) { 1859 if (*ptr == '%') { 1860 file_magwarn(ms, 1861 "Too many format strings (should have at most one) " 1862 "for `%s' with description `%s'", 1863 file_names[m->type], m->desc); 1864 return -1; 1865 } 1866 } 1867 return 0; 1868 } 1869 1870 /* 1871 * Read a numeric value from a pointer, into the value union of a magic 1872 * pointer, according to the magic type. Update the string pointer to point 1873 * just after the number read. Return 0 for success, non-zero for failure. 1874 */ 1875 private int 1876 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action) 1877 { 1878 switch (m->type) { 1879 case FILE_BESTRING16: 1880 case FILE_LESTRING16: 1881 case FILE_STRING: 1882 case FILE_PSTRING: 1883 case FILE_REGEX: 1884 case FILE_SEARCH: 1885 *p = getstr(ms, m, *p, action == FILE_COMPILE); 1886 if (*p == NULL) { 1887 if (ms->flags & MAGIC_CHECK) 1888 file_magwarn(ms, "cannot get string from `%s'", 1889 m->value.s); 1890 return -1; 1891 } 1892 return 0; 1893 case FILE_FLOAT: 1894 case FILE_BEFLOAT: 1895 case FILE_LEFLOAT: 1896 if (m->reln != 'x') { 1897 char *ep; 1898 #if defined(HAVE_STRTOF) && !defined(COMPILE_ONLY) 1899 m->value.f = strtof(*p, &ep); 1900 #else 1901 m->value.f = (float)strtod(*p, &ep); 1902 #endif 1903 *p = ep; 1904 } 1905 return 0; 1906 case FILE_DOUBLE: 1907 case FILE_BEDOUBLE: 1908 case FILE_LEDOUBLE: 1909 if (m->reln != 'x') { 1910 char *ep; 1911 m->value.d = strtod(*p, &ep); 1912 *p = ep; 1913 } 1914 return 0; 1915 default: 1916 if (m->reln != 'x') { 1917 char *ep; 1918 m->value.q = file_signextend(ms, m, 1919 (uint64_t)strtoull(*p, &ep, 0)); 1920 *p = ep; 1921 eatsize(p); 1922 } 1923 return 0; 1924 } 1925 } 1926 1927 /* 1928 * Convert a string containing C character escapes. Stop at an unescaped 1929 * space or tab. 1930 * Copy the converted version to "m->value.s", and the length in m->vallen. 1931 * Return updated scan pointer as function result. Warn if set. 1932 */ 1933 private const char * 1934 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn) 1935 { 1936 const char *origs = s; 1937 char *p = m->value.s; 1938 size_t plen = sizeof(m->value.s); 1939 char *origp = p; 1940 char *pmax = p + plen - 1; 1941 int c; 1942 int val; 1943 1944 while ((c = *s++) != '\0') { 1945 if (isspace((unsigned char) c)) 1946 break; 1947 if (p >= pmax) { 1948 file_error(ms, 0, "string too long: `%s'", origs); 1949 return NULL; 1950 } 1951 if (c == '\\') { 1952 switch(c = *s++) { 1953 1954 case '\0': 1955 if (warn) 1956 file_magwarn(ms, "incomplete escape"); 1957 goto out; 1958 1959 case '\t': 1960 if (warn) { 1961 file_magwarn(ms, 1962 "escaped tab found, use \\t instead"); 1963 warn = 0; /* already did */ 1964 } 1965 /*FALLTHROUGH*/ 1966 default: 1967 if (warn) { 1968 if (isprint((unsigned char)c)) { 1969 /* Allow escaping of 1970 * ``relations'' */ 1971 if (strchr("<>&^=!", c) == NULL 1972 && (m->type != FILE_REGEX || 1973 strchr("[]().*?^$|{}", c) 1974 == NULL)) { 1975 file_magwarn(ms, "no " 1976 "need to escape " 1977 "`%c'", c); 1978 } 1979 } else { 1980 file_magwarn(ms, 1981 "unknown escape sequence: " 1982 "\\%03o", c); 1983 } 1984 } 1985 /*FALLTHROUGH*/ 1986 /* space, perhaps force people to use \040? */ 1987 case ' ': 1988 #if 0 1989 /* 1990 * Other things people escape, but shouldn't need to, 1991 * so we disallow them 1992 */ 1993 case '\'': 1994 case '"': 1995 case '?': 1996 #endif 1997 /* Relations */ 1998 case '>': 1999 case '<': 2000 case '&': 2001 case '^': 2002 case '=': 2003 case '!': 2004 /* and baskslash itself */ 2005 case '\\': 2006 *p++ = (char) c; 2007 break; 2008 2009 case 'a': 2010 *p++ = '\a'; 2011 break; 2012 2013 case 'b': 2014 *p++ = '\b'; 2015 break; 2016 2017 case 'f': 2018 *p++ = '\f'; 2019 break; 2020 2021 case 'n': 2022 *p++ = '\n'; 2023 break; 2024 2025 case 'r': 2026 *p++ = '\r'; 2027 break; 2028 2029 case 't': 2030 *p++ = '\t'; 2031 break; 2032 2033 case 'v': 2034 *p++ = '\v'; 2035 break; 2036 2037 /* \ and up to 3 octal digits */ 2038 case '0': 2039 case '1': 2040 case '2': 2041 case '3': 2042 case '4': 2043 case '5': 2044 case '6': 2045 case '7': 2046 val = c - '0'; 2047 c = *s++; /* try for 2 */ 2048 if (c >= '0' && c <= '7') { 2049 val = (val << 3) | (c - '0'); 2050 c = *s++; /* try for 3 */ 2051 if (c >= '0' && c <= '7') 2052 val = (val << 3) | (c-'0'); 2053 else 2054 --s; 2055 } 2056 else 2057 --s; 2058 *p++ = (char)val; 2059 break; 2060 2061 /* \x and up to 2 hex digits */ 2062 case 'x': 2063 val = 'x'; /* Default if no digits */ 2064 c = hextoint(*s++); /* Get next char */ 2065 if (c >= 0) { 2066 val = c; 2067 c = hextoint(*s++); 2068 if (c >= 0) 2069 val = (val << 4) + c; 2070 else 2071 --s; 2072 } else 2073 --s; 2074 *p++ = (char)val; 2075 break; 2076 } 2077 } else 2078 *p++ = (char)c; 2079 } 2080 out: 2081 *p = '\0'; 2082 m->vallen = CAST(unsigned char, (p - origp)); 2083 if (m->type == FILE_PSTRING) 2084 m->vallen += file_pstring_length_size(m); 2085 return s; 2086 } 2087 2088 2089 /* Single hex char to int; -1 if not a hex char. */ 2090 private int 2091 hextoint(int c) 2092 { 2093 if (!isascii((unsigned char) c)) 2094 return -1; 2095 if (isdigit((unsigned char) c)) 2096 return c - '0'; 2097 if ((c >= 'a') && (c <= 'f')) 2098 return c + 10 - 'a'; 2099 if (( c>= 'A') && (c <= 'F')) 2100 return c + 10 - 'A'; 2101 return -1; 2102 } 2103 2104 2105 /* 2106 * Print a string containing C character escapes. 2107 */ 2108 protected void 2109 file_showstr(FILE *fp, const char *s, size_t len) 2110 { 2111 char c; 2112 2113 for (;;) { 2114 if (len == ~0U) { 2115 c = *s++; 2116 if (c == '\0') 2117 break; 2118 } 2119 else { 2120 if (len-- == 0) 2121 break; 2122 c = *s++; 2123 } 2124 if (c >= 040 && c <= 0176) /* TODO isprint && !iscntrl */ 2125 (void) fputc(c, fp); 2126 else { 2127 (void) fputc('\\', fp); 2128 switch (c) { 2129 case '\a': 2130 (void) fputc('a', fp); 2131 break; 2132 2133 case '\b': 2134 (void) fputc('b', fp); 2135 break; 2136 2137 case '\f': 2138 (void) fputc('f', fp); 2139 break; 2140 2141 case '\n': 2142 (void) fputc('n', fp); 2143 break; 2144 2145 case '\r': 2146 (void) fputc('r', fp); 2147 break; 2148 2149 case '\t': 2150 (void) fputc('t', fp); 2151 break; 2152 2153 case '\v': 2154 (void) fputc('v', fp); 2155 break; 2156 2157 default: 2158 (void) fprintf(fp, "%.3o", c & 0377); 2159 break; 2160 } 2161 } 2162 } 2163 } 2164 2165 /* 2166 * eatsize(): Eat the size spec from a number [eg. 10UL] 2167 */ 2168 private void 2169 eatsize(const char **p) 2170 { 2171 const char *l = *p; 2172 2173 if (LOWCASE(*l) == 'u') 2174 l++; 2175 2176 switch (LOWCASE(*l)) { 2177 case 'l': /* long */ 2178 case 's': /* short */ 2179 case 'h': /* short */ 2180 case 'b': /* char/byte */ 2181 case 'c': /* char/byte */ 2182 l++; 2183 /*FALLTHROUGH*/ 2184 default: 2185 break; 2186 } 2187 2188 *p = l; 2189 } 2190 2191 /* 2192 * handle a compiled file. 2193 */ 2194 private int 2195 apprentice_map(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp, 2196 const char *fn) 2197 { 2198 int fd; 2199 struct stat st; 2200 uint32_t *ptr; 2201 uint32_t version; 2202 int needsbyteswap; 2203 char *dbname = NULL; 2204 void *mm = NULL; 2205 2206 dbname = mkdbname(ms, fn, 0); 2207 if (dbname == NULL) 2208 goto error2; 2209 2210 if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1) 2211 goto error2; 2212 2213 if (fstat(fd, &st) == -1) { 2214 file_error(ms, errno, "cannot stat `%s'", dbname); 2215 goto error1; 2216 } 2217 if (st.st_size < 8) { 2218 file_error(ms, 0, "file `%s' is too small", dbname); 2219 goto error1; 2220 } 2221 2222 #ifdef QUICK 2223 if ((mm = mmap(0, (size_t)st.st_size, PROT_READ|PROT_WRITE, 2224 MAP_PRIVATE|MAP_FILE, fd, (off_t)0)) == MAP_FAILED) { 2225 file_error(ms, errno, "cannot map `%s'", dbname); 2226 goto error1; 2227 } 2228 #define RET 2 2229 #else 2230 if ((mm = CAST(void *, malloc((size_t)st.st_size))) == NULL) { 2231 file_oomem(ms, (size_t)st.st_size); 2232 goto error1; 2233 } 2234 if (read(fd, mm, (size_t)st.st_size) != (ssize_t)st.st_size) { 2235 file_badread(ms); 2236 goto error1; 2237 } 2238 #define RET 1 2239 #endif 2240 *magicp = CAST(struct magic *, mm); 2241 (void)close(fd); 2242 fd = -1; 2243 ptr = (uint32_t *)(void *)*magicp; 2244 if (*ptr != MAGICNO) { 2245 if (swap4(*ptr) != MAGICNO) { 2246 file_error(ms, 0, "bad magic in `%s'", dbname); 2247 goto error1; 2248 } 2249 needsbyteswap = 1; 2250 } else 2251 needsbyteswap = 0; 2252 if (needsbyteswap) 2253 version = swap4(ptr[1]); 2254 else 2255 version = ptr[1]; 2256 if (version != VERSIONNO) { 2257 file_error(ms, 0, "File %d.%d supports only version %d magic " 2258 "files. `%s' is version %d", FILE_VERSION_MAJOR, patchlevel, 2259 VERSIONNO, dbname, version); 2260 goto error1; 2261 } 2262 *nmagicp = (uint32_t)(st.st_size / sizeof(struct magic)); 2263 if (*nmagicp > 0) 2264 (*nmagicp)--; 2265 (*magicp)++; 2266 if (needsbyteswap) 2267 byteswap(*magicp, *nmagicp); 2268 free(dbname); 2269 return RET; 2270 2271 error1: 2272 if (fd != -1) 2273 (void)close(fd); 2274 if (mm) { 2275 #ifdef QUICK 2276 (void)munmap((void *)mm, (size_t)st.st_size); 2277 #else 2278 free(mm); 2279 #endif 2280 } else { 2281 *magicp = NULL; 2282 *nmagicp = 0; 2283 } 2284 error2: 2285 free(dbname); 2286 return -1; 2287 } 2288 2289 private const uint32_t ar[] = { 2290 MAGICNO, VERSIONNO 2291 }; 2292 /* 2293 * handle an mmaped file. 2294 */ 2295 private int 2296 apprentice_compile(struct magic_set *ms, struct magic **magicp, 2297 uint32_t *nmagicp, const char *fn) 2298 { 2299 int fd; 2300 char *dbname; 2301 int rv = -1; 2302 2303 dbname = mkdbname(ms, fn, 1); 2304 2305 if (dbname == NULL) 2306 goto out; 2307 2308 if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) { 2309 file_error(ms, errno, "cannot open `%s'", dbname); 2310 goto out; 2311 } 2312 2313 if (write(fd, ar, sizeof(ar)) != (ssize_t)sizeof(ar)) { 2314 file_error(ms, errno, "error writing `%s'", dbname); 2315 goto out; 2316 } 2317 2318 if (lseek(fd, (off_t)sizeof(struct magic), SEEK_SET) 2319 != sizeof(struct magic)) { 2320 file_error(ms, errno, "error seeking `%s'", dbname); 2321 goto out; 2322 } 2323 2324 if (write(fd, *magicp, (sizeof(struct magic) * *nmagicp)) 2325 != (ssize_t)(sizeof(struct magic) * *nmagicp)) { 2326 file_error(ms, errno, "error writing `%s'", dbname); 2327 goto out; 2328 } 2329 2330 (void)close(fd); 2331 rv = 0; 2332 out: 2333 free(dbname); 2334 return rv; 2335 } 2336 2337 private const char ext[] = ".mgc"; 2338 /* 2339 * make a dbname 2340 */ 2341 private char * 2342 mkdbname(struct magic_set *ms, const char *fn, int strip) 2343 { 2344 const char *p, *q; 2345 char *buf; 2346 2347 if (strip) { 2348 if ((p = strrchr(fn, '/')) != NULL) 2349 fn = ++p; 2350 } 2351 2352 for (q = fn; *q; q++) 2353 continue; 2354 /* Look for .mgc */ 2355 for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--) 2356 if (*p != *q) 2357 break; 2358 2359 /* Did not find .mgc, restore q */ 2360 if (p >= ext) 2361 while (*q) 2362 q++; 2363 2364 q++; 2365 /* Compatibility with old code that looked in .mime */ 2366 if (ms->flags & MAGIC_MIME) { 2367 asprintf(&buf, "%.*s.mime%s", (int)(q - fn), fn, ext); 2368 if (access(buf, R_OK) != -1) { 2369 ms->flags &= MAGIC_MIME_TYPE; 2370 return buf; 2371 } 2372 free(buf); 2373 } 2374 asprintf(&buf, "%.*s%s", (int)(q - fn), fn, ext); 2375 2376 /* Compatibility with old code that looked in .mime */ 2377 if (strstr(p, ".mime") != NULL) 2378 ms->flags &= MAGIC_MIME_TYPE; 2379 return buf; 2380 } 2381 2382 /* 2383 * Byteswap an mmap'ed file if needed 2384 */ 2385 private void 2386 byteswap(struct magic *magic, uint32_t nmagic) 2387 { 2388 uint32_t i; 2389 for (i = 0; i < nmagic; i++) 2390 bs1(&magic[i]); 2391 } 2392 2393 /* 2394 * swap a short 2395 */ 2396 private uint16_t 2397 swap2(uint16_t sv) 2398 { 2399 uint16_t rv; 2400 uint8_t *s = (uint8_t *)(void *)&sv; 2401 uint8_t *d = (uint8_t *)(void *)&rv; 2402 d[0] = s[1]; 2403 d[1] = s[0]; 2404 return rv; 2405 } 2406 2407 /* 2408 * swap an int 2409 */ 2410 private uint32_t 2411 swap4(uint32_t sv) 2412 { 2413 uint32_t rv; 2414 uint8_t *s = (uint8_t *)(void *)&sv; 2415 uint8_t *d = (uint8_t *)(void *)&rv; 2416 d[0] = s[3]; 2417 d[1] = s[2]; 2418 d[2] = s[1]; 2419 d[3] = s[0]; 2420 return rv; 2421 } 2422 2423 /* 2424 * swap a quad 2425 */ 2426 private uint64_t 2427 swap8(uint64_t sv) 2428 { 2429 uint64_t rv; 2430 uint8_t *s = (uint8_t *)(void *)&sv; 2431 uint8_t *d = (uint8_t *)(void *)&rv; 2432 #if 0 2433 d[0] = s[3]; 2434 d[1] = s[2]; 2435 d[2] = s[1]; 2436 d[3] = s[0]; 2437 d[4] = s[7]; 2438 d[5] = s[6]; 2439 d[6] = s[5]; 2440 d[7] = s[4]; 2441 #else 2442 d[0] = s[7]; 2443 d[1] = s[6]; 2444 d[2] = s[5]; 2445 d[3] = s[4]; 2446 d[4] = s[3]; 2447 d[5] = s[2]; 2448 d[6] = s[1]; 2449 d[7] = s[0]; 2450 #endif 2451 return rv; 2452 } 2453 2454 /* 2455 * byteswap a single magic entry 2456 */ 2457 private void 2458 bs1(struct magic *m) 2459 { 2460 m->cont_level = swap2(m->cont_level); 2461 m->offset = swap4((uint32_t)m->offset); 2462 m->in_offset = swap4((uint32_t)m->in_offset); 2463 m->lineno = swap4((uint32_t)m->lineno); 2464 if (IS_STRING(m->type)) { 2465 if (m->type == FILE_PSTRING) 2466 printf("flags! %d\n", m->str_flags); 2467 m->str_range = swap4(m->str_range); 2468 m->str_flags = swap4(m->str_flags); 2469 } 2470 else { 2471 m->value.q = swap8(m->value.q); 2472 m->num_mask = swap8(m->num_mask); 2473 } 2474 } 2475 2476 protected size_t 2477 file_pstring_length_size(const struct magic *m) 2478 { 2479 switch (m->str_flags & PSTRING_LEN) { 2480 case PSTRING_1_LE: 2481 return 1; 2482 case PSTRING_2_LE: 2483 case PSTRING_2_BE: 2484 return 2; 2485 case PSTRING_4_LE: 2486 case PSTRING_4_BE: 2487 return 4; 2488 default: 2489 abort(); /* Impossible */ 2490 return 1; 2491 } 2492 } 2493 protected size_t 2494 file_pstring_get_length(const struct magic *m, const char *s) 2495 { 2496 size_t len = 0; 2497 2498 switch (m->str_flags & PSTRING_LEN) { 2499 case PSTRING_1_LE: 2500 len = *s; 2501 break; 2502 case PSTRING_2_LE: 2503 len = (s[1] << 8) | s[0]; 2504 break; 2505 case PSTRING_2_BE: 2506 len = (s[0] << 8) | s[1]; 2507 break; 2508 case PSTRING_4_LE: 2509 len = (s[3] << 24) | (s[2] << 16) | (s[1] << 8) | s[0]; 2510 break; 2511 case PSTRING_4_BE: 2512 len = (s[0] << 24) | (s[1] << 16) | (s[2] << 8) | s[3]; 2513 break; 2514 default: 2515 abort(); /* Impossible */ 2516 } 2517 2518 if (m->str_flags & PSTRING_LENGTH_INCLUDES_ITSELF) 2519 len -= file_pstring_length_size(m); 2520 2521 return len; 2522 } 2523