1 /* $NetBSD: apprentice.c,v 1.25 2020/06/15 00:37:24 christos Exp $ */ 2 3 /* 4 * Copyright (c) Ian F. Darwin 1986-1995. 5 * Software written by Ian F. Darwin and others; 6 * maintained 1995-present by Christos Zoulas and others. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice immediately at the beginning of the file, without modification, 13 * this list of conditions, and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 18 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 19 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 21 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR 22 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 23 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 24 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 25 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 28 * SUCH DAMAGE. 29 */ 30 /* 31 * apprentice - make one pass through /etc/magic, learning its secrets. 32 */ 33 34 #include "file.h" 35 36 #ifndef lint 37 #if 0 38 FILE_RCSID("@(#)$File: apprentice.c,v 1.297 2020/05/09 18:57:15 christos Exp $") 39 #else 40 __RCSID("$NetBSD: apprentice.c,v 1.25 2020/06/15 00:37:24 christos Exp $"); 41 #endif 42 #endif /* lint */ 43 44 #include "magic.h" 45 #include <stdlib.h> 46 #ifdef HAVE_UNISTD_H 47 #include <unistd.h> 48 #endif 49 #include <stddef.h> 50 #include <string.h> 51 #include <assert.h> 52 #include <ctype.h> 53 #include <fcntl.h> 54 #ifdef QUICK 55 #include <sys/mman.h> 56 #endif 57 #include <dirent.h> 58 #include <limits.h> 59 60 61 #define EATAB {while (isascii(CAST(unsigned char, *l)) && \ 62 isspace(CAST(unsigned char, *l))) ++l;} 63 #define LOWCASE(l) (isupper(CAST(unsigned char, l)) ? \ 64 tolower(CAST(unsigned char, l)) : (l)) 65 /* 66 * Work around a bug in headers on Digital Unix. 67 * At least confirmed for: OSF1 V4.0 878 68 */ 69 #if defined(__osf__) && defined(__DECC) 70 #ifdef MAP_FAILED 71 #undef MAP_FAILED 72 #endif 73 #endif 74 75 #ifndef MAP_FAILED 76 #define MAP_FAILED (void *) -1 77 #endif 78 79 #ifndef MAP_FILE 80 #define MAP_FILE 0 81 #endif 82 83 #define ALLOC_CHUNK CAST(size_t, 10) 84 #define ALLOC_INCR CAST(size_t, 200) 85 86 #define MAP_TYPE_USER 0 87 #define MAP_TYPE_MALLOC 1 88 #define MAP_TYPE_MMAP 2 89 90 struct magic_entry { 91 struct magic *mp; 92 uint32_t cont_count; 93 uint32_t max_count; 94 }; 95 96 struct magic_entry_set { 97 struct magic_entry *me; 98 uint32_t count; 99 uint32_t max; 100 }; 101 102 struct magic_map { 103 void *p; 104 size_t len; 105 int type; 106 struct magic *magic[MAGIC_SETS]; 107 uint32_t nmagic[MAGIC_SETS]; 108 }; 109 110 int file_formats[FILE_NAMES_SIZE]; 111 const size_t file_nformats = FILE_NAMES_SIZE; 112 const char *file_names[FILE_NAMES_SIZE]; 113 const size_t file_nnames = FILE_NAMES_SIZE; 114 115 private int getvalue(struct magic_set *ms, struct magic *, const char **, int); 116 private int hextoint(int); 117 private const char *getstr(struct magic_set *, struct magic *, const char *, 118 int); 119 private int parse(struct magic_set *, struct magic_entry *, const char *, 120 size_t, int); 121 private void eatsize(const char **); 122 private int apprentice_1(struct magic_set *, const char *, int); 123 private size_t apprentice_magic_strength(const struct magic *); 124 private int apprentice_sort(const void *, const void *); 125 private void apprentice_list(struct mlist *, int ); 126 private struct magic_map *apprentice_load(struct magic_set *, 127 const char *, int); 128 private struct mlist *mlist_alloc(void); 129 private void mlist_free_all(struct magic_set *); 130 private void mlist_free(struct mlist *); 131 private void byteswap(struct magic *, uint32_t); 132 private void bs1(struct magic *); 133 private uint16_t swap2(uint16_t); 134 private uint32_t swap4(uint32_t); 135 private uint64_t swap8(uint64_t); 136 private char *mkdbname(struct magic_set *, const char *, int); 137 private struct magic_map *apprentice_buf(struct magic_set *, struct magic *, 138 size_t); 139 private struct magic_map *apprentice_map(struct magic_set *, const char *); 140 private int check_buffer(struct magic_set *, struct magic_map *, const char *); 141 private void apprentice_unmap(struct magic_map *); 142 private int apprentice_compile(struct magic_set *, struct magic_map *, 143 const char *); 144 private int check_format_type(const char *, int, const char **); 145 private int check_format(struct magic_set *, struct magic *); 146 private int get_op(char); 147 private int parse_mime(struct magic_set *, struct magic_entry *, const char *, 148 size_t); 149 private int parse_strength(struct magic_set *, struct magic_entry *, 150 const char *, size_t); 151 private int parse_apple(struct magic_set *, struct magic_entry *, const char *, 152 size_t); 153 private int parse_ext(struct magic_set *, struct magic_entry *, const char *, 154 size_t); 155 156 157 private size_t magicsize = sizeof(struct magic); 158 159 private const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc"; 160 161 private struct { 162 const char *name; 163 size_t len; 164 int (*fun)(struct magic_set *, struct magic_entry *, const char *, 165 size_t); 166 } bang[] = { 167 #define DECLARE_FIELD(name) { # name, sizeof(# name) - 1, parse_ ## name } 168 DECLARE_FIELD(mime), 169 DECLARE_FIELD(apple), 170 DECLARE_FIELD(ext), 171 DECLARE_FIELD(strength), 172 #undef DECLARE_FIELD 173 { NULL, 0, NULL } 174 }; 175 176 #ifdef COMPILE_ONLY 177 178 int main(int, char *[]); 179 180 int 181 main(int argc, char *argv[]) 182 { 183 int ret; 184 struct magic_set *ms; 185 char *progname; 186 187 if ((progname = strrchr(argv[0], '/')) != NULL) 188 progname++; 189 else 190 progname = argv[0]; 191 192 if (argc != 2) { 193 (void)fprintf(stderr, "Usage: %s file\n", progname); 194 return 1; 195 } 196 197 if ((ms = magic_open(MAGIC_CHECK)) == NULL) { 198 (void)fprintf(stderr, "%s: %s\n", progname, strerror(errno)); 199 return 1; 200 } 201 ret = magic_compile(ms, argv[1]) == -1 ? 1 : 0; 202 if (ret == 1) 203 (void)fprintf(stderr, "%s: %s\n", progname, magic_error(ms)); 204 magic_close(ms); 205 return ret; 206 } 207 #endif /* COMPILE_ONLY */ 208 209 struct type_tbl_s { 210 const char name[16]; 211 const size_t len; 212 const int type; 213 const int format; 214 }; 215 216 /* 217 * XXX - the actual Single UNIX Specification says that "long" means "long", 218 * as in the C data type, but we treat it as meaning "4-byte integer". 219 * Given that the OS X version of file 5.04 did the same, I guess that passes 220 * the actual test; having "long" be dependent on how big a "long" is on 221 * the machine running "file" is silly. 222 */ 223 static const struct type_tbl_s type_tbl[] = { 224 # define XX(s) s, (sizeof(s) - 1) 225 # define XX_NULL "", 0 226 { XX("invalid"), FILE_INVALID, FILE_FMT_NONE }, 227 { XX("byte"), FILE_BYTE, FILE_FMT_NUM }, 228 { XX("short"), FILE_SHORT, FILE_FMT_NUM }, 229 { XX("default"), FILE_DEFAULT, FILE_FMT_NONE }, 230 { XX("long"), FILE_LONG, FILE_FMT_NUM }, 231 { XX("string"), FILE_STRING, FILE_FMT_STR }, 232 { XX("date"), FILE_DATE, FILE_FMT_STR }, 233 { XX("beshort"), FILE_BESHORT, FILE_FMT_NUM }, 234 { XX("belong"), FILE_BELONG, FILE_FMT_NUM }, 235 { XX("bedate"), FILE_BEDATE, FILE_FMT_STR }, 236 { XX("leshort"), FILE_LESHORT, FILE_FMT_NUM }, 237 { XX("lelong"), FILE_LELONG, FILE_FMT_NUM }, 238 { XX("ledate"), FILE_LEDATE, FILE_FMT_STR }, 239 { XX("pstring"), FILE_PSTRING, FILE_FMT_STR }, 240 { XX("ldate"), FILE_LDATE, FILE_FMT_STR }, 241 { XX("beldate"), FILE_BELDATE, FILE_FMT_STR }, 242 { XX("leldate"), FILE_LELDATE, FILE_FMT_STR }, 243 { XX("regex"), FILE_REGEX, FILE_FMT_STR }, 244 { XX("bestring16"), FILE_BESTRING16, FILE_FMT_STR }, 245 { XX("lestring16"), FILE_LESTRING16, FILE_FMT_STR }, 246 { XX("search"), FILE_SEARCH, FILE_FMT_STR }, 247 { XX("medate"), FILE_MEDATE, FILE_FMT_STR }, 248 { XX("meldate"), FILE_MELDATE, FILE_FMT_STR }, 249 { XX("melong"), FILE_MELONG, FILE_FMT_NUM }, 250 { XX("quad"), FILE_QUAD, FILE_FMT_QUAD }, 251 { XX("lequad"), FILE_LEQUAD, FILE_FMT_QUAD }, 252 { XX("bequad"), FILE_BEQUAD, FILE_FMT_QUAD }, 253 { XX("qdate"), FILE_QDATE, FILE_FMT_STR }, 254 { XX("leqdate"), FILE_LEQDATE, FILE_FMT_STR }, 255 { XX("beqdate"), FILE_BEQDATE, FILE_FMT_STR }, 256 { XX("qldate"), FILE_QLDATE, FILE_FMT_STR }, 257 { XX("leqldate"), FILE_LEQLDATE, FILE_FMT_STR }, 258 { XX("beqldate"), FILE_BEQLDATE, FILE_FMT_STR }, 259 { XX("float"), FILE_FLOAT, FILE_FMT_FLOAT }, 260 { XX("befloat"), FILE_BEFLOAT, FILE_FMT_FLOAT }, 261 { XX("lefloat"), FILE_LEFLOAT, FILE_FMT_FLOAT }, 262 { XX("double"), FILE_DOUBLE, FILE_FMT_DOUBLE }, 263 { XX("bedouble"), FILE_BEDOUBLE, FILE_FMT_DOUBLE }, 264 { XX("ledouble"), FILE_LEDOUBLE, FILE_FMT_DOUBLE }, 265 { XX("leid3"), FILE_LEID3, FILE_FMT_NUM }, 266 { XX("beid3"), FILE_BEID3, FILE_FMT_NUM }, 267 { XX("indirect"), FILE_INDIRECT, FILE_FMT_NUM }, 268 { XX("qwdate"), FILE_QWDATE, FILE_FMT_STR }, 269 { XX("leqwdate"), FILE_LEQWDATE, FILE_FMT_STR }, 270 { XX("beqwdate"), FILE_BEQWDATE, FILE_FMT_STR }, 271 { XX("name"), FILE_NAME, FILE_FMT_NONE }, 272 { XX("use"), FILE_USE, FILE_FMT_NONE }, 273 { XX("clear"), FILE_CLEAR, FILE_FMT_NONE }, 274 { XX("der"), FILE_DER, FILE_FMT_STR }, 275 { XX("guid"), FILE_GUID, FILE_FMT_STR }, 276 { XX("offset"), FILE_OFFSET, FILE_FMT_QUAD }, 277 { XX_NULL, FILE_INVALID, FILE_FMT_NONE }, 278 }; 279 280 /* 281 * These are not types, and cannot be preceded by "u" to make them 282 * unsigned. 283 */ 284 static const struct type_tbl_s special_tbl[] = { 285 { XX("der"), FILE_DER, FILE_FMT_STR }, 286 { XX("name"), FILE_NAME, FILE_FMT_STR }, 287 { XX("use"), FILE_USE, FILE_FMT_STR }, 288 { XX_NULL, FILE_INVALID, FILE_FMT_NONE }, 289 }; 290 # undef XX 291 # undef XX_NULL 292 293 private int 294 get_type(const struct type_tbl_s *tbl, const char *l, const char **t) 295 { 296 const struct type_tbl_s *p; 297 298 for (p = tbl; p->len; p++) { 299 if (strncmp(l, p->name, p->len) == 0) { 300 if (t) 301 *t = l + p->len; 302 break; 303 } 304 } 305 return p->type; 306 } 307 308 private off_t 309 maxoff_t(void) { 310 if (/*CONSTCOND*/sizeof(off_t) == sizeof(int)) 311 return CAST(off_t, INT_MAX); 312 if (/*CONSTCOND*/sizeof(off_t) == sizeof(long)) 313 return CAST(off_t, LONG_MAX); 314 return 0x7fffffff; 315 } 316 317 private int 318 get_standard_integer_type(const char *l, const char **t) 319 { 320 int type; 321 322 if (isalpha(CAST(unsigned char, l[1]))) { 323 switch (l[1]) { 324 case 'C': 325 /* "dC" and "uC" */ 326 type = FILE_BYTE; 327 break; 328 case 'S': 329 /* "dS" and "uS" */ 330 type = FILE_SHORT; 331 break; 332 case 'I': 333 case 'L': 334 /* 335 * "dI", "dL", "uI", and "uL". 336 * 337 * XXX - the actual Single UNIX Specification says 338 * that "L" means "long", as in the C data type, 339 * but we treat it as meaning "4-byte integer". 340 * Given that the OS X version of file 5.04 did 341 * the same, I guess that passes the actual SUS 342 * validation suite; having "dL" be dependent on 343 * how big a "long" is on the machine running 344 * "file" is silly. 345 */ 346 type = FILE_LONG; 347 break; 348 case 'Q': 349 /* "dQ" and "uQ" */ 350 type = FILE_QUAD; 351 break; 352 default: 353 /* "d{anything else}", "u{anything else}" */ 354 return FILE_INVALID; 355 } 356 l += 2; 357 } else if (isdigit(CAST(unsigned char, l[1]))) { 358 /* 359 * "d{num}" and "u{num}"; we only support {num} values 360 * of 1, 2, 4, and 8 - the Single UNIX Specification 361 * doesn't say anything about whether arbitrary 362 * values should be supported, but both the Solaris 10 363 * and OS X Mountain Lion versions of file passed the 364 * Single UNIX Specification validation suite, and 365 * neither of them support values bigger than 8 or 366 * non-power-of-2 values. 367 */ 368 if (isdigit(CAST(unsigned char, l[2]))) { 369 /* Multi-digit, so > 9 */ 370 return FILE_INVALID; 371 } 372 switch (l[1]) { 373 case '1': 374 type = FILE_BYTE; 375 break; 376 case '2': 377 type = FILE_SHORT; 378 break; 379 case '4': 380 type = FILE_LONG; 381 break; 382 case '8': 383 type = FILE_QUAD; 384 break; 385 default: 386 /* XXX - what about 3, 5, 6, or 7? */ 387 return FILE_INVALID; 388 } 389 l += 2; 390 } else { 391 /* 392 * "d" or "u" by itself. 393 */ 394 type = FILE_LONG; 395 ++l; 396 } 397 if (t) 398 *t = l; 399 return type; 400 } 401 402 private void 403 init_file_tables(void) 404 { 405 static int done = 0; 406 const struct type_tbl_s *p; 407 408 if (done) 409 return; 410 done++; 411 412 for (p = type_tbl; p->len; p++) { 413 assert(p->type < FILE_NAMES_SIZE); 414 file_names[p->type] = p->name; 415 file_formats[p->type] = p->format; 416 } 417 assert(p - type_tbl == FILE_NAMES_SIZE); 418 } 419 420 private int 421 add_mlist(struct mlist *mlp, struct magic_map *map, size_t idx) 422 { 423 struct mlist *ml; 424 425 mlp->map = NULL; 426 if ((ml = CAST(struct mlist *, malloc(sizeof(*ml)))) == NULL) 427 return -1; 428 429 ml->map = idx == 0 ? map : NULL; 430 ml->magic = map->magic[idx]; 431 ml->nmagic = map->nmagic[idx]; 432 433 mlp->prev->next = ml; 434 ml->prev = mlp->prev; 435 ml->next = mlp; 436 mlp->prev = ml; 437 return 0; 438 } 439 440 /* 441 * Handle one file or directory. 442 */ 443 private int 444 apprentice_1(struct magic_set *ms, const char *fn, int action) 445 { 446 struct magic_map *map; 447 #ifndef COMPILE_ONLY 448 struct mlist *ml; 449 size_t i; 450 #endif 451 452 if (magicsize != FILE_MAGICSIZE) { 453 file_error(ms, 0, "magic element size %lu != %lu", 454 CAST(unsigned long, sizeof(*map->magic[0])), 455 CAST(unsigned long, FILE_MAGICSIZE)); 456 return -1; 457 } 458 459 if (action == FILE_COMPILE) { 460 map = apprentice_load(ms, fn, action); 461 if (map == NULL) 462 return -1; 463 return apprentice_compile(ms, map, fn); 464 } 465 466 #ifndef COMPILE_ONLY 467 map = apprentice_map(ms, fn); 468 if (map == NULL) { 469 if (ms->flags & MAGIC_CHECK) 470 file_magwarn(ms, "using regular magic file `%s'", fn); 471 map = apprentice_load(ms, fn, action); 472 if (map == NULL) 473 return -1; 474 } 475 476 for (i = 0; i < MAGIC_SETS; i++) { 477 if (add_mlist(ms->mlist[i], map, i) == -1) { 478 /* failed to add to any list, free explicitly */ 479 if (i == 0) 480 apprentice_unmap(map); 481 else 482 mlist_free_all(ms); 483 file_oomem(ms, sizeof(*ml)); 484 return -1; 485 } 486 } 487 488 if (action == FILE_LIST) { 489 for (i = 0; i < MAGIC_SETS; i++) { 490 printf("Set %" SIZE_T_FORMAT "u:\nBinary patterns:\n", 491 i); 492 apprentice_list(ms->mlist[i], BINTEST); 493 printf("Text patterns:\n"); 494 apprentice_list(ms->mlist[i], TEXTTEST); 495 } 496 } 497 return 0; 498 #else 499 return 0; 500 #endif /* COMPILE_ONLY */ 501 } 502 503 protected void 504 file_ms_free(struct magic_set *ms) 505 { 506 size_t i; 507 if (ms == NULL) 508 return; 509 for (i = 0; i < MAGIC_SETS; i++) 510 mlist_free(ms->mlist[i]); 511 free(ms->o.pbuf); 512 free(ms->o.buf); 513 free(ms->c.li); 514 free(ms); 515 } 516 517 protected struct magic_set * 518 file_ms_alloc(int flags) 519 { 520 struct magic_set *ms; 521 size_t i, len; 522 523 if ((ms = CAST(struct magic_set *, calloc(CAST(size_t, 1u), 524 sizeof(struct magic_set)))) == NULL) 525 return NULL; 526 527 if (magic_setflags(ms, flags) == -1) { 528 errno = EINVAL; 529 goto free; 530 } 531 532 ms->o.buf = ms->o.pbuf = NULL; 533 ms->o.blen = 0; 534 len = (ms->c.len = 10) * sizeof(*ms->c.li); 535 536 if ((ms->c.li = CAST(struct level_info *, malloc(len))) == NULL) 537 goto free; 538 539 ms->event_flags = 0; 540 ms->error = -1; 541 for (i = 0; i < MAGIC_SETS; i++) 542 ms->mlist[i] = NULL; 543 ms->file = "unknown"; 544 ms->line = 0; 545 ms->indir_max = FILE_INDIR_MAX; 546 ms->name_max = FILE_NAME_MAX; 547 ms->elf_shnum_max = FILE_ELF_SHNUM_MAX; 548 ms->elf_phnum_max = FILE_ELF_PHNUM_MAX; 549 ms->elf_notes_max = FILE_ELF_NOTES_MAX; 550 ms->regex_max = FILE_REGEX_MAX; 551 ms->bytes_max = FILE_BYTES_MAX; 552 return ms; 553 free: 554 free(ms); 555 return NULL; 556 } 557 558 private void 559 apprentice_unmap(struct magic_map *map) 560 { 561 size_t i; 562 if (map == NULL) 563 return; 564 565 switch (map->type) { 566 case MAP_TYPE_USER: 567 break; 568 case MAP_TYPE_MALLOC: 569 for (i = 0; i < MAGIC_SETS; i++) { 570 void *b = map->magic[i]; 571 void *p = map->p; 572 if (CAST(char *, b) >= CAST(char *, p) && 573 CAST(char *, b) <= CAST(char *, p) + map->len) 574 continue; 575 free(map->magic[i]); 576 } 577 free(map->p); 578 break; 579 #ifdef QUICK 580 case MAP_TYPE_MMAP: 581 if (map->p && map->p != MAP_FAILED) 582 (void)munmap(map->p, map->len); 583 break; 584 #endif 585 default: 586 abort(); 587 } 588 free(map); 589 } 590 591 private struct mlist * 592 mlist_alloc(void) 593 { 594 struct mlist *mlist; 595 if ((mlist = CAST(struct mlist *, calloc(1, sizeof(*mlist)))) == NULL) { 596 return NULL; 597 } 598 mlist->next = mlist->prev = mlist; 599 return mlist; 600 } 601 602 private void 603 mlist_free_all(struct magic_set *ms) 604 { 605 size_t i; 606 607 for (i = 0; i < MAGIC_SETS; i++) { 608 mlist_free(ms->mlist[i]); 609 ms->mlist[i] = NULL; 610 } 611 } 612 613 private void 614 mlist_free_one(struct mlist *ml) 615 { 616 if (ml->map) 617 apprentice_unmap(CAST(struct magic_map *, ml->map)); 618 free(ml); 619 } 620 621 private void 622 mlist_free(struct mlist *mlist) 623 { 624 struct mlist *ml, *next; 625 626 if (mlist == NULL) 627 return; 628 629 for (ml = mlist->next; ml != mlist;) { 630 next = ml->next; 631 mlist_free_one(ml); 632 ml = next; 633 } 634 mlist_free_one(mlist); 635 } 636 637 #ifndef COMPILE_ONLY 638 /* void **bufs: an array of compiled magic files */ 639 protected int 640 buffer_apprentice(struct magic_set *ms, struct magic **bufs, 641 size_t *sizes, size_t nbufs) 642 { 643 size_t i, j; 644 struct mlist *ml; 645 struct magic_map *map; 646 647 if (nbufs == 0) 648 return -1; 649 650 (void)file_reset(ms, 0); 651 652 init_file_tables(); 653 654 for (i = 0; i < MAGIC_SETS; i++) { 655 mlist_free(ms->mlist[i]); 656 if ((ms->mlist[i] = mlist_alloc()) == NULL) { 657 file_oomem(ms, sizeof(*ms->mlist[i])); 658 goto fail; 659 } 660 } 661 662 for (i = 0; i < nbufs; i++) { 663 map = apprentice_buf(ms, bufs[i], sizes[i]); 664 if (map == NULL) 665 goto fail; 666 667 for (j = 0; j < MAGIC_SETS; j++) { 668 if (add_mlist(ms->mlist[j], map, j) == -1) { 669 file_oomem(ms, sizeof(*ml)); 670 goto fail; 671 } 672 } 673 } 674 675 return 0; 676 fail: 677 mlist_free_all(ms); 678 return -1; 679 } 680 #endif 681 682 /* const char *fn: list of magic files and directories */ 683 protected int 684 file_apprentice(struct magic_set *ms, const char *fn, int action) 685 { 686 char *p, *mfn; 687 int fileerr, errs = -1; 688 size_t i, j; 689 690 (void)file_reset(ms, 0); 691 692 if ((fn = magic_getpath(fn, action)) == NULL) 693 return -1; 694 695 init_file_tables(); 696 697 if ((mfn = strdup(fn)) == NULL) { 698 file_oomem(ms, strlen(fn)); 699 return -1; 700 } 701 702 for (i = 0; i < MAGIC_SETS; i++) { 703 mlist_free(ms->mlist[i]); 704 if ((ms->mlist[i] = mlist_alloc()) == NULL) { 705 file_oomem(ms, sizeof(*ms->mlist[i])); 706 for (j = 0; j < i; j++) { 707 mlist_free(ms->mlist[j]); 708 ms->mlist[j] = NULL; 709 } 710 free(mfn); 711 return -1; 712 } 713 } 714 fn = mfn; 715 716 while (fn) { 717 p = strchr(fn, PATHSEP); 718 if (p) 719 *p++ = '\0'; 720 if (*fn == '\0') 721 break; 722 fileerr = apprentice_1(ms, fn, action); 723 errs = MAX(errs, fileerr); 724 fn = p; 725 } 726 727 free(mfn); 728 729 if (errs == -1) { 730 for (i = 0; i < MAGIC_SETS; i++) { 731 mlist_free(ms->mlist[i]); 732 ms->mlist[i] = NULL; 733 } 734 file_error(ms, 0, "could not find any valid magic files!"); 735 return -1; 736 } 737 738 #if 0 739 /* 740 * Always leave the database loaded 741 */ 742 if (action == FILE_LOAD) 743 return 0; 744 745 for (i = 0; i < MAGIC_SETS; i++) { 746 mlist_free(ms->mlist[i]); 747 ms->mlist[i] = NULL; 748 } 749 #endif 750 751 switch (action) { 752 case FILE_LOAD: 753 case FILE_COMPILE: 754 case FILE_CHECK: 755 case FILE_LIST: 756 return 0; 757 default: 758 file_error(ms, 0, "Invalid action %d", action); 759 return -1; 760 } 761 } 762 763 /* 764 * Compute the real length of a magic expression, for the purposes 765 * of determining how "strong" a magic expression is (approximating 766 * how specific its matches are): 767 * - magic characters count 0 unless escaped. 768 * - [] expressions count 1 769 * - {} expressions count 0 770 * - regular characters or escaped magic characters count 1 771 * - 0 length expressions count as one 772 */ 773 private size_t 774 nonmagic(const char *str) 775 { 776 const char *p; 777 size_t rv = 0; 778 779 for (p = str; *p; p++) 780 switch (*p) { 781 case '\\': /* Escaped anything counts 1 */ 782 if (!*++p) 783 p--; 784 rv++; 785 continue; 786 case '?': /* Magic characters count 0 */ 787 case '*': 788 case '.': 789 case '+': 790 case '^': 791 case '$': 792 continue; 793 case '[': /* Bracketed expressions count 1 the ']' */ 794 while (*p && *p != ']') 795 p++; 796 p--; 797 continue; 798 case '{': /* Braced expressions count 0 */ 799 while (*p && *p != '}') 800 p++; 801 if (!*p) 802 p--; 803 continue; 804 default: /* Anything else counts 1 */ 805 rv++; 806 continue; 807 } 808 809 return rv == 0 ? 1 : rv; /* Return at least 1 */ 810 } 811 812 813 private size_t 814 typesize(int type) 815 { 816 switch (type) { 817 case FILE_BYTE: 818 return 1; 819 820 case FILE_SHORT: 821 case FILE_LESHORT: 822 case FILE_BESHORT: 823 return 2; 824 825 case FILE_LONG: 826 case FILE_LELONG: 827 case FILE_BELONG: 828 case FILE_MELONG: 829 return 4; 830 831 case FILE_DATE: 832 case FILE_LEDATE: 833 case FILE_BEDATE: 834 case FILE_MEDATE: 835 case FILE_LDATE: 836 case FILE_LELDATE: 837 case FILE_BELDATE: 838 case FILE_MELDATE: 839 case FILE_FLOAT: 840 case FILE_BEFLOAT: 841 case FILE_LEFLOAT: 842 return 4; 843 844 case FILE_QUAD: 845 case FILE_BEQUAD: 846 case FILE_LEQUAD: 847 case FILE_QDATE: 848 case FILE_LEQDATE: 849 case FILE_BEQDATE: 850 case FILE_QLDATE: 851 case FILE_LEQLDATE: 852 case FILE_BEQLDATE: 853 case FILE_QWDATE: 854 case FILE_LEQWDATE: 855 case FILE_BEQWDATE: 856 case FILE_DOUBLE: 857 case FILE_BEDOUBLE: 858 case FILE_LEDOUBLE: 859 case FILE_OFFSET: 860 return 8; 861 862 case FILE_GUID: 863 return 16; 864 865 default: 866 return FILE_BADSIZE; 867 } 868 } 869 870 /* 871 * Get weight of this magic entry, for sorting purposes. 872 */ 873 private size_t 874 apprentice_magic_strength(const struct magic *m) 875 { 876 #define MULT 10U 877 size_t ts, v; 878 ssize_t val = 2 * MULT; /* baseline strength */ 879 880 switch (m->type) { 881 case FILE_DEFAULT: /* make sure this sorts last */ 882 if (m->factor_op != FILE_FACTOR_OP_NONE) 883 abort(); 884 return 0; 885 886 case FILE_BYTE: 887 case FILE_SHORT: 888 case FILE_LESHORT: 889 case FILE_BESHORT: 890 case FILE_LONG: 891 case FILE_LELONG: 892 case FILE_BELONG: 893 case FILE_MELONG: 894 case FILE_DATE: 895 case FILE_LEDATE: 896 case FILE_BEDATE: 897 case FILE_MEDATE: 898 case FILE_LDATE: 899 case FILE_LELDATE: 900 case FILE_BELDATE: 901 case FILE_MELDATE: 902 case FILE_FLOAT: 903 case FILE_BEFLOAT: 904 case FILE_LEFLOAT: 905 case FILE_QUAD: 906 case FILE_BEQUAD: 907 case FILE_LEQUAD: 908 case FILE_QDATE: 909 case FILE_LEQDATE: 910 case FILE_BEQDATE: 911 case FILE_QLDATE: 912 case FILE_LEQLDATE: 913 case FILE_BEQLDATE: 914 case FILE_QWDATE: 915 case FILE_LEQWDATE: 916 case FILE_BEQWDATE: 917 case FILE_DOUBLE: 918 case FILE_BEDOUBLE: 919 case FILE_LEDOUBLE: 920 case FILE_GUID: 921 case FILE_OFFSET: 922 ts = typesize(m->type); 923 if (ts == FILE_BADSIZE) 924 abort(); 925 val += ts * MULT; 926 break; 927 928 case FILE_PSTRING: 929 case FILE_STRING: 930 val += m->vallen * MULT; 931 break; 932 933 case FILE_BESTRING16: 934 case FILE_LESTRING16: 935 val += m->vallen * MULT / 2; 936 break; 937 938 case FILE_SEARCH: 939 if (m->vallen == 0) 940 break; 941 val += m->vallen * MAX(MULT / m->vallen, 1); 942 break; 943 944 case FILE_REGEX: 945 v = nonmagic(m->value.s); 946 val += v * MAX(MULT / v, 1); 947 break; 948 949 case FILE_INDIRECT: 950 case FILE_NAME: 951 case FILE_USE: 952 break; 953 954 case FILE_DER: 955 val += MULT; 956 break; 957 958 default: 959 (void)fprintf(stderr, "Bad type %d\n", m->type); 960 abort(); 961 } 962 963 switch (m->reln) { 964 case 'x': /* matches anything penalize */ 965 case '!': /* matches almost anything penalize */ 966 val = 0; 967 break; 968 969 case '=': /* Exact match, prefer */ 970 val += MULT; 971 break; 972 973 case '>': 974 case '<': /* comparison match reduce strength */ 975 val -= 2 * MULT; 976 break; 977 978 case '^': 979 case '&': /* masking bits, we could count them too */ 980 val -= MULT; 981 break; 982 983 default: 984 (void)fprintf(stderr, "Bad relation %c\n", m->reln); 985 abort(); 986 } 987 988 switch (m->factor_op) { 989 case FILE_FACTOR_OP_NONE: 990 break; 991 case FILE_FACTOR_OP_PLUS: 992 val += m->factor; 993 break; 994 case FILE_FACTOR_OP_MINUS: 995 val -= m->factor; 996 break; 997 case FILE_FACTOR_OP_TIMES: 998 val *= m->factor; 999 break; 1000 case FILE_FACTOR_OP_DIV: 1001 val /= m->factor; 1002 break; 1003 default: 1004 abort(); 1005 } 1006 1007 if (val <= 0) /* ensure we only return 0 for FILE_DEFAULT */ 1008 val = 1; 1009 1010 /* 1011 * Magic entries with no description get a bonus because they depend 1012 * on subsequent magic entries to print something. 1013 */ 1014 if (m->desc[0] == '\0') 1015 val++; 1016 return val; 1017 } 1018 1019 /* 1020 * Sort callback for sorting entries by "strength" (basically length) 1021 */ 1022 private int 1023 apprentice_sort(const void *a, const void *b) 1024 { 1025 const struct magic_entry *ma = CAST(const struct magic_entry *, a); 1026 const struct magic_entry *mb = CAST(const struct magic_entry *, b); 1027 size_t sa = apprentice_magic_strength(ma->mp); 1028 size_t sb = apprentice_magic_strength(mb->mp); 1029 if (sa == sb) 1030 return 0; 1031 else if (sa > sb) 1032 return -1; 1033 else 1034 return 1; 1035 } 1036 1037 /* 1038 * Shows sorted patterns list in the order which is used for the matching 1039 */ 1040 private void 1041 apprentice_list(struct mlist *mlist, int mode) 1042 { 1043 uint32_t magindex = 0; 1044 struct mlist *ml; 1045 for (ml = mlist->next; ml != mlist; ml = ml->next) { 1046 for (magindex = 0; magindex < ml->nmagic; magindex++) { 1047 struct magic *m = &ml->magic[magindex]; 1048 if ((m->flag & mode) != mode) { 1049 /* Skip sub-tests */ 1050 while (magindex + 1 < ml->nmagic && 1051 ml->magic[magindex + 1].cont_level != 0) 1052 ++magindex; 1053 continue; /* Skip to next top-level test*/ 1054 } 1055 1056 /* 1057 * Try to iterate over the tree until we find item with 1058 * description/mimetype. 1059 */ 1060 while (magindex + 1 < ml->nmagic && 1061 ml->magic[magindex + 1].cont_level != 0 && 1062 *ml->magic[magindex].desc == '\0' && 1063 *ml->magic[magindex].mimetype == '\0') 1064 magindex++; 1065 1066 printf("Strength = %3" SIZE_T_FORMAT "u@%u: %s [%s]\n", 1067 apprentice_magic_strength(m), 1068 ml->magic[magindex].lineno, 1069 ml->magic[magindex].desc, 1070 ml->magic[magindex].mimetype); 1071 } 1072 } 1073 } 1074 1075 private void 1076 set_test_type(struct magic *mstart, struct magic *m) 1077 { 1078 switch (m->type) { 1079 case FILE_BYTE: 1080 case FILE_SHORT: 1081 case FILE_LONG: 1082 case FILE_DATE: 1083 case FILE_BESHORT: 1084 case FILE_BELONG: 1085 case FILE_BEDATE: 1086 case FILE_LESHORT: 1087 case FILE_LELONG: 1088 case FILE_LEDATE: 1089 case FILE_LDATE: 1090 case FILE_BELDATE: 1091 case FILE_LELDATE: 1092 case FILE_MEDATE: 1093 case FILE_MELDATE: 1094 case FILE_MELONG: 1095 case FILE_QUAD: 1096 case FILE_LEQUAD: 1097 case FILE_BEQUAD: 1098 case FILE_QDATE: 1099 case FILE_LEQDATE: 1100 case FILE_BEQDATE: 1101 case FILE_QLDATE: 1102 case FILE_LEQLDATE: 1103 case FILE_BEQLDATE: 1104 case FILE_QWDATE: 1105 case FILE_LEQWDATE: 1106 case FILE_BEQWDATE: 1107 case FILE_FLOAT: 1108 case FILE_BEFLOAT: 1109 case FILE_LEFLOAT: 1110 case FILE_DOUBLE: 1111 case FILE_BEDOUBLE: 1112 case FILE_LEDOUBLE: 1113 case FILE_DER: 1114 case FILE_GUID: 1115 case FILE_OFFSET: 1116 mstart->flag |= BINTEST; 1117 break; 1118 case FILE_STRING: 1119 case FILE_PSTRING: 1120 case FILE_BESTRING16: 1121 case FILE_LESTRING16: 1122 /* Allow text overrides */ 1123 if (mstart->str_flags & STRING_TEXTTEST) 1124 mstart->flag |= TEXTTEST; 1125 else 1126 mstart->flag |= BINTEST; 1127 break; 1128 case FILE_REGEX: 1129 case FILE_SEARCH: 1130 /* Check for override */ 1131 if (mstart->str_flags & STRING_BINTEST) 1132 mstart->flag |= BINTEST; 1133 if (mstart->str_flags & STRING_TEXTTEST) 1134 mstart->flag |= TEXTTEST; 1135 1136 if (mstart->flag & (TEXTTEST|BINTEST)) 1137 break; 1138 1139 /* binary test if pattern is not text */ 1140 if (file_looks_utf8(m->value.us, CAST(size_t, m->vallen), NULL, 1141 NULL) <= 0) 1142 mstart->flag |= BINTEST; 1143 else 1144 mstart->flag |= TEXTTEST; 1145 break; 1146 case FILE_DEFAULT: 1147 /* can't deduce anything; we shouldn't see this at the 1148 top level anyway */ 1149 break; 1150 case FILE_INVALID: 1151 default: 1152 /* invalid search type, but no need to complain here */ 1153 break; 1154 } 1155 } 1156 1157 private int 1158 addentry(struct magic_set *ms, struct magic_entry *me, 1159 struct magic_entry_set *mset) 1160 { 1161 size_t i = me->mp->type == FILE_NAME ? 1 : 0; 1162 if (mset[i].count == mset[i].max) { 1163 struct magic_entry *mp; 1164 1165 mset[i].max += ALLOC_INCR; 1166 if ((mp = CAST(struct magic_entry *, 1167 realloc(mset[i].me, sizeof(*mp) * mset[i].max))) == 1168 NULL) { 1169 file_oomem(ms, sizeof(*mp) * mset[i].max); 1170 return -1; 1171 } 1172 (void)memset(&mp[mset[i].count], 0, sizeof(*mp) * 1173 ALLOC_INCR); 1174 mset[i].me = mp; 1175 } 1176 mset[i].me[mset[i].count++] = *me; 1177 memset(me, 0, sizeof(*me)); 1178 return 0; 1179 } 1180 1181 /* 1182 * Load and parse one file. 1183 */ 1184 private void 1185 load_1(struct magic_set *ms, int action, const char *fn, int *errs, 1186 struct magic_entry_set *mset) 1187 { 1188 size_t lineno = 0, llen = 0; 1189 char *line = NULL; 1190 ssize_t len; 1191 struct magic_entry me; 1192 1193 FILE *f = fopen(ms->file = fn, "r"); 1194 if (f == NULL) { 1195 if (errno != ENOENT) 1196 file_error(ms, errno, "cannot read magic file `%s'", 1197 fn); 1198 (*errs)++; 1199 return; 1200 } 1201 1202 memset(&me, 0, sizeof(me)); 1203 /* read and parse this file */ 1204 for (ms->line = 1; (len = getline(&line, &llen, f)) != -1; 1205 ms->line++) { 1206 if (len == 0) /* null line, garbage, etc */ 1207 continue; 1208 if (line[len - 1] == '\n') { 1209 lineno++; 1210 line[len - 1] = '\0'; /* delete newline */ 1211 } 1212 switch (line[0]) { 1213 case '\0': /* empty, do not parse */ 1214 case '#': /* comment, do not parse */ 1215 continue; 1216 case '!': 1217 if (line[1] == ':') { 1218 size_t i; 1219 1220 for (i = 0; bang[i].name != NULL; i++) { 1221 if (CAST(size_t, len - 2) > bang[i].len && 1222 memcmp(bang[i].name, line + 2, 1223 bang[i].len) == 0) 1224 break; 1225 } 1226 if (bang[i].name == NULL) { 1227 file_error(ms, 0, 1228 "Unknown !: entry `%s'", line); 1229 (*errs)++; 1230 continue; 1231 } 1232 if (me.mp == NULL) { 1233 file_error(ms, 0, 1234 "No current entry for :!%s type", 1235 bang[i].name); 1236 (*errs)++; 1237 continue; 1238 } 1239 if ((*bang[i].fun)(ms, &me, 1240 line + bang[i].len + 2, 1241 len - bang[i].len - 2) != 0) { 1242 (*errs)++; 1243 continue; 1244 } 1245 continue; 1246 } 1247 /*FALLTHROUGH*/ 1248 default: 1249 again: 1250 switch (parse(ms, &me, line, lineno, action)) { 1251 case 0: 1252 continue; 1253 case 1: 1254 (void)addentry(ms, &me, mset); 1255 goto again; 1256 default: 1257 (*errs)++; 1258 break; 1259 } 1260 } 1261 } 1262 if (me.mp) 1263 (void)addentry(ms, &me, mset); 1264 free(line); 1265 (void)fclose(f); 1266 } 1267 1268 /* 1269 * parse a file or directory of files 1270 * const char *fn: name of magic file or directory 1271 */ 1272 private int 1273 cmpstrp(const void *p1, const void *p2) 1274 { 1275 return strcmp(*RCAST(char *const *, p1), *RCAST(char *const *, p2)); 1276 } 1277 1278 1279 private uint32_t 1280 set_text_binary(struct magic_set *ms, struct magic_entry *me, uint32_t nme, 1281 uint32_t starttest) 1282 { 1283 static const char text[] = "text"; 1284 static const char binary[] = "binary"; 1285 static const size_t len = sizeof(text); 1286 1287 uint32_t i = starttest; 1288 1289 do { 1290 set_test_type(me[starttest].mp, me[i].mp); 1291 if ((ms->flags & MAGIC_DEBUG) == 0) 1292 continue; 1293 (void)fprintf(stderr, "%s%s%s: %s\n", 1294 me[i].mp->mimetype, 1295 me[i].mp->mimetype[0] == '\0' ? "" : "; ", 1296 me[i].mp->desc[0] ? me[i].mp->desc : "(no description)", 1297 me[i].mp->flag & BINTEST ? binary : text); 1298 if (me[i].mp->flag & BINTEST) { 1299 char *p = strstr(me[i].mp->desc, text); 1300 if (p && (p == me[i].mp->desc || 1301 isspace(CAST(unsigned char, p[-1]))) && 1302 (p + len - me[i].mp->desc == MAXstring 1303 || (p[len] == '\0' || 1304 isspace(CAST(unsigned char, p[len]))))) 1305 (void)fprintf(stderr, "*** Possible " 1306 "binary test for text type\n"); 1307 } 1308 } while (++i < nme && me[i].mp->cont_level != 0); 1309 return i; 1310 } 1311 1312 private void 1313 set_last_default(struct magic_set *ms, struct magic_entry *me, uint32_t nme) 1314 { 1315 uint32_t i; 1316 for (i = 0; i < nme; i++) { 1317 if (me[i].mp->cont_level == 0 && 1318 me[i].mp->type == FILE_DEFAULT) { 1319 while (++i < nme) 1320 if (me[i].mp->cont_level == 0) 1321 break; 1322 if (i != nme) { 1323 /* XXX - Ugh! */ 1324 ms->line = me[i].mp->lineno; 1325 file_magwarn(ms, 1326 "level 0 \"default\" did not sort last"); 1327 } 1328 return; 1329 } 1330 } 1331 } 1332 1333 private int 1334 coalesce_entries(struct magic_set *ms, struct magic_entry *me, uint32_t nme, 1335 struct magic **ma, uint32_t *nma) 1336 { 1337 uint32_t i, mentrycount = 0; 1338 size_t slen; 1339 1340 for (i = 0; i < nme; i++) 1341 mentrycount += me[i].cont_count; 1342 1343 slen = sizeof(**ma) * mentrycount; 1344 if ((*ma = CAST(struct magic *, malloc(slen))) == NULL) { 1345 file_oomem(ms, slen); 1346 return -1; 1347 } 1348 1349 mentrycount = 0; 1350 for (i = 0; i < nme; i++) { 1351 (void)memcpy(*ma + mentrycount, me[i].mp, 1352 me[i].cont_count * sizeof(**ma)); 1353 mentrycount += me[i].cont_count; 1354 } 1355 *nma = mentrycount; 1356 return 0; 1357 } 1358 1359 private void 1360 magic_entry_free(struct magic_entry *me, uint32_t nme) 1361 { 1362 uint32_t i; 1363 if (me == NULL) 1364 return; 1365 for (i = 0; i < nme; i++) 1366 free(me[i].mp); 1367 free(me); 1368 } 1369 1370 private struct magic_map * 1371 apprentice_load(struct magic_set *ms, const char *fn, int action) 1372 { 1373 int errs = 0; 1374 uint32_t i, j; 1375 size_t files = 0, maxfiles = 0; 1376 char **filearr = NULL, *mfn; 1377 struct stat st; 1378 struct magic_map *map; 1379 struct magic_entry_set mset[MAGIC_SETS]; 1380 DIR *dir; 1381 struct dirent *d; 1382 1383 memset(mset, 0, sizeof(mset)); 1384 ms->flags |= MAGIC_CHECK; /* Enable checks for parsed files */ 1385 1386 1387 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) 1388 { 1389 file_oomem(ms, sizeof(*map)); 1390 return NULL; 1391 } 1392 map->type = MAP_TYPE_MALLOC; 1393 1394 /* print silly verbose header for USG compat. */ 1395 if (action == FILE_CHECK) 1396 (void)fprintf(stderr, "%s\n", usg_hdr); 1397 1398 /* load directory or file */ 1399 if (stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) { 1400 dir = opendir(fn); 1401 if (!dir) { 1402 errs++; 1403 goto out; 1404 } 1405 while ((d = readdir(dir)) != NULL) { 1406 if (d->d_name[0] == '.') 1407 continue; 1408 if (asprintf(&mfn, "%s/%s", fn, d->d_name) < 0) { 1409 file_oomem(ms, 1410 strlen(fn) + strlen(d->d_name) + 2); 1411 errs++; 1412 closedir(dir); 1413 goto out; 1414 } 1415 if (stat(mfn, &st) == -1 || !S_ISREG(st.st_mode)) { 1416 free(mfn); 1417 continue; 1418 } 1419 if (files >= maxfiles) { 1420 size_t mlen; 1421 char **nfilearr; 1422 maxfiles = (maxfiles + 1) * 2; 1423 mlen = maxfiles * sizeof(*filearr); 1424 if ((nfilearr = CAST(char **, 1425 realloc(filearr, mlen))) == NULL) { 1426 file_oomem(ms, mlen); 1427 free(mfn); 1428 closedir(dir); 1429 errs++; 1430 goto out; 1431 } 1432 filearr = nfilearr; 1433 } 1434 filearr[files++] = mfn; 1435 } 1436 closedir(dir); 1437 if (filearr) { 1438 qsort(filearr, files, sizeof(*filearr), cmpstrp); 1439 for (i = 0; i < files; i++) { 1440 load_1(ms, action, filearr[i], &errs, mset); 1441 free(filearr[i]); 1442 } 1443 free(filearr); 1444 filearr = NULL; 1445 } 1446 } else 1447 load_1(ms, action, fn, &errs, mset); 1448 if (errs) 1449 goto out; 1450 1451 for (j = 0; j < MAGIC_SETS; j++) { 1452 /* Set types of tests */ 1453 for (i = 0; i < mset[j].count; ) { 1454 if (mset[j].me[i].mp->cont_level != 0) { 1455 i++; 1456 continue; 1457 } 1458 i = set_text_binary(ms, mset[j].me, mset[j].count, i); 1459 } 1460 if (mset[j].me) 1461 qsort(mset[j].me, mset[j].count, sizeof(*mset[j].me), 1462 apprentice_sort); 1463 1464 /* 1465 * Make sure that any level 0 "default" line is last 1466 * (if one exists). 1467 */ 1468 set_last_default(ms, mset[j].me, mset[j].count); 1469 1470 /* coalesce per file arrays into a single one */ 1471 if (coalesce_entries(ms, mset[j].me, mset[j].count, 1472 &map->magic[j], &map->nmagic[j]) == -1) { 1473 errs++; 1474 goto out; 1475 } 1476 } 1477 1478 out: 1479 free(filearr); 1480 for (j = 0; j < MAGIC_SETS; j++) 1481 magic_entry_free(mset[j].me, mset[j].count); 1482 1483 if (errs) { 1484 apprentice_unmap(map); 1485 return NULL; 1486 } 1487 return map; 1488 } 1489 1490 /* 1491 * extend the sign bit if the comparison is to be signed 1492 */ 1493 protected uint64_t 1494 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v) 1495 { 1496 if (!(m->flag & UNSIGNED)) { 1497 switch(m->type) { 1498 /* 1499 * Do not remove the casts below. They are 1500 * vital. When later compared with the data, 1501 * the sign extension must have happened. 1502 */ 1503 case FILE_BYTE: 1504 v = CAST(signed char, v); 1505 break; 1506 case FILE_SHORT: 1507 case FILE_BESHORT: 1508 case FILE_LESHORT: 1509 v = CAST(short, v); 1510 break; 1511 case FILE_DATE: 1512 case FILE_BEDATE: 1513 case FILE_LEDATE: 1514 case FILE_MEDATE: 1515 case FILE_LDATE: 1516 case FILE_BELDATE: 1517 case FILE_LELDATE: 1518 case FILE_MELDATE: 1519 case FILE_LONG: 1520 case FILE_BELONG: 1521 case FILE_LELONG: 1522 case FILE_MELONG: 1523 case FILE_FLOAT: 1524 case FILE_BEFLOAT: 1525 case FILE_LEFLOAT: 1526 v = CAST(int32_t, v); 1527 break; 1528 case FILE_QUAD: 1529 case FILE_BEQUAD: 1530 case FILE_LEQUAD: 1531 case FILE_QDATE: 1532 case FILE_QLDATE: 1533 case FILE_QWDATE: 1534 case FILE_BEQDATE: 1535 case FILE_BEQLDATE: 1536 case FILE_BEQWDATE: 1537 case FILE_LEQDATE: 1538 case FILE_LEQLDATE: 1539 case FILE_LEQWDATE: 1540 case FILE_DOUBLE: 1541 case FILE_BEDOUBLE: 1542 case FILE_LEDOUBLE: 1543 case FILE_OFFSET: 1544 v = CAST(int64_t, v); 1545 break; 1546 case FILE_STRING: 1547 case FILE_PSTRING: 1548 case FILE_BESTRING16: 1549 case FILE_LESTRING16: 1550 case FILE_REGEX: 1551 case FILE_SEARCH: 1552 case FILE_DEFAULT: 1553 case FILE_INDIRECT: 1554 case FILE_NAME: 1555 case FILE_USE: 1556 case FILE_CLEAR: 1557 case FILE_DER: 1558 case FILE_GUID: 1559 break; 1560 default: 1561 if (ms->flags & MAGIC_CHECK) 1562 file_magwarn(ms, "cannot happen: m->type=%d\n", 1563 m->type); 1564 return FILE_BADSIZE; 1565 } 1566 } 1567 return v; 1568 } 1569 1570 private int 1571 string_modifier_check(struct magic_set *ms, struct magic *m) 1572 { 1573 if ((ms->flags & MAGIC_CHECK) == 0) 1574 return 0; 1575 1576 if ((m->type != FILE_REGEX || (m->str_flags & REGEX_LINE_COUNT) == 0) && 1577 (m->type != FILE_PSTRING && (m->str_flags & PSTRING_LEN) != 0)) { 1578 file_magwarn(ms, 1579 "'/BHhLl' modifiers are only allowed for pascal strings\n"); 1580 return -1; 1581 } 1582 switch (m->type) { 1583 case FILE_BESTRING16: 1584 case FILE_LESTRING16: 1585 if (m->str_flags != 0) { 1586 file_magwarn(ms, 1587 "no modifiers allowed for 16-bit strings\n"); 1588 return -1; 1589 } 1590 break; 1591 case FILE_STRING: 1592 case FILE_PSTRING: 1593 if ((m->str_flags & REGEX_OFFSET_START) != 0) { 1594 file_magwarn(ms, 1595 "'/%c' only allowed on regex and search\n", 1596 CHAR_REGEX_OFFSET_START); 1597 return -1; 1598 } 1599 break; 1600 case FILE_SEARCH: 1601 if (m->str_range == 0) { 1602 file_magwarn(ms, 1603 "missing range; defaulting to %d\n", 1604 STRING_DEFAULT_RANGE); 1605 m->str_range = STRING_DEFAULT_RANGE; 1606 return -1; 1607 } 1608 break; 1609 case FILE_REGEX: 1610 if ((m->str_flags & STRING_COMPACT_WHITESPACE) != 0) { 1611 file_magwarn(ms, "'/%c' not allowed on regex\n", 1612 CHAR_COMPACT_WHITESPACE); 1613 return -1; 1614 } 1615 if ((m->str_flags & STRING_COMPACT_OPTIONAL_WHITESPACE) != 0) { 1616 file_magwarn(ms, "'/%c' not allowed on regex\n", 1617 CHAR_COMPACT_OPTIONAL_WHITESPACE); 1618 return -1; 1619 } 1620 break; 1621 default: 1622 file_magwarn(ms, "coding error: m->type=%d\n", 1623 m->type); 1624 return -1; 1625 } 1626 return 0; 1627 } 1628 1629 private int 1630 get_op(char c) 1631 { 1632 switch (c) { 1633 case '&': 1634 return FILE_OPAND; 1635 case '|': 1636 return FILE_OPOR; 1637 case '^': 1638 return FILE_OPXOR; 1639 case '+': 1640 return FILE_OPADD; 1641 case '-': 1642 return FILE_OPMINUS; 1643 case '*': 1644 return FILE_OPMULTIPLY; 1645 case '/': 1646 return FILE_OPDIVIDE; 1647 case '%': 1648 return FILE_OPMODULO; 1649 default: 1650 return -1; 1651 } 1652 } 1653 1654 #ifdef ENABLE_CONDITIONALS 1655 private int 1656 get_cond(const char *l, const char **t) 1657 { 1658 static const struct cond_tbl_s { 1659 char name[8]; 1660 size_t len; 1661 int cond; 1662 } cond_tbl[] = { 1663 { "if", 2, COND_IF }, 1664 { "elif", 4, COND_ELIF }, 1665 { "else", 4, COND_ELSE }, 1666 { "", 0, COND_NONE }, 1667 }; 1668 const struct cond_tbl_s *p; 1669 1670 for (p = cond_tbl; p->len; p++) { 1671 if (strncmp(l, p->name, p->len) == 0 && 1672 isspace(CAST(unsigned char, l[p->len]))) { 1673 if (t) 1674 *t = l + p->len; 1675 break; 1676 } 1677 } 1678 return p->cond; 1679 } 1680 1681 private int 1682 check_cond(struct magic_set *ms, int cond, uint32_t cont_level) 1683 { 1684 int last_cond; 1685 last_cond = ms->c.li[cont_level].last_cond; 1686 1687 switch (cond) { 1688 case COND_IF: 1689 if (last_cond != COND_NONE && last_cond != COND_ELIF) { 1690 if (ms->flags & MAGIC_CHECK) 1691 file_magwarn(ms, "syntax error: `if'"); 1692 return -1; 1693 } 1694 last_cond = COND_IF; 1695 break; 1696 1697 case COND_ELIF: 1698 if (last_cond != COND_IF && last_cond != COND_ELIF) { 1699 if (ms->flags & MAGIC_CHECK) 1700 file_magwarn(ms, "syntax error: `elif'"); 1701 return -1; 1702 } 1703 last_cond = COND_ELIF; 1704 break; 1705 1706 case COND_ELSE: 1707 if (last_cond != COND_IF && last_cond != COND_ELIF) { 1708 if (ms->flags & MAGIC_CHECK) 1709 file_magwarn(ms, "syntax error: `else'"); 1710 return -1; 1711 } 1712 last_cond = COND_NONE; 1713 break; 1714 1715 case COND_NONE: 1716 last_cond = COND_NONE; 1717 break; 1718 } 1719 1720 ms->c.li[cont_level].last_cond = last_cond; 1721 return 0; 1722 } 1723 #endif /* ENABLE_CONDITIONALS */ 1724 1725 private int 1726 parse_indirect_modifier(struct magic_set *ms, struct magic *m, const char **lp) 1727 { 1728 const char *l = *lp; 1729 1730 while (!isspace(CAST(unsigned char, *++l))) 1731 switch (*l) { 1732 case CHAR_INDIRECT_RELATIVE: 1733 m->str_flags |= INDIRECT_RELATIVE; 1734 break; 1735 default: 1736 if (ms->flags & MAGIC_CHECK) 1737 file_magwarn(ms, "indirect modifier `%c' " 1738 "invalid", *l); 1739 *lp = l; 1740 return -1; 1741 } 1742 *lp = l; 1743 return 0; 1744 } 1745 1746 private void 1747 parse_op_modifier(struct magic_set *ms, struct magic *m, const char **lp, 1748 int op) 1749 { 1750 const char *l = *lp; 1751 char *t; 1752 uint64_t val; 1753 1754 ++l; 1755 m->mask_op |= op; 1756 val = CAST(uint64_t, strtoull(l, &t, 0)); 1757 l = t; 1758 m->num_mask = file_signextend(ms, m, val); 1759 eatsize(&l); 1760 *lp = l; 1761 } 1762 1763 private int 1764 parse_string_modifier(struct magic_set *ms, struct magic *m, const char **lp) 1765 { 1766 const char *l = *lp; 1767 char *t; 1768 int have_range = 0; 1769 1770 while (!isspace(CAST(unsigned char, *++l))) { 1771 switch (*l) { 1772 case '0': case '1': case '2': 1773 case '3': case '4': case '5': 1774 case '6': case '7': case '8': 1775 case '9': 1776 if (have_range && (ms->flags & MAGIC_CHECK)) 1777 file_magwarn(ms, "multiple ranges"); 1778 have_range = 1; 1779 m->str_range = CAST(uint32_t, strtoul(l, &t, 0)); 1780 if (m->str_range == 0) 1781 file_magwarn(ms, "zero range"); 1782 l = t - 1; 1783 break; 1784 case CHAR_COMPACT_WHITESPACE: 1785 m->str_flags |= STRING_COMPACT_WHITESPACE; 1786 break; 1787 case CHAR_COMPACT_OPTIONAL_WHITESPACE: 1788 m->str_flags |= STRING_COMPACT_OPTIONAL_WHITESPACE; 1789 break; 1790 case CHAR_IGNORE_LOWERCASE: 1791 m->str_flags |= STRING_IGNORE_LOWERCASE; 1792 break; 1793 case CHAR_IGNORE_UPPERCASE: 1794 m->str_flags |= STRING_IGNORE_UPPERCASE; 1795 break; 1796 case CHAR_REGEX_OFFSET_START: 1797 m->str_flags |= REGEX_OFFSET_START; 1798 break; 1799 case CHAR_BINTEST: 1800 m->str_flags |= STRING_BINTEST; 1801 break; 1802 case CHAR_TEXTTEST: 1803 m->str_flags |= STRING_TEXTTEST; 1804 break; 1805 case CHAR_TRIM: 1806 m->str_flags |= STRING_TRIM; 1807 break; 1808 case CHAR_PSTRING_1_LE: 1809 #define SET_LENGTH(a) m->str_flags = (m->str_flags & ~PSTRING_LEN) | (a) 1810 if (m->type != FILE_PSTRING) 1811 goto bad; 1812 SET_LENGTH(PSTRING_1_LE); 1813 break; 1814 case CHAR_PSTRING_2_BE: 1815 if (m->type != FILE_PSTRING) 1816 goto bad; 1817 SET_LENGTH(PSTRING_2_BE); 1818 break; 1819 case CHAR_PSTRING_2_LE: 1820 if (m->type != FILE_PSTRING) 1821 goto bad; 1822 SET_LENGTH(PSTRING_2_LE); 1823 break; 1824 case CHAR_PSTRING_4_BE: 1825 if (m->type != FILE_PSTRING) 1826 goto bad; 1827 SET_LENGTH(PSTRING_4_BE); 1828 break; 1829 case CHAR_PSTRING_4_LE: 1830 switch (m->type) { 1831 case FILE_PSTRING: 1832 case FILE_REGEX: 1833 break; 1834 default: 1835 goto bad; 1836 } 1837 SET_LENGTH(PSTRING_4_LE); 1838 break; 1839 case CHAR_PSTRING_LENGTH_INCLUDES_ITSELF: 1840 if (m->type != FILE_PSTRING) 1841 goto bad; 1842 m->str_flags |= PSTRING_LENGTH_INCLUDES_ITSELF; 1843 break; 1844 default: 1845 bad: 1846 if (ms->flags & MAGIC_CHECK) 1847 file_magwarn(ms, "string modifier `%c' " 1848 "invalid", *l); 1849 goto out; 1850 } 1851 /* allow multiple '/' for readability */ 1852 if (l[1] == '/' && !isspace(CAST(unsigned char, l[2]))) 1853 l++; 1854 } 1855 if (string_modifier_check(ms, m) == -1) 1856 goto out; 1857 *lp = l; 1858 return 0; 1859 out: 1860 *lp = l; 1861 return -1; 1862 } 1863 1864 /* 1865 * parse one line from magic file, put into magic[index++] if valid 1866 */ 1867 private int 1868 parse(struct magic_set *ms, struct magic_entry *me, const char *line, 1869 size_t lineno, int action) 1870 { 1871 #ifdef ENABLE_CONDITIONALS 1872 static uint32_t last_cont_level = 0; 1873 #endif 1874 size_t i; 1875 struct magic *m; 1876 const char *l = line; 1877 char *t; 1878 int op; 1879 uint32_t cont_level; 1880 int32_t diff; 1881 1882 cont_level = 0; 1883 1884 /* 1885 * Parse the offset. 1886 */ 1887 while (*l == '>') { 1888 ++l; /* step over */ 1889 cont_level++; 1890 } 1891 #ifdef ENABLE_CONDITIONALS 1892 if (cont_level == 0 || cont_level > last_cont_level) 1893 if (file_check_mem(ms, cont_level) == -1) 1894 return -1; 1895 last_cont_level = cont_level; 1896 #endif 1897 if (cont_level != 0) { 1898 if (me->mp == NULL) { 1899 file_magerror(ms, "No current entry for continuation"); 1900 return -1; 1901 } 1902 if (me->cont_count == 0) { 1903 file_magerror(ms, "Continuations present with 0 count"); 1904 return -1; 1905 } 1906 m = &me->mp[me->cont_count - 1]; 1907 diff = CAST(int32_t, cont_level) - CAST(int32_t, m->cont_level); 1908 if (diff > 1) 1909 file_magwarn(ms, "New continuation level %u is more " 1910 "than one larger than current level %u", cont_level, 1911 m->cont_level); 1912 if (me->cont_count == me->max_count) { 1913 struct magic *nm; 1914 size_t cnt = me->max_count + ALLOC_CHUNK; 1915 if ((nm = CAST(struct magic *, realloc(me->mp, 1916 sizeof(*nm) * cnt))) == NULL) { 1917 file_oomem(ms, sizeof(*nm) * cnt); 1918 return -1; 1919 } 1920 me->mp = nm; 1921 me->max_count = CAST(uint32_t, cnt); 1922 } 1923 m = &me->mp[me->cont_count++]; 1924 (void)memset(m, 0, sizeof(*m)); 1925 m->cont_level = cont_level; 1926 } else { 1927 static const size_t len = sizeof(*m) * ALLOC_CHUNK; 1928 if (me->mp != NULL) 1929 return 1; 1930 if ((m = CAST(struct magic *, malloc(len))) == NULL) { 1931 file_oomem(ms, len); 1932 return -1; 1933 } 1934 me->mp = m; 1935 me->max_count = ALLOC_CHUNK; 1936 (void)memset(m, 0, sizeof(*m)); 1937 m->factor_op = FILE_FACTOR_OP_NONE; 1938 m->cont_level = 0; 1939 me->cont_count = 1; 1940 } 1941 m->lineno = CAST(uint32_t, lineno); 1942 1943 if (*l == '&') { /* m->cont_level == 0 checked below. */ 1944 ++l; /* step over */ 1945 m->flag |= OFFADD; 1946 } 1947 if (*l == '(') { 1948 ++l; /* step over */ 1949 m->flag |= INDIR; 1950 if (m->flag & OFFADD) 1951 m->flag = (m->flag & ~OFFADD) | INDIROFFADD; 1952 1953 if (*l == '&') { /* m->cont_level == 0 checked below */ 1954 ++l; /* step over */ 1955 m->flag |= OFFADD; 1956 } 1957 } 1958 /* Indirect offsets are not valid at level 0. */ 1959 if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD))) { 1960 if (ms->flags & MAGIC_CHECK) 1961 file_magwarn(ms, "relative offset at level 0"); 1962 return -1; 1963 } 1964 1965 /* get offset, then skip over it */ 1966 if (*l == '-') { 1967 ++l; /* step over */ 1968 m->flag |= OFFNEGATIVE; 1969 } 1970 m->offset = CAST(int32_t, strtol(l, &t, 0)); 1971 if (l == t) { 1972 if (ms->flags & MAGIC_CHECK) 1973 file_magwarn(ms, "offset `%s' invalid", l); 1974 return -1; 1975 } 1976 1977 l = t; 1978 1979 if (m->flag & INDIR) { 1980 m->in_type = FILE_LONG; 1981 m->in_offset = 0; 1982 m->in_op = 0; 1983 /* 1984 * read [.,lbs][+-]nnnnn) 1985 */ 1986 if (*l == '.' || *l == ',') { 1987 if (*l == ',') 1988 m->in_op |= FILE_OPSIGNED; 1989 l++; 1990 switch (*l) { 1991 case 'l': 1992 m->in_type = FILE_LELONG; 1993 break; 1994 case 'L': 1995 m->in_type = FILE_BELONG; 1996 break; 1997 case 'm': 1998 m->in_type = FILE_MELONG; 1999 break; 2000 case 'h': 2001 case 's': 2002 m->in_type = FILE_LESHORT; 2003 break; 2004 case 'H': 2005 case 'S': 2006 m->in_type = FILE_BESHORT; 2007 break; 2008 case 'c': 2009 case 'b': 2010 case 'C': 2011 case 'B': 2012 m->in_type = FILE_BYTE; 2013 break; 2014 case 'e': 2015 case 'f': 2016 case 'g': 2017 m->in_type = FILE_LEDOUBLE; 2018 break; 2019 case 'E': 2020 case 'F': 2021 case 'G': 2022 m->in_type = FILE_BEDOUBLE; 2023 break; 2024 case 'i': 2025 m->in_type = FILE_LEID3; 2026 break; 2027 case 'I': 2028 m->in_type = FILE_BEID3; 2029 break; 2030 case 'q': 2031 m->in_type = FILE_LEQUAD; 2032 break; 2033 case 'Q': 2034 m->in_type = FILE_BEQUAD; 2035 break; 2036 default: 2037 if (ms->flags & MAGIC_CHECK) 2038 file_magwarn(ms, 2039 "indirect offset type `%c' invalid", 2040 *l); 2041 return -1; 2042 } 2043 l++; 2044 } 2045 2046 if (*l == '~') { 2047 m->in_op |= FILE_OPINVERSE; 2048 l++; 2049 } 2050 if ((op = get_op(*l)) != -1) { 2051 m->in_op |= op; 2052 l++; 2053 } 2054 if (*l == '(') { 2055 m->in_op |= FILE_OPINDIRECT; 2056 l++; 2057 } 2058 if (isdigit(CAST(unsigned char, *l)) || *l == '-') { 2059 m->in_offset = CAST(int32_t, strtol(l, &t, 0)); 2060 if (l == t) { 2061 if (ms->flags & MAGIC_CHECK) 2062 file_magwarn(ms, 2063 "in_offset `%s' invalid", l); 2064 return -1; 2065 } 2066 l = t; 2067 } 2068 if (*l++ != ')' || 2069 ((m->in_op & FILE_OPINDIRECT) && *l++ != ')')) { 2070 if (ms->flags & MAGIC_CHECK) 2071 file_magwarn(ms, 2072 "missing ')' in indirect offset"); 2073 return -1; 2074 } 2075 } 2076 EATAB; 2077 2078 #ifdef ENABLE_CONDITIONALS 2079 m->cond = get_cond(l, &l); 2080 if (check_cond(ms, m->cond, cont_level) == -1) 2081 return -1; 2082 2083 EATAB; 2084 #endif 2085 2086 /* 2087 * Parse the type. 2088 */ 2089 if (*l == 'u') { 2090 /* 2091 * Try it as a keyword type prefixed by "u"; match what 2092 * follows the "u". If that fails, try it as an SUS 2093 * integer type. 2094 */ 2095 m->type = get_type(type_tbl, l + 1, &l); 2096 if (m->type == FILE_INVALID) { 2097 /* 2098 * Not a keyword type; parse it as an SUS type, 2099 * 'u' possibly followed by a number or C/S/L. 2100 */ 2101 m->type = get_standard_integer_type(l, &l); 2102 } 2103 /* It's unsigned. */ 2104 if (m->type != FILE_INVALID) 2105 m->flag |= UNSIGNED; 2106 } else { 2107 /* 2108 * Try it as a keyword type. If that fails, try it as 2109 * an SUS integer type if it begins with "d" or as an 2110 * SUS string type if it begins with "s". In any case, 2111 * it's not unsigned. 2112 */ 2113 m->type = get_type(type_tbl, l, &l); 2114 if (m->type == FILE_INVALID) { 2115 /* 2116 * Not a keyword type; parse it as an SUS type, 2117 * either 'd' possibly followed by a number or 2118 * C/S/L, or just 's'. 2119 */ 2120 if (*l == 'd') 2121 m->type = get_standard_integer_type(l, &l); 2122 else if (*l == 's' 2123 && !isalpha(CAST(unsigned char, l[1]))) { 2124 m->type = FILE_STRING; 2125 ++l; 2126 } 2127 } 2128 } 2129 2130 if (m->type == FILE_INVALID) { 2131 /* Not found - try it as a special keyword. */ 2132 m->type = get_type(special_tbl, l, &l); 2133 } 2134 2135 if (m->type == FILE_INVALID) { 2136 if (ms->flags & MAGIC_CHECK) 2137 file_magwarn(ms, "type `%s' invalid", l); 2138 return -1; 2139 } 2140 2141 /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */ 2142 /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */ 2143 2144 m->mask_op = 0; 2145 if (*l == '~') { 2146 if (!IS_STRING(m->type)) 2147 m->mask_op |= FILE_OPINVERSE; 2148 else if (ms->flags & MAGIC_CHECK) 2149 file_magwarn(ms, "'~' invalid for string types"); 2150 ++l; 2151 } 2152 m->str_range = 0; 2153 m->str_flags = m->type == FILE_PSTRING ? PSTRING_1_LE : 0; 2154 if ((op = get_op(*l)) != -1) { 2155 if (IS_STRING(m->type)) { 2156 int r; 2157 2158 if (op != FILE_OPDIVIDE) { 2159 if (ms->flags & MAGIC_CHECK) 2160 file_magwarn(ms, 2161 "invalid string/indirect op: " 2162 "`%c'", *t); 2163 return -1; 2164 } 2165 2166 if (m->type == FILE_INDIRECT) 2167 r = parse_indirect_modifier(ms, m, &l); 2168 else 2169 r = parse_string_modifier(ms, m, &l); 2170 if (r == -1) 2171 return -1; 2172 } else 2173 parse_op_modifier(ms, m, &l, op); 2174 } 2175 2176 /* 2177 * We used to set mask to all 1's here, instead let's just not do 2178 * anything if mask = 0 (unless you have a better idea) 2179 */ 2180 EATAB; 2181 2182 switch (*l) { 2183 case '>': 2184 case '<': 2185 m->reln = *l; 2186 ++l; 2187 if (*l == '=') { 2188 if (ms->flags & MAGIC_CHECK) { 2189 file_magwarn(ms, "%c= not supported", 2190 m->reln); 2191 return -1; 2192 } 2193 ++l; 2194 } 2195 break; 2196 /* Old-style anding: "0 byte &0x80 dynamically linked" */ 2197 case '&': 2198 case '^': 2199 case '=': 2200 m->reln = *l; 2201 ++l; 2202 if (*l == '=') { 2203 /* HP compat: ignore &= etc. */ 2204 ++l; 2205 } 2206 break; 2207 case '!': 2208 m->reln = *l; 2209 ++l; 2210 break; 2211 default: 2212 m->reln = '='; /* the default relation */ 2213 if (*l == 'x' && ((isascii(CAST(unsigned char, l[1])) && 2214 isspace(CAST(unsigned char, l[1]))) || !l[1])) { 2215 m->reln = *l; 2216 ++l; 2217 } 2218 break; 2219 } 2220 /* 2221 * Grab the value part, except for an 'x' reln. 2222 */ 2223 if (m->reln != 'x' && getvalue(ms, m, &l, action)) 2224 return -1; 2225 2226 /* 2227 * TODO finish this macro and start using it! 2228 * #define offsetcheck {if (offset > ms->bytes_max -1) 2229 * magwarn("offset too big"); } 2230 */ 2231 2232 /* 2233 * Now get last part - the description 2234 */ 2235 EATAB; 2236 if (l[0] == '\b') { 2237 ++l; 2238 m->flag |= NOSPACE; 2239 } else if ((l[0] == '\\') && (l[1] == 'b')) { 2240 ++l; 2241 ++l; 2242 m->flag |= NOSPACE; 2243 } 2244 for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); ) 2245 continue; 2246 if (i == sizeof(m->desc)) { 2247 m->desc[sizeof(m->desc) - 1] = '\0'; 2248 if (ms->flags & MAGIC_CHECK) 2249 file_magwarn(ms, "description `%s' truncated", m->desc); 2250 } 2251 2252 /* 2253 * We only do this check while compiling, or if any of the magic 2254 * files were not compiled. 2255 */ 2256 if (ms->flags & MAGIC_CHECK) { 2257 if (check_format(ms, m) == -1) 2258 return -1; 2259 } 2260 #ifndef COMPILE_ONLY 2261 if (action == FILE_CHECK) { 2262 file_mdump(m); 2263 } 2264 #endif 2265 m->mimetype[0] = '\0'; /* initialise MIME type to none */ 2266 return 0; 2267 } 2268 2269 /* 2270 * parse a STRENGTH annotation line from magic file, put into magic[index - 1] 2271 * if valid 2272 */ 2273 private int 2274 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line, 2275 size_t len __attribute__((__unused__))) 2276 { 2277 const char *l = line; 2278 char *el; 2279 unsigned long factor; 2280 struct magic *m = &me->mp[0]; 2281 2282 if (m->factor_op != FILE_FACTOR_OP_NONE) { 2283 file_magwarn(ms, 2284 "Current entry already has a strength type: %c %d", 2285 m->factor_op, m->factor); 2286 return -1; 2287 } 2288 if (m->type == FILE_NAME) { 2289 file_magwarn(ms, "%s: Strength setting is not supported in " 2290 "\"name\" magic entries", m->value.s); 2291 return -1; 2292 } 2293 EATAB; 2294 switch (*l) { 2295 case FILE_FACTOR_OP_NONE: 2296 case FILE_FACTOR_OP_PLUS: 2297 case FILE_FACTOR_OP_MINUS: 2298 case FILE_FACTOR_OP_TIMES: 2299 case FILE_FACTOR_OP_DIV: 2300 m->factor_op = *l++; 2301 break; 2302 default: 2303 file_magwarn(ms, "Unknown factor op `%c'", *l); 2304 return -1; 2305 } 2306 EATAB; 2307 factor = strtoul(l, &el, 0); 2308 if (factor > 255) { 2309 file_magwarn(ms, "Too large factor `%lu'", factor); 2310 goto out; 2311 } 2312 if (*el && !isspace(CAST(unsigned char, *el))) { 2313 file_magwarn(ms, "Bad factor `%s'", l); 2314 goto out; 2315 } 2316 m->factor = CAST(uint8_t, factor); 2317 if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) { 2318 file_magwarn(ms, "Cannot have factor op `%c' and factor %u", 2319 m->factor_op, m->factor); 2320 goto out; 2321 } 2322 return 0; 2323 out: 2324 m->factor_op = FILE_FACTOR_OP_NONE; 2325 m->factor = 0; 2326 return -1; 2327 } 2328 2329 private int 2330 goodchar(unsigned char x, const char *extra) 2331 { 2332 return (isascii(x) && isalnum(x)) || strchr(extra, x); 2333 } 2334 2335 private int 2336 parse_extra(struct magic_set *ms, struct magic_entry *me, const char *line, 2337 size_t llen, off_t off, size_t len, const char *name, const char *extra, 2338 int nt) 2339 { 2340 size_t i; 2341 const char *l = line; 2342 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1]; 2343 char *buf = CAST(char *, CAST(void *, m)) + off; 2344 2345 if (buf[0] != '\0') { 2346 len = nt ? strlen(buf) : len; 2347 file_magwarn(ms, "Current entry already has a %s type " 2348 "`%.*s', new type `%s'", name, CAST(int, len), buf, l); 2349 return -1; 2350 } 2351 2352 if (*m->desc == '\0') { 2353 file_magwarn(ms, "Current entry does not yet have a " 2354 "description for adding a %s type", name); 2355 return -1; 2356 } 2357 2358 EATAB; 2359 for (i = 0; *l && i < llen && i < len && goodchar(*l, extra); 2360 buf[i++] = *l++) 2361 continue; 2362 2363 if (i == len && *l) { 2364 if (nt) 2365 buf[len - 1] = '\0'; 2366 if (ms->flags & MAGIC_CHECK) 2367 file_magwarn(ms, "%s type `%s' truncated %" 2368 SIZE_T_FORMAT "u", name, line, i); 2369 } else { 2370 if (!isspace(CAST(unsigned char, *l)) && !goodchar(*l, extra)) 2371 file_magwarn(ms, "%s type `%s' has bad char '%c'", 2372 name, line, *l); 2373 if (nt) 2374 buf[i] = '\0'; 2375 } 2376 2377 if (i > 0) 2378 return 0; 2379 2380 file_magerror(ms, "Bad magic entry '%s'", line); 2381 return -1; 2382 } 2383 2384 /* 2385 * Parse an Apple CREATOR/TYPE annotation from magic file and put it into 2386 * magic[index - 1] 2387 */ 2388 private int 2389 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line, 2390 size_t len) 2391 { 2392 struct magic *m = &me->mp[0]; 2393 2394 return parse_extra(ms, me, line, len, 2395 CAST(off_t, offsetof(struct magic, apple)), 2396 sizeof(m->apple), "APPLE", "!+-./?", 0); 2397 } 2398 2399 /* 2400 * Parse a comma-separated list of extensions 2401 */ 2402 private int 2403 parse_ext(struct magic_set *ms, struct magic_entry *me, const char *line, 2404 size_t len) 2405 { 2406 struct magic *m = &me->mp[0]; 2407 2408 return parse_extra(ms, me, line, len, 2409 CAST(off_t, offsetof(struct magic, ext)), 2410 sizeof(m->ext), "EXTENSION", ",!+-/@?_$", 0); 2411 } 2412 2413 /* 2414 * parse a MIME annotation line from magic file, put into magic[index - 1] 2415 * if valid 2416 */ 2417 private int 2418 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line, 2419 size_t len) 2420 { 2421 struct magic *m = &me->mp[0]; 2422 2423 return parse_extra(ms, me, line, len, 2424 CAST(off_t, offsetof(struct magic, mimetype)), 2425 sizeof(m->mimetype), "MIME", "+-/.$?:{}", 1); 2426 } 2427 2428 private int 2429 check_format_type(const char *ptr, int type, const char **estr) 2430 { 2431 int quad = 0, h; 2432 size_t len, cnt; 2433 if (*ptr == '\0') { 2434 /* Missing format string; bad */ 2435 *estr = "missing format spec"; 2436 return -1; 2437 } 2438 2439 switch (file_formats[type]) { 2440 case FILE_FMT_QUAD: 2441 quad = 1; 2442 /*FALLTHROUGH*/ 2443 case FILE_FMT_NUM: 2444 if (quad == 0) { 2445 switch (type) { 2446 case FILE_BYTE: 2447 h = 2; 2448 break; 2449 case FILE_SHORT: 2450 case FILE_BESHORT: 2451 case FILE_LESHORT: 2452 h = 1; 2453 break; 2454 case FILE_LONG: 2455 case FILE_BELONG: 2456 case FILE_LELONG: 2457 case FILE_MELONG: 2458 case FILE_LEID3: 2459 case FILE_BEID3: 2460 case FILE_INDIRECT: 2461 h = 0; 2462 break; 2463 default: 2464 abort(); 2465 } 2466 } else 2467 h = 0; 2468 if (*ptr == '-') 2469 ptr++; 2470 if (*ptr == '.') 2471 ptr++; 2472 if (*ptr == '#') 2473 ptr++; 2474 #define CHECKLEN() do { \ 2475 for (len = cnt = 0; isdigit(CAST(unsigned char, *ptr)); ptr++, cnt++) \ 2476 len = len * 10 + (*ptr - '0'); \ 2477 if (cnt > 5 || len > 1024) \ 2478 goto toolong; \ 2479 } while (/*CONSTCOND*/0) 2480 2481 CHECKLEN(); 2482 if (*ptr == '.') 2483 ptr++; 2484 CHECKLEN(); 2485 if (quad) { 2486 if (*ptr++ != 'l') 2487 goto invalid; 2488 if (*ptr++ != 'l') 2489 goto invalid; 2490 } 2491 2492 switch (*ptr++) { 2493 #ifdef STRICT_FORMAT /* "long" formats are int formats for us */ 2494 /* so don't accept the 'l' modifier */ 2495 case 'l': 2496 switch (*ptr++) { 2497 case 'i': 2498 case 'd': 2499 case 'u': 2500 case 'o': 2501 case 'x': 2502 case 'X': 2503 if (h == 0) 2504 return 0; 2505 /*FALLTHROUGH*/ 2506 default: 2507 goto invalid; 2508 } 2509 2510 /* 2511 * Don't accept h and hh modifiers. They make writing 2512 * magic entries more complicated, for very little benefit 2513 */ 2514 case 'h': 2515 if (h-- <= 0) 2516 goto invalid; 2517 switch (*ptr++) { 2518 case 'h': 2519 if (h-- <= 0) 2520 goto invalid; 2521 switch (*ptr++) { 2522 case 'i': 2523 case 'd': 2524 case 'u': 2525 case 'o': 2526 case 'x': 2527 case 'X': 2528 return 0; 2529 default: 2530 goto invalid; 2531 } 2532 case 'i': 2533 case 'd': 2534 case 'u': 2535 case 'o': 2536 case 'x': 2537 case 'X': 2538 if (h == 0) 2539 return 0; 2540 /*FALLTHROUGH*/ 2541 default: 2542 goto invalid; 2543 } 2544 #endif 2545 case 'c': 2546 if (h == 2) 2547 return 0; 2548 goto invalid; 2549 case 'i': 2550 case 'd': 2551 case 'u': 2552 case 'o': 2553 case 'x': 2554 case 'X': 2555 #ifdef STRICT_FORMAT 2556 if (h == 0) 2557 return 0; 2558 /*FALLTHROUGH*/ 2559 #else 2560 return 0; 2561 #endif 2562 default: 2563 goto invalid; 2564 } 2565 2566 case FILE_FMT_FLOAT: 2567 case FILE_FMT_DOUBLE: 2568 if (*ptr == '-') 2569 ptr++; 2570 if (*ptr == '.') 2571 ptr++; 2572 CHECKLEN(); 2573 if (*ptr == '.') 2574 ptr++; 2575 CHECKLEN(); 2576 switch (*ptr++) { 2577 case 'e': 2578 case 'E': 2579 case 'f': 2580 case 'F': 2581 case 'g': 2582 case 'G': 2583 return 0; 2584 2585 default: 2586 goto invalid; 2587 } 2588 2589 2590 case FILE_FMT_STR: 2591 if (*ptr == '-') 2592 ptr++; 2593 while (isdigit(CAST(unsigned char, *ptr))) 2594 ptr++; 2595 if (*ptr == '.') { 2596 ptr++; 2597 while (isdigit(CAST(unsigned char , *ptr))) 2598 ptr++; 2599 } 2600 2601 switch (*ptr++) { 2602 case 's': 2603 return 0; 2604 default: 2605 goto invalid; 2606 } 2607 2608 default: 2609 /* internal error */ 2610 abort(); 2611 } 2612 invalid: 2613 *estr = "not valid"; 2614 toolong: 2615 *estr = "too long"; 2616 return -1; 2617 } 2618 2619 /* 2620 * Check that the optional printf format in description matches 2621 * the type of the magic. 2622 */ 2623 private int 2624 check_format(struct magic_set *ms, struct magic *m) 2625 { 2626 char *ptr; 2627 const char *estr; 2628 2629 for (ptr = m->desc; *ptr; ptr++) 2630 if (*ptr == '%') 2631 break; 2632 if (*ptr == '\0') { 2633 /* No format string; ok */ 2634 return 1; 2635 } 2636 2637 assert(file_nformats == file_nnames); 2638 2639 if (m->type >= file_nformats) { 2640 file_magwarn(ms, "Internal error inconsistency between " 2641 "m->type and format strings"); 2642 return -1; 2643 } 2644 if (file_formats[m->type] == FILE_FMT_NONE) { 2645 file_magwarn(ms, "No format string for `%s' with description " 2646 "`%s'", m->desc, file_names[m->type]); 2647 return -1; 2648 } 2649 2650 ptr++; 2651 if (check_format_type(ptr, m->type, &estr) == -1) { 2652 /* 2653 * TODO: this error message is unhelpful if the format 2654 * string is not one character long 2655 */ 2656 file_magwarn(ms, "Printf format is %s for type " 2657 "`%s' in description `%s'", estr, 2658 file_names[m->type], m->desc); 2659 return -1; 2660 } 2661 2662 for (; *ptr; ptr++) { 2663 if (*ptr == '%') { 2664 file_magwarn(ms, 2665 "Too many format strings (should have at most one) " 2666 "for `%s' with description `%s'", 2667 file_names[m->type], m->desc); 2668 return -1; 2669 } 2670 } 2671 return 0; 2672 } 2673 2674 /* 2675 * Read a numeric value from a pointer, into the value union of a magic 2676 * pointer, according to the magic type. Update the string pointer to point 2677 * just after the number read. Return 0 for success, non-zero for failure. 2678 */ 2679 private int 2680 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action) 2681 { 2682 char *ep; 2683 uint64_t ull; 2684 2685 switch (m->type) { 2686 case FILE_BESTRING16: 2687 case FILE_LESTRING16: 2688 case FILE_STRING: 2689 case FILE_PSTRING: 2690 case FILE_REGEX: 2691 case FILE_SEARCH: 2692 case FILE_NAME: 2693 case FILE_USE: 2694 case FILE_DER: 2695 *p = getstr(ms, m, *p, action == FILE_COMPILE); 2696 if (*p == NULL) { 2697 if (ms->flags & MAGIC_CHECK) 2698 file_magwarn(ms, "cannot get string from `%s'", 2699 m->value.s); 2700 return -1; 2701 } 2702 if (m->type == FILE_REGEX) { 2703 file_regex_t rx; 2704 int rc = file_regcomp(&rx, m->value.s, REG_EXTENDED); 2705 if (rc) { 2706 if (ms->flags & MAGIC_CHECK) 2707 file_regerror(&rx, rc, ms); 2708 } 2709 file_regfree(&rx); 2710 return rc ? -1 : 0; 2711 } 2712 return 0; 2713 default: 2714 if (m->reln == 'x') 2715 return 0; 2716 break; 2717 } 2718 2719 switch (m->type) { 2720 case FILE_FLOAT: 2721 case FILE_BEFLOAT: 2722 case FILE_LEFLOAT: 2723 errno = 0; 2724 #ifdef HAVE_STRTOF 2725 m->value.f = strtof(*p, &ep); 2726 #else 2727 m->value.f = (float)strtod(*p, &ep); 2728 #endif 2729 if (errno == 0) 2730 *p = ep; 2731 return 0; 2732 case FILE_DOUBLE: 2733 case FILE_BEDOUBLE: 2734 case FILE_LEDOUBLE: 2735 errno = 0; 2736 m->value.d = strtod(*p, &ep); 2737 if (errno == 0) 2738 *p = ep; 2739 return 0; 2740 case FILE_GUID: 2741 if (file_parse_guid(*p, m->value.guid) == -1) 2742 return -1; 2743 *p += FILE_GUID_SIZE - 1; 2744 return 0; 2745 default: 2746 errno = 0; 2747 ull = CAST(uint64_t, strtoull(*p, &ep, 0)); 2748 m->value.q = file_signextend(ms, m, ull); 2749 if (*p == ep) { 2750 file_magwarn(ms, "Unparseable number `%s'", *p); 2751 } else { 2752 size_t ts = typesize(m->type); 2753 uint64_t x; 2754 const char *q; 2755 2756 if (ts == FILE_BADSIZE) { 2757 file_magwarn(ms, 2758 "Expected numeric type got `%s'", 2759 type_tbl[m->type].name); 2760 } 2761 for (q = *p; isspace(CAST(unsigned char, *q)); q++) 2762 continue; 2763 if (*q == '-') 2764 ull = -CAST(int64_t, ull); 2765 switch (ts) { 2766 case 1: 2767 x = CAST(uint64_t, ull & ~0xffULL); 2768 break; 2769 case 2: 2770 x = CAST(uint64_t, ull & ~0xffffULL); 2771 break; 2772 case 4: 2773 x = CAST(uint64_t, ull & ~0xffffffffULL); 2774 break; 2775 case 8: 2776 x = 0; 2777 break; 2778 default: 2779 abort(); 2780 } 2781 if (x) { 2782 file_magwarn(ms, "Overflow for numeric" 2783 " type `%s' value %#" PRIx64, 2784 type_tbl[m->type].name, ull); 2785 } 2786 } 2787 if (errno == 0) { 2788 *p = ep; 2789 eatsize(p); 2790 } 2791 return 0; 2792 } 2793 } 2794 2795 /* 2796 * Convert a string containing C character escapes. Stop at an unescaped 2797 * space or tab. 2798 * Copy the converted version to "m->value.s", and the length in m->vallen. 2799 * Return updated scan pointer as function result. Warn if set. 2800 */ 2801 private const char * 2802 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn) 2803 { 2804 const char *origs = s; 2805 char *p = m->value.s; 2806 size_t plen = sizeof(m->value.s); 2807 char *origp = p; 2808 char *pmax = p + plen - 1; 2809 int c; 2810 int val; 2811 2812 while ((c = *s++) != '\0') { 2813 if (isspace(CAST(unsigned char, c))) 2814 break; 2815 if (p >= pmax) { 2816 file_error(ms, 0, "string too long: `%s'", origs); 2817 return NULL; 2818 } 2819 if (c == '\\') { 2820 switch(c = *s++) { 2821 2822 case '\0': 2823 if (warn) 2824 file_magwarn(ms, "incomplete escape"); 2825 s--; 2826 goto out; 2827 2828 case '\t': 2829 if (warn) { 2830 file_magwarn(ms, 2831 "escaped tab found, use \\t instead"); 2832 warn = 0; /* already did */ 2833 } 2834 /*FALLTHROUGH*/ 2835 default: 2836 if (warn) { 2837 if (isprint(CAST(unsigned char, c))) { 2838 /* Allow escaping of 2839 * ``relations'' */ 2840 if (strchr("<>&^=!", c) == NULL 2841 && (m->type != FILE_REGEX || 2842 strchr("[]().*?^$|{}", c) 2843 == NULL)) { 2844 file_magwarn(ms, "no " 2845 "need to escape " 2846 "`%c'", c); 2847 } 2848 } else { 2849 file_magwarn(ms, 2850 "unknown escape sequence: " 2851 "\\%03o", c); 2852 } 2853 } 2854 /*FALLTHROUGH*/ 2855 /* space, perhaps force people to use \040? */ 2856 case ' ': 2857 #if 0 2858 /* 2859 * Other things people escape, but shouldn't need to, 2860 * so we disallow them 2861 */ 2862 case '\'': 2863 case '"': 2864 case '?': 2865 #endif 2866 /* Relations */ 2867 case '>': 2868 case '<': 2869 case '&': 2870 case '^': 2871 case '=': 2872 case '!': 2873 /* and baskslash itself */ 2874 case '\\': 2875 *p++ = CAST(char, c); 2876 break; 2877 2878 case 'a': 2879 *p++ = '\a'; 2880 break; 2881 2882 case 'b': 2883 *p++ = '\b'; 2884 break; 2885 2886 case 'f': 2887 *p++ = '\f'; 2888 break; 2889 2890 case 'n': 2891 *p++ = '\n'; 2892 break; 2893 2894 case 'r': 2895 *p++ = '\r'; 2896 break; 2897 2898 case 't': 2899 *p++ = '\t'; 2900 break; 2901 2902 case 'v': 2903 *p++ = '\v'; 2904 break; 2905 2906 /* \ and up to 3 octal digits */ 2907 case '0': 2908 case '1': 2909 case '2': 2910 case '3': 2911 case '4': 2912 case '5': 2913 case '6': 2914 case '7': 2915 val = c - '0'; 2916 c = *s++; /* try for 2 */ 2917 if (c >= '0' && c <= '7') { 2918 val = (val << 3) | (c - '0'); 2919 c = *s++; /* try for 3 */ 2920 if (c >= '0' && c <= '7') 2921 val = (val << 3) | (c-'0'); 2922 else 2923 --s; 2924 } 2925 else 2926 --s; 2927 *p++ = CAST(char, val); 2928 break; 2929 2930 /* \x and up to 2 hex digits */ 2931 case 'x': 2932 val = 'x'; /* Default if no digits */ 2933 c = hextoint(*s++); /* Get next char */ 2934 if (c >= 0) { 2935 val = c; 2936 c = hextoint(*s++); 2937 if (c >= 0) 2938 val = (val << 4) + c; 2939 else 2940 --s; 2941 } else 2942 --s; 2943 *p++ = CAST(char, val); 2944 break; 2945 } 2946 } else 2947 *p++ = CAST(char, c); 2948 } 2949 --s; 2950 out: 2951 *p = '\0'; 2952 m->vallen = CAST(unsigned char, (p - origp)); 2953 if (m->type == FILE_PSTRING) { 2954 size_t l = file_pstring_length_size(ms, m); 2955 if (l == FILE_BADSIZE) 2956 return NULL; 2957 m->vallen += CAST(unsigned char, l); 2958 } 2959 return s; 2960 } 2961 2962 2963 /* Single hex char to int; -1 if not a hex char. */ 2964 private int 2965 hextoint(int c) 2966 { 2967 if (!isascii(CAST(unsigned char, c))) 2968 return -1; 2969 if (isdigit(CAST(unsigned char, c))) 2970 return c - '0'; 2971 if ((c >= 'a') && (c <= 'f')) 2972 return c + 10 - 'a'; 2973 if (( c>= 'A') && (c <= 'F')) 2974 return c + 10 - 'A'; 2975 return -1; 2976 } 2977 2978 2979 /* 2980 * Print a string containing C character escapes. 2981 */ 2982 protected void 2983 file_showstr(FILE *fp, const char *s, size_t len) 2984 { 2985 char c; 2986 2987 for (;;) { 2988 if (len == FILE_BADSIZE) { 2989 c = *s++; 2990 if (c == '\0') 2991 break; 2992 } 2993 else { 2994 if (len-- == 0) 2995 break; 2996 c = *s++; 2997 } 2998 if (c >= 040 && c <= 0176) /* TODO isprint && !iscntrl */ 2999 (void) fputc(c, fp); 3000 else { 3001 (void) fputc('\\', fp); 3002 switch (c) { 3003 case '\a': 3004 (void) fputc('a', fp); 3005 break; 3006 3007 case '\b': 3008 (void) fputc('b', fp); 3009 break; 3010 3011 case '\f': 3012 (void) fputc('f', fp); 3013 break; 3014 3015 case '\n': 3016 (void) fputc('n', fp); 3017 break; 3018 3019 case '\r': 3020 (void) fputc('r', fp); 3021 break; 3022 3023 case '\t': 3024 (void) fputc('t', fp); 3025 break; 3026 3027 case '\v': 3028 (void) fputc('v', fp); 3029 break; 3030 3031 default: 3032 (void) fprintf(fp, "%.3o", c & 0377); 3033 break; 3034 } 3035 } 3036 } 3037 } 3038 3039 /* 3040 * eatsize(): Eat the size spec from a number [eg. 10UL] 3041 */ 3042 private void 3043 eatsize(const char **p) 3044 { 3045 const char *l = *p; 3046 3047 if (LOWCASE(*l) == 'u') 3048 l++; 3049 3050 switch (LOWCASE(*l)) { 3051 case 'l': /* long */ 3052 case 's': /* short */ 3053 case 'h': /* short */ 3054 case 'b': /* char/byte */ 3055 case 'c': /* char/byte */ 3056 l++; 3057 /*FALLTHROUGH*/ 3058 default: 3059 break; 3060 } 3061 3062 *p = l; 3063 } 3064 3065 /* 3066 * handle a buffer containing a compiled file. 3067 */ 3068 private struct magic_map * 3069 apprentice_buf(struct magic_set *ms, struct magic *buf, size_t len) 3070 { 3071 struct magic_map *map; 3072 3073 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) { 3074 file_oomem(ms, sizeof(*map)); 3075 return NULL; 3076 } 3077 map->len = len; 3078 map->p = buf; 3079 map->type = MAP_TYPE_USER; 3080 if (check_buffer(ms, map, "buffer") != 0) { 3081 apprentice_unmap(map); 3082 return NULL; 3083 } 3084 return map; 3085 } 3086 3087 /* 3088 * handle a compiled file. 3089 */ 3090 3091 private struct magic_map * 3092 apprentice_map(struct magic_set *ms, const char *fn) 3093 { 3094 int fd; 3095 struct stat st; 3096 char *dbname = NULL; 3097 struct magic_map *map; 3098 struct magic_map *rv = NULL; 3099 3100 fd = -1; 3101 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) { 3102 file_oomem(ms, sizeof(*map)); 3103 goto error; 3104 } 3105 map->type = MAP_TYPE_USER; /* unspecified */ 3106 3107 dbname = mkdbname(ms, fn, 0); 3108 if (dbname == NULL) 3109 goto error; 3110 3111 if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1) 3112 goto error; 3113 3114 if (fstat(fd, &st) == -1) { 3115 file_error(ms, errno, "cannot stat `%s'", dbname); 3116 goto error; 3117 } 3118 if (st.st_size < 8 || st.st_size > maxoff_t()) { 3119 file_error(ms, 0, "file `%s' is too %s", dbname, 3120 st.st_size < 8 ? "small" : "large"); 3121 goto error; 3122 } 3123 3124 map->len = CAST(size_t, st.st_size); 3125 #ifdef QUICK 3126 map->type = MAP_TYPE_MMAP; 3127 if ((map->p = mmap(0, CAST(size_t, st.st_size), PROT_READ|PROT_WRITE, 3128 MAP_PRIVATE|MAP_FILE, fd, CAST(off_t, 0))) == MAP_FAILED) { 3129 file_error(ms, errno, "cannot map `%s'", dbname); 3130 goto error; 3131 } 3132 #else 3133 map->type = MAP_TYPE_MALLOC; 3134 if ((map->p = CAST(void *, malloc(map->len))) == NULL) { 3135 file_oomem(ms, map->len); 3136 goto error; 3137 } 3138 if (read(fd, map->p, map->len) != (ssize_t)map->len) { 3139 file_badread(ms); 3140 goto error; 3141 } 3142 #endif 3143 (void)close(fd); 3144 fd = -1; 3145 3146 if (check_buffer(ms, map, dbname) != 0) { 3147 goto error; 3148 } 3149 #ifdef QUICK 3150 if (mprotect(map->p, CAST(size_t, st.st_size), PROT_READ) == -1) { 3151 file_error(ms, errno, "cannot mprotect `%s'", dbname); 3152 goto error; 3153 } 3154 #endif 3155 3156 free(dbname); 3157 return map; 3158 3159 error: 3160 if (fd != -1) 3161 (void)close(fd); 3162 apprentice_unmap(map); 3163 free(dbname); 3164 return rv; 3165 } 3166 3167 private int 3168 check_buffer(struct magic_set *ms, struct magic_map *map, const char *dbname) 3169 { 3170 uint32_t *ptr; 3171 uint32_t entries, nentries; 3172 uint32_t version; 3173 int i, needsbyteswap; 3174 3175 ptr = CAST(uint32_t *, map->p); 3176 if (*ptr != MAGICNO) { 3177 if (swap4(*ptr) != MAGICNO) { 3178 file_error(ms, 0, "bad magic in `%s'", dbname); 3179 return -1; 3180 } 3181 needsbyteswap = 1; 3182 } else 3183 needsbyteswap = 0; 3184 if (needsbyteswap) 3185 version = swap4(ptr[1]); 3186 else 3187 version = ptr[1]; 3188 if (version != VERSIONNO) { 3189 file_error(ms, 0, "File %s supports only version %d magic " 3190 "files. `%s' is version %d", VERSION, 3191 VERSIONNO, dbname, version); 3192 return -1; 3193 } 3194 entries = CAST(uint32_t, map->len / sizeof(struct magic)); 3195 if ((entries * sizeof(struct magic)) != map->len) { 3196 file_error(ms, 0, "Size of `%s' %" SIZE_T_FORMAT "u is not " 3197 "a multiple of %" SIZE_T_FORMAT "u", 3198 dbname, map->len, sizeof(struct magic)); 3199 return -1; 3200 } 3201 map->magic[0] = CAST(struct magic *, map->p) + 1; 3202 nentries = 0; 3203 for (i = 0; i < MAGIC_SETS; i++) { 3204 if (needsbyteswap) 3205 map->nmagic[i] = swap4(ptr[i + 2]); 3206 else 3207 map->nmagic[i] = ptr[i + 2]; 3208 if (i != MAGIC_SETS - 1) 3209 map->magic[i + 1] = map->magic[i] + map->nmagic[i]; 3210 nentries += map->nmagic[i]; 3211 } 3212 if (entries != nentries + 1) { 3213 file_error(ms, 0, "Inconsistent entries in `%s' %u != %u", 3214 dbname, entries, nentries + 1); 3215 return -1; 3216 } 3217 if (needsbyteswap) 3218 for (i = 0; i < MAGIC_SETS; i++) 3219 byteswap(map->magic[i], map->nmagic[i]); 3220 return 0; 3221 } 3222 3223 /* 3224 * handle an mmaped file. 3225 */ 3226 private int 3227 apprentice_compile(struct magic_set *ms, struct magic_map *map, const char *fn) 3228 { 3229 static const size_t nm = sizeof(*map->nmagic) * MAGIC_SETS; 3230 static const size_t m = sizeof(**map->magic); 3231 int fd = -1; 3232 size_t len; 3233 char *dbname; 3234 int rv = -1; 3235 uint32_t i; 3236 union { 3237 struct magic m; 3238 uint32_t h[2 + MAGIC_SETS]; 3239 } hdr; 3240 3241 dbname = mkdbname(ms, fn, 1); 3242 3243 if (dbname == NULL) 3244 goto out; 3245 3246 if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) 3247 { 3248 file_error(ms, errno, "cannot open `%s'", dbname); 3249 goto out; 3250 } 3251 memset(&hdr, 0, sizeof(hdr)); 3252 hdr.h[0] = MAGICNO; 3253 hdr.h[1] = VERSIONNO; 3254 memcpy(hdr.h + 2, map->nmagic, nm); 3255 3256 if (write(fd, &hdr, sizeof(hdr)) != CAST(ssize_t, sizeof(hdr))) { 3257 file_error(ms, errno, "error writing `%s'", dbname); 3258 goto out2; 3259 } 3260 3261 for (i = 0; i < MAGIC_SETS; i++) { 3262 len = m * map->nmagic[i]; 3263 if (write(fd, map->magic[i], len) != CAST(ssize_t, len)) { 3264 file_error(ms, errno, "error writing `%s'", dbname); 3265 goto out2; 3266 } 3267 } 3268 3269 rv = 0; 3270 out2: 3271 if (fd != -1) 3272 (void)close(fd); 3273 out: 3274 apprentice_unmap(map); 3275 free(dbname); 3276 return rv; 3277 } 3278 3279 private const char ext[] = ".mgc"; 3280 /* 3281 * make a dbname 3282 */ 3283 private char * 3284 mkdbname(struct magic_set *ms, const char *fn, int strip) 3285 { 3286 const char *p, *q; 3287 char *buf; 3288 3289 if (strip) { 3290 if ((p = strrchr(fn, '/')) != NULL) 3291 fn = ++p; 3292 } 3293 3294 for (q = fn; *q; q++) 3295 continue; 3296 /* Look for .mgc */ 3297 for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--) 3298 if (*p != *q) 3299 break; 3300 3301 /* Did not find .mgc, restore q */ 3302 if (p >= ext) 3303 while (*q) 3304 q++; 3305 3306 q++; 3307 /* Compatibility with old code that looked in .mime */ 3308 if (ms->flags & MAGIC_MIME) { 3309 if (asprintf(&buf, "%.*s.mime%s", CAST(int, q - fn), fn, ext) 3310 < 0) 3311 return NULL; 3312 if (access(buf, R_OK) != -1) { 3313 ms->flags &= MAGIC_MIME_TYPE; 3314 return buf; 3315 } 3316 free(buf); 3317 } 3318 if (asprintf(&buf, "%.*s%s", CAST(int, q - fn), fn, ext) < 0) 3319 return NULL; 3320 3321 /* Compatibility with old code that looked in .mime */ 3322 if (strstr(fn, ".mime") != NULL) 3323 ms->flags &= MAGIC_MIME_TYPE; 3324 return buf; 3325 } 3326 3327 /* 3328 * Byteswap an mmap'ed file if needed 3329 */ 3330 private void 3331 byteswap(struct magic *magic, uint32_t nmagic) 3332 { 3333 uint32_t i; 3334 for (i = 0; i < nmagic; i++) 3335 bs1(&magic[i]); 3336 } 3337 3338 /* 3339 * swap a short 3340 */ 3341 private uint16_t 3342 swap2(uint16_t sv) 3343 { 3344 uint16_t rv; 3345 uint8_t *s = RCAST(uint8_t *, RCAST(void *, &sv)); 3346 uint8_t *d = RCAST(uint8_t *, RCAST(void *, &rv)); 3347 d[0] = s[1]; 3348 d[1] = s[0]; 3349 return rv; 3350 } 3351 3352 /* 3353 * swap an int 3354 */ 3355 private uint32_t 3356 swap4(uint32_t sv) 3357 { 3358 uint32_t rv; 3359 uint8_t *s = RCAST(uint8_t *, RCAST(void *, &sv)); 3360 uint8_t *d = RCAST(uint8_t *, RCAST(void *, &rv)); 3361 d[0] = s[3]; 3362 d[1] = s[2]; 3363 d[2] = s[1]; 3364 d[3] = s[0]; 3365 return rv; 3366 } 3367 3368 /* 3369 * swap a quad 3370 */ 3371 private uint64_t 3372 swap8(uint64_t sv) 3373 { 3374 uint64_t rv; 3375 uint8_t *s = RCAST(uint8_t *, RCAST(void *, &sv)); 3376 uint8_t *d = RCAST(uint8_t *, RCAST(void *, &rv)); 3377 #if 0 3378 d[0] = s[3]; 3379 d[1] = s[2]; 3380 d[2] = s[1]; 3381 d[3] = s[0]; 3382 d[4] = s[7]; 3383 d[5] = s[6]; 3384 d[6] = s[5]; 3385 d[7] = s[4]; 3386 #else 3387 d[0] = s[7]; 3388 d[1] = s[6]; 3389 d[2] = s[5]; 3390 d[3] = s[4]; 3391 d[4] = s[3]; 3392 d[5] = s[2]; 3393 d[6] = s[1]; 3394 d[7] = s[0]; 3395 #endif 3396 return rv; 3397 } 3398 3399 /* 3400 * byteswap a single magic entry 3401 */ 3402 private void 3403 bs1(struct magic *m) 3404 { 3405 m->cont_level = swap2(m->cont_level); 3406 m->offset = swap4(CAST(uint32_t, m->offset)); 3407 m->in_offset = swap4(CAST(uint32_t, m->in_offset)); 3408 m->lineno = swap4(CAST(uint32_t, m->lineno)); 3409 if (IS_STRING(m->type)) { 3410 m->str_range = swap4(m->str_range); 3411 m->str_flags = swap4(m->str_flags); 3412 } 3413 else { 3414 m->value.q = swap8(m->value.q); 3415 m->num_mask = swap8(m->num_mask); 3416 } 3417 } 3418 3419 protected size_t 3420 file_pstring_length_size(struct magic_set *ms, const struct magic *m) 3421 { 3422 switch (m->str_flags & PSTRING_LEN) { 3423 case PSTRING_1_LE: 3424 return 1; 3425 case PSTRING_2_LE: 3426 case PSTRING_2_BE: 3427 return 2; 3428 case PSTRING_4_LE: 3429 case PSTRING_4_BE: 3430 return 4; 3431 default: 3432 file_error(ms, 0, "corrupt magic file " 3433 "(bad pascal string length %d)", 3434 m->str_flags & PSTRING_LEN); 3435 return FILE_BADSIZE; 3436 } 3437 } 3438 protected size_t 3439 file_pstring_get_length(struct magic_set *ms, const struct magic *m, 3440 const char *ss) 3441 { 3442 size_t len = 0; 3443 const unsigned char *s = RCAST(const unsigned char *, ss); 3444 unsigned int s3, s2, s1, s0; 3445 3446 switch (m->str_flags & PSTRING_LEN) { 3447 case PSTRING_1_LE: 3448 len = *s; 3449 break; 3450 case PSTRING_2_LE: 3451 s0 = s[0]; 3452 s1 = s[1]; 3453 len = (s1 << 8) | s0; 3454 break; 3455 case PSTRING_2_BE: 3456 s0 = s[0]; 3457 s1 = s[1]; 3458 len = (s0 << 8) | s1; 3459 break; 3460 case PSTRING_4_LE: 3461 s0 = s[0]; 3462 s1 = s[1]; 3463 s2 = s[2]; 3464 s3 = s[3]; 3465 len = (s3 << 24) | (s2 << 16) | (s1 << 8) | s0; 3466 break; 3467 case PSTRING_4_BE: 3468 s0 = s[0]; 3469 s1 = s[1]; 3470 s2 = s[2]; 3471 s3 = s[3]; 3472 len = (s0 << 24) | (s1 << 16) | (s2 << 8) | s3; 3473 break; 3474 default: 3475 file_error(ms, 0, "corrupt magic file " 3476 "(bad pascal string length %d)", 3477 m->str_flags & PSTRING_LEN); 3478 return FILE_BADSIZE; 3479 } 3480 3481 if (m->str_flags & PSTRING_LENGTH_INCLUDES_ITSELF) { 3482 size_t l = file_pstring_length_size(ms, m); 3483 if (l == FILE_BADSIZE) 3484 return l; 3485 len -= l; 3486 } 3487 3488 return len; 3489 } 3490 3491 protected int 3492 file_magicfind(struct magic_set *ms, const char *name, struct mlist *v) 3493 { 3494 uint32_t i, j; 3495 struct mlist *mlist, *ml; 3496 3497 mlist = ms->mlist[1]; 3498 3499 for (ml = mlist->next; ml != mlist; ml = ml->next) { 3500 struct magic *ma = ml->magic; 3501 uint32_t nma = ml->nmagic; 3502 for (i = 0; i < nma; i++) { 3503 if (ma[i].type != FILE_NAME) 3504 continue; 3505 if (strcmp(ma[i].value.s, name) == 0) { 3506 v->magic = &ma[i]; 3507 for (j = i + 1; j < nma; j++) 3508 if (ma[j].cont_level == 0) 3509 break; 3510 v->nmagic = j - i; 3511 return 0; 3512 } 3513 } 3514 } 3515 return -1; 3516 } 3517