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 1.229 2015/01/01 17:07:34 christos 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 #ifdef HAVE_STDDEF_H 44 #include <stddef.h> 45 #endif 46 #include <string.h> 47 #include <assert.h> 48 #include <ctype.h> 49 #include <fcntl.h> 50 #ifdef QUICK 51 #include <sys/mman.h> 52 #endif 53 #include <dirent.h> 54 #if defined(HAVE_LIMITS_H) 55 #include <limits.h> 56 #endif 57 58 #ifndef SSIZE_MAX 59 #define MAXMAGIC_SIZE ((ssize_t)0x7fffffff) 60 #else 61 #define MAXMAGIC_SIZE SSIZE_MAX 62 #endif 63 64 #define EATAB {while (isascii((unsigned char) *l) && \ 65 isspace((unsigned char) *l)) ++l;} 66 #define LOWCASE(l) (isupper((unsigned char) (l)) ? \ 67 tolower((unsigned char) (l)) : (l)) 68 /* 69 * Work around a bug in headers on Digital Unix. 70 * At least confirmed for: OSF1 V4.0 878 71 */ 72 #if defined(__osf__) && defined(__DECC) 73 #ifdef MAP_FAILED 74 #undef MAP_FAILED 75 #endif 76 #endif 77 78 #ifndef MAP_FAILED 79 #define MAP_FAILED (void *) -1 80 #endif 81 82 #ifndef MAP_FILE 83 #define MAP_FILE 0 84 #endif 85 86 #define ALLOC_CHUNK (size_t)10 87 #define ALLOC_INCR (size_t)200 88 89 #define MAP_TYPE_MMAP 0 90 #define MAP_TYPE_MALLOC 1 91 #define MAP_TYPE_USER 2 92 93 struct magic_entry { 94 struct magic *mp; 95 uint32_t cont_count; 96 uint32_t max_count; 97 }; 98 99 struct magic_entry_set { 100 struct magic_entry *me; 101 uint32_t count; 102 uint32_t max; 103 }; 104 105 struct magic_map { 106 void *p; 107 size_t len; 108 int type; 109 struct magic *magic[MAGIC_SETS]; 110 uint32_t nmagic[MAGIC_SETS]; 111 }; 112 113 int file_formats[FILE_NAMES_SIZE]; 114 const size_t file_nformats = FILE_NAMES_SIZE; 115 const char *file_names[FILE_NAMES_SIZE]; 116 const size_t file_nnames = FILE_NAMES_SIZE; 117 118 private int getvalue(struct magic_set *ms, struct magic *, const char **, int); 119 private int hextoint(int); 120 private const char *getstr(struct magic_set *, struct magic *, const char *, 121 int); 122 private int parse(struct magic_set *, struct magic_entry *, const char *, 123 size_t, int); 124 private void eatsize(const char **); 125 private int apprentice_1(struct magic_set *, const char *, int); 126 private size_t apprentice_magic_strength(const struct magic *); 127 private int apprentice_sort(const void *, const void *); 128 private void apprentice_list(struct mlist *, int ); 129 private struct magic_map *apprentice_load(struct magic_set *, 130 const char *, int); 131 private struct mlist *mlist_alloc(void); 132 private void mlist_free(struct mlist *); 133 private void byteswap(struct magic *, uint32_t); 134 private void bs1(struct magic *); 135 private uint16_t swap2(uint16_t); 136 private uint32_t swap4(uint32_t); 137 private uint64_t swap8(uint64_t); 138 private char *mkdbname(struct magic_set *, const char *, int); 139 private struct magic_map *apprentice_buf(struct magic_set *, struct magic *, 140 size_t); 141 private struct magic_map *apprentice_map(struct magic_set *, const char *); 142 private int check_buffer(struct magic_set *, struct magic_map *, const char *); 143 private void apprentice_unmap(struct magic_map *); 144 private int apprentice_compile(struct magic_set *, struct magic_map *, 145 const char *); 146 private int check_format_type(const char *, int); 147 private int check_format(struct magic_set *, struct magic *); 148 private int get_op(char); 149 private int parse_mime(struct magic_set *, struct magic_entry *, const char *); 150 private int parse_strength(struct magic_set *, struct magic_entry *, const char *); 151 private int parse_apple(struct magic_set *, struct magic_entry *, const char *); 152 153 154 private size_t magicsize = sizeof(struct magic); 155 156 private const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc"; 157 158 private struct { 159 const char *name; 160 size_t len; 161 int (*fun)(struct magic_set *, struct magic_entry *, const char *); 162 } bang[] = { 163 #define DECLARE_FIELD(name) { # name, sizeof(# name) - 1, parse_ ## name } 164 DECLARE_FIELD(mime), 165 DECLARE_FIELD(apple), 166 DECLARE_FIELD(strength), 167 #undef DECLARE_FIELD 168 { NULL, 0, NULL } 169 }; 170 171 #ifdef COMPILE_ONLY 172 173 int main(int, char *[]); 174 175 int 176 main(int argc, char *argv[]) 177 { 178 int ret; 179 struct magic_set *ms; 180 char *progname; 181 182 if ((progname = strrchr(argv[0], '/')) != NULL) 183 progname++; 184 else 185 progname = argv[0]; 186 187 if (argc != 2) { 188 (void)fprintf(stderr, "Usage: %s file\n", progname); 189 return 1; 190 } 191 192 if ((ms = magic_open(MAGIC_CHECK)) == NULL) { 193 (void)fprintf(stderr, "%s: %s\n", progname, strerror(errno)); 194 return 1; 195 } 196 ret = magic_compile(ms, argv[1]) == -1 ? 1 : 0; 197 if (ret == 1) 198 (void)fprintf(stderr, "%s: %s\n", progname, magic_error(ms)); 199 magic_close(ms); 200 return ret; 201 } 202 #endif /* COMPILE_ONLY */ 203 204 struct type_tbl_s { 205 const char name[16]; 206 const size_t len; 207 const int type; 208 const int format; 209 }; 210 211 /* 212 * XXX - the actual Single UNIX Specification says that "long" means "long", 213 * as in the C data type, but we treat it as meaning "4-byte integer". 214 * Given that the OS X version of file 5.04 did the same, I guess that passes 215 * the actual test; having "long" be dependent on how big a "long" is on 216 * the machine running "file" is silly. 217 */ 218 static const struct type_tbl_s type_tbl[] = { 219 # define XX(s) s, (sizeof(s) - 1) 220 # define XX_NULL "", 0 221 { XX("invalid"), FILE_INVALID, FILE_FMT_NONE }, 222 { XX("byte"), FILE_BYTE, FILE_FMT_NUM }, 223 { XX("short"), FILE_SHORT, FILE_FMT_NUM }, 224 { XX("default"), FILE_DEFAULT, FILE_FMT_NONE }, 225 { XX("long"), FILE_LONG, FILE_FMT_NUM }, 226 { XX("string"), FILE_STRING, FILE_FMT_STR }, 227 { XX("date"), FILE_DATE, FILE_FMT_STR }, 228 { XX("beshort"), FILE_BESHORT, FILE_FMT_NUM }, 229 { XX("belong"), FILE_BELONG, FILE_FMT_NUM }, 230 { XX("bedate"), FILE_BEDATE, FILE_FMT_STR }, 231 { XX("leshort"), FILE_LESHORT, FILE_FMT_NUM }, 232 { XX("lelong"), FILE_LELONG, FILE_FMT_NUM }, 233 { XX("ledate"), FILE_LEDATE, FILE_FMT_STR }, 234 { XX("pstring"), FILE_PSTRING, FILE_FMT_STR }, 235 { XX("ldate"), FILE_LDATE, FILE_FMT_STR }, 236 { XX("beldate"), FILE_BELDATE, FILE_FMT_STR }, 237 { XX("leldate"), FILE_LELDATE, FILE_FMT_STR }, 238 { XX("regex"), FILE_REGEX, FILE_FMT_STR }, 239 { XX("bestring16"), FILE_BESTRING16, FILE_FMT_STR }, 240 { XX("lestring16"), FILE_LESTRING16, FILE_FMT_STR }, 241 { XX("search"), FILE_SEARCH, FILE_FMT_STR }, 242 { XX("medate"), FILE_MEDATE, FILE_FMT_STR }, 243 { XX("meldate"), FILE_MELDATE, FILE_FMT_STR }, 244 { XX("melong"), FILE_MELONG, FILE_FMT_NUM }, 245 { XX("quad"), FILE_QUAD, FILE_FMT_QUAD }, 246 { XX("lequad"), FILE_LEQUAD, FILE_FMT_QUAD }, 247 { XX("bequad"), FILE_BEQUAD, FILE_FMT_QUAD }, 248 { XX("qdate"), FILE_QDATE, FILE_FMT_STR }, 249 { XX("leqdate"), FILE_LEQDATE, FILE_FMT_STR }, 250 { XX("beqdate"), FILE_BEQDATE, FILE_FMT_STR }, 251 { XX("qldate"), FILE_QLDATE, FILE_FMT_STR }, 252 { XX("leqldate"), FILE_LEQLDATE, FILE_FMT_STR }, 253 { XX("beqldate"), FILE_BEQLDATE, FILE_FMT_STR }, 254 { XX("float"), FILE_FLOAT, FILE_FMT_FLOAT }, 255 { XX("befloat"), FILE_BEFLOAT, FILE_FMT_FLOAT }, 256 { XX("lefloat"), FILE_LEFLOAT, FILE_FMT_FLOAT }, 257 { XX("double"), FILE_DOUBLE, FILE_FMT_DOUBLE }, 258 { XX("bedouble"), FILE_BEDOUBLE, FILE_FMT_DOUBLE }, 259 { XX("ledouble"), FILE_LEDOUBLE, FILE_FMT_DOUBLE }, 260 { XX("leid3"), FILE_LEID3, FILE_FMT_NUM }, 261 { XX("beid3"), FILE_BEID3, FILE_FMT_NUM }, 262 { XX("indirect"), FILE_INDIRECT, FILE_FMT_NUM }, 263 { XX("qwdate"), FILE_QWDATE, FILE_FMT_STR }, 264 { XX("leqwdate"), FILE_LEQWDATE, FILE_FMT_STR }, 265 { XX("beqwdate"), FILE_BEQWDATE, FILE_FMT_STR }, 266 { XX("name"), FILE_NAME, FILE_FMT_NONE }, 267 { XX("use"), FILE_USE, FILE_FMT_NONE }, 268 { XX("clear"), FILE_CLEAR, FILE_FMT_NONE }, 269 { XX_NULL, FILE_INVALID, FILE_FMT_NONE }, 270 }; 271 272 /* 273 * These are not types, and cannot be preceded by "u" to make them 274 * unsigned. 275 */ 276 static const struct type_tbl_s special_tbl[] = { 277 { XX("name"), FILE_NAME, FILE_FMT_STR }, 278 { XX("use"), FILE_USE, FILE_FMT_STR }, 279 { XX_NULL, FILE_INVALID, FILE_FMT_NONE }, 280 }; 281 # undef XX 282 # undef XX_NULL 283 284 private int 285 get_type(const struct type_tbl_s *tbl, const char *l, const char **t) 286 { 287 const struct type_tbl_s *p; 288 289 for (p = tbl; p->len; p++) { 290 if (strncmp(l, p->name, p->len) == 0) { 291 if (t) 292 *t = l + p->len; 293 break; 294 } 295 } 296 return p->type; 297 } 298 299 private int 300 get_standard_integer_type(const char *l, const char **t) 301 { 302 int type; 303 304 if (isalpha((unsigned char)l[1])) { 305 switch (l[1]) { 306 case 'C': 307 /* "dC" and "uC" */ 308 type = FILE_BYTE; 309 break; 310 case 'S': 311 /* "dS" and "uS" */ 312 type = FILE_SHORT; 313 break; 314 case 'I': 315 case 'L': 316 /* 317 * "dI", "dL", "uI", and "uL". 318 * 319 * XXX - the actual Single UNIX Specification says 320 * that "L" means "long", as in the C data type, 321 * but we treat it as meaning "4-byte integer". 322 * Given that the OS X version of file 5.04 did 323 * the same, I guess that passes the actual SUS 324 * validation suite; having "dL" be dependent on 325 * how big a "long" is on the machine running 326 * "file" is silly. 327 */ 328 type = FILE_LONG; 329 break; 330 case 'Q': 331 /* "dQ" and "uQ" */ 332 type = FILE_QUAD; 333 break; 334 default: 335 /* "d{anything else}", "u{anything else}" */ 336 return FILE_INVALID; 337 } 338 l += 2; 339 } else if (isdigit((unsigned char)l[1])) { 340 /* 341 * "d{num}" and "u{num}"; we only support {num} values 342 * of 1, 2, 4, and 8 - the Single UNIX Specification 343 * doesn't say anything about whether arbitrary 344 * values should be supported, but both the Solaris 10 345 * and OS X Mountain Lion versions of file passed the 346 * Single UNIX Specification validation suite, and 347 * neither of them support values bigger than 8 or 348 * non-power-of-2 values. 349 */ 350 if (isdigit((unsigned char)l[2])) { 351 /* Multi-digit, so > 9 */ 352 return FILE_INVALID; 353 } 354 switch (l[1]) { 355 case '1': 356 type = FILE_BYTE; 357 break; 358 case '2': 359 type = FILE_SHORT; 360 break; 361 case '4': 362 type = FILE_LONG; 363 break; 364 case '8': 365 type = FILE_QUAD; 366 break; 367 default: 368 /* XXX - what about 3, 5, 6, or 7? */ 369 return FILE_INVALID; 370 } 371 l += 2; 372 } else { 373 /* 374 * "d" or "u" by itself. 375 */ 376 type = FILE_LONG; 377 ++l; 378 } 379 if (t) 380 *t = l; 381 return type; 382 } 383 384 private void 385 init_file_tables(void) 386 { 387 static int done = 0; 388 const struct type_tbl_s *p; 389 390 if (done) 391 return; 392 done++; 393 394 for (p = type_tbl; p->len; p++) { 395 assert(p->type < FILE_NAMES_SIZE); 396 file_names[p->type] = p->name; 397 file_formats[p->type] = p->format; 398 } 399 assert(p - type_tbl == FILE_NAMES_SIZE); 400 } 401 402 private int 403 add_mlist(struct mlist *mlp, struct magic_map *map, size_t idx) 404 { 405 struct mlist *ml; 406 407 mlp->map = idx == 0 ? map : NULL; 408 if ((ml = CAST(struct mlist *, malloc(sizeof(*ml)))) == NULL) 409 return -1; 410 411 ml->map = NULL; 412 ml->magic = map->magic[idx]; 413 ml->nmagic = map->nmagic[idx]; 414 415 mlp->prev->next = ml; 416 ml->prev = mlp->prev; 417 ml->next = mlp; 418 mlp->prev = ml; 419 return 0; 420 } 421 422 /* 423 * Handle one file or directory. 424 */ 425 private int 426 apprentice_1(struct magic_set *ms, const char *fn, int action) 427 { 428 struct magic_map *map; 429 #ifndef COMPILE_ONLY 430 struct mlist *ml; 431 size_t i; 432 #endif 433 434 if (magicsize != FILE_MAGICSIZE) { 435 file_error(ms, 0, "magic element size %lu != %lu", 436 (unsigned long)sizeof(*map->magic[0]), 437 (unsigned long)FILE_MAGICSIZE); 438 return -1; 439 } 440 441 if (action == FILE_COMPILE) { 442 map = apprentice_load(ms, fn, action); 443 if (map == NULL) 444 return -1; 445 return apprentice_compile(ms, map, fn); 446 } 447 448 #ifndef COMPILE_ONLY 449 map = apprentice_map(ms, fn); 450 if (map == NULL) { 451 if (ms->flags & MAGIC_CHECK) 452 file_magwarn(ms, "using regular magic file `%s'", fn); 453 map = apprentice_load(ms, fn, action); 454 if (map == NULL) 455 return -1; 456 } 457 458 for (i = 0; i < MAGIC_SETS; i++) { 459 if (add_mlist(ms->mlist[i], map, i) == -1) { 460 file_oomem(ms, sizeof(*ml)); 461 goto fail; 462 } 463 } 464 465 if (action == FILE_LIST) { 466 for (i = 0; i < MAGIC_SETS; i++) { 467 printf("Set %" SIZE_T_FORMAT "u:\nBinary patterns:\n", 468 i); 469 apprentice_list(ms->mlist[i], BINTEST); 470 printf("Text patterns:\n"); 471 apprentice_list(ms->mlist[i], TEXTTEST); 472 } 473 } 474 return 0; 475 fail: 476 for (i = 0; i < MAGIC_SETS; i++) { 477 mlist_free(ms->mlist[i]); 478 ms->mlist[i] = NULL; 479 } 480 return -1; 481 #else 482 return 0; 483 #endif /* COMPILE_ONLY */ 484 } 485 486 protected void 487 file_ms_free(struct magic_set *ms) 488 { 489 size_t i; 490 if (ms == NULL) 491 return; 492 for (i = 0; i < MAGIC_SETS; i++) 493 mlist_free(ms->mlist[i]); 494 free(ms->o.pbuf); 495 free(ms->o.buf); 496 free(ms->c.li); 497 free(ms); 498 } 499 500 protected struct magic_set * 501 file_ms_alloc(int flags) 502 { 503 struct magic_set *ms; 504 size_t i, len; 505 506 if ((ms = CAST(struct magic_set *, calloc((size_t)1, 507 sizeof(struct magic_set)))) == NULL) 508 return NULL; 509 510 if (magic_setflags(ms, flags) == -1) { 511 errno = EINVAL; 512 goto free; 513 } 514 515 ms->o.buf = ms->o.pbuf = NULL; 516 len = (ms->c.len = 10) * sizeof(*ms->c.li); 517 518 if ((ms->c.li = CAST(struct level_info *, malloc(len))) == NULL) 519 goto free; 520 521 ms->event_flags = 0; 522 ms->error = -1; 523 for (i = 0; i < MAGIC_SETS; i++) 524 ms->mlist[i] = NULL; 525 ms->file = "unknown"; 526 ms->line = 0; 527 ms->indir_max = FILE_INDIR_MAX; 528 ms->name_max = FILE_NAME_MAX; 529 ms->elf_shnum_max = FILE_ELF_SHNUM_MAX; 530 ms->elf_phnum_max = FILE_ELF_PHNUM_MAX; 531 ms->elf_notes_max = FILE_ELF_NOTES_MAX; 532 return ms; 533 free: 534 free(ms); 535 return NULL; 536 } 537 538 private void 539 apprentice_unmap(struct magic_map *map) 540 { 541 if (map == NULL) 542 return; 543 544 switch (map->type) { 545 #ifdef QUICK 546 case MAP_TYPE_MMAP: 547 if (map->p) 548 (void)munmap(map->p, map->len); 549 break; 550 #endif 551 case MAP_TYPE_MALLOC: 552 free(map->p); 553 break; 554 case MAP_TYPE_USER: 555 break; 556 default: 557 abort(); 558 } 559 free(map); 560 } 561 562 private struct mlist * 563 mlist_alloc(void) 564 { 565 struct mlist *mlist; 566 if ((mlist = CAST(struct mlist *, calloc(1, sizeof(*mlist)))) == NULL) { 567 return NULL; 568 } 569 mlist->next = mlist->prev = mlist; 570 return mlist; 571 } 572 573 private void 574 mlist_free(struct mlist *mlist) 575 { 576 struct mlist *ml, *next; 577 578 if (mlist == NULL) 579 return; 580 581 ml = mlist->next; 582 for (ml = mlist->next; (next = ml->next) != NULL; ml = next) { 583 if (ml->map) 584 apprentice_unmap(ml->map); 585 free(ml); 586 if (ml == mlist) 587 break; 588 } 589 } 590 591 #ifndef COMPILE_ONLY 592 /* void **bufs: an array of compiled magic files */ 593 protected int 594 buffer_apprentice(struct magic_set *ms, struct magic **bufs, 595 size_t *sizes, size_t nbufs) 596 { 597 size_t i, j; 598 struct mlist *ml; 599 struct magic_map *map; 600 601 if (nbufs == 0) 602 return -1; 603 604 if (ms->mlist[0] != NULL) 605 file_reset(ms); 606 607 init_file_tables(); 608 609 for (i = 0; i < MAGIC_SETS; i++) { 610 mlist_free(ms->mlist[i]); 611 if ((ms->mlist[i] = mlist_alloc()) == NULL) { 612 file_oomem(ms, sizeof(*ms->mlist[i])); 613 goto fail; 614 } 615 } 616 617 for (i = 0; i < nbufs; i++) { 618 map = apprentice_buf(ms, bufs[i], sizes[i]); 619 if (map == NULL) 620 goto fail; 621 622 for (j = 0; j < MAGIC_SETS; j++) { 623 if (add_mlist(ms->mlist[j], map, j) == -1) { 624 file_oomem(ms, sizeof(*ml)); 625 goto fail; 626 } 627 } 628 } 629 630 return 0; 631 fail: 632 for (i = 0; i < MAGIC_SETS; i++) { 633 mlist_free(ms->mlist[i]); 634 ms->mlist[i] = NULL; 635 } 636 return -1; 637 } 638 #endif 639 640 /* const char *fn: list of magic files and directories */ 641 protected int 642 file_apprentice(struct magic_set *ms, const char *fn, int action) 643 { 644 char *p, *mfn; 645 int file_err, errs = -1; 646 size_t i; 647 648 if (ms->mlist[0] != NULL) 649 file_reset(ms); 650 651 if ((fn = magic_getpath(fn, action)) == NULL) 652 return -1; 653 654 init_file_tables(); 655 656 if ((mfn = strdup(fn)) == NULL) { 657 file_oomem(ms, strlen(fn)); 658 return -1; 659 } 660 661 for (i = 0; i < MAGIC_SETS; i++) { 662 mlist_free(ms->mlist[i]); 663 if ((ms->mlist[i] = mlist_alloc()) == NULL) { 664 file_oomem(ms, sizeof(*ms->mlist[i])); 665 while (i-- > 0) { 666 mlist_free(ms->mlist[i]); 667 ms->mlist[i] = NULL; 668 } 669 free(mfn); 670 return -1; 671 } 672 } 673 fn = mfn; 674 675 while (fn) { 676 p = strchr(fn, PATHSEP); 677 if (p) 678 *p++ = '\0'; 679 if (*fn == '\0') 680 break; 681 file_err = apprentice_1(ms, fn, action); 682 errs = MAX(errs, file_err); 683 fn = p; 684 } 685 686 free(mfn); 687 688 if (errs == -1) { 689 for (i = 0; i < MAGIC_SETS; i++) { 690 mlist_free(ms->mlist[i]); 691 ms->mlist[i] = NULL; 692 } 693 file_error(ms, 0, "could not find any valid magic files!"); 694 return -1; 695 } 696 697 #if 0 698 /* 699 * Always leave the database loaded 700 */ 701 if (action == FILE_LOAD) 702 return 0; 703 704 for (i = 0; i < MAGIC_SETS; i++) { 705 mlist_free(ms->mlist[i]); 706 ms->mlist[i] = NULL; 707 } 708 #endif 709 710 switch (action) { 711 case FILE_LOAD: 712 case FILE_COMPILE: 713 case FILE_CHECK: 714 case FILE_LIST: 715 return 0; 716 default: 717 file_error(ms, 0, "Invalid action %d", action); 718 return -1; 719 } 720 } 721 722 /* 723 * Compute the real length of a magic expression, for the purposes 724 * of determining how "strong" a magic expression is (approximating 725 * how specific its matches are): 726 * - magic characters count 0 unless escaped. 727 * - [] expressions count 1 728 * - {} expressions count 0 729 * - regular characters or escaped magic characters count 1 730 * - 0 length expressions count as one 731 */ 732 private size_t 733 nonmagic(const char *str) 734 { 735 const char *p; 736 size_t rv = 0; 737 738 for (p = str; *p; p++) 739 switch (*p) { 740 case '\\': /* Escaped anything counts 1 */ 741 if (!*++p) 742 p--; 743 rv++; 744 continue; 745 case '?': /* Magic characters count 0 */ 746 case '*': 747 case '.': 748 case '+': 749 case '^': 750 case '$': 751 continue; 752 case '[': /* Bracketed expressions count 1 the ']' */ 753 while (*p && *p != ']') 754 p++; 755 p--; 756 continue; 757 case '{': /* Braced expressions count 0 */ 758 while (*p && *p != '}') 759 p++; 760 if (!*p) 761 p--; 762 continue; 763 default: /* Anything else counts 1 */ 764 rv++; 765 continue; 766 } 767 768 return rv == 0 ? 1 : rv; /* Return at least 1 */ 769 } 770 771 /* 772 * Get weight of this magic entry, for sorting purposes. 773 */ 774 private size_t 775 apprentice_magic_strength(const struct magic *m) 776 { 777 #define MULT 10 778 size_t v, val = 2 * MULT; /* baseline strength */ 779 780 switch (m->type) { 781 case FILE_DEFAULT: /* make sure this sorts last */ 782 if (m->factor_op != FILE_FACTOR_OP_NONE) 783 abort(); 784 return 0; 785 786 case FILE_BYTE: 787 val += 1 * MULT; 788 break; 789 790 case FILE_SHORT: 791 case FILE_LESHORT: 792 case FILE_BESHORT: 793 val += 2 * MULT; 794 break; 795 796 case FILE_LONG: 797 case FILE_LELONG: 798 case FILE_BELONG: 799 case FILE_MELONG: 800 val += 4 * MULT; 801 break; 802 803 case FILE_PSTRING: 804 case FILE_STRING: 805 val += m->vallen * MULT; 806 break; 807 808 case FILE_BESTRING16: 809 case FILE_LESTRING16: 810 val += m->vallen * MULT / 2; 811 break; 812 813 case FILE_SEARCH: 814 val += m->vallen * MAX(MULT / m->vallen, 1); 815 break; 816 817 case FILE_REGEX: 818 v = nonmagic(m->value.s); 819 val += v * MAX(MULT / v, 1); 820 break; 821 822 case FILE_DATE: 823 case FILE_LEDATE: 824 case FILE_BEDATE: 825 case FILE_MEDATE: 826 case FILE_LDATE: 827 case FILE_LELDATE: 828 case FILE_BELDATE: 829 case FILE_MELDATE: 830 case FILE_FLOAT: 831 case FILE_BEFLOAT: 832 case FILE_LEFLOAT: 833 val += 4 * MULT; 834 break; 835 836 case FILE_QUAD: 837 case FILE_BEQUAD: 838 case FILE_LEQUAD: 839 case FILE_QDATE: 840 case FILE_LEQDATE: 841 case FILE_BEQDATE: 842 case FILE_QLDATE: 843 case FILE_LEQLDATE: 844 case FILE_BEQLDATE: 845 case FILE_QWDATE: 846 case FILE_LEQWDATE: 847 case FILE_BEQWDATE: 848 case FILE_DOUBLE: 849 case FILE_BEDOUBLE: 850 case FILE_LEDOUBLE: 851 val += 8 * MULT; 852 break; 853 854 case FILE_INDIRECT: 855 case FILE_NAME: 856 case FILE_USE: 857 break; 858 859 default: 860 (void)fprintf(stderr, "Bad type %d\n", m->type); 861 abort(); 862 } 863 864 switch (m->reln) { 865 case 'x': /* matches anything penalize */ 866 case '!': /* matches almost anything penalize */ 867 val = 0; 868 break; 869 870 case '=': /* Exact match, prefer */ 871 val += MULT; 872 break; 873 874 case '>': 875 case '<': /* comparison match reduce strength */ 876 val -= 2 * MULT; 877 break; 878 879 case '^': 880 case '&': /* masking bits, we could count them too */ 881 val -= MULT; 882 break; 883 884 default: 885 (void)fprintf(stderr, "Bad relation %c\n", m->reln); 886 abort(); 887 } 888 889 if (val == 0) /* ensure we only return 0 for FILE_DEFAULT */ 890 val = 1; 891 892 switch (m->factor_op) { 893 case FILE_FACTOR_OP_NONE: 894 break; 895 case FILE_FACTOR_OP_PLUS: 896 val += m->factor; 897 break; 898 case FILE_FACTOR_OP_MINUS: 899 val -= m->factor; 900 break; 901 case FILE_FACTOR_OP_TIMES: 902 val *= m->factor; 903 break; 904 case FILE_FACTOR_OP_DIV: 905 val /= m->factor; 906 break; 907 default: 908 abort(); 909 } 910 911 /* 912 * Magic entries with no description get a bonus because they depend 913 * on subsequent magic entries to print something. 914 */ 915 if (m->desc[0] == '\0') 916 val++; 917 return val; 918 } 919 920 /* 921 * Sort callback for sorting entries by "strength" (basically length) 922 */ 923 private int 924 apprentice_sort(const void *a, const void *b) 925 { 926 const struct magic_entry *ma = CAST(const struct magic_entry *, a); 927 const struct magic_entry *mb = CAST(const struct magic_entry *, b); 928 size_t sa = apprentice_magic_strength(ma->mp); 929 size_t sb = apprentice_magic_strength(mb->mp); 930 if (sa == sb) 931 return 0; 932 else if (sa > sb) 933 return -1; 934 else 935 return 1; 936 } 937 938 /* 939 * Shows sorted patterns list in the order which is used for the matching 940 */ 941 private void 942 apprentice_list(struct mlist *mlist, int mode) 943 { 944 uint32_t magindex = 0; 945 struct mlist *ml; 946 for (ml = mlist->next; ml != mlist; ml = ml->next) { 947 for (magindex = 0; magindex < ml->nmagic; magindex++) { 948 struct magic *m = &ml->magic[magindex]; 949 if ((m->flag & mode) != mode) { 950 /* Skip sub-tests */ 951 while (magindex + 1 < ml->nmagic && 952 ml->magic[magindex + 1].cont_level != 0) 953 ++magindex; 954 continue; /* Skip to next top-level test*/ 955 } 956 957 /* 958 * Try to iterate over the tree until we find item with 959 * description/mimetype. 960 */ 961 while (magindex + 1 < ml->nmagic && 962 ml->magic[magindex + 1].cont_level != 0 && 963 *ml->magic[magindex].desc == '\0' && 964 *ml->magic[magindex].mimetype == '\0') 965 magindex++; 966 967 printf("Strength = %3" SIZE_T_FORMAT "u : %s [%s]\n", 968 apprentice_magic_strength(m), 969 ml->magic[magindex].desc, 970 ml->magic[magindex].mimetype); 971 } 972 } 973 } 974 975 private void 976 set_test_type(struct magic *mstart, struct magic *m) 977 { 978 switch (m->type) { 979 case FILE_BYTE: 980 case FILE_SHORT: 981 case FILE_LONG: 982 case FILE_DATE: 983 case FILE_BESHORT: 984 case FILE_BELONG: 985 case FILE_BEDATE: 986 case FILE_LESHORT: 987 case FILE_LELONG: 988 case FILE_LEDATE: 989 case FILE_LDATE: 990 case FILE_BELDATE: 991 case FILE_LELDATE: 992 case FILE_MEDATE: 993 case FILE_MELDATE: 994 case FILE_MELONG: 995 case FILE_QUAD: 996 case FILE_LEQUAD: 997 case FILE_BEQUAD: 998 case FILE_QDATE: 999 case FILE_LEQDATE: 1000 case FILE_BEQDATE: 1001 case FILE_QLDATE: 1002 case FILE_LEQLDATE: 1003 case FILE_BEQLDATE: 1004 case FILE_QWDATE: 1005 case FILE_LEQWDATE: 1006 case FILE_BEQWDATE: 1007 case FILE_FLOAT: 1008 case FILE_BEFLOAT: 1009 case FILE_LEFLOAT: 1010 case FILE_DOUBLE: 1011 case FILE_BEDOUBLE: 1012 case FILE_LEDOUBLE: 1013 mstart->flag |= BINTEST; 1014 break; 1015 case FILE_STRING: 1016 case FILE_PSTRING: 1017 case FILE_BESTRING16: 1018 case FILE_LESTRING16: 1019 /* Allow text overrides */ 1020 if (mstart->str_flags & STRING_TEXTTEST) 1021 mstart->flag |= TEXTTEST; 1022 else 1023 mstart->flag |= BINTEST; 1024 break; 1025 case FILE_REGEX: 1026 case FILE_SEARCH: 1027 /* Check for override */ 1028 if (mstart->str_flags & STRING_BINTEST) 1029 mstart->flag |= BINTEST; 1030 if (mstart->str_flags & STRING_TEXTTEST) 1031 mstart->flag |= TEXTTEST; 1032 1033 if (mstart->flag & (TEXTTEST|BINTEST)) 1034 break; 1035 1036 /* binary test if pattern is not text */ 1037 if (file_looks_utf8(m->value.us, (size_t)m->vallen, NULL, 1038 NULL) <= 0) 1039 mstart->flag |= BINTEST; 1040 else 1041 mstart->flag |= TEXTTEST; 1042 break; 1043 case FILE_DEFAULT: 1044 /* can't deduce anything; we shouldn't see this at the 1045 top level anyway */ 1046 break; 1047 case FILE_INVALID: 1048 default: 1049 /* invalid search type, but no need to complain here */ 1050 break; 1051 } 1052 } 1053 1054 private int 1055 addentry(struct magic_set *ms, struct magic_entry *me, 1056 struct magic_entry_set *mset) 1057 { 1058 size_t i = me->mp->type == FILE_NAME ? 1 : 0; 1059 if (mset[i].count == mset[i].max) { 1060 struct magic_entry *mp; 1061 1062 mset[i].max += ALLOC_INCR; 1063 if ((mp = CAST(struct magic_entry *, 1064 realloc(mset[i].me, sizeof(*mp) * mset[i].max))) == 1065 NULL) { 1066 file_oomem(ms, sizeof(*mp) * mset[i].max); 1067 return -1; 1068 } 1069 (void)memset(&mp[mset[i].count], 0, sizeof(*mp) * 1070 ALLOC_INCR); 1071 mset[i].me = mp; 1072 } 1073 mset[i].me[mset[i].count++] = *me; 1074 memset(me, 0, sizeof(*me)); 1075 return 0; 1076 } 1077 1078 /* 1079 * Load and parse one file. 1080 */ 1081 private void 1082 load_1(struct magic_set *ms, int action, const char *fn, int *errs, 1083 struct magic_entry_set *mset) 1084 { 1085 size_t lineno = 0, llen = 0; 1086 char *line = NULL; 1087 ssize_t len; 1088 struct magic_entry me; 1089 1090 FILE *f = fopen(ms->file = fn, "r"); 1091 if (f == NULL) { 1092 if (errno != ENOENT) 1093 file_error(ms, errno, "cannot read magic file `%s'", 1094 fn); 1095 (*errs)++; 1096 return; 1097 } 1098 1099 memset(&me, 0, sizeof(me)); 1100 /* read and parse this file */ 1101 for (ms->line = 1; (len = getline(&line, &llen, f)) != -1; 1102 ms->line++) { 1103 if (len == 0) /* null line, garbage, etc */ 1104 continue; 1105 if (line[len - 1] == '\n') { 1106 lineno++; 1107 line[len - 1] = '\0'; /* delete newline */ 1108 } 1109 switch (line[0]) { 1110 case '\0': /* empty, do not parse */ 1111 case '#': /* comment, do not parse */ 1112 continue; 1113 case '!': 1114 if (line[1] == ':') { 1115 size_t i; 1116 1117 for (i = 0; bang[i].name != NULL; i++) { 1118 if ((size_t)(len - 2) > bang[i].len && 1119 memcmp(bang[i].name, line + 2, 1120 bang[i].len) == 0) 1121 break; 1122 } 1123 if (bang[i].name == NULL) { 1124 file_error(ms, 0, 1125 "Unknown !: entry `%s'", line); 1126 (*errs)++; 1127 continue; 1128 } 1129 if (me.mp == NULL) { 1130 file_error(ms, 0, 1131 "No current entry for :!%s type", 1132 bang[i].name); 1133 (*errs)++; 1134 continue; 1135 } 1136 if ((*bang[i].fun)(ms, &me, 1137 line + bang[i].len + 2) != 0) { 1138 (*errs)++; 1139 continue; 1140 } 1141 continue; 1142 } 1143 /*FALLTHROUGH*/ 1144 default: 1145 again: 1146 switch (parse(ms, &me, line, lineno, action)) { 1147 case 0: 1148 continue; 1149 case 1: 1150 (void)addentry(ms, &me, mset); 1151 goto again; 1152 default: 1153 (*errs)++; 1154 break; 1155 } 1156 } 1157 } 1158 if (me.mp) 1159 (void)addentry(ms, &me, mset); 1160 free(line); 1161 (void)fclose(f); 1162 } 1163 1164 /* 1165 * parse a file or directory of files 1166 * const char *fn: name of magic file or directory 1167 */ 1168 private int 1169 cmpstrp(const void *p1, const void *p2) 1170 { 1171 return strcmp(*(char *const *)p1, *(char *const *)p2); 1172 } 1173 1174 1175 private uint32_t 1176 set_text_binary(struct magic_set *ms, struct magic_entry *me, uint32_t nme, 1177 uint32_t starttest) 1178 { 1179 static const char text[] = "text"; 1180 static const char binary[] = "binary"; 1181 static const size_t len = sizeof(text); 1182 1183 uint32_t i = starttest; 1184 1185 do { 1186 set_test_type(me[starttest].mp, me[i].mp); 1187 if ((ms->flags & MAGIC_DEBUG) == 0) 1188 continue; 1189 (void)fprintf(stderr, "%s%s%s: %s\n", 1190 me[i].mp->mimetype, 1191 me[i].mp->mimetype[0] == '\0' ? "" : "; ", 1192 me[i].mp->desc[0] ? me[i].mp->desc : "(no description)", 1193 me[i].mp->flag & BINTEST ? binary : text); 1194 if (me[i].mp->flag & BINTEST) { 1195 char *p = strstr(me[i].mp->desc, text); 1196 if (p && (p == me[i].mp->desc || 1197 isspace((unsigned char)p[-1])) && 1198 (p + len - me[i].mp->desc == MAXstring 1199 || (p[len] == '\0' || 1200 isspace((unsigned char)p[len])))) 1201 (void)fprintf(stderr, "*** Possible " 1202 "binary test for text type\n"); 1203 } 1204 } while (++i < nme && me[i].mp->cont_level != 0); 1205 return i; 1206 } 1207 1208 private void 1209 set_last_default(struct magic_set *ms, struct magic_entry *me, uint32_t nme) 1210 { 1211 uint32_t i; 1212 for (i = 0; i < nme; i++) { 1213 if (me[i].mp->cont_level == 0 && 1214 me[i].mp->type == FILE_DEFAULT) { 1215 while (++i < nme) 1216 if (me[i].mp->cont_level == 0) 1217 break; 1218 if (i != nme) { 1219 /* XXX - Ugh! */ 1220 ms->line = me[i].mp->lineno; 1221 file_magwarn(ms, 1222 "level 0 \"default\" did not sort last"); 1223 } 1224 return; 1225 } 1226 } 1227 } 1228 1229 private int 1230 coalesce_entries(struct magic_set *ms, struct magic_entry *me, uint32_t nme, 1231 struct magic **ma, uint32_t *nma) 1232 { 1233 uint32_t i, mentrycount = 0; 1234 size_t slen; 1235 1236 for (i = 0; i < nme; i++) 1237 mentrycount += me[i].cont_count; 1238 1239 slen = sizeof(**ma) * mentrycount; 1240 if ((*ma = CAST(struct magic *, malloc(slen))) == NULL) { 1241 file_oomem(ms, slen); 1242 return -1; 1243 } 1244 1245 mentrycount = 0; 1246 for (i = 0; i < nme; i++) { 1247 (void)memcpy(*ma + mentrycount, me[i].mp, 1248 me[i].cont_count * sizeof(**ma)); 1249 mentrycount += me[i].cont_count; 1250 } 1251 *nma = mentrycount; 1252 return 0; 1253 } 1254 1255 private void 1256 magic_entry_free(struct magic_entry *me, uint32_t nme) 1257 { 1258 uint32_t i; 1259 if (me == NULL) 1260 return; 1261 for (i = 0; i < nme; i++) 1262 free(me[i].mp); 1263 free(me); 1264 } 1265 1266 private struct magic_map * 1267 apprentice_load(struct magic_set *ms, const char *fn, int action) 1268 { 1269 int errs = 0; 1270 uint32_t i, j; 1271 size_t files = 0, maxfiles = 0; 1272 char **filearr = NULL, *mfn; 1273 struct stat st; 1274 struct magic_map *map; 1275 struct magic_entry_set mset[MAGIC_SETS]; 1276 DIR *dir; 1277 struct dirent *d; 1278 1279 memset(mset, 0, sizeof(mset)); 1280 ms->flags |= MAGIC_CHECK; /* Enable checks for parsed files */ 1281 1282 1283 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) 1284 { 1285 file_oomem(ms, sizeof(*map)); 1286 return NULL; 1287 } 1288 1289 /* print silly verbose header for USG compat. */ 1290 if (action == FILE_CHECK) 1291 (void)fprintf(stderr, "%s\n", usg_hdr); 1292 1293 /* load directory or file */ 1294 if (stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) { 1295 dir = opendir(fn); 1296 if (!dir) { 1297 errs++; 1298 goto out; 1299 } 1300 while ((d = readdir(dir)) != NULL) { 1301 if (asprintf(&mfn, "%s/%s", fn, d->d_name) < 0) { 1302 file_oomem(ms, 1303 strlen(fn) + strlen(d->d_name) + 2); 1304 errs++; 1305 closedir(dir); 1306 goto out; 1307 } 1308 if (stat(mfn, &st) == -1 || !S_ISREG(st.st_mode)) { 1309 free(mfn); 1310 continue; 1311 } 1312 if (files >= maxfiles) { 1313 size_t mlen; 1314 maxfiles = (maxfiles + 1) * 2; 1315 mlen = maxfiles * sizeof(*filearr); 1316 if ((filearr = CAST(char **, 1317 realloc(filearr, mlen))) == NULL) { 1318 file_oomem(ms, mlen); 1319 free(mfn); 1320 closedir(dir); 1321 errs++; 1322 goto out; 1323 } 1324 } 1325 filearr[files++] = mfn; 1326 } 1327 closedir(dir); 1328 qsort(filearr, files, sizeof(*filearr), cmpstrp); 1329 for (i = 0; i < files; i++) { 1330 load_1(ms, action, filearr[i], &errs, mset); 1331 free(filearr[i]); 1332 } 1333 free(filearr); 1334 } else 1335 load_1(ms, action, fn, &errs, mset); 1336 if (errs) 1337 goto out; 1338 1339 for (j = 0; j < MAGIC_SETS; j++) { 1340 /* Set types of tests */ 1341 for (i = 0; i < mset[j].count; ) { 1342 if (mset[j].me[i].mp->cont_level != 0) { 1343 i++; 1344 continue; 1345 } 1346 i = set_text_binary(ms, mset[j].me, mset[j].count, i); 1347 } 1348 qsort(mset[j].me, mset[j].count, sizeof(*mset[j].me), 1349 apprentice_sort); 1350 1351 /* 1352 * Make sure that any level 0 "default" line is last 1353 * (if one exists). 1354 */ 1355 set_last_default(ms, mset[j].me, mset[j].count); 1356 1357 /* coalesce per file arrays into a single one */ 1358 if (coalesce_entries(ms, mset[j].me, mset[j].count, 1359 &map->magic[j], &map->nmagic[j]) == -1) { 1360 errs++; 1361 goto out; 1362 } 1363 } 1364 1365 out: 1366 for (j = 0; j < MAGIC_SETS; j++) 1367 magic_entry_free(mset[j].me, mset[j].count); 1368 1369 if (errs) { 1370 apprentice_unmap(map); 1371 return NULL; 1372 } 1373 return map; 1374 } 1375 1376 /* 1377 * extend the sign bit if the comparison is to be signed 1378 */ 1379 protected uint64_t 1380 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v) 1381 { 1382 if (!(m->flag & UNSIGNED)) { 1383 switch(m->type) { 1384 /* 1385 * Do not remove the casts below. They are 1386 * vital. When later compared with the data, 1387 * the sign extension must have happened. 1388 */ 1389 case FILE_BYTE: 1390 v = (signed char) v; 1391 break; 1392 case FILE_SHORT: 1393 case FILE_BESHORT: 1394 case FILE_LESHORT: 1395 v = (short) v; 1396 break; 1397 case FILE_DATE: 1398 case FILE_BEDATE: 1399 case FILE_LEDATE: 1400 case FILE_MEDATE: 1401 case FILE_LDATE: 1402 case FILE_BELDATE: 1403 case FILE_LELDATE: 1404 case FILE_MELDATE: 1405 case FILE_LONG: 1406 case FILE_BELONG: 1407 case FILE_LELONG: 1408 case FILE_MELONG: 1409 case FILE_FLOAT: 1410 case FILE_BEFLOAT: 1411 case FILE_LEFLOAT: 1412 v = (int32_t) v; 1413 break; 1414 case FILE_QUAD: 1415 case FILE_BEQUAD: 1416 case FILE_LEQUAD: 1417 case FILE_QDATE: 1418 case FILE_QLDATE: 1419 case FILE_QWDATE: 1420 case FILE_BEQDATE: 1421 case FILE_BEQLDATE: 1422 case FILE_BEQWDATE: 1423 case FILE_LEQDATE: 1424 case FILE_LEQLDATE: 1425 case FILE_LEQWDATE: 1426 case FILE_DOUBLE: 1427 case FILE_BEDOUBLE: 1428 case FILE_LEDOUBLE: 1429 v = (int64_t) v; 1430 break; 1431 case FILE_STRING: 1432 case FILE_PSTRING: 1433 case FILE_BESTRING16: 1434 case FILE_LESTRING16: 1435 case FILE_REGEX: 1436 case FILE_SEARCH: 1437 case FILE_DEFAULT: 1438 case FILE_INDIRECT: 1439 case FILE_NAME: 1440 case FILE_USE: 1441 case FILE_CLEAR: 1442 break; 1443 default: 1444 if (ms->flags & MAGIC_CHECK) 1445 file_magwarn(ms, "cannot happen: m->type=%d\n", 1446 m->type); 1447 return ~0U; 1448 } 1449 } 1450 return v; 1451 } 1452 1453 private int 1454 string_modifier_check(struct magic_set *ms, struct magic *m) 1455 { 1456 if ((ms->flags & MAGIC_CHECK) == 0) 1457 return 0; 1458 1459 if ((m->type != FILE_REGEX || (m->str_flags & REGEX_LINE_COUNT) == 0) && 1460 (m->type != FILE_PSTRING && (m->str_flags & PSTRING_LEN) != 0)) { 1461 file_magwarn(ms, 1462 "'/BHhLl' modifiers are only allowed for pascal strings\n"); 1463 return -1; 1464 } 1465 switch (m->type) { 1466 case FILE_BESTRING16: 1467 case FILE_LESTRING16: 1468 if (m->str_flags != 0) { 1469 file_magwarn(ms, 1470 "no modifiers allowed for 16-bit strings\n"); 1471 return -1; 1472 } 1473 break; 1474 case FILE_STRING: 1475 case FILE_PSTRING: 1476 if ((m->str_flags & REGEX_OFFSET_START) != 0) { 1477 file_magwarn(ms, 1478 "'/%c' only allowed on regex and search\n", 1479 CHAR_REGEX_OFFSET_START); 1480 return -1; 1481 } 1482 break; 1483 case FILE_SEARCH: 1484 if (m->str_range == 0) { 1485 file_magwarn(ms, 1486 "missing range; defaulting to %d\n", 1487 STRING_DEFAULT_RANGE); 1488 m->str_range = STRING_DEFAULT_RANGE; 1489 return -1; 1490 } 1491 break; 1492 case FILE_REGEX: 1493 if ((m->str_flags & STRING_COMPACT_WHITESPACE) != 0) { 1494 file_magwarn(ms, "'/%c' not allowed on regex\n", 1495 CHAR_COMPACT_WHITESPACE); 1496 return -1; 1497 } 1498 if ((m->str_flags & STRING_COMPACT_OPTIONAL_WHITESPACE) != 0) { 1499 file_magwarn(ms, "'/%c' not allowed on regex\n", 1500 CHAR_COMPACT_OPTIONAL_WHITESPACE); 1501 return -1; 1502 } 1503 break; 1504 default: 1505 file_magwarn(ms, "coding error: m->type=%d\n", 1506 m->type); 1507 return -1; 1508 } 1509 return 0; 1510 } 1511 1512 private int 1513 get_op(char c) 1514 { 1515 switch (c) { 1516 case '&': 1517 return FILE_OPAND; 1518 case '|': 1519 return FILE_OPOR; 1520 case '^': 1521 return FILE_OPXOR; 1522 case '+': 1523 return FILE_OPADD; 1524 case '-': 1525 return FILE_OPMINUS; 1526 case '*': 1527 return FILE_OPMULTIPLY; 1528 case '/': 1529 return FILE_OPDIVIDE; 1530 case '%': 1531 return FILE_OPMODULO; 1532 default: 1533 return -1; 1534 } 1535 } 1536 1537 #ifdef ENABLE_CONDITIONALS 1538 private int 1539 get_cond(const char *l, const char **t) 1540 { 1541 static const struct cond_tbl_s { 1542 char name[8]; 1543 size_t len; 1544 int cond; 1545 } cond_tbl[] = { 1546 { "if", 2, COND_IF }, 1547 { "elif", 4, COND_ELIF }, 1548 { "else", 4, COND_ELSE }, 1549 { "", 0, COND_NONE }, 1550 }; 1551 const struct cond_tbl_s *p; 1552 1553 for (p = cond_tbl; p->len; p++) { 1554 if (strncmp(l, p->name, p->len) == 0 && 1555 isspace((unsigned char)l[p->len])) { 1556 if (t) 1557 *t = l + p->len; 1558 break; 1559 } 1560 } 1561 return p->cond; 1562 } 1563 1564 private int 1565 check_cond(struct magic_set *ms, int cond, uint32_t cont_level) 1566 { 1567 int last_cond; 1568 last_cond = ms->c.li[cont_level].last_cond; 1569 1570 switch (cond) { 1571 case COND_IF: 1572 if (last_cond != COND_NONE && last_cond != COND_ELIF) { 1573 if (ms->flags & MAGIC_CHECK) 1574 file_magwarn(ms, "syntax error: `if'"); 1575 return -1; 1576 } 1577 last_cond = COND_IF; 1578 break; 1579 1580 case COND_ELIF: 1581 if (last_cond != COND_IF && last_cond != COND_ELIF) { 1582 if (ms->flags & MAGIC_CHECK) 1583 file_magwarn(ms, "syntax error: `elif'"); 1584 return -1; 1585 } 1586 last_cond = COND_ELIF; 1587 break; 1588 1589 case COND_ELSE: 1590 if (last_cond != COND_IF && last_cond != COND_ELIF) { 1591 if (ms->flags & MAGIC_CHECK) 1592 file_magwarn(ms, "syntax error: `else'"); 1593 return -1; 1594 } 1595 last_cond = COND_NONE; 1596 break; 1597 1598 case COND_NONE: 1599 last_cond = COND_NONE; 1600 break; 1601 } 1602 1603 ms->c.li[cont_level].last_cond = last_cond; 1604 return 0; 1605 } 1606 #endif /* ENABLE_CONDITIONALS */ 1607 1608 private int 1609 parse_indirect_modifier(struct magic_set *ms, struct magic *m, const char **lp) 1610 { 1611 const char *l = *lp; 1612 1613 while (!isspace((unsigned char)*++l)) 1614 switch (*l) { 1615 case CHAR_INDIRECT_RELATIVE: 1616 m->str_flags |= INDIRECT_RELATIVE; 1617 break; 1618 default: 1619 if (ms->flags & MAGIC_CHECK) 1620 file_magwarn(ms, "indirect modifier `%c' " 1621 "invalid", *l); 1622 *lp = l; 1623 return -1; 1624 } 1625 *lp = l; 1626 return 0; 1627 } 1628 1629 private void 1630 parse_op_modifier(struct magic_set *ms, struct magic *m, const char **lp, 1631 int op) 1632 { 1633 const char *l = *lp; 1634 char *t; 1635 uint64_t val; 1636 1637 ++l; 1638 m->mask_op |= op; 1639 val = (uint64_t)strtoull(l, &t, 0); 1640 l = t; 1641 m->num_mask = file_signextend(ms, m, val); 1642 eatsize(&l); 1643 *lp = l; 1644 } 1645 1646 private int 1647 parse_string_modifier(struct magic_set *ms, struct magic *m, const char **lp) 1648 { 1649 const char *l = *lp; 1650 char *t; 1651 int have_range = 0; 1652 1653 while (!isspace((unsigned char)*++l)) { 1654 switch (*l) { 1655 case '0': case '1': case '2': 1656 case '3': case '4': case '5': 1657 case '6': case '7': case '8': 1658 case '9': 1659 if (have_range && (ms->flags & MAGIC_CHECK)) 1660 file_magwarn(ms, "multiple ranges"); 1661 have_range = 1; 1662 m->str_range = CAST(uint32_t, strtoul(l, &t, 0)); 1663 if (m->str_range == 0) 1664 file_magwarn(ms, "zero range"); 1665 l = t - 1; 1666 break; 1667 case CHAR_COMPACT_WHITESPACE: 1668 m->str_flags |= STRING_COMPACT_WHITESPACE; 1669 break; 1670 case CHAR_COMPACT_OPTIONAL_WHITESPACE: 1671 m->str_flags |= STRING_COMPACT_OPTIONAL_WHITESPACE; 1672 break; 1673 case CHAR_IGNORE_LOWERCASE: 1674 m->str_flags |= STRING_IGNORE_LOWERCASE; 1675 break; 1676 case CHAR_IGNORE_UPPERCASE: 1677 m->str_flags |= STRING_IGNORE_UPPERCASE; 1678 break; 1679 case CHAR_REGEX_OFFSET_START: 1680 m->str_flags |= REGEX_OFFSET_START; 1681 break; 1682 case CHAR_BINTEST: 1683 m->str_flags |= STRING_BINTEST; 1684 break; 1685 case CHAR_TEXTTEST: 1686 m->str_flags |= STRING_TEXTTEST; 1687 break; 1688 case CHAR_TRIM: 1689 m->str_flags |= STRING_TRIM; 1690 break; 1691 case CHAR_PSTRING_1_LE: 1692 #define SET_LENGTH(a) m->str_flags = (m->str_flags & ~PSTRING_LEN) | (a) 1693 if (m->type != FILE_PSTRING) 1694 goto bad; 1695 SET_LENGTH(PSTRING_1_LE); 1696 break; 1697 case CHAR_PSTRING_2_BE: 1698 if (m->type != FILE_PSTRING) 1699 goto bad; 1700 SET_LENGTH(PSTRING_2_BE); 1701 break; 1702 case CHAR_PSTRING_2_LE: 1703 if (m->type != FILE_PSTRING) 1704 goto bad; 1705 SET_LENGTH(PSTRING_2_LE); 1706 break; 1707 case CHAR_PSTRING_4_BE: 1708 if (m->type != FILE_PSTRING) 1709 goto bad; 1710 SET_LENGTH(PSTRING_4_BE); 1711 break; 1712 case CHAR_PSTRING_4_LE: 1713 switch (m->type) { 1714 case FILE_PSTRING: 1715 case FILE_REGEX: 1716 break; 1717 default: 1718 goto bad; 1719 } 1720 SET_LENGTH(PSTRING_4_LE); 1721 break; 1722 case CHAR_PSTRING_LENGTH_INCLUDES_ITSELF: 1723 if (m->type != FILE_PSTRING) 1724 goto bad; 1725 m->str_flags |= PSTRING_LENGTH_INCLUDES_ITSELF; 1726 break; 1727 default: 1728 bad: 1729 if (ms->flags & MAGIC_CHECK) 1730 file_magwarn(ms, "string modifier `%c' " 1731 "invalid", *l); 1732 goto out; 1733 } 1734 /* allow multiple '/' for readability */ 1735 if (l[1] == '/' && !isspace((unsigned char)l[2])) 1736 l++; 1737 } 1738 if (string_modifier_check(ms, m) == -1) 1739 goto out; 1740 *lp = l; 1741 return 0; 1742 out: 1743 *lp = l; 1744 return -1; 1745 } 1746 1747 /* 1748 * parse one line from magic file, put into magic[index++] if valid 1749 */ 1750 private int 1751 parse(struct magic_set *ms, struct magic_entry *me, const char *line, 1752 size_t lineno, int action) 1753 { 1754 #ifdef ENABLE_CONDITIONALS 1755 static uint32_t last_cont_level = 0; 1756 #endif 1757 size_t i; 1758 struct magic *m; 1759 const char *l = line; 1760 char *t; 1761 int op; 1762 uint32_t cont_level; 1763 int32_t diff; 1764 1765 cont_level = 0; 1766 1767 /* 1768 * Parse the offset. 1769 */ 1770 while (*l == '>') { 1771 ++l; /* step over */ 1772 cont_level++; 1773 } 1774 #ifdef ENABLE_CONDITIONALS 1775 if (cont_level == 0 || cont_level > last_cont_level) 1776 if (file_check_mem(ms, cont_level) == -1) 1777 return -1; 1778 last_cont_level = cont_level; 1779 #endif 1780 if (cont_level != 0) { 1781 if (me->mp == NULL) { 1782 file_magerror(ms, "No current entry for continuation"); 1783 return -1; 1784 } 1785 if (me->cont_count == 0) { 1786 file_magerror(ms, "Continuations present with 0 count"); 1787 return -1; 1788 } 1789 m = &me->mp[me->cont_count - 1]; 1790 diff = (int32_t)cont_level - (int32_t)m->cont_level; 1791 if (diff > 1) 1792 file_magwarn(ms, "New continuation level %u is more " 1793 "than one larger than current level %u", cont_level, 1794 m->cont_level); 1795 if (me->cont_count == me->max_count) { 1796 struct magic *nm; 1797 size_t cnt = me->max_count + ALLOC_CHUNK; 1798 if ((nm = CAST(struct magic *, realloc(me->mp, 1799 sizeof(*nm) * cnt))) == NULL) { 1800 file_oomem(ms, sizeof(*nm) * cnt); 1801 return -1; 1802 } 1803 me->mp = m = nm; 1804 me->max_count = CAST(uint32_t, cnt); 1805 } 1806 m = &me->mp[me->cont_count++]; 1807 (void)memset(m, 0, sizeof(*m)); 1808 m->cont_level = cont_level; 1809 } else { 1810 static const size_t len = sizeof(*m) * ALLOC_CHUNK; 1811 if (me->mp != NULL) 1812 return 1; 1813 if ((m = CAST(struct magic *, malloc(len))) == NULL) { 1814 file_oomem(ms, len); 1815 return -1; 1816 } 1817 me->mp = m; 1818 me->max_count = ALLOC_CHUNK; 1819 (void)memset(m, 0, sizeof(*m)); 1820 m->factor_op = FILE_FACTOR_OP_NONE; 1821 m->cont_level = 0; 1822 me->cont_count = 1; 1823 } 1824 m->lineno = CAST(uint32_t, lineno); 1825 1826 if (*l == '&') { /* m->cont_level == 0 checked below. */ 1827 ++l; /* step over */ 1828 m->flag |= OFFADD; 1829 } 1830 if (*l == '(') { 1831 ++l; /* step over */ 1832 m->flag |= INDIR; 1833 if (m->flag & OFFADD) 1834 m->flag = (m->flag & ~OFFADD) | INDIROFFADD; 1835 1836 if (*l == '&') { /* m->cont_level == 0 checked below */ 1837 ++l; /* step over */ 1838 m->flag |= OFFADD; 1839 } 1840 } 1841 /* Indirect offsets are not valid at level 0. */ 1842 if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD))) 1843 if (ms->flags & MAGIC_CHECK) 1844 file_magwarn(ms, "relative offset at level 0"); 1845 1846 /* get offset, then skip over it */ 1847 m->offset = (uint32_t)strtoul(l, &t, 0); 1848 if (l == t) 1849 if (ms->flags & MAGIC_CHECK) 1850 file_magwarn(ms, "offset `%s' invalid", l); 1851 l = t; 1852 1853 if (m->flag & INDIR) { 1854 m->in_type = FILE_LONG; 1855 m->in_offset = 0; 1856 /* 1857 * read [.lbs][+-]nnnnn) 1858 */ 1859 if (*l == '.') { 1860 l++; 1861 switch (*l) { 1862 case 'l': 1863 m->in_type = FILE_LELONG; 1864 break; 1865 case 'L': 1866 m->in_type = FILE_BELONG; 1867 break; 1868 case 'm': 1869 m->in_type = FILE_MELONG; 1870 break; 1871 case 'h': 1872 case 's': 1873 m->in_type = FILE_LESHORT; 1874 break; 1875 case 'H': 1876 case 'S': 1877 m->in_type = FILE_BESHORT; 1878 break; 1879 case 'c': 1880 case 'b': 1881 case 'C': 1882 case 'B': 1883 m->in_type = FILE_BYTE; 1884 break; 1885 case 'e': 1886 case 'f': 1887 case 'g': 1888 m->in_type = FILE_LEDOUBLE; 1889 break; 1890 case 'E': 1891 case 'F': 1892 case 'G': 1893 m->in_type = FILE_BEDOUBLE; 1894 break; 1895 case 'i': 1896 m->in_type = FILE_LEID3; 1897 break; 1898 case 'I': 1899 m->in_type = FILE_BEID3; 1900 break; 1901 default: 1902 if (ms->flags & MAGIC_CHECK) 1903 file_magwarn(ms, 1904 "indirect offset type `%c' invalid", 1905 *l); 1906 break; 1907 } 1908 l++; 1909 } 1910 1911 m->in_op = 0; 1912 if (*l == '~') { 1913 m->in_op |= FILE_OPINVERSE; 1914 l++; 1915 } 1916 if ((op = get_op(*l)) != -1) { 1917 m->in_op |= op; 1918 l++; 1919 } 1920 if (*l == '(') { 1921 m->in_op |= FILE_OPINDIRECT; 1922 l++; 1923 } 1924 if (isdigit((unsigned char)*l) || *l == '-') { 1925 m->in_offset = (int32_t)strtol(l, &t, 0); 1926 if (l == t) 1927 if (ms->flags & MAGIC_CHECK) 1928 file_magwarn(ms, 1929 "in_offset `%s' invalid", l); 1930 l = t; 1931 } 1932 if (*l++ != ')' || 1933 ((m->in_op & FILE_OPINDIRECT) && *l++ != ')')) 1934 if (ms->flags & MAGIC_CHECK) 1935 file_magwarn(ms, 1936 "missing ')' in indirect offset"); 1937 } 1938 EATAB; 1939 1940 #ifdef ENABLE_CONDITIONALS 1941 m->cond = get_cond(l, &l); 1942 if (check_cond(ms, m->cond, cont_level) == -1) 1943 return -1; 1944 1945 EATAB; 1946 #endif 1947 1948 /* 1949 * Parse the type. 1950 */ 1951 if (*l == 'u') { 1952 /* 1953 * Try it as a keyword type prefixed by "u"; match what 1954 * follows the "u". If that fails, try it as an SUS 1955 * integer type. 1956 */ 1957 m->type = get_type(type_tbl, l + 1, &l); 1958 if (m->type == FILE_INVALID) { 1959 /* 1960 * Not a keyword type; parse it as an SUS type, 1961 * 'u' possibly followed by a number or C/S/L. 1962 */ 1963 m->type = get_standard_integer_type(l, &l); 1964 } 1965 /* It's unsigned. */ 1966 if (m->type != FILE_INVALID) 1967 m->flag |= UNSIGNED; 1968 } else { 1969 /* 1970 * Try it as a keyword type. If that fails, try it as 1971 * an SUS integer type if it begins with "d" or as an 1972 * SUS string type if it begins with "s". In any case, 1973 * it's not unsigned. 1974 */ 1975 m->type = get_type(type_tbl, l, &l); 1976 if (m->type == FILE_INVALID) { 1977 /* 1978 * Not a keyword type; parse it as an SUS type, 1979 * either 'd' possibly followed by a number or 1980 * C/S/L, or just 's'. 1981 */ 1982 if (*l == 'd') 1983 m->type = get_standard_integer_type(l, &l); 1984 else if (*l == 's' && !isalpha((unsigned char)l[1])) { 1985 m->type = FILE_STRING; 1986 ++l; 1987 } 1988 } 1989 } 1990 1991 if (m->type == FILE_INVALID) { 1992 /* Not found - try it as a special keyword. */ 1993 m->type = get_type(special_tbl, l, &l); 1994 } 1995 1996 if (m->type == FILE_INVALID) { 1997 if (ms->flags & MAGIC_CHECK) 1998 file_magwarn(ms, "type `%s' invalid", l); 1999 return -1; 2000 } 2001 2002 /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */ 2003 /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */ 2004 2005 m->mask_op = 0; 2006 if (*l == '~') { 2007 if (!IS_STRING(m->type)) 2008 m->mask_op |= FILE_OPINVERSE; 2009 else if (ms->flags & MAGIC_CHECK) 2010 file_magwarn(ms, "'~' invalid for string types"); 2011 ++l; 2012 } 2013 m->str_range = 0; 2014 m->str_flags = m->type == FILE_PSTRING ? PSTRING_1_LE : 0; 2015 if ((op = get_op(*l)) != -1) { 2016 if (IS_STRING(m->type)) { 2017 int r; 2018 2019 if (op != FILE_OPDIVIDE) { 2020 if (ms->flags & MAGIC_CHECK) 2021 file_magwarn(ms, 2022 "invalid string/indirect op: " 2023 "`%c'", *t); 2024 return -1; 2025 } 2026 2027 if (m->type == FILE_INDIRECT) 2028 r = parse_indirect_modifier(ms, m, &l); 2029 else 2030 r = parse_string_modifier(ms, m, &l); 2031 if (r == -1) 2032 return -1; 2033 } else 2034 parse_op_modifier(ms, m, &l, op); 2035 } 2036 2037 /* 2038 * We used to set mask to all 1's here, instead let's just not do 2039 * anything if mask = 0 (unless you have a better idea) 2040 */ 2041 EATAB; 2042 2043 switch (*l) { 2044 case '>': 2045 case '<': 2046 m->reln = *l; 2047 ++l; 2048 if (*l == '=') { 2049 if (ms->flags & MAGIC_CHECK) { 2050 file_magwarn(ms, "%c= not supported", 2051 m->reln); 2052 return -1; 2053 } 2054 ++l; 2055 } 2056 break; 2057 /* Old-style anding: "0 byte &0x80 dynamically linked" */ 2058 case '&': 2059 case '^': 2060 case '=': 2061 m->reln = *l; 2062 ++l; 2063 if (*l == '=') { 2064 /* HP compat: ignore &= etc. */ 2065 ++l; 2066 } 2067 break; 2068 case '!': 2069 m->reln = *l; 2070 ++l; 2071 break; 2072 default: 2073 m->reln = '='; /* the default relation */ 2074 if (*l == 'x' && ((isascii((unsigned char)l[1]) && 2075 isspace((unsigned char)l[1])) || !l[1])) { 2076 m->reln = *l; 2077 ++l; 2078 } 2079 break; 2080 } 2081 /* 2082 * Grab the value part, except for an 'x' reln. 2083 */ 2084 if (m->reln != 'x' && getvalue(ms, m, &l, action)) 2085 return -1; 2086 2087 /* 2088 * TODO finish this macro and start using it! 2089 * #define offsetcheck {if (offset > HOWMANY-1) 2090 * magwarn("offset too big"); } 2091 */ 2092 2093 /* 2094 * Now get last part - the description 2095 */ 2096 EATAB; 2097 if (l[0] == '\b') { 2098 ++l; 2099 m->flag |= NOSPACE; 2100 } else if ((l[0] == '\\') && (l[1] == 'b')) { 2101 ++l; 2102 ++l; 2103 m->flag |= NOSPACE; 2104 } 2105 for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); ) 2106 continue; 2107 if (i == sizeof(m->desc)) { 2108 m->desc[sizeof(m->desc) - 1] = '\0'; 2109 if (ms->flags & MAGIC_CHECK) 2110 file_magwarn(ms, "description `%s' truncated", m->desc); 2111 } 2112 2113 /* 2114 * We only do this check while compiling, or if any of the magic 2115 * files were not compiled. 2116 */ 2117 if (ms->flags & MAGIC_CHECK) { 2118 if (check_format(ms, m) == -1) 2119 return -1; 2120 } 2121 #ifndef COMPILE_ONLY 2122 if (action == FILE_CHECK) { 2123 file_mdump(m); 2124 } 2125 #endif 2126 m->mimetype[0] = '\0'; /* initialise MIME type to none */ 2127 return 0; 2128 } 2129 2130 /* 2131 * parse a STRENGTH annotation line from magic file, put into magic[index - 1] 2132 * if valid 2133 */ 2134 private int 2135 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line) 2136 { 2137 const char *l = line; 2138 char *el; 2139 unsigned long factor; 2140 struct magic *m = &me->mp[0]; 2141 2142 if (m->factor_op != FILE_FACTOR_OP_NONE) { 2143 file_magwarn(ms, 2144 "Current entry already has a strength type: %c %d", 2145 m->factor_op, m->factor); 2146 return -1; 2147 } 2148 if (m->type == FILE_NAME) { 2149 file_magwarn(ms, "%s: Strength setting is not supported in " 2150 "\"name\" magic entries", m->value.s); 2151 return -1; 2152 } 2153 EATAB; 2154 switch (*l) { 2155 case FILE_FACTOR_OP_NONE: 2156 case FILE_FACTOR_OP_PLUS: 2157 case FILE_FACTOR_OP_MINUS: 2158 case FILE_FACTOR_OP_TIMES: 2159 case FILE_FACTOR_OP_DIV: 2160 m->factor_op = *l++; 2161 break; 2162 default: 2163 file_magwarn(ms, "Unknown factor op `%c'", *l); 2164 return -1; 2165 } 2166 EATAB; 2167 factor = strtoul(l, &el, 0); 2168 if (factor > 255) { 2169 file_magwarn(ms, "Too large factor `%lu'", factor); 2170 goto out; 2171 } 2172 if (*el && !isspace((unsigned char)*el)) { 2173 file_magwarn(ms, "Bad factor `%s'", l); 2174 goto out; 2175 } 2176 m->factor = (uint8_t)factor; 2177 if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) { 2178 file_magwarn(ms, "Cannot have factor op `%c' and factor %u", 2179 m->factor_op, m->factor); 2180 goto out; 2181 } 2182 return 0; 2183 out: 2184 m->factor_op = FILE_FACTOR_OP_NONE; 2185 m->factor = 0; 2186 return -1; 2187 } 2188 2189 private int 2190 goodchar(unsigned char x, const char *extra) 2191 { 2192 return (isascii(x) && isalnum(x)) || strchr(extra, x); 2193 } 2194 2195 private int 2196 parse_extra(struct magic_set *ms, struct magic_entry *me, const char *line, 2197 off_t off, size_t len, const char *name, const char *extra, int nt) 2198 { 2199 size_t i; 2200 const char *l = line; 2201 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1]; 2202 char *buf = (char *)m + off; 2203 2204 if (buf[0] != '\0') { 2205 len = nt ? strlen(buf) : len; 2206 file_magwarn(ms, "Current entry already has a %s type " 2207 "`%.*s', new type `%s'", name, (int)len, buf, l); 2208 return -1; 2209 } 2210 2211 if (*m->desc == '\0') { 2212 file_magwarn(ms, "Current entry does not yet have a " 2213 "description for adding a %s type", name); 2214 return -1; 2215 } 2216 2217 EATAB; 2218 for (i = 0; *l && i < len && goodchar(*l, extra); buf[i++] = *l++) 2219 continue; 2220 2221 if (i == len && *l) { 2222 if (nt) 2223 buf[len - 1] = '\0'; 2224 if (ms->flags & MAGIC_CHECK) 2225 file_magwarn(ms, "%s type `%s' truncated %" 2226 SIZE_T_FORMAT "u", name, line, i); 2227 } else { 2228 if (!isspace((unsigned char)*l) && !goodchar(*l, extra)) 2229 file_magwarn(ms, "%s type `%s' has bad char '%c'", 2230 name, line, *l); 2231 if (nt) 2232 buf[i] = '\0'; 2233 } 2234 2235 if (i > 0) 2236 return 0; 2237 2238 file_magerror(ms, "Bad magic entry '%s'", line); 2239 return -1; 2240 } 2241 2242 /* 2243 * Parse an Apple CREATOR/TYPE annotation from magic file and put it into 2244 * magic[index - 1] 2245 */ 2246 private int 2247 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line) 2248 { 2249 struct magic *m = &me->mp[0]; 2250 2251 return parse_extra(ms, me, line, offsetof(struct magic, apple), 2252 sizeof(m->apple), "APPLE", "!+-./", 0); 2253 } 2254 2255 /* 2256 * parse a MIME annotation line from magic file, put into magic[index - 1] 2257 * if valid 2258 */ 2259 private int 2260 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line) 2261 { 2262 struct magic *m = &me->mp[0]; 2263 2264 return parse_extra(ms, me, line, offsetof(struct magic, mimetype), 2265 sizeof(m->mimetype), "MIME", "+-/.", 1); 2266 } 2267 2268 private int 2269 check_format_type(const char *ptr, int type) 2270 { 2271 int quad = 0, h; 2272 if (*ptr == '\0') { 2273 /* Missing format string; bad */ 2274 return -1; 2275 } 2276 2277 switch (file_formats[type]) { 2278 case FILE_FMT_QUAD: 2279 quad = 1; 2280 /*FALLTHROUGH*/ 2281 case FILE_FMT_NUM: 2282 if (quad == 0) { 2283 switch (type) { 2284 case FILE_BYTE: 2285 h = 2; 2286 break; 2287 case FILE_SHORT: 2288 case FILE_BESHORT: 2289 case FILE_LESHORT: 2290 h = 1; 2291 break; 2292 case FILE_LONG: 2293 case FILE_BELONG: 2294 case FILE_LELONG: 2295 case FILE_MELONG: 2296 case FILE_LEID3: 2297 case FILE_BEID3: 2298 case FILE_INDIRECT: 2299 h = 0; 2300 break; 2301 default: 2302 abort(); 2303 } 2304 } else 2305 h = 0; 2306 if (*ptr == '-') 2307 ptr++; 2308 if (*ptr == '.') 2309 ptr++; 2310 while (isdigit((unsigned char)*ptr)) ptr++; 2311 if (*ptr == '.') 2312 ptr++; 2313 while (isdigit((unsigned char)*ptr)) ptr++; 2314 if (quad) { 2315 if (*ptr++ != 'l') 2316 return -1; 2317 if (*ptr++ != 'l') 2318 return -1; 2319 } 2320 2321 switch (*ptr++) { 2322 #ifdef STRICT_FORMAT /* "long" formats are int formats for us */ 2323 /* so don't accept the 'l' modifier */ 2324 case 'l': 2325 switch (*ptr++) { 2326 case 'i': 2327 case 'd': 2328 case 'u': 2329 case 'o': 2330 case 'x': 2331 case 'X': 2332 return h != 0 ? -1 : 0; 2333 default: 2334 return -1; 2335 } 2336 2337 /* 2338 * Don't accept h and hh modifiers. They make writing 2339 * magic entries more complicated, for very little benefit 2340 */ 2341 case 'h': 2342 if (h-- <= 0) 2343 return -1; 2344 switch (*ptr++) { 2345 case 'h': 2346 if (h-- <= 0) 2347 return -1; 2348 switch (*ptr++) { 2349 case 'i': 2350 case 'd': 2351 case 'u': 2352 case 'o': 2353 case 'x': 2354 case 'X': 2355 return 0; 2356 default: 2357 return -1; 2358 } 2359 case 'i': 2360 case 'd': 2361 case 'u': 2362 case 'o': 2363 case 'x': 2364 case 'X': 2365 return h != 0 ? -1 : 0; 2366 default: 2367 return -1; 2368 } 2369 #endif 2370 case 'c': 2371 return h != 2 ? -1 : 0; 2372 case 'i': 2373 case 'd': 2374 case 'u': 2375 case 'o': 2376 case 'x': 2377 case 'X': 2378 #ifdef STRICT_FORMAT 2379 return h != 0 ? -1 : 0; 2380 #else 2381 return 0; 2382 #endif 2383 default: 2384 return -1; 2385 } 2386 2387 case FILE_FMT_FLOAT: 2388 case FILE_FMT_DOUBLE: 2389 if (*ptr == '-') 2390 ptr++; 2391 if (*ptr == '.') 2392 ptr++; 2393 while (isdigit((unsigned char)*ptr)) ptr++; 2394 if (*ptr == '.') 2395 ptr++; 2396 while (isdigit((unsigned char)*ptr)) ptr++; 2397 2398 switch (*ptr++) { 2399 case 'e': 2400 case 'E': 2401 case 'f': 2402 case 'F': 2403 case 'g': 2404 case 'G': 2405 return 0; 2406 2407 default: 2408 return -1; 2409 } 2410 2411 2412 case FILE_FMT_STR: 2413 if (*ptr == '-') 2414 ptr++; 2415 while (isdigit((unsigned char )*ptr)) 2416 ptr++; 2417 if (*ptr == '.') { 2418 ptr++; 2419 while (isdigit((unsigned char )*ptr)) 2420 ptr++; 2421 } 2422 2423 switch (*ptr++) { 2424 case 's': 2425 return 0; 2426 default: 2427 return -1; 2428 } 2429 2430 default: 2431 /* internal error */ 2432 abort(); 2433 } 2434 /*NOTREACHED*/ 2435 return -1; 2436 } 2437 2438 /* 2439 * Check that the optional printf format in description matches 2440 * the type of the magic. 2441 */ 2442 private int 2443 check_format(struct magic_set *ms, struct magic *m) 2444 { 2445 char *ptr; 2446 2447 for (ptr = m->desc; *ptr; ptr++) 2448 if (*ptr == '%') 2449 break; 2450 if (*ptr == '\0') { 2451 /* No format string; ok */ 2452 return 1; 2453 } 2454 2455 assert(file_nformats == file_nnames); 2456 2457 if (m->type >= file_nformats) { 2458 file_magwarn(ms, "Internal error inconsistency between " 2459 "m->type and format strings"); 2460 return -1; 2461 } 2462 if (file_formats[m->type] == FILE_FMT_NONE) { 2463 file_magwarn(ms, "No format string for `%s' with description " 2464 "`%s'", m->desc, file_names[m->type]); 2465 return -1; 2466 } 2467 2468 ptr++; 2469 if (check_format_type(ptr, m->type) == -1) { 2470 /* 2471 * TODO: this error message is unhelpful if the format 2472 * string is not one character long 2473 */ 2474 file_magwarn(ms, "Printf format `%c' is not valid for type " 2475 "`%s' in description `%s'", *ptr ? *ptr : '?', 2476 file_names[m->type], m->desc); 2477 return -1; 2478 } 2479 2480 for (; *ptr; ptr++) { 2481 if (*ptr == '%') { 2482 file_magwarn(ms, 2483 "Too many format strings (should have at most one) " 2484 "for `%s' with description `%s'", 2485 file_names[m->type], m->desc); 2486 return -1; 2487 } 2488 } 2489 return 0; 2490 } 2491 2492 /* 2493 * Read a numeric value from a pointer, into the value union of a magic 2494 * pointer, according to the magic type. Update the string pointer to point 2495 * just after the number read. Return 0 for success, non-zero for failure. 2496 */ 2497 private int 2498 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action) 2499 { 2500 switch (m->type) { 2501 case FILE_BESTRING16: 2502 case FILE_LESTRING16: 2503 case FILE_STRING: 2504 case FILE_PSTRING: 2505 case FILE_REGEX: 2506 case FILE_SEARCH: 2507 case FILE_NAME: 2508 case FILE_USE: 2509 *p = getstr(ms, m, *p, action == FILE_COMPILE); 2510 if (*p == NULL) { 2511 if (ms->flags & MAGIC_CHECK) 2512 file_magwarn(ms, "cannot get string from `%s'", 2513 m->value.s); 2514 return -1; 2515 } 2516 if (m->type == FILE_REGEX) { 2517 file_regex_t rx; 2518 int rc = file_regcomp(&rx, m->value.s, REG_EXTENDED); 2519 if (rc) { 2520 if (ms->flags & MAGIC_CHECK) 2521 file_regerror(&rx, rc, ms); 2522 } 2523 file_regfree(&rx); 2524 return rc ? -1 : 0; 2525 } 2526 return 0; 2527 case FILE_FLOAT: 2528 case FILE_BEFLOAT: 2529 case FILE_LEFLOAT: 2530 if (m->reln != 'x') { 2531 char *ep; 2532 #ifdef HAVE_STRTOF 2533 m->value.f = strtof(*p, &ep); 2534 #else 2535 m->value.f = (float)strtod(*p, &ep); 2536 #endif 2537 *p = ep; 2538 } 2539 return 0; 2540 case FILE_DOUBLE: 2541 case FILE_BEDOUBLE: 2542 case FILE_LEDOUBLE: 2543 if (m->reln != 'x') { 2544 char *ep; 2545 m->value.d = strtod(*p, &ep); 2546 *p = ep; 2547 } 2548 return 0; 2549 default: 2550 if (m->reln != 'x') { 2551 char *ep; 2552 m->value.q = file_signextend(ms, m, 2553 (uint64_t)strtoull(*p, &ep, 0)); 2554 *p = ep; 2555 eatsize(p); 2556 } 2557 return 0; 2558 } 2559 } 2560 2561 /* 2562 * Convert a string containing C character escapes. Stop at an unescaped 2563 * space or tab. 2564 * Copy the converted version to "m->value.s", and the length in m->vallen. 2565 * Return updated scan pointer as function result. Warn if set. 2566 */ 2567 private const char * 2568 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn) 2569 { 2570 const char *origs = s; 2571 char *p = m->value.s; 2572 size_t plen = sizeof(m->value.s); 2573 char *origp = p; 2574 char *pmax = p + plen - 1; 2575 int c; 2576 int val; 2577 2578 while ((c = *s++) != '\0') { 2579 if (isspace((unsigned char) c)) 2580 break; 2581 if (p >= pmax) { 2582 file_error(ms, 0, "string too long: `%s'", origs); 2583 return NULL; 2584 } 2585 if (c == '\\') { 2586 switch(c = *s++) { 2587 2588 case '\0': 2589 if (warn) 2590 file_magwarn(ms, "incomplete escape"); 2591 goto out; 2592 2593 case '\t': 2594 if (warn) { 2595 file_magwarn(ms, 2596 "escaped tab found, use \\t instead"); 2597 warn = 0; /* already did */ 2598 } 2599 /*FALLTHROUGH*/ 2600 default: 2601 if (warn) { 2602 if (isprint((unsigned char)c)) { 2603 /* Allow escaping of 2604 * ``relations'' */ 2605 if (strchr("<>&^=!", c) == NULL 2606 && (m->type != FILE_REGEX || 2607 strchr("[]().*?^$|{}", c) 2608 == NULL)) { 2609 file_magwarn(ms, "no " 2610 "need to escape " 2611 "`%c'", c); 2612 } 2613 } else { 2614 file_magwarn(ms, 2615 "unknown escape sequence: " 2616 "\\%03o", c); 2617 } 2618 } 2619 /*FALLTHROUGH*/ 2620 /* space, perhaps force people to use \040? */ 2621 case ' ': 2622 #if 0 2623 /* 2624 * Other things people escape, but shouldn't need to, 2625 * so we disallow them 2626 */ 2627 case '\'': 2628 case '"': 2629 case '?': 2630 #endif 2631 /* Relations */ 2632 case '>': 2633 case '<': 2634 case '&': 2635 case '^': 2636 case '=': 2637 case '!': 2638 /* and baskslash itself */ 2639 case '\\': 2640 *p++ = (char) c; 2641 break; 2642 2643 case 'a': 2644 *p++ = '\a'; 2645 break; 2646 2647 case 'b': 2648 *p++ = '\b'; 2649 break; 2650 2651 case 'f': 2652 *p++ = '\f'; 2653 break; 2654 2655 case 'n': 2656 *p++ = '\n'; 2657 break; 2658 2659 case 'r': 2660 *p++ = '\r'; 2661 break; 2662 2663 case 't': 2664 *p++ = '\t'; 2665 break; 2666 2667 case 'v': 2668 *p++ = '\v'; 2669 break; 2670 2671 /* \ and up to 3 octal digits */ 2672 case '0': 2673 case '1': 2674 case '2': 2675 case '3': 2676 case '4': 2677 case '5': 2678 case '6': 2679 case '7': 2680 val = c - '0'; 2681 c = *s++; /* try for 2 */ 2682 if (c >= '0' && c <= '7') { 2683 val = (val << 3) | (c - '0'); 2684 c = *s++; /* try for 3 */ 2685 if (c >= '0' && c <= '7') 2686 val = (val << 3) | (c-'0'); 2687 else 2688 --s; 2689 } 2690 else 2691 --s; 2692 *p++ = (char)val; 2693 break; 2694 2695 /* \x and up to 2 hex digits */ 2696 case 'x': 2697 val = 'x'; /* Default if no digits */ 2698 c = hextoint(*s++); /* Get next char */ 2699 if (c >= 0) { 2700 val = c; 2701 c = hextoint(*s++); 2702 if (c >= 0) 2703 val = (val << 4) + c; 2704 else 2705 --s; 2706 } else 2707 --s; 2708 *p++ = (char)val; 2709 break; 2710 } 2711 } else 2712 *p++ = (char)c; 2713 } 2714 out: 2715 *p = '\0'; 2716 m->vallen = CAST(unsigned char, (p - origp)); 2717 if (m->type == FILE_PSTRING) 2718 m->vallen += (unsigned char)file_pstring_length_size(m); 2719 return s; 2720 } 2721 2722 2723 /* Single hex char to int; -1 if not a hex char. */ 2724 private int 2725 hextoint(int c) 2726 { 2727 if (!isascii((unsigned char) c)) 2728 return -1; 2729 if (isdigit((unsigned char) c)) 2730 return c - '0'; 2731 if ((c >= 'a') && (c <= 'f')) 2732 return c + 10 - 'a'; 2733 if (( c>= 'A') && (c <= 'F')) 2734 return c + 10 - 'A'; 2735 return -1; 2736 } 2737 2738 2739 /* 2740 * Print a string containing C character escapes. 2741 */ 2742 protected void 2743 file_showstr(FILE *fp, const char *s, size_t len) 2744 { 2745 char c; 2746 2747 for (;;) { 2748 if (len == ~0U) { 2749 c = *s++; 2750 if (c == '\0') 2751 break; 2752 } 2753 else { 2754 if (len-- == 0) 2755 break; 2756 c = *s++; 2757 } 2758 if (c >= 040 && c <= 0176) /* TODO isprint && !iscntrl */ 2759 (void) fputc(c, fp); 2760 else { 2761 (void) fputc('\\', fp); 2762 switch (c) { 2763 case '\a': 2764 (void) fputc('a', fp); 2765 break; 2766 2767 case '\b': 2768 (void) fputc('b', fp); 2769 break; 2770 2771 case '\f': 2772 (void) fputc('f', fp); 2773 break; 2774 2775 case '\n': 2776 (void) fputc('n', fp); 2777 break; 2778 2779 case '\r': 2780 (void) fputc('r', fp); 2781 break; 2782 2783 case '\t': 2784 (void) fputc('t', fp); 2785 break; 2786 2787 case '\v': 2788 (void) fputc('v', fp); 2789 break; 2790 2791 default: 2792 (void) fprintf(fp, "%.3o", c & 0377); 2793 break; 2794 } 2795 } 2796 } 2797 } 2798 2799 /* 2800 * eatsize(): Eat the size spec from a number [eg. 10UL] 2801 */ 2802 private void 2803 eatsize(const char **p) 2804 { 2805 const char *l = *p; 2806 2807 if (LOWCASE(*l) == 'u') 2808 l++; 2809 2810 switch (LOWCASE(*l)) { 2811 case 'l': /* long */ 2812 case 's': /* short */ 2813 case 'h': /* short */ 2814 case 'b': /* char/byte */ 2815 case 'c': /* char/byte */ 2816 l++; 2817 /*FALLTHROUGH*/ 2818 default: 2819 break; 2820 } 2821 2822 *p = l; 2823 } 2824 2825 /* 2826 * handle a buffer containing a compiled file. 2827 */ 2828 private struct magic_map * 2829 apprentice_buf(struct magic_set *ms, struct magic *buf, size_t len) 2830 { 2831 struct magic_map *map; 2832 2833 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) { 2834 file_oomem(ms, sizeof(*map)); 2835 return NULL; 2836 } 2837 map->len = len; 2838 map->p = buf; 2839 map->type = MAP_TYPE_USER; 2840 if (check_buffer(ms, map, "buffer") != 0) { 2841 apprentice_unmap(map); 2842 return NULL; 2843 } 2844 return map; 2845 } 2846 2847 /* 2848 * handle a compiled file. 2849 */ 2850 2851 private struct magic_map * 2852 apprentice_map(struct magic_set *ms, const char *fn) 2853 { 2854 int fd; 2855 struct stat st; 2856 char *dbname = NULL; 2857 struct magic_map *map; 2858 2859 fd = -1; 2860 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) { 2861 file_oomem(ms, sizeof(*map)); 2862 goto error; 2863 } 2864 2865 dbname = mkdbname(ms, fn, 0); 2866 if (dbname == NULL) 2867 goto error; 2868 2869 if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1) 2870 goto error; 2871 2872 if (fstat(fd, &st) == -1) { 2873 file_error(ms, errno, "cannot stat `%s'", dbname); 2874 goto error; 2875 } 2876 if (st.st_size < 8 || st.st_size > MAXMAGIC_SIZE) { 2877 file_error(ms, 0, "file `%s' is too %s", dbname, 2878 st.st_size < 8 ? "small" : "large"); 2879 goto error; 2880 } 2881 2882 map->len = (size_t)st.st_size; 2883 #ifdef QUICK 2884 if ((map->p = mmap(0, (size_t)st.st_size, PROT_READ|PROT_WRITE, 2885 MAP_PRIVATE|MAP_FILE, fd, (off_t)0)) == MAP_FAILED) { 2886 file_error(ms, errno, "cannot map `%s'", dbname); 2887 goto error; 2888 } 2889 map->type = MAP_TYPE_MMAP; 2890 #else 2891 if ((map->p = CAST(void *, malloc(map->len))) == NULL) { 2892 file_oomem(ms, map->len); 2893 goto error; 2894 } 2895 if (read(fd, map->p, map->len) != (ssize_t)map->len) { 2896 file_badread(ms); 2897 goto error; 2898 } 2899 map->type = MAP_TYPE_MALLOC; 2900 #define RET 1 2901 #endif 2902 (void)close(fd); 2903 fd = -1; 2904 2905 if (check_buffer(ms, map, dbname) != 0) 2906 goto error; 2907 2908 free(dbname); 2909 return map; 2910 2911 error: 2912 if (fd != -1) 2913 (void)close(fd); 2914 apprentice_unmap(map); 2915 free(dbname); 2916 return NULL; 2917 } 2918 2919 private int 2920 check_buffer(struct magic_set *ms, struct magic_map *map, const char *dbname) 2921 { 2922 uint32_t *ptr; 2923 uint32_t entries, nentries; 2924 uint32_t version; 2925 int i, needsbyteswap; 2926 2927 ptr = CAST(uint32_t *, map->p); 2928 if (*ptr != MAGICNO) { 2929 if (swap4(*ptr) != MAGICNO) { 2930 file_error(ms, 0, "bad magic in `%s'", dbname); 2931 return -1; 2932 } 2933 needsbyteswap = 1; 2934 } else 2935 needsbyteswap = 0; 2936 if (needsbyteswap) 2937 version = swap4(ptr[1]); 2938 else 2939 version = ptr[1]; 2940 if (version != VERSIONNO) { 2941 file_error(ms, 0, "File %s supports only version %d magic " 2942 "files. `%s' is version %d", VERSION, 2943 VERSIONNO, dbname, version); 2944 return -1; 2945 } 2946 entries = (uint32_t)(map->len / sizeof(struct magic)); 2947 if ((entries * sizeof(struct magic)) != map->len) { 2948 file_error(ms, 0, "Size of `%s' %" SIZE_T_FORMAT "u is not " 2949 "a multiple of %" SIZE_T_FORMAT "u", 2950 dbname, map->len, sizeof(struct magic)); 2951 return -1; 2952 } 2953 map->magic[0] = CAST(struct magic *, map->p) + 1; 2954 nentries = 0; 2955 for (i = 0; i < MAGIC_SETS; i++) { 2956 if (needsbyteswap) 2957 map->nmagic[i] = swap4(ptr[i + 2]); 2958 else 2959 map->nmagic[i] = ptr[i + 2]; 2960 if (i != MAGIC_SETS - 1) 2961 map->magic[i + 1] = map->magic[i] + map->nmagic[i]; 2962 nentries += map->nmagic[i]; 2963 } 2964 if (entries != nentries + 1) { 2965 file_error(ms, 0, "Inconsistent entries in `%s' %u != %u", 2966 dbname, entries, nentries + 1); 2967 return -1; 2968 } 2969 if (needsbyteswap) 2970 for (i = 0; i < MAGIC_SETS; i++) 2971 byteswap(map->magic[i], map->nmagic[i]); 2972 return 0; 2973 } 2974 2975 /* 2976 * handle an mmaped file. 2977 */ 2978 private int 2979 apprentice_compile(struct magic_set *ms, struct magic_map *map, const char *fn) 2980 { 2981 static const size_t nm = sizeof(*map->nmagic) * MAGIC_SETS; 2982 static const size_t m = sizeof(**map->magic); 2983 int fd = -1; 2984 size_t len; 2985 char *dbname; 2986 int rv = -1; 2987 uint32_t i; 2988 union { 2989 struct magic m; 2990 uint32_t h[2 + MAGIC_SETS]; 2991 } hdr; 2992 2993 dbname = mkdbname(ms, fn, 1); 2994 2995 if (dbname == NULL) 2996 goto out; 2997 2998 if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) 2999 { 3000 file_error(ms, errno, "cannot open `%s'", dbname); 3001 goto out; 3002 } 3003 memset(&hdr, 0, sizeof(hdr)); 3004 hdr.h[0] = MAGICNO; 3005 hdr.h[1] = VERSIONNO; 3006 memcpy(hdr.h + 2, map->nmagic, nm); 3007 3008 if (write(fd, &hdr, sizeof(hdr)) != (ssize_t)sizeof(hdr)) { 3009 file_error(ms, errno, "error writing `%s'", dbname); 3010 goto out; 3011 } 3012 3013 for (i = 0; i < MAGIC_SETS; i++) { 3014 len = m * map->nmagic[i]; 3015 if (write(fd, map->magic[i], len) != (ssize_t)len) { 3016 file_error(ms, errno, "error writing `%s'", dbname); 3017 goto out; 3018 } 3019 } 3020 3021 if (fd != -1) 3022 (void)close(fd); 3023 rv = 0; 3024 out: 3025 free(dbname); 3026 return rv; 3027 } 3028 3029 private const char ext[] = ".mgc"; 3030 /* 3031 * make a dbname 3032 */ 3033 private char * 3034 mkdbname(struct magic_set *ms, const char *fn, int strip) 3035 { 3036 const char *p, *q; 3037 char *buf; 3038 3039 if (strip) { 3040 if ((p = strrchr(fn, '/')) != NULL) 3041 fn = ++p; 3042 } 3043 3044 for (q = fn; *q; q++) 3045 continue; 3046 /* Look for .mgc */ 3047 for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--) 3048 if (*p != *q) 3049 break; 3050 3051 /* Did not find .mgc, restore q */ 3052 if (p >= ext) 3053 while (*q) 3054 q++; 3055 3056 q++; 3057 /* Compatibility with old code that looked in .mime */ 3058 if (ms->flags & MAGIC_MIME) { 3059 if (asprintf(&buf, "%.*s.mime%s", (int)(q - fn), fn, ext) < 0) 3060 return NULL; 3061 if (access(buf, R_OK) != -1) { 3062 ms->flags &= MAGIC_MIME_TYPE; 3063 return buf; 3064 } 3065 free(buf); 3066 } 3067 if (asprintf(&buf, "%.*s%s", (int)(q - fn), fn, ext) < 0) 3068 return NULL; 3069 3070 /* Compatibility with old code that looked in .mime */ 3071 if (strstr(p, ".mime") != NULL) 3072 ms->flags &= MAGIC_MIME_TYPE; 3073 return buf; 3074 } 3075 3076 /* 3077 * Byteswap an mmap'ed file if needed 3078 */ 3079 private void 3080 byteswap(struct magic *magic, uint32_t nmagic) 3081 { 3082 uint32_t i; 3083 for (i = 0; i < nmagic; i++) 3084 bs1(&magic[i]); 3085 } 3086 3087 /* 3088 * swap a short 3089 */ 3090 private uint16_t 3091 swap2(uint16_t sv) 3092 { 3093 uint16_t rv; 3094 uint8_t *s = (uint8_t *)(void *)&sv; 3095 uint8_t *d = (uint8_t *)(void *)&rv; 3096 d[0] = s[1]; 3097 d[1] = s[0]; 3098 return rv; 3099 } 3100 3101 /* 3102 * swap an int 3103 */ 3104 private uint32_t 3105 swap4(uint32_t sv) 3106 { 3107 uint32_t rv; 3108 uint8_t *s = (uint8_t *)(void *)&sv; 3109 uint8_t *d = (uint8_t *)(void *)&rv; 3110 d[0] = s[3]; 3111 d[1] = s[2]; 3112 d[2] = s[1]; 3113 d[3] = s[0]; 3114 return rv; 3115 } 3116 3117 /* 3118 * swap a quad 3119 */ 3120 private uint64_t 3121 swap8(uint64_t sv) 3122 { 3123 uint64_t rv; 3124 uint8_t *s = (uint8_t *)(void *)&sv; 3125 uint8_t *d = (uint8_t *)(void *)&rv; 3126 #if 0 3127 d[0] = s[3]; 3128 d[1] = s[2]; 3129 d[2] = s[1]; 3130 d[3] = s[0]; 3131 d[4] = s[7]; 3132 d[5] = s[6]; 3133 d[6] = s[5]; 3134 d[7] = s[4]; 3135 #else 3136 d[0] = s[7]; 3137 d[1] = s[6]; 3138 d[2] = s[5]; 3139 d[3] = s[4]; 3140 d[4] = s[3]; 3141 d[5] = s[2]; 3142 d[6] = s[1]; 3143 d[7] = s[0]; 3144 #endif 3145 return rv; 3146 } 3147 3148 /* 3149 * byteswap a single magic entry 3150 */ 3151 private void 3152 bs1(struct magic *m) 3153 { 3154 m->cont_level = swap2(m->cont_level); 3155 m->offset = swap4((uint32_t)m->offset); 3156 m->in_offset = swap4((uint32_t)m->in_offset); 3157 m->lineno = swap4((uint32_t)m->lineno); 3158 if (IS_STRING(m->type)) { 3159 m->str_range = swap4(m->str_range); 3160 m->str_flags = swap4(m->str_flags); 3161 } 3162 else { 3163 m->value.q = swap8(m->value.q); 3164 m->num_mask = swap8(m->num_mask); 3165 } 3166 } 3167 3168 protected size_t 3169 file_pstring_length_size(const struct magic *m) 3170 { 3171 switch (m->str_flags & PSTRING_LEN) { 3172 case PSTRING_1_LE: 3173 return 1; 3174 case PSTRING_2_LE: 3175 case PSTRING_2_BE: 3176 return 2; 3177 case PSTRING_4_LE: 3178 case PSTRING_4_BE: 3179 return 4; 3180 default: 3181 abort(); /* Impossible */ 3182 return 1; 3183 } 3184 } 3185 protected size_t 3186 file_pstring_get_length(const struct magic *m, const char *s) 3187 { 3188 size_t len = 0; 3189 3190 switch (m->str_flags & PSTRING_LEN) { 3191 case PSTRING_1_LE: 3192 len = *s; 3193 break; 3194 case PSTRING_2_LE: 3195 len = (s[1] << 8) | s[0]; 3196 break; 3197 case PSTRING_2_BE: 3198 len = (s[0] << 8) | s[1]; 3199 break; 3200 case PSTRING_4_LE: 3201 len = (s[3] << 24) | (s[2] << 16) | (s[1] << 8) | s[0]; 3202 break; 3203 case PSTRING_4_BE: 3204 len = (s[0] << 24) | (s[1] << 16) | (s[2] << 8) | s[3]; 3205 break; 3206 default: 3207 abort(); /* Impossible */ 3208 } 3209 3210 if (m->str_flags & PSTRING_LENGTH_INCLUDES_ITSELF) 3211 len -= file_pstring_length_size(m); 3212 3213 return len; 3214 } 3215 3216 protected int 3217 file_magicfind(struct magic_set *ms, const char *name, struct mlist *v) 3218 { 3219 uint32_t i, j; 3220 struct mlist *mlist, *ml; 3221 3222 mlist = ms->mlist[1]; 3223 3224 for (ml = mlist->next; ml != mlist; ml = ml->next) { 3225 struct magic *ma = ml->magic; 3226 uint32_t nma = ml->nmagic; 3227 for (i = 0; i < nma; i++) { 3228 if (ma[i].type != FILE_NAME) 3229 continue; 3230 if (strcmp(ma[i].value.s, name) == 0) { 3231 v->magic = &ma[i]; 3232 for (j = i + 1; j < nma; j++) 3233 if (ma[j].cont_level == 0) 3234 break; 3235 v->nmagic = j - i; 3236 return 0; 3237 } 3238 } 3239 } 3240 return -1; 3241 } 3242