1 /* $NetBSD: apprentice.c,v 1.24 2019/12/17 02:31:05 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 1.284 2019/06/29 22:31:04 christos Exp $") 39 #else 40 __RCSID("$NetBSD: apprentice.c,v 1.24 2019/12/17 02:31:05 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(struct mlist *); 130 private void byteswap(struct magic *, uint32_t); 131 private void bs1(struct magic *); 132 private uint16_t swap2(uint16_t); 133 private uint32_t swap4(uint32_t); 134 private uint64_t swap8(uint64_t); 135 private char *mkdbname(struct magic_set *, const char *, int); 136 private struct magic_map *apprentice_buf(struct magic_set *, struct magic *, 137 size_t); 138 private struct magic_map *apprentice_map(struct magic_set *, const char *); 139 private int check_buffer(struct magic_set *, struct magic_map *, const char *); 140 private void apprentice_unmap(struct magic_map *); 141 private int apprentice_compile(struct magic_set *, struct magic_map *, 142 const char *); 143 private int check_format_type(const char *, int, const char **); 144 private int check_format(struct magic_set *, struct magic *); 145 private int get_op(char); 146 private int parse_mime(struct magic_set *, struct magic_entry *, const char *); 147 private int parse_strength(struct magic_set *, struct magic_entry *, const char *); 148 private int parse_apple(struct magic_set *, struct magic_entry *, const char *); 149 private int parse_ext(struct magic_set *, struct magic_entry *, const char *); 150 151 152 private size_t magicsize = sizeof(struct magic); 153 154 private const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc"; 155 156 private struct { 157 const char *name; 158 size_t len; 159 int (*fun)(struct magic_set *, struct magic_entry *, const char *); 160 } bang[] = { 161 #define DECLARE_FIELD(name) { # name, sizeof(# name) - 1, parse_ ## name } 162 DECLARE_FIELD(mime), 163 DECLARE_FIELD(apple), 164 DECLARE_FIELD(ext), 165 DECLARE_FIELD(strength), 166 #undef DECLARE_FIELD 167 { NULL, 0, NULL } 168 }; 169 170 #ifdef COMPILE_ONLY 171 172 int main(int, char *[]); 173 174 int 175 main(int argc, char *argv[]) 176 { 177 int ret; 178 struct magic_set *ms; 179 char *progname; 180 181 if ((progname = strrchr(argv[0], '/')) != NULL) 182 progname++; 183 else 184 progname = argv[0]; 185 186 if (argc != 2) { 187 (void)fprintf(stderr, "Usage: %s file\n", progname); 188 return 1; 189 } 190 191 if ((ms = magic_open(MAGIC_CHECK)) == NULL) { 192 (void)fprintf(stderr, "%s: %s\n", progname, strerror(errno)); 193 return 1; 194 } 195 ret = magic_compile(ms, argv[1]) == -1 ? 1 : 0; 196 if (ret == 1) 197 (void)fprintf(stderr, "%s: %s\n", progname, magic_error(ms)); 198 magic_close(ms); 199 return ret; 200 } 201 #endif /* COMPILE_ONLY */ 202 203 struct type_tbl_s { 204 const char name[16]; 205 const size_t len; 206 const int type; 207 const int format; 208 }; 209 210 /* 211 * XXX - the actual Single UNIX Specification says that "long" means "long", 212 * as in the C data type, but we treat it as meaning "4-byte integer". 213 * Given that the OS X version of file 5.04 did the same, I guess that passes 214 * the actual test; having "long" be dependent on how big a "long" is on 215 * the machine running "file" is silly. 216 */ 217 static const struct type_tbl_s type_tbl[] = { 218 # define XX(s) s, (sizeof(s) - 1) 219 # define XX_NULL "", 0 220 { XX("invalid"), FILE_INVALID, FILE_FMT_NONE }, 221 { XX("byte"), FILE_BYTE, FILE_FMT_NUM }, 222 { XX("short"), FILE_SHORT, FILE_FMT_NUM }, 223 { XX("default"), FILE_DEFAULT, FILE_FMT_NONE }, 224 { XX("long"), FILE_LONG, FILE_FMT_NUM }, 225 { XX("string"), FILE_STRING, FILE_FMT_STR }, 226 { XX("date"), FILE_DATE, FILE_FMT_STR }, 227 { XX("beshort"), FILE_BESHORT, FILE_FMT_NUM }, 228 { XX("belong"), FILE_BELONG, FILE_FMT_NUM }, 229 { XX("bedate"), FILE_BEDATE, FILE_FMT_STR }, 230 { XX("leshort"), FILE_LESHORT, FILE_FMT_NUM }, 231 { XX("lelong"), FILE_LELONG, FILE_FMT_NUM }, 232 { XX("ledate"), FILE_LEDATE, FILE_FMT_STR }, 233 { XX("pstring"), FILE_PSTRING, FILE_FMT_STR }, 234 { XX("ldate"), FILE_LDATE, FILE_FMT_STR }, 235 { XX("beldate"), FILE_BELDATE, FILE_FMT_STR }, 236 { XX("leldate"), FILE_LELDATE, FILE_FMT_STR }, 237 { XX("regex"), FILE_REGEX, FILE_FMT_STR }, 238 { XX("bestring16"), FILE_BESTRING16, FILE_FMT_STR }, 239 { XX("lestring16"), FILE_LESTRING16, FILE_FMT_STR }, 240 { XX("search"), FILE_SEARCH, FILE_FMT_STR }, 241 { XX("medate"), FILE_MEDATE, FILE_FMT_STR }, 242 { XX("meldate"), FILE_MELDATE, FILE_FMT_STR }, 243 { XX("melong"), FILE_MELONG, FILE_FMT_NUM }, 244 { XX("quad"), FILE_QUAD, FILE_FMT_QUAD }, 245 { XX("lequad"), FILE_LEQUAD, FILE_FMT_QUAD }, 246 { XX("bequad"), FILE_BEQUAD, FILE_FMT_QUAD }, 247 { XX("qdate"), FILE_QDATE, FILE_FMT_STR }, 248 { XX("leqdate"), FILE_LEQDATE, FILE_FMT_STR }, 249 { XX("beqdate"), FILE_BEQDATE, FILE_FMT_STR }, 250 { XX("qldate"), FILE_QLDATE, FILE_FMT_STR }, 251 { XX("leqldate"), FILE_LEQLDATE, FILE_FMT_STR }, 252 { XX("beqldate"), FILE_BEQLDATE, FILE_FMT_STR }, 253 { XX("float"), FILE_FLOAT, FILE_FMT_FLOAT }, 254 { XX("befloat"), FILE_BEFLOAT, FILE_FMT_FLOAT }, 255 { XX("lefloat"), FILE_LEFLOAT, FILE_FMT_FLOAT }, 256 { XX("double"), FILE_DOUBLE, FILE_FMT_DOUBLE }, 257 { XX("bedouble"), FILE_BEDOUBLE, FILE_FMT_DOUBLE }, 258 { XX("ledouble"), FILE_LEDOUBLE, FILE_FMT_DOUBLE }, 259 { XX("leid3"), FILE_LEID3, FILE_FMT_NUM }, 260 { XX("beid3"), FILE_BEID3, FILE_FMT_NUM }, 261 { XX("indirect"), FILE_INDIRECT, FILE_FMT_NUM }, 262 { XX("qwdate"), FILE_QWDATE, FILE_FMT_STR }, 263 { XX("leqwdate"), FILE_LEQWDATE, FILE_FMT_STR }, 264 { XX("beqwdate"), FILE_BEQWDATE, FILE_FMT_STR }, 265 { XX("name"), FILE_NAME, FILE_FMT_NONE }, 266 { XX("use"), FILE_USE, FILE_FMT_NONE }, 267 { XX("clear"), FILE_CLEAR, FILE_FMT_NONE }, 268 { XX("der"), FILE_DER, FILE_FMT_STR }, 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("der"), FILE_DER, FILE_FMT_STR }, 278 { XX("name"), FILE_NAME, FILE_FMT_STR }, 279 { XX("use"), FILE_USE, FILE_FMT_STR }, 280 { XX_NULL, FILE_INVALID, FILE_FMT_NONE }, 281 }; 282 # undef XX 283 # undef XX_NULL 284 285 private int 286 get_type(const struct type_tbl_s *tbl, const char *l, const char **t) 287 { 288 const struct type_tbl_s *p; 289 290 for (p = tbl; p->len; p++) { 291 if (strncmp(l, p->name, p->len) == 0) { 292 if (t) 293 *t = l + p->len; 294 break; 295 } 296 } 297 return p->type; 298 } 299 300 private off_t 301 maxoff_t(void) { 302 if (/*CONSTCOND*/sizeof(off_t) == sizeof(int)) 303 return CAST(off_t, INT_MAX); 304 if (/*CONSTCOND*/sizeof(off_t) == sizeof(long)) 305 return CAST(off_t, LONG_MAX); 306 return 0x7fffffff; 307 } 308 309 private int 310 get_standard_integer_type(const char *l, const char **t) 311 { 312 int type; 313 314 if (isalpha(CAST(unsigned char, l[1]))) { 315 switch (l[1]) { 316 case 'C': 317 /* "dC" and "uC" */ 318 type = FILE_BYTE; 319 break; 320 case 'S': 321 /* "dS" and "uS" */ 322 type = FILE_SHORT; 323 break; 324 case 'I': 325 case 'L': 326 /* 327 * "dI", "dL", "uI", and "uL". 328 * 329 * XXX - the actual Single UNIX Specification says 330 * that "L" means "long", as in the C data type, 331 * but we treat it as meaning "4-byte integer". 332 * Given that the OS X version of file 5.04 did 333 * the same, I guess that passes the actual SUS 334 * validation suite; having "dL" be dependent on 335 * how big a "long" is on the machine running 336 * "file" is silly. 337 */ 338 type = FILE_LONG; 339 break; 340 case 'Q': 341 /* "dQ" and "uQ" */ 342 type = FILE_QUAD; 343 break; 344 default: 345 /* "d{anything else}", "u{anything else}" */ 346 return FILE_INVALID; 347 } 348 l += 2; 349 } else if (isdigit(CAST(unsigned char, l[1]))) { 350 /* 351 * "d{num}" and "u{num}"; we only support {num} values 352 * of 1, 2, 4, and 8 - the Single UNIX Specification 353 * doesn't say anything about whether arbitrary 354 * values should be supported, but both the Solaris 10 355 * and OS X Mountain Lion versions of file passed the 356 * Single UNIX Specification validation suite, and 357 * neither of them support values bigger than 8 or 358 * non-power-of-2 values. 359 */ 360 if (isdigit(CAST(unsigned char, l[2]))) { 361 /* Multi-digit, so > 9 */ 362 return FILE_INVALID; 363 } 364 switch (l[1]) { 365 case '1': 366 type = FILE_BYTE; 367 break; 368 case '2': 369 type = FILE_SHORT; 370 break; 371 case '4': 372 type = FILE_LONG; 373 break; 374 case '8': 375 type = FILE_QUAD; 376 break; 377 default: 378 /* XXX - what about 3, 5, 6, or 7? */ 379 return FILE_INVALID; 380 } 381 l += 2; 382 } else { 383 /* 384 * "d" or "u" by itself. 385 */ 386 type = FILE_LONG; 387 ++l; 388 } 389 if (t) 390 *t = l; 391 return type; 392 } 393 394 private void 395 init_file_tables(void) 396 { 397 static int done = 0; 398 const struct type_tbl_s *p; 399 400 if (done) 401 return; 402 done++; 403 404 for (p = type_tbl; p->len; p++) { 405 assert(p->type < FILE_NAMES_SIZE); 406 file_names[p->type] = p->name; 407 file_formats[p->type] = p->format; 408 } 409 assert(p - type_tbl == FILE_NAMES_SIZE); 410 } 411 412 private int 413 add_mlist(struct mlist *mlp, struct magic_map *map, size_t idx) 414 { 415 struct mlist *ml; 416 417 mlp->map = NULL; 418 if ((ml = CAST(struct mlist *, malloc(sizeof(*ml)))) == NULL) 419 return -1; 420 421 ml->map = idx == 0 ? map : NULL; 422 ml->magic = map->magic[idx]; 423 ml->nmagic = map->nmagic[idx]; 424 425 mlp->prev->next = ml; 426 ml->prev = mlp->prev; 427 ml->next = mlp; 428 mlp->prev = ml; 429 return 0; 430 } 431 432 /* 433 * Handle one file or directory. 434 */ 435 private int 436 apprentice_1(struct magic_set *ms, const char *fn, int action) 437 { 438 struct magic_map *map; 439 #ifndef COMPILE_ONLY 440 struct mlist *ml; 441 size_t i; 442 #endif 443 444 if (magicsize != FILE_MAGICSIZE) { 445 file_error(ms, 0, "magic element size %lu != %lu", 446 CAST(unsigned long, sizeof(*map->magic[0])), 447 CAST(unsigned long, FILE_MAGICSIZE)); 448 return -1; 449 } 450 451 if (action == FILE_COMPILE) { 452 map = apprentice_load(ms, fn, action); 453 if (map == NULL) 454 return -1; 455 return apprentice_compile(ms, map, fn); 456 } 457 458 #ifndef COMPILE_ONLY 459 map = apprentice_map(ms, fn); 460 if (map == RCAST(struct magic_map *, -1)) 461 return -1; 462 if (map == NULL) { 463 if (ms->flags & MAGIC_CHECK) 464 file_magwarn(ms, "using regular magic file `%s'", fn); 465 map = apprentice_load(ms, fn, action); 466 if (map == NULL) 467 return -1; 468 } 469 470 for (i = 0; i < MAGIC_SETS; i++) { 471 if (add_mlist(ms->mlist[i], map, i) == -1) { 472 file_oomem(ms, sizeof(*ml)); 473 return -1; 474 } 475 } 476 477 if (action == FILE_LIST) { 478 for (i = 0; i < MAGIC_SETS; i++) { 479 printf("Set %" SIZE_T_FORMAT "u:\nBinary patterns:\n", 480 i); 481 apprentice_list(ms->mlist[i], BINTEST); 482 printf("Text patterns:\n"); 483 apprentice_list(ms->mlist[i], TEXTTEST); 484 } 485 } 486 return 0; 487 #else 488 return 0; 489 #endif /* COMPILE_ONLY */ 490 } 491 492 protected void 493 file_ms_free(struct magic_set *ms) 494 { 495 size_t i; 496 if (ms == NULL) 497 return; 498 for (i = 0; i < MAGIC_SETS; i++) 499 mlist_free(ms->mlist[i]); 500 free(ms->o.pbuf); 501 free(ms->o.buf); 502 free(ms->c.li); 503 free(ms); 504 } 505 506 protected struct magic_set * 507 file_ms_alloc(int flags) 508 { 509 struct magic_set *ms; 510 size_t i, len; 511 512 if ((ms = CAST(struct magic_set *, calloc(CAST(size_t, 1u), 513 sizeof(struct magic_set)))) == NULL) 514 return NULL; 515 516 if (magic_setflags(ms, flags) == -1) { 517 errno = EINVAL; 518 goto free; 519 } 520 521 ms->o.buf = ms->o.pbuf = NULL; 522 len = (ms->c.len = 10) * sizeof(*ms->c.li); 523 524 if ((ms->c.li = CAST(struct level_info *, malloc(len))) == NULL) 525 goto free; 526 527 ms->event_flags = 0; 528 ms->error = -1; 529 for (i = 0; i < MAGIC_SETS; i++) 530 ms->mlist[i] = NULL; 531 ms->file = "unknown"; 532 ms->line = 0; 533 ms->indir_max = FILE_INDIR_MAX; 534 ms->name_max = FILE_NAME_MAX; 535 ms->elf_shnum_max = FILE_ELF_SHNUM_MAX; 536 ms->elf_phnum_max = FILE_ELF_PHNUM_MAX; 537 ms->elf_notes_max = FILE_ELF_NOTES_MAX; 538 ms->regex_max = FILE_REGEX_MAX; 539 ms->bytes_max = FILE_BYTES_MAX; 540 return ms; 541 free: 542 free(ms); 543 return NULL; 544 } 545 546 private void 547 apprentice_unmap(struct magic_map *map) 548 { 549 size_t i; 550 if (map == NULL) 551 return; 552 553 switch (map->type) { 554 case MAP_TYPE_USER: 555 break; 556 case MAP_TYPE_MALLOC: 557 for (i = 0; i < MAGIC_SETS; i++) { 558 void *b = map->magic[i]; 559 void *p = map->p; 560 if (CAST(char *, b) >= CAST(char *, p) && 561 CAST(char *, b) <= CAST(char *, p) + map->len) 562 continue; 563 free(map->magic[i]); 564 } 565 free(map->p); 566 break; 567 #ifdef QUICK 568 case MAP_TYPE_MMAP: 569 if (map->p && map->p != MAP_FAILED) 570 (void)munmap(map->p, map->len); 571 break; 572 #endif 573 default: 574 abort(); 575 } 576 free(map); 577 } 578 579 private struct mlist * 580 mlist_alloc(void) 581 { 582 struct mlist *mlist; 583 if ((mlist = CAST(struct mlist *, calloc(1, sizeof(*mlist)))) == NULL) { 584 return NULL; 585 } 586 mlist->next = mlist->prev = mlist; 587 return mlist; 588 } 589 590 private void 591 mlist_free_one(struct mlist *ml) 592 { 593 if (ml->map) 594 apprentice_unmap(CAST(struct magic_map *, ml->map)); 595 free(ml); 596 } 597 598 private void 599 mlist_free(struct mlist *mlist) 600 { 601 struct mlist *ml, *next; 602 603 if (mlist == NULL) 604 return; 605 606 for (ml = mlist->next; ml != mlist; ml = next) { 607 next = ml->next; 608 mlist_free_one(ml); 609 } 610 mlist_free_one(mlist); 611 } 612 613 #ifndef COMPILE_ONLY 614 /* void **bufs: an array of compiled magic files */ 615 protected int 616 buffer_apprentice(struct magic_set *ms, struct magic **bufs, 617 size_t *sizes, size_t nbufs) 618 { 619 size_t i, j; 620 struct mlist *ml; 621 struct magic_map *map; 622 623 if (nbufs == 0) 624 return -1; 625 626 (void)file_reset(ms, 0); 627 628 init_file_tables(); 629 630 for (i = 0; i < MAGIC_SETS; i++) { 631 mlist_free(ms->mlist[i]); 632 if ((ms->mlist[i] = mlist_alloc()) == NULL) { 633 file_oomem(ms, sizeof(*ms->mlist[i])); 634 goto fail; 635 } 636 } 637 638 for (i = 0; i < nbufs; i++) { 639 map = apprentice_buf(ms, bufs[i], sizes[i]); 640 if (map == NULL) 641 goto fail; 642 643 for (j = 0; j < MAGIC_SETS; j++) { 644 if (add_mlist(ms->mlist[j], map, j) == -1) { 645 file_oomem(ms, sizeof(*ml)); 646 goto fail; 647 } 648 } 649 } 650 651 return 0; 652 fail: 653 for (i = 0; i < MAGIC_SETS; i++) { 654 mlist_free(ms->mlist[i]); 655 ms->mlist[i] = NULL; 656 } 657 return -1; 658 } 659 #endif 660 661 /* const char *fn: list of magic files and directories */ 662 protected int 663 file_apprentice(struct magic_set *ms, const char *fn, int action) 664 { 665 char *p, *mfn; 666 int fileerr, errs = -1; 667 size_t i; 668 669 (void)file_reset(ms, 0); 670 671 if ((fn = magic_getpath(fn, action)) == NULL) 672 return -1; 673 674 init_file_tables(); 675 676 if ((mfn = strdup(fn)) == NULL) { 677 file_oomem(ms, strlen(fn)); 678 return -1; 679 } 680 681 for (i = 0; i < MAGIC_SETS; i++) { 682 mlist_free(ms->mlist[i]); 683 if ((ms->mlist[i] = mlist_alloc()) == NULL) { 684 file_oomem(ms, sizeof(*ms->mlist[i])); 685 while (i-- > 0) { 686 mlist_free(ms->mlist[i]); 687 ms->mlist[i] = NULL; 688 } 689 free(mfn); 690 return -1; 691 } 692 } 693 fn = mfn; 694 695 while (fn) { 696 p = strchr(fn, PATHSEP); 697 if (p) 698 *p++ = '\0'; 699 if (*fn == '\0') 700 break; 701 fileerr = apprentice_1(ms, fn, action); 702 errs = MAX(errs, fileerr); 703 fn = p; 704 } 705 706 free(mfn); 707 708 if (errs == -1) { 709 for (i = 0; i < MAGIC_SETS; i++) { 710 mlist_free(ms->mlist[i]); 711 ms->mlist[i] = NULL; 712 } 713 file_error(ms, 0, "could not find any valid magic files!"); 714 return -1; 715 } 716 717 #if 0 718 /* 719 * Always leave the database loaded 720 */ 721 if (action == FILE_LOAD) 722 return 0; 723 724 for (i = 0; i < MAGIC_SETS; i++) { 725 mlist_free(ms->mlist[i]); 726 ms->mlist[i] = NULL; 727 } 728 #endif 729 730 switch (action) { 731 case FILE_LOAD: 732 case FILE_COMPILE: 733 case FILE_CHECK: 734 case FILE_LIST: 735 return 0; 736 default: 737 file_error(ms, 0, "Invalid action %d", action); 738 return -1; 739 } 740 } 741 742 /* 743 * Compute the real length of a magic expression, for the purposes 744 * of determining how "strong" a magic expression is (approximating 745 * how specific its matches are): 746 * - magic characters count 0 unless escaped. 747 * - [] expressions count 1 748 * - {} expressions count 0 749 * - regular characters or escaped magic characters count 1 750 * - 0 length expressions count as one 751 */ 752 private size_t 753 nonmagic(const char *str) 754 { 755 const char *p; 756 size_t rv = 0; 757 758 for (p = str; *p; p++) 759 switch (*p) { 760 case '\\': /* Escaped anything counts 1 */ 761 if (!*++p) 762 p--; 763 rv++; 764 continue; 765 case '?': /* Magic characters count 0 */ 766 case '*': 767 case '.': 768 case '+': 769 case '^': 770 case '$': 771 continue; 772 case '[': /* Bracketed expressions count 1 the ']' */ 773 while (*p && *p != ']') 774 p++; 775 p--; 776 continue; 777 case '{': /* Braced expressions count 0 */ 778 while (*p && *p != '}') 779 p++; 780 if (!*p) 781 p--; 782 continue; 783 default: /* Anything else counts 1 */ 784 rv++; 785 continue; 786 } 787 788 return rv == 0 ? 1 : rv; /* Return at least 1 */ 789 } 790 791 792 private size_t 793 typesize(int type) 794 { 795 switch (type) { 796 case FILE_BYTE: 797 return 1; 798 799 case FILE_SHORT: 800 case FILE_LESHORT: 801 case FILE_BESHORT: 802 return 2; 803 804 case FILE_LONG: 805 case FILE_LELONG: 806 case FILE_BELONG: 807 case FILE_MELONG: 808 return 4; 809 810 case FILE_DATE: 811 case FILE_LEDATE: 812 case FILE_BEDATE: 813 case FILE_MEDATE: 814 case FILE_LDATE: 815 case FILE_LELDATE: 816 case FILE_BELDATE: 817 case FILE_MELDATE: 818 case FILE_FLOAT: 819 case FILE_BEFLOAT: 820 case FILE_LEFLOAT: 821 return 4; 822 823 case FILE_QUAD: 824 case FILE_BEQUAD: 825 case FILE_LEQUAD: 826 case FILE_QDATE: 827 case FILE_LEQDATE: 828 case FILE_BEQDATE: 829 case FILE_QLDATE: 830 case FILE_LEQLDATE: 831 case FILE_BEQLDATE: 832 case FILE_QWDATE: 833 case FILE_LEQWDATE: 834 case FILE_BEQWDATE: 835 case FILE_DOUBLE: 836 case FILE_BEDOUBLE: 837 case FILE_LEDOUBLE: 838 return 8; 839 default: 840 return CAST(size_t, ~0); 841 } 842 } 843 844 /* 845 * Get weight of this magic entry, for sorting purposes. 846 */ 847 private size_t 848 apprentice_magic_strength(const struct magic *m) 849 { 850 #define MULT 10U 851 size_t ts, v; 852 ssize_t val = 2 * MULT; /* baseline strength */ 853 854 switch (m->type) { 855 case FILE_DEFAULT: /* make sure this sorts last */ 856 if (m->factor_op != FILE_FACTOR_OP_NONE) 857 abort(); 858 return 0; 859 860 case FILE_BYTE: 861 case FILE_SHORT: 862 case FILE_LESHORT: 863 case FILE_BESHORT: 864 case FILE_LONG: 865 case FILE_LELONG: 866 case FILE_BELONG: 867 case FILE_MELONG: 868 case FILE_DATE: 869 case FILE_LEDATE: 870 case FILE_BEDATE: 871 case FILE_MEDATE: 872 case FILE_LDATE: 873 case FILE_LELDATE: 874 case FILE_BELDATE: 875 case FILE_MELDATE: 876 case FILE_FLOAT: 877 case FILE_BEFLOAT: 878 case FILE_LEFLOAT: 879 case FILE_QUAD: 880 case FILE_BEQUAD: 881 case FILE_LEQUAD: 882 case FILE_QDATE: 883 case FILE_LEQDATE: 884 case FILE_BEQDATE: 885 case FILE_QLDATE: 886 case FILE_LEQLDATE: 887 case FILE_BEQLDATE: 888 case FILE_QWDATE: 889 case FILE_LEQWDATE: 890 case FILE_BEQWDATE: 891 case FILE_DOUBLE: 892 case FILE_BEDOUBLE: 893 case FILE_LEDOUBLE: 894 ts = typesize(m->type); 895 if (ts == CAST(size_t, ~0)) 896 abort(); 897 val += ts * MULT; 898 break; 899 900 case FILE_PSTRING: 901 case FILE_STRING: 902 val += m->vallen * MULT; 903 break; 904 905 case FILE_BESTRING16: 906 case FILE_LESTRING16: 907 val += m->vallen * MULT / 2; 908 break; 909 910 case FILE_SEARCH: 911 if (m->vallen == 0) 912 break; 913 val += m->vallen * MAX(MULT / m->vallen, 1); 914 break; 915 916 case FILE_REGEX: 917 v = nonmagic(m->value.s); 918 val += v * MAX(MULT / v, 1); 919 break; 920 921 case FILE_INDIRECT: 922 case FILE_NAME: 923 case FILE_USE: 924 break; 925 926 case FILE_DER: 927 val += MULT; 928 break; 929 930 default: 931 (void)fprintf(stderr, "Bad type %d\n", m->type); 932 abort(); 933 } 934 935 switch (m->reln) { 936 case 'x': /* matches anything penalize */ 937 case '!': /* matches almost anything penalize */ 938 val = 0; 939 break; 940 941 case '=': /* Exact match, prefer */ 942 val += MULT; 943 break; 944 945 case '>': 946 case '<': /* comparison match reduce strength */ 947 val -= 2 * MULT; 948 break; 949 950 case '^': 951 case '&': /* masking bits, we could count them too */ 952 val -= MULT; 953 break; 954 955 default: 956 (void)fprintf(stderr, "Bad relation %c\n", m->reln); 957 abort(); 958 } 959 960 switch (m->factor_op) { 961 case FILE_FACTOR_OP_NONE: 962 break; 963 case FILE_FACTOR_OP_PLUS: 964 val += m->factor; 965 break; 966 case FILE_FACTOR_OP_MINUS: 967 val -= m->factor; 968 break; 969 case FILE_FACTOR_OP_TIMES: 970 val *= m->factor; 971 break; 972 case FILE_FACTOR_OP_DIV: 973 val /= m->factor; 974 break; 975 default: 976 abort(); 977 } 978 979 if (val <= 0) /* ensure we only return 0 for FILE_DEFAULT */ 980 val = 1; 981 982 /* 983 * Magic entries with no description get a bonus because they depend 984 * on subsequent magic entries to print something. 985 */ 986 if (m->desc[0] == '\0') 987 val++; 988 return val; 989 } 990 991 /* 992 * Sort callback for sorting entries by "strength" (basically length) 993 */ 994 private int 995 apprentice_sort(const void *a, const void *b) 996 { 997 const struct magic_entry *ma = CAST(const struct magic_entry *, a); 998 const struct magic_entry *mb = CAST(const struct magic_entry *, b); 999 size_t sa = apprentice_magic_strength(ma->mp); 1000 size_t sb = apprentice_magic_strength(mb->mp); 1001 if (sa == sb) 1002 return 0; 1003 else if (sa > sb) 1004 return -1; 1005 else 1006 return 1; 1007 } 1008 1009 /* 1010 * Shows sorted patterns list in the order which is used for the matching 1011 */ 1012 private void 1013 apprentice_list(struct mlist *mlist, int mode) 1014 { 1015 uint32_t magindex = 0; 1016 struct mlist *ml; 1017 for (ml = mlist->next; ml != mlist; ml = ml->next) { 1018 for (magindex = 0; magindex < ml->nmagic; magindex++) { 1019 struct magic *m = &ml->magic[magindex]; 1020 if ((m->flag & mode) != mode) { 1021 /* Skip sub-tests */ 1022 while (magindex + 1 < ml->nmagic && 1023 ml->magic[magindex + 1].cont_level != 0) 1024 ++magindex; 1025 continue; /* Skip to next top-level test*/ 1026 } 1027 1028 /* 1029 * Try to iterate over the tree until we find item with 1030 * description/mimetype. 1031 */ 1032 while (magindex + 1 < ml->nmagic && 1033 ml->magic[magindex + 1].cont_level != 0 && 1034 *ml->magic[magindex].desc == '\0' && 1035 *ml->magic[magindex].mimetype == '\0') 1036 magindex++; 1037 1038 printf("Strength = %3" SIZE_T_FORMAT "u@%u: %s [%s]\n", 1039 apprentice_magic_strength(m), 1040 ml->magic[magindex].lineno, 1041 ml->magic[magindex].desc, 1042 ml->magic[magindex].mimetype); 1043 } 1044 } 1045 } 1046 1047 private void 1048 set_test_type(struct magic *mstart, struct magic *m) 1049 { 1050 switch (m->type) { 1051 case FILE_BYTE: 1052 case FILE_SHORT: 1053 case FILE_LONG: 1054 case FILE_DATE: 1055 case FILE_BESHORT: 1056 case FILE_BELONG: 1057 case FILE_BEDATE: 1058 case FILE_LESHORT: 1059 case FILE_LELONG: 1060 case FILE_LEDATE: 1061 case FILE_LDATE: 1062 case FILE_BELDATE: 1063 case FILE_LELDATE: 1064 case FILE_MEDATE: 1065 case FILE_MELDATE: 1066 case FILE_MELONG: 1067 case FILE_QUAD: 1068 case FILE_LEQUAD: 1069 case FILE_BEQUAD: 1070 case FILE_QDATE: 1071 case FILE_LEQDATE: 1072 case FILE_BEQDATE: 1073 case FILE_QLDATE: 1074 case FILE_LEQLDATE: 1075 case FILE_BEQLDATE: 1076 case FILE_QWDATE: 1077 case FILE_LEQWDATE: 1078 case FILE_BEQWDATE: 1079 case FILE_FLOAT: 1080 case FILE_BEFLOAT: 1081 case FILE_LEFLOAT: 1082 case FILE_DOUBLE: 1083 case FILE_BEDOUBLE: 1084 case FILE_LEDOUBLE: 1085 case FILE_DER: 1086 mstart->flag |= BINTEST; 1087 break; 1088 case FILE_STRING: 1089 case FILE_PSTRING: 1090 case FILE_BESTRING16: 1091 case FILE_LESTRING16: 1092 /* Allow text overrides */ 1093 if (mstart->str_flags & STRING_TEXTTEST) 1094 mstart->flag |= TEXTTEST; 1095 else 1096 mstart->flag |= BINTEST; 1097 break; 1098 case FILE_REGEX: 1099 case FILE_SEARCH: 1100 /* Check for override */ 1101 if (mstart->str_flags & STRING_BINTEST) 1102 mstart->flag |= BINTEST; 1103 if (mstart->str_flags & STRING_TEXTTEST) 1104 mstart->flag |= TEXTTEST; 1105 1106 if (mstart->flag & (TEXTTEST|BINTEST)) 1107 break; 1108 1109 /* binary test if pattern is not text */ 1110 if (file_looks_utf8(m->value.us, CAST(size_t, m->vallen), NULL, 1111 NULL) <= 0) 1112 mstart->flag |= BINTEST; 1113 else 1114 mstart->flag |= TEXTTEST; 1115 break; 1116 case FILE_DEFAULT: 1117 /* can't deduce anything; we shouldn't see this at the 1118 top level anyway */ 1119 break; 1120 case FILE_INVALID: 1121 default: 1122 /* invalid search type, but no need to complain here */ 1123 break; 1124 } 1125 } 1126 1127 private int 1128 addentry(struct magic_set *ms, struct magic_entry *me, 1129 struct magic_entry_set *mset) 1130 { 1131 size_t i = me->mp->type == FILE_NAME ? 1 : 0; 1132 if (mset[i].count == mset[i].max) { 1133 struct magic_entry *mp; 1134 1135 mset[i].max += ALLOC_INCR; 1136 if ((mp = CAST(struct magic_entry *, 1137 realloc(mset[i].me, sizeof(*mp) * mset[i].max))) == 1138 NULL) { 1139 file_oomem(ms, sizeof(*mp) * mset[i].max); 1140 return -1; 1141 } 1142 (void)memset(&mp[mset[i].count], 0, sizeof(*mp) * 1143 ALLOC_INCR); 1144 mset[i].me = mp; 1145 } 1146 mset[i].me[mset[i].count++] = *me; 1147 memset(me, 0, sizeof(*me)); 1148 return 0; 1149 } 1150 1151 /* 1152 * Load and parse one file. 1153 */ 1154 private void 1155 load_1(struct magic_set *ms, int action, const char *fn, int *errs, 1156 struct magic_entry_set *mset) 1157 { 1158 size_t lineno = 0, llen = 0; 1159 char *line = NULL; 1160 ssize_t len; 1161 struct magic_entry me; 1162 1163 FILE *f = fopen(ms->file = fn, "r"); 1164 if (f == NULL) { 1165 if (errno != ENOENT) 1166 file_error(ms, errno, "cannot read magic file `%s'", 1167 fn); 1168 (*errs)++; 1169 return; 1170 } 1171 1172 memset(&me, 0, sizeof(me)); 1173 /* read and parse this file */ 1174 for (ms->line = 1; (len = getline(&line, &llen, f)) != -1; 1175 ms->line++) { 1176 if (len == 0) /* null line, garbage, etc */ 1177 continue; 1178 if (line[len - 1] == '\n') { 1179 lineno++; 1180 line[len - 1] = '\0'; /* delete newline */ 1181 } 1182 switch (line[0]) { 1183 case '\0': /* empty, do not parse */ 1184 case '#': /* comment, do not parse */ 1185 continue; 1186 case '!': 1187 if (line[1] == ':') { 1188 size_t i; 1189 1190 for (i = 0; bang[i].name != NULL; i++) { 1191 if (CAST(size_t, len - 2) > bang[i].len && 1192 memcmp(bang[i].name, line + 2, 1193 bang[i].len) == 0) 1194 break; 1195 } 1196 if (bang[i].name == NULL) { 1197 file_error(ms, 0, 1198 "Unknown !: entry `%s'", line); 1199 (*errs)++; 1200 continue; 1201 } 1202 if (me.mp == NULL) { 1203 file_error(ms, 0, 1204 "No current entry for :!%s type", 1205 bang[i].name); 1206 (*errs)++; 1207 continue; 1208 } 1209 if ((*bang[i].fun)(ms, &me, 1210 line + bang[i].len + 2) != 0) { 1211 (*errs)++; 1212 continue; 1213 } 1214 continue; 1215 } 1216 /*FALLTHROUGH*/ 1217 default: 1218 again: 1219 switch (parse(ms, &me, line, lineno, action)) { 1220 case 0: 1221 continue; 1222 case 1: 1223 (void)addentry(ms, &me, mset); 1224 goto again; 1225 default: 1226 (*errs)++; 1227 break; 1228 } 1229 } 1230 } 1231 if (me.mp) 1232 (void)addentry(ms, &me, mset); 1233 free(line); 1234 (void)fclose(f); 1235 } 1236 1237 /* 1238 * parse a file or directory of files 1239 * const char *fn: name of magic file or directory 1240 */ 1241 private int 1242 cmpstrp(const void *p1, const void *p2) 1243 { 1244 return strcmp(*RCAST(char *const *, p1), *RCAST(char *const *, p2)); 1245 } 1246 1247 1248 private uint32_t 1249 set_text_binary(struct magic_set *ms, struct magic_entry *me, uint32_t nme, 1250 uint32_t starttest) 1251 { 1252 static const char text[] = "text"; 1253 static const char binary[] = "binary"; 1254 static const size_t len = sizeof(text); 1255 1256 uint32_t i = starttest; 1257 1258 do { 1259 set_test_type(me[starttest].mp, me[i].mp); 1260 if ((ms->flags & MAGIC_DEBUG) == 0) 1261 continue; 1262 (void)fprintf(stderr, "%s%s%s: %s\n", 1263 me[i].mp->mimetype, 1264 me[i].mp->mimetype[0] == '\0' ? "" : "; ", 1265 me[i].mp->desc[0] ? me[i].mp->desc : "(no description)", 1266 me[i].mp->flag & BINTEST ? binary : text); 1267 if (me[i].mp->flag & BINTEST) { 1268 char *p = strstr(me[i].mp->desc, text); 1269 if (p && (p == me[i].mp->desc || 1270 isspace(CAST(unsigned char, p[-1]))) && 1271 (p + len - me[i].mp->desc == MAXstring 1272 || (p[len] == '\0' || 1273 isspace(CAST(unsigned char, p[len]))))) 1274 (void)fprintf(stderr, "*** Possible " 1275 "binary test for text type\n"); 1276 } 1277 } while (++i < nme && me[i].mp->cont_level != 0); 1278 return i; 1279 } 1280 1281 private void 1282 set_last_default(struct magic_set *ms, struct magic_entry *me, uint32_t nme) 1283 { 1284 uint32_t i; 1285 for (i = 0; i < nme; i++) { 1286 if (me[i].mp->cont_level == 0 && 1287 me[i].mp->type == FILE_DEFAULT) { 1288 while (++i < nme) 1289 if (me[i].mp->cont_level == 0) 1290 break; 1291 if (i != nme) { 1292 /* XXX - Ugh! */ 1293 ms->line = me[i].mp->lineno; 1294 file_magwarn(ms, 1295 "level 0 \"default\" did not sort last"); 1296 } 1297 return; 1298 } 1299 } 1300 } 1301 1302 private int 1303 coalesce_entries(struct magic_set *ms, struct magic_entry *me, uint32_t nme, 1304 struct magic **ma, uint32_t *nma) 1305 { 1306 uint32_t i, mentrycount = 0; 1307 size_t slen; 1308 1309 for (i = 0; i < nme; i++) 1310 mentrycount += me[i].cont_count; 1311 1312 slen = sizeof(**ma) * mentrycount; 1313 if ((*ma = CAST(struct magic *, malloc(slen))) == NULL) { 1314 file_oomem(ms, slen); 1315 return -1; 1316 } 1317 1318 mentrycount = 0; 1319 for (i = 0; i < nme; i++) { 1320 (void)memcpy(*ma + mentrycount, me[i].mp, 1321 me[i].cont_count * sizeof(**ma)); 1322 mentrycount += me[i].cont_count; 1323 } 1324 *nma = mentrycount; 1325 return 0; 1326 } 1327 1328 private void 1329 magic_entry_free(struct magic_entry *me, uint32_t nme) 1330 { 1331 uint32_t i; 1332 if (me == NULL) 1333 return; 1334 for (i = 0; i < nme; i++) 1335 free(me[i].mp); 1336 free(me); 1337 } 1338 1339 private struct magic_map * 1340 apprentice_load(struct magic_set *ms, const char *fn, int action) 1341 { 1342 int errs = 0; 1343 uint32_t i, j; 1344 size_t files = 0, maxfiles = 0; 1345 char **filearr = NULL, *mfn; 1346 struct stat st; 1347 struct magic_map *map; 1348 struct magic_entry_set mset[MAGIC_SETS]; 1349 DIR *dir; 1350 struct dirent *d; 1351 1352 memset(mset, 0, sizeof(mset)); 1353 ms->flags |= MAGIC_CHECK; /* Enable checks for parsed files */ 1354 1355 1356 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) 1357 { 1358 file_oomem(ms, sizeof(*map)); 1359 return NULL; 1360 } 1361 map->type = MAP_TYPE_MALLOC; 1362 1363 /* print silly verbose header for USG compat. */ 1364 if (action == FILE_CHECK) 1365 (void)fprintf(stderr, "%s\n", usg_hdr); 1366 1367 /* load directory or file */ 1368 if (stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) { 1369 dir = opendir(fn); 1370 if (!dir) { 1371 errs++; 1372 goto out; 1373 } 1374 while ((d = readdir(dir)) != NULL) { 1375 if (d->d_name[0] == '.') 1376 continue; 1377 if (asprintf(&mfn, "%s/%s", fn, d->d_name) < 0) { 1378 file_oomem(ms, 1379 strlen(fn) + strlen(d->d_name) + 2); 1380 errs++; 1381 closedir(dir); 1382 goto out; 1383 } 1384 if (stat(mfn, &st) == -1 || !S_ISREG(st.st_mode)) { 1385 free(mfn); 1386 continue; 1387 } 1388 if (files >= maxfiles) { 1389 size_t mlen; 1390 maxfiles = (maxfiles + 1) * 2; 1391 mlen = maxfiles * sizeof(*filearr); 1392 if ((filearr = CAST(char **, 1393 realloc(filearr, mlen))) == NULL) { 1394 file_oomem(ms, mlen); 1395 free(mfn); 1396 closedir(dir); 1397 errs++; 1398 goto out; 1399 } 1400 } 1401 filearr[files++] = mfn; 1402 } 1403 closedir(dir); 1404 if (filearr) { 1405 qsort(filearr, files, sizeof(*filearr), cmpstrp); 1406 for (i = 0; i < files; i++) { 1407 load_1(ms, action, filearr[i], &errs, mset); 1408 free(filearr[i]); 1409 } 1410 free(filearr); 1411 } 1412 } else 1413 load_1(ms, action, fn, &errs, mset); 1414 if (errs) 1415 goto out; 1416 1417 for (j = 0; j < MAGIC_SETS; j++) { 1418 /* Set types of tests */ 1419 for (i = 0; i < mset[j].count; ) { 1420 if (mset[j].me[i].mp->cont_level != 0) { 1421 i++; 1422 continue; 1423 } 1424 i = set_text_binary(ms, mset[j].me, mset[j].count, i); 1425 } 1426 if (mset[j].me) 1427 qsort(mset[j].me, mset[j].count, sizeof(*mset[j].me), 1428 apprentice_sort); 1429 1430 /* 1431 * Make sure that any level 0 "default" line is last 1432 * (if one exists). 1433 */ 1434 set_last_default(ms, mset[j].me, mset[j].count); 1435 1436 /* coalesce per file arrays into a single one */ 1437 if (coalesce_entries(ms, mset[j].me, mset[j].count, 1438 &map->magic[j], &map->nmagic[j]) == -1) { 1439 errs++; 1440 goto out; 1441 } 1442 } 1443 1444 out: 1445 for (j = 0; j < MAGIC_SETS; j++) 1446 magic_entry_free(mset[j].me, mset[j].count); 1447 1448 if (errs) { 1449 apprentice_unmap(map); 1450 return NULL; 1451 } 1452 return map; 1453 } 1454 1455 /* 1456 * extend the sign bit if the comparison is to be signed 1457 */ 1458 protected uint64_t 1459 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v) 1460 { 1461 if (!(m->flag & UNSIGNED)) { 1462 switch(m->type) { 1463 /* 1464 * Do not remove the casts below. They are 1465 * vital. When later compared with the data, 1466 * the sign extension must have happened. 1467 */ 1468 case FILE_BYTE: 1469 v = CAST(signed char, v); 1470 break; 1471 case FILE_SHORT: 1472 case FILE_BESHORT: 1473 case FILE_LESHORT: 1474 v = CAST(short, v); 1475 break; 1476 case FILE_DATE: 1477 case FILE_BEDATE: 1478 case FILE_LEDATE: 1479 case FILE_MEDATE: 1480 case FILE_LDATE: 1481 case FILE_BELDATE: 1482 case FILE_LELDATE: 1483 case FILE_MELDATE: 1484 case FILE_LONG: 1485 case FILE_BELONG: 1486 case FILE_LELONG: 1487 case FILE_MELONG: 1488 case FILE_FLOAT: 1489 case FILE_BEFLOAT: 1490 case FILE_LEFLOAT: 1491 v = CAST(int32_t, v); 1492 break; 1493 case FILE_QUAD: 1494 case FILE_BEQUAD: 1495 case FILE_LEQUAD: 1496 case FILE_QDATE: 1497 case FILE_QLDATE: 1498 case FILE_QWDATE: 1499 case FILE_BEQDATE: 1500 case FILE_BEQLDATE: 1501 case FILE_BEQWDATE: 1502 case FILE_LEQDATE: 1503 case FILE_LEQLDATE: 1504 case FILE_LEQWDATE: 1505 case FILE_DOUBLE: 1506 case FILE_BEDOUBLE: 1507 case FILE_LEDOUBLE: 1508 v = CAST(int64_t, v); 1509 break; 1510 case FILE_STRING: 1511 case FILE_PSTRING: 1512 case FILE_BESTRING16: 1513 case FILE_LESTRING16: 1514 case FILE_REGEX: 1515 case FILE_SEARCH: 1516 case FILE_DEFAULT: 1517 case FILE_INDIRECT: 1518 case FILE_NAME: 1519 case FILE_USE: 1520 case FILE_CLEAR: 1521 case FILE_DER: 1522 break; 1523 default: 1524 if (ms->flags & MAGIC_CHECK) 1525 file_magwarn(ms, "cannot happen: m->type=%d\n", 1526 m->type); 1527 return ~0U; 1528 } 1529 } 1530 return v; 1531 } 1532 1533 private int 1534 string_modifier_check(struct magic_set *ms, struct magic *m) 1535 { 1536 if ((ms->flags & MAGIC_CHECK) == 0) 1537 return 0; 1538 1539 if ((m->type != FILE_REGEX || (m->str_flags & REGEX_LINE_COUNT) == 0) && 1540 (m->type != FILE_PSTRING && (m->str_flags & PSTRING_LEN) != 0)) { 1541 file_magwarn(ms, 1542 "'/BHhLl' modifiers are only allowed for pascal strings\n"); 1543 return -1; 1544 } 1545 switch (m->type) { 1546 case FILE_BESTRING16: 1547 case FILE_LESTRING16: 1548 if (m->str_flags != 0) { 1549 file_magwarn(ms, 1550 "no modifiers allowed for 16-bit strings\n"); 1551 return -1; 1552 } 1553 break; 1554 case FILE_STRING: 1555 case FILE_PSTRING: 1556 if ((m->str_flags & REGEX_OFFSET_START) != 0) { 1557 file_magwarn(ms, 1558 "'/%c' only allowed on regex and search\n", 1559 CHAR_REGEX_OFFSET_START); 1560 return -1; 1561 } 1562 break; 1563 case FILE_SEARCH: 1564 if (m->str_range == 0) { 1565 file_magwarn(ms, 1566 "missing range; defaulting to %d\n", 1567 STRING_DEFAULT_RANGE); 1568 m->str_range = STRING_DEFAULT_RANGE; 1569 return -1; 1570 } 1571 break; 1572 case FILE_REGEX: 1573 if ((m->str_flags & STRING_COMPACT_WHITESPACE) != 0) { 1574 file_magwarn(ms, "'/%c' not allowed on regex\n", 1575 CHAR_COMPACT_WHITESPACE); 1576 return -1; 1577 } 1578 if ((m->str_flags & STRING_COMPACT_OPTIONAL_WHITESPACE) != 0) { 1579 file_magwarn(ms, "'/%c' not allowed on regex\n", 1580 CHAR_COMPACT_OPTIONAL_WHITESPACE); 1581 return -1; 1582 } 1583 break; 1584 default: 1585 file_magwarn(ms, "coding error: m->type=%d\n", 1586 m->type); 1587 return -1; 1588 } 1589 return 0; 1590 } 1591 1592 private int 1593 get_op(char c) 1594 { 1595 switch (c) { 1596 case '&': 1597 return FILE_OPAND; 1598 case '|': 1599 return FILE_OPOR; 1600 case '^': 1601 return FILE_OPXOR; 1602 case '+': 1603 return FILE_OPADD; 1604 case '-': 1605 return FILE_OPMINUS; 1606 case '*': 1607 return FILE_OPMULTIPLY; 1608 case '/': 1609 return FILE_OPDIVIDE; 1610 case '%': 1611 return FILE_OPMODULO; 1612 default: 1613 return -1; 1614 } 1615 } 1616 1617 #ifdef ENABLE_CONDITIONALS 1618 private int 1619 get_cond(const char *l, const char **t) 1620 { 1621 static const struct cond_tbl_s { 1622 char name[8]; 1623 size_t len; 1624 int cond; 1625 } cond_tbl[] = { 1626 { "if", 2, COND_IF }, 1627 { "elif", 4, COND_ELIF }, 1628 { "else", 4, COND_ELSE }, 1629 { "", 0, COND_NONE }, 1630 }; 1631 const struct cond_tbl_s *p; 1632 1633 for (p = cond_tbl; p->len; p++) { 1634 if (strncmp(l, p->name, p->len) == 0 && 1635 isspace(CAST(unsigned char, l[p->len]))) { 1636 if (t) 1637 *t = l + p->len; 1638 break; 1639 } 1640 } 1641 return p->cond; 1642 } 1643 1644 private int 1645 check_cond(struct magic_set *ms, int cond, uint32_t cont_level) 1646 { 1647 int last_cond; 1648 last_cond = ms->c.li[cont_level].last_cond; 1649 1650 switch (cond) { 1651 case COND_IF: 1652 if (last_cond != COND_NONE && last_cond != COND_ELIF) { 1653 if (ms->flags & MAGIC_CHECK) 1654 file_magwarn(ms, "syntax error: `if'"); 1655 return -1; 1656 } 1657 last_cond = COND_IF; 1658 break; 1659 1660 case COND_ELIF: 1661 if (last_cond != COND_IF && last_cond != COND_ELIF) { 1662 if (ms->flags & MAGIC_CHECK) 1663 file_magwarn(ms, "syntax error: `elif'"); 1664 return -1; 1665 } 1666 last_cond = COND_ELIF; 1667 break; 1668 1669 case COND_ELSE: 1670 if (last_cond != COND_IF && last_cond != COND_ELIF) { 1671 if (ms->flags & MAGIC_CHECK) 1672 file_magwarn(ms, "syntax error: `else'"); 1673 return -1; 1674 } 1675 last_cond = COND_NONE; 1676 break; 1677 1678 case COND_NONE: 1679 last_cond = COND_NONE; 1680 break; 1681 } 1682 1683 ms->c.li[cont_level].last_cond = last_cond; 1684 return 0; 1685 } 1686 #endif /* ENABLE_CONDITIONALS */ 1687 1688 private int 1689 parse_indirect_modifier(struct magic_set *ms, struct magic *m, const char **lp) 1690 { 1691 const char *l = *lp; 1692 1693 while (!isspace(CAST(unsigned char, *++l))) 1694 switch (*l) { 1695 case CHAR_INDIRECT_RELATIVE: 1696 m->str_flags |= INDIRECT_RELATIVE; 1697 break; 1698 default: 1699 if (ms->flags & MAGIC_CHECK) 1700 file_magwarn(ms, "indirect modifier `%c' " 1701 "invalid", *l); 1702 *lp = l; 1703 return -1; 1704 } 1705 *lp = l; 1706 return 0; 1707 } 1708 1709 private void 1710 parse_op_modifier(struct magic_set *ms, struct magic *m, const char **lp, 1711 int op) 1712 { 1713 const char *l = *lp; 1714 char *t; 1715 uint64_t val; 1716 1717 ++l; 1718 m->mask_op |= op; 1719 val = CAST(uint64_t, strtoull(l, &t, 0)); 1720 l = t; 1721 m->num_mask = file_signextend(ms, m, val); 1722 eatsize(&l); 1723 *lp = l; 1724 } 1725 1726 private int 1727 parse_string_modifier(struct magic_set *ms, struct magic *m, const char **lp) 1728 { 1729 const char *l = *lp; 1730 char *t; 1731 int have_range = 0; 1732 1733 while (!isspace(CAST(unsigned char, *++l))) { 1734 switch (*l) { 1735 case '0': case '1': case '2': 1736 case '3': case '4': case '5': 1737 case '6': case '7': case '8': 1738 case '9': 1739 if (have_range && (ms->flags & MAGIC_CHECK)) 1740 file_magwarn(ms, "multiple ranges"); 1741 have_range = 1; 1742 m->str_range = CAST(uint32_t, strtoul(l, &t, 0)); 1743 if (m->str_range == 0) 1744 file_magwarn(ms, "zero range"); 1745 l = t - 1; 1746 break; 1747 case CHAR_COMPACT_WHITESPACE: 1748 m->str_flags |= STRING_COMPACT_WHITESPACE; 1749 break; 1750 case CHAR_COMPACT_OPTIONAL_WHITESPACE: 1751 m->str_flags |= STRING_COMPACT_OPTIONAL_WHITESPACE; 1752 break; 1753 case CHAR_IGNORE_LOWERCASE: 1754 m->str_flags |= STRING_IGNORE_LOWERCASE; 1755 break; 1756 case CHAR_IGNORE_UPPERCASE: 1757 m->str_flags |= STRING_IGNORE_UPPERCASE; 1758 break; 1759 case CHAR_REGEX_OFFSET_START: 1760 m->str_flags |= REGEX_OFFSET_START; 1761 break; 1762 case CHAR_BINTEST: 1763 m->str_flags |= STRING_BINTEST; 1764 break; 1765 case CHAR_TEXTTEST: 1766 m->str_flags |= STRING_TEXTTEST; 1767 break; 1768 case CHAR_TRIM: 1769 m->str_flags |= STRING_TRIM; 1770 break; 1771 case CHAR_PSTRING_1_LE: 1772 #define SET_LENGTH(a) m->str_flags = (m->str_flags & ~PSTRING_LEN) | (a) 1773 if (m->type != FILE_PSTRING) 1774 goto bad; 1775 SET_LENGTH(PSTRING_1_LE); 1776 break; 1777 case CHAR_PSTRING_2_BE: 1778 if (m->type != FILE_PSTRING) 1779 goto bad; 1780 SET_LENGTH(PSTRING_2_BE); 1781 break; 1782 case CHAR_PSTRING_2_LE: 1783 if (m->type != FILE_PSTRING) 1784 goto bad; 1785 SET_LENGTH(PSTRING_2_LE); 1786 break; 1787 case CHAR_PSTRING_4_BE: 1788 if (m->type != FILE_PSTRING) 1789 goto bad; 1790 SET_LENGTH(PSTRING_4_BE); 1791 break; 1792 case CHAR_PSTRING_4_LE: 1793 switch (m->type) { 1794 case FILE_PSTRING: 1795 case FILE_REGEX: 1796 break; 1797 default: 1798 goto bad; 1799 } 1800 SET_LENGTH(PSTRING_4_LE); 1801 break; 1802 case CHAR_PSTRING_LENGTH_INCLUDES_ITSELF: 1803 if (m->type != FILE_PSTRING) 1804 goto bad; 1805 m->str_flags |= PSTRING_LENGTH_INCLUDES_ITSELF; 1806 break; 1807 default: 1808 bad: 1809 if (ms->flags & MAGIC_CHECK) 1810 file_magwarn(ms, "string modifier `%c' " 1811 "invalid", *l); 1812 goto out; 1813 } 1814 /* allow multiple '/' for readability */ 1815 if (l[1] == '/' && !isspace(CAST(unsigned char, l[2]))) 1816 l++; 1817 } 1818 if (string_modifier_check(ms, m) == -1) 1819 goto out; 1820 *lp = l; 1821 return 0; 1822 out: 1823 *lp = l; 1824 return -1; 1825 } 1826 1827 /* 1828 * parse one line from magic file, put into magic[index++] if valid 1829 */ 1830 private int 1831 parse(struct magic_set *ms, struct magic_entry *me, const char *line, 1832 size_t lineno, int action) 1833 { 1834 #ifdef ENABLE_CONDITIONALS 1835 static uint32_t last_cont_level = 0; 1836 #endif 1837 size_t i; 1838 struct magic *m; 1839 const char *l = line; 1840 char *t; 1841 int op; 1842 uint32_t cont_level; 1843 int32_t diff; 1844 1845 cont_level = 0; 1846 1847 /* 1848 * Parse the offset. 1849 */ 1850 while (*l == '>') { 1851 ++l; /* step over */ 1852 cont_level++; 1853 } 1854 #ifdef ENABLE_CONDITIONALS 1855 if (cont_level == 0 || cont_level > last_cont_level) 1856 if (file_check_mem(ms, cont_level) == -1) 1857 return -1; 1858 last_cont_level = cont_level; 1859 #endif 1860 if (cont_level != 0) { 1861 if (me->mp == NULL) { 1862 file_magerror(ms, "No current entry for continuation"); 1863 return -1; 1864 } 1865 if (me->cont_count == 0) { 1866 file_magerror(ms, "Continuations present with 0 count"); 1867 return -1; 1868 } 1869 m = &me->mp[me->cont_count - 1]; 1870 diff = CAST(int32_t, cont_level) - CAST(int32_t, m->cont_level); 1871 if (diff > 1) 1872 file_magwarn(ms, "New continuation level %u is more " 1873 "than one larger than current level %u", cont_level, 1874 m->cont_level); 1875 if (me->cont_count == me->max_count) { 1876 struct magic *nm; 1877 size_t cnt = me->max_count + ALLOC_CHUNK; 1878 if ((nm = CAST(struct magic *, realloc(me->mp, 1879 sizeof(*nm) * cnt))) == NULL) { 1880 file_oomem(ms, sizeof(*nm) * cnt); 1881 return -1; 1882 } 1883 me->mp = nm; 1884 me->max_count = CAST(uint32_t, cnt); 1885 } 1886 m = &me->mp[me->cont_count++]; 1887 (void)memset(m, 0, sizeof(*m)); 1888 m->cont_level = cont_level; 1889 } else { 1890 static const size_t len = sizeof(*m) * ALLOC_CHUNK; 1891 if (me->mp != NULL) 1892 return 1; 1893 if ((m = CAST(struct magic *, malloc(len))) == NULL) { 1894 file_oomem(ms, len); 1895 return -1; 1896 } 1897 me->mp = m; 1898 me->max_count = ALLOC_CHUNK; 1899 (void)memset(m, 0, sizeof(*m)); 1900 m->factor_op = FILE_FACTOR_OP_NONE; 1901 m->cont_level = 0; 1902 me->cont_count = 1; 1903 } 1904 m->lineno = CAST(uint32_t, lineno); 1905 1906 if (*l == '&') { /* m->cont_level == 0 checked below. */ 1907 ++l; /* step over */ 1908 m->flag |= OFFADD; 1909 } 1910 if (*l == '(') { 1911 ++l; /* step over */ 1912 m->flag |= INDIR; 1913 if (m->flag & OFFADD) 1914 m->flag = (m->flag & ~OFFADD) | INDIROFFADD; 1915 1916 if (*l == '&') { /* m->cont_level == 0 checked below */ 1917 ++l; /* step over */ 1918 m->flag |= OFFADD; 1919 } 1920 } 1921 /* Indirect offsets are not valid at level 0. */ 1922 if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD))) { 1923 if (ms->flags & MAGIC_CHECK) 1924 file_magwarn(ms, "relative offset at level 0"); 1925 return -1; 1926 } 1927 1928 /* get offset, then skip over it */ 1929 m->offset = CAST(int32_t, strtol(l, &t, 0)); 1930 if (l == t) { 1931 if (ms->flags & MAGIC_CHECK) 1932 file_magwarn(ms, "offset `%s' invalid", l); 1933 return -1; 1934 } 1935 1936 l = t; 1937 1938 if (m->flag & INDIR) { 1939 m->in_type = FILE_LONG; 1940 m->in_offset = 0; 1941 m->in_op = 0; 1942 /* 1943 * read [.,lbs][+-]nnnnn) 1944 */ 1945 if (*l == '.' || *l == ',') { 1946 if (*l == ',') 1947 m->in_op |= FILE_OPSIGNED; 1948 l++; 1949 switch (*l) { 1950 case 'l': 1951 m->in_type = FILE_LELONG; 1952 break; 1953 case 'L': 1954 m->in_type = FILE_BELONG; 1955 break; 1956 case 'm': 1957 m->in_type = FILE_MELONG; 1958 break; 1959 case 'h': 1960 case 's': 1961 m->in_type = FILE_LESHORT; 1962 break; 1963 case 'H': 1964 case 'S': 1965 m->in_type = FILE_BESHORT; 1966 break; 1967 case 'c': 1968 case 'b': 1969 case 'C': 1970 case 'B': 1971 m->in_type = FILE_BYTE; 1972 break; 1973 case 'e': 1974 case 'f': 1975 case 'g': 1976 m->in_type = FILE_LEDOUBLE; 1977 break; 1978 case 'E': 1979 case 'F': 1980 case 'G': 1981 m->in_type = FILE_BEDOUBLE; 1982 break; 1983 case 'i': 1984 m->in_type = FILE_LEID3; 1985 break; 1986 case 'I': 1987 m->in_type = FILE_BEID3; 1988 break; 1989 case 'q': 1990 m->in_type = FILE_LEQUAD; 1991 break; 1992 case 'Q': 1993 m->in_type = FILE_BEQUAD; 1994 break; 1995 default: 1996 if (ms->flags & MAGIC_CHECK) 1997 file_magwarn(ms, 1998 "indirect offset type `%c' invalid", 1999 *l); 2000 return -1; 2001 } 2002 l++; 2003 } 2004 2005 if (*l == '~') { 2006 m->in_op |= FILE_OPINVERSE; 2007 l++; 2008 } 2009 if ((op = get_op(*l)) != -1) { 2010 m->in_op |= op; 2011 l++; 2012 } 2013 if (*l == '(') { 2014 m->in_op |= FILE_OPINDIRECT; 2015 l++; 2016 } 2017 if (isdigit(CAST(unsigned char, *l)) || *l == '-') { 2018 m->in_offset = CAST(int32_t, strtol(l, &t, 0)); 2019 if (l == t) { 2020 if (ms->flags & MAGIC_CHECK) 2021 file_magwarn(ms, 2022 "in_offset `%s' invalid", l); 2023 return -1; 2024 } 2025 l = t; 2026 } 2027 if (*l++ != ')' || 2028 ((m->in_op & FILE_OPINDIRECT) && *l++ != ')')) { 2029 if (ms->flags & MAGIC_CHECK) 2030 file_magwarn(ms, 2031 "missing ')' in indirect offset"); 2032 return -1; 2033 } 2034 } 2035 EATAB; 2036 2037 #ifdef ENABLE_CONDITIONALS 2038 m->cond = get_cond(l, &l); 2039 if (check_cond(ms, m->cond, cont_level) == -1) 2040 return -1; 2041 2042 EATAB; 2043 #endif 2044 2045 /* 2046 * Parse the type. 2047 */ 2048 if (*l == 'u') { 2049 /* 2050 * Try it as a keyword type prefixed by "u"; match what 2051 * follows the "u". If that fails, try it as an SUS 2052 * integer type. 2053 */ 2054 m->type = get_type(type_tbl, l + 1, &l); 2055 if (m->type == FILE_INVALID) { 2056 /* 2057 * Not a keyword type; parse it as an SUS type, 2058 * 'u' possibly followed by a number or C/S/L. 2059 */ 2060 m->type = get_standard_integer_type(l, &l); 2061 } 2062 /* It's unsigned. */ 2063 if (m->type != FILE_INVALID) 2064 m->flag |= UNSIGNED; 2065 } else { 2066 /* 2067 * Try it as a keyword type. If that fails, try it as 2068 * an SUS integer type if it begins with "d" or as an 2069 * SUS string type if it begins with "s". In any case, 2070 * it's not unsigned. 2071 */ 2072 m->type = get_type(type_tbl, l, &l); 2073 if (m->type == FILE_INVALID) { 2074 /* 2075 * Not a keyword type; parse it as an SUS type, 2076 * either 'd' possibly followed by a number or 2077 * C/S/L, or just 's'. 2078 */ 2079 if (*l == 'd') 2080 m->type = get_standard_integer_type(l, &l); 2081 else if (*l == 's' 2082 && !isalpha(CAST(unsigned char, l[1]))) { 2083 m->type = FILE_STRING; 2084 ++l; 2085 } 2086 } 2087 } 2088 2089 if (m->type == FILE_INVALID) { 2090 /* Not found - try it as a special keyword. */ 2091 m->type = get_type(special_tbl, l, &l); 2092 } 2093 2094 if (m->type == FILE_INVALID) { 2095 if (ms->flags & MAGIC_CHECK) 2096 file_magwarn(ms, "type `%s' invalid", l); 2097 return -1; 2098 } 2099 2100 /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */ 2101 /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */ 2102 2103 m->mask_op = 0; 2104 if (*l == '~') { 2105 if (!IS_STRING(m->type)) 2106 m->mask_op |= FILE_OPINVERSE; 2107 else if (ms->flags & MAGIC_CHECK) 2108 file_magwarn(ms, "'~' invalid for string types"); 2109 ++l; 2110 } 2111 m->str_range = 0; 2112 m->str_flags = m->type == FILE_PSTRING ? PSTRING_1_LE : 0; 2113 if ((op = get_op(*l)) != -1) { 2114 if (IS_STRING(m->type)) { 2115 int r; 2116 2117 if (op != FILE_OPDIVIDE) { 2118 if (ms->flags & MAGIC_CHECK) 2119 file_magwarn(ms, 2120 "invalid string/indirect op: " 2121 "`%c'", *t); 2122 return -1; 2123 } 2124 2125 if (m->type == FILE_INDIRECT) 2126 r = parse_indirect_modifier(ms, m, &l); 2127 else 2128 r = parse_string_modifier(ms, m, &l); 2129 if (r == -1) 2130 return -1; 2131 } else 2132 parse_op_modifier(ms, m, &l, op); 2133 } 2134 2135 /* 2136 * We used to set mask to all 1's here, instead let's just not do 2137 * anything if mask = 0 (unless you have a better idea) 2138 */ 2139 EATAB; 2140 2141 switch (*l) { 2142 case '>': 2143 case '<': 2144 m->reln = *l; 2145 ++l; 2146 if (*l == '=') { 2147 if (ms->flags & MAGIC_CHECK) { 2148 file_magwarn(ms, "%c= not supported", 2149 m->reln); 2150 return -1; 2151 } 2152 ++l; 2153 } 2154 break; 2155 /* Old-style anding: "0 byte &0x80 dynamically linked" */ 2156 case '&': 2157 case '^': 2158 case '=': 2159 m->reln = *l; 2160 ++l; 2161 if (*l == '=') { 2162 /* HP compat: ignore &= etc. */ 2163 ++l; 2164 } 2165 break; 2166 case '!': 2167 m->reln = *l; 2168 ++l; 2169 break; 2170 default: 2171 m->reln = '='; /* the default relation */ 2172 if (*l == 'x' && ((isascii(CAST(unsigned char, l[1])) && 2173 isspace(CAST(unsigned char, l[1]))) || !l[1])) { 2174 m->reln = *l; 2175 ++l; 2176 } 2177 break; 2178 } 2179 /* 2180 * Grab the value part, except for an 'x' reln. 2181 */ 2182 if (m->reln != 'x' && getvalue(ms, m, &l, action)) 2183 return -1; 2184 2185 /* 2186 * TODO finish this macro and start using it! 2187 * #define offsetcheck {if (offset > ms->bytes_max -1) 2188 * magwarn("offset too big"); } 2189 */ 2190 2191 /* 2192 * Now get last part - the description 2193 */ 2194 EATAB; 2195 if (l[0] == '\b') { 2196 ++l; 2197 m->flag |= NOSPACE; 2198 } else if ((l[0] == '\\') && (l[1] == 'b')) { 2199 ++l; 2200 ++l; 2201 m->flag |= NOSPACE; 2202 } 2203 for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); ) 2204 continue; 2205 if (i == sizeof(m->desc)) { 2206 m->desc[sizeof(m->desc) - 1] = '\0'; 2207 if (ms->flags & MAGIC_CHECK) 2208 file_magwarn(ms, "description `%s' truncated", m->desc); 2209 } 2210 2211 /* 2212 * We only do this check while compiling, or if any of the magic 2213 * files were not compiled. 2214 */ 2215 if (ms->flags & MAGIC_CHECK) { 2216 if (check_format(ms, m) == -1) 2217 return -1; 2218 } 2219 #ifndef COMPILE_ONLY 2220 if (action == FILE_CHECK) { 2221 file_mdump(m); 2222 } 2223 #endif 2224 m->mimetype[0] = '\0'; /* initialise MIME type to none */ 2225 return 0; 2226 } 2227 2228 /* 2229 * parse a STRENGTH annotation line from magic file, put into magic[index - 1] 2230 * if valid 2231 */ 2232 private int 2233 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line) 2234 { 2235 const char *l = line; 2236 char *el; 2237 unsigned long factor; 2238 struct magic *m = &me->mp[0]; 2239 2240 if (m->factor_op != FILE_FACTOR_OP_NONE) { 2241 file_magwarn(ms, 2242 "Current entry already has a strength type: %c %d", 2243 m->factor_op, m->factor); 2244 return -1; 2245 } 2246 if (m->type == FILE_NAME) { 2247 file_magwarn(ms, "%s: Strength setting is not supported in " 2248 "\"name\" magic entries", m->value.s); 2249 return -1; 2250 } 2251 EATAB; 2252 switch (*l) { 2253 case FILE_FACTOR_OP_NONE: 2254 case FILE_FACTOR_OP_PLUS: 2255 case FILE_FACTOR_OP_MINUS: 2256 case FILE_FACTOR_OP_TIMES: 2257 case FILE_FACTOR_OP_DIV: 2258 m->factor_op = *l++; 2259 break; 2260 default: 2261 file_magwarn(ms, "Unknown factor op `%c'", *l); 2262 return -1; 2263 } 2264 EATAB; 2265 factor = strtoul(l, &el, 0); 2266 if (factor > 255) { 2267 file_magwarn(ms, "Too large factor `%lu'", factor); 2268 goto out; 2269 } 2270 if (*el && !isspace(CAST(unsigned char, *el))) { 2271 file_magwarn(ms, "Bad factor `%s'", l); 2272 goto out; 2273 } 2274 m->factor = CAST(uint8_t, factor); 2275 if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) { 2276 file_magwarn(ms, "Cannot have factor op `%c' and factor %u", 2277 m->factor_op, m->factor); 2278 goto out; 2279 } 2280 return 0; 2281 out: 2282 m->factor_op = FILE_FACTOR_OP_NONE; 2283 m->factor = 0; 2284 return -1; 2285 } 2286 2287 private int 2288 goodchar(unsigned char x, const char *extra) 2289 { 2290 return (isascii(x) && isalnum(x)) || strchr(extra, x); 2291 } 2292 2293 private int 2294 parse_extra(struct magic_set *ms, struct magic_entry *me, const char *line, 2295 off_t off, size_t len, const char *name, const char *extra, int nt) 2296 { 2297 size_t i; 2298 const char *l = line; 2299 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1]; 2300 char *buf = CAST(char *, CAST(void *, m)) + off; 2301 2302 if (buf[0] != '\0') { 2303 len = nt ? strlen(buf) : len; 2304 file_magwarn(ms, "Current entry already has a %s type " 2305 "`%.*s', new type `%s'", name, CAST(int, len), buf, l); 2306 return -1; 2307 } 2308 2309 if (*m->desc == '\0') { 2310 file_magwarn(ms, "Current entry does not yet have a " 2311 "description for adding a %s type", name); 2312 return -1; 2313 } 2314 2315 EATAB; 2316 for (i = 0; *l && i < len && goodchar(*l, extra); buf[i++] = *l++) 2317 continue; 2318 2319 if (i == len && *l) { 2320 if (nt) 2321 buf[len - 1] = '\0'; 2322 if (ms->flags & MAGIC_CHECK) 2323 file_magwarn(ms, "%s type `%s' truncated %" 2324 SIZE_T_FORMAT "u", name, line, i); 2325 } else { 2326 if (!isspace(CAST(unsigned char, *l)) && !goodchar(*l, extra)) 2327 file_magwarn(ms, "%s type `%s' has bad char '%c'", 2328 name, line, *l); 2329 if (nt) 2330 buf[i] = '\0'; 2331 } 2332 2333 if (i > 0) 2334 return 0; 2335 2336 file_magerror(ms, "Bad magic entry '%s'", line); 2337 return -1; 2338 } 2339 2340 /* 2341 * Parse an Apple CREATOR/TYPE annotation from magic file and put it into 2342 * magic[index - 1] 2343 */ 2344 private int 2345 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line) 2346 { 2347 struct magic *m = &me->mp[0]; 2348 2349 return parse_extra(ms, me, line, 2350 CAST(off_t, offsetof(struct magic, apple)), 2351 sizeof(m->apple), "APPLE", "!+-./?", 0); 2352 } 2353 2354 /* 2355 * Parse a comma-separated list of extensions 2356 */ 2357 private int 2358 parse_ext(struct magic_set *ms, struct magic_entry *me, const char *line) 2359 { 2360 struct magic *m = &me->mp[0]; 2361 2362 return parse_extra(ms, me, line, 2363 CAST(off_t, offsetof(struct magic, ext)), 2364 sizeof(m->ext), "EXTENSION", ",!+-/@?_$", 0); 2365 } 2366 2367 /* 2368 * parse a MIME annotation line from magic file, put into magic[index - 1] 2369 * if valid 2370 */ 2371 private int 2372 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line) 2373 { 2374 struct magic *m = &me->mp[0]; 2375 2376 return parse_extra(ms, me, line, 2377 CAST(off_t, offsetof(struct magic, mimetype)), 2378 sizeof(m->mimetype), "MIME", "+-/.$?:{}", 1); 2379 } 2380 2381 private int 2382 check_format_type(const char *ptr, int type, const char **estr) 2383 { 2384 int quad = 0, h; 2385 size_t len, cnt; 2386 if (*ptr == '\0') { 2387 /* Missing format string; bad */ 2388 *estr = "missing format spec"; 2389 return -1; 2390 } 2391 2392 switch (file_formats[type]) { 2393 case FILE_FMT_QUAD: 2394 quad = 1; 2395 /*FALLTHROUGH*/ 2396 case FILE_FMT_NUM: 2397 if (quad == 0) { 2398 switch (type) { 2399 case FILE_BYTE: 2400 h = 2; 2401 break; 2402 case FILE_SHORT: 2403 case FILE_BESHORT: 2404 case FILE_LESHORT: 2405 h = 1; 2406 break; 2407 case FILE_LONG: 2408 case FILE_BELONG: 2409 case FILE_LELONG: 2410 case FILE_MELONG: 2411 case FILE_LEID3: 2412 case FILE_BEID3: 2413 case FILE_INDIRECT: 2414 h = 0; 2415 break; 2416 default: 2417 abort(); 2418 } 2419 } else 2420 h = 0; 2421 if (*ptr == '-') 2422 ptr++; 2423 if (*ptr == '.') 2424 ptr++; 2425 if (*ptr == '#') 2426 ptr++; 2427 #define CHECKLEN() do { \ 2428 for (len = cnt = 0; isdigit(CAST(unsigned char, *ptr)); ptr++, cnt++) \ 2429 len = len * 10 + (*ptr - '0'); \ 2430 if (cnt > 5 || len > 1024) \ 2431 goto toolong; \ 2432 } while (/*CONSTCOND*/0) 2433 2434 CHECKLEN(); 2435 if (*ptr == '.') 2436 ptr++; 2437 CHECKLEN(); 2438 if (quad) { 2439 if (*ptr++ != 'l') 2440 goto invalid; 2441 if (*ptr++ != 'l') 2442 goto invalid; 2443 } 2444 2445 switch (*ptr++) { 2446 #ifdef STRICT_FORMAT /* "long" formats are int formats for us */ 2447 /* so don't accept the 'l' modifier */ 2448 case 'l': 2449 switch (*ptr++) { 2450 case 'i': 2451 case 'd': 2452 case 'u': 2453 case 'o': 2454 case 'x': 2455 case 'X': 2456 if (h == 0) 2457 return 0; 2458 /*FALLTHROUGH*/ 2459 default: 2460 goto invalid; 2461 } 2462 2463 /* 2464 * Don't accept h and hh modifiers. They make writing 2465 * magic entries more complicated, for very little benefit 2466 */ 2467 case 'h': 2468 if (h-- <= 0) 2469 goto invalid; 2470 switch (*ptr++) { 2471 case 'h': 2472 if (h-- <= 0) 2473 goto invalid; 2474 switch (*ptr++) { 2475 case 'i': 2476 case 'd': 2477 case 'u': 2478 case 'o': 2479 case 'x': 2480 case 'X': 2481 return 0; 2482 default: 2483 goto invalid; 2484 } 2485 case 'i': 2486 case 'd': 2487 case 'u': 2488 case 'o': 2489 case 'x': 2490 case 'X': 2491 if (h == 0) 2492 return 0; 2493 /*FALLTHROUGH*/ 2494 default: 2495 goto invalid; 2496 } 2497 #endif 2498 case 'c': 2499 if (h == 2) 2500 return 0; 2501 goto invalid; 2502 case 'i': 2503 case 'd': 2504 case 'u': 2505 case 'o': 2506 case 'x': 2507 case 'X': 2508 #ifdef STRICT_FORMAT 2509 if (h == 0) 2510 return 0; 2511 /*FALLTHROUGH*/ 2512 #else 2513 return 0; 2514 #endif 2515 default: 2516 goto invalid; 2517 } 2518 2519 case FILE_FMT_FLOAT: 2520 case FILE_FMT_DOUBLE: 2521 if (*ptr == '-') 2522 ptr++; 2523 if (*ptr == '.') 2524 ptr++; 2525 CHECKLEN(); 2526 if (*ptr == '.') 2527 ptr++; 2528 CHECKLEN(); 2529 switch (*ptr++) { 2530 case 'e': 2531 case 'E': 2532 case 'f': 2533 case 'F': 2534 case 'g': 2535 case 'G': 2536 return 0; 2537 2538 default: 2539 goto invalid; 2540 } 2541 2542 2543 case FILE_FMT_STR: 2544 if (*ptr == '-') 2545 ptr++; 2546 while (isdigit(CAST(unsigned char, *ptr))) 2547 ptr++; 2548 if (*ptr == '.') { 2549 ptr++; 2550 while (isdigit(CAST(unsigned char , *ptr))) 2551 ptr++; 2552 } 2553 2554 switch (*ptr++) { 2555 case 's': 2556 return 0; 2557 default: 2558 goto invalid; 2559 } 2560 2561 default: 2562 /* internal error */ 2563 abort(); 2564 } 2565 invalid: 2566 *estr = "not valid"; 2567 toolong: 2568 *estr = "too long"; 2569 return -1; 2570 } 2571 2572 /* 2573 * Check that the optional printf format in description matches 2574 * the type of the magic. 2575 */ 2576 private int 2577 check_format(struct magic_set *ms, struct magic *m) 2578 { 2579 char *ptr; 2580 const char *estr; 2581 2582 for (ptr = m->desc; *ptr; ptr++) 2583 if (*ptr == '%') 2584 break; 2585 if (*ptr == '\0') { 2586 /* No format string; ok */ 2587 return 1; 2588 } 2589 2590 assert(file_nformats == file_nnames); 2591 2592 if (m->type >= file_nformats) { 2593 file_magwarn(ms, "Internal error inconsistency between " 2594 "m->type and format strings"); 2595 return -1; 2596 } 2597 if (file_formats[m->type] == FILE_FMT_NONE) { 2598 file_magwarn(ms, "No format string for `%s' with description " 2599 "`%s'", m->desc, file_names[m->type]); 2600 return -1; 2601 } 2602 2603 ptr++; 2604 if (check_format_type(ptr, m->type, &estr) == -1) { 2605 /* 2606 * TODO: this error message is unhelpful if the format 2607 * string is not one character long 2608 */ 2609 file_magwarn(ms, "Printf format is %s for type " 2610 "`%s' in description `%s'", estr, 2611 file_names[m->type], m->desc); 2612 return -1; 2613 } 2614 2615 for (; *ptr; ptr++) { 2616 if (*ptr == '%') { 2617 file_magwarn(ms, 2618 "Too many format strings (should have at most one) " 2619 "for `%s' with description `%s'", 2620 file_names[m->type], m->desc); 2621 return -1; 2622 } 2623 } 2624 return 0; 2625 } 2626 2627 /* 2628 * Read a numeric value from a pointer, into the value union of a magic 2629 * pointer, according to the magic type. Update the string pointer to point 2630 * just after the number read. Return 0 for success, non-zero for failure. 2631 */ 2632 private int 2633 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action) 2634 { 2635 char *ep; 2636 uint64_t ull; 2637 2638 switch (m->type) { 2639 case FILE_BESTRING16: 2640 case FILE_LESTRING16: 2641 case FILE_STRING: 2642 case FILE_PSTRING: 2643 case FILE_REGEX: 2644 case FILE_SEARCH: 2645 case FILE_NAME: 2646 case FILE_USE: 2647 case FILE_DER: 2648 *p = getstr(ms, m, *p, action == FILE_COMPILE); 2649 if (*p == NULL) { 2650 if (ms->flags & MAGIC_CHECK) 2651 file_magwarn(ms, "cannot get string from `%s'", 2652 m->value.s); 2653 return -1; 2654 } 2655 if (m->type == FILE_REGEX) { 2656 file_regex_t rx; 2657 int rc = file_regcomp(&rx, m->value.s, REG_EXTENDED); 2658 if (rc) { 2659 if (ms->flags & MAGIC_CHECK) 2660 file_regerror(&rx, rc, ms); 2661 } 2662 file_regfree(&rx); 2663 return rc ? -1 : 0; 2664 } 2665 return 0; 2666 default: 2667 if (m->reln == 'x') 2668 return 0; 2669 break; 2670 } 2671 2672 switch (m->type) { 2673 case FILE_FLOAT: 2674 case FILE_BEFLOAT: 2675 case FILE_LEFLOAT: 2676 errno = 0; 2677 #ifdef HAVE_STRTOF 2678 m->value.f = strtof(*p, &ep); 2679 #else 2680 m->value.f = (float)strtod(*p, &ep); 2681 #endif 2682 if (errno == 0) 2683 *p = ep; 2684 return 0; 2685 case FILE_DOUBLE: 2686 case FILE_BEDOUBLE: 2687 case FILE_LEDOUBLE: 2688 errno = 0; 2689 m->value.d = strtod(*p, &ep); 2690 if (errno == 0) 2691 *p = ep; 2692 return 0; 2693 default: 2694 errno = 0; 2695 ull = CAST(uint64_t, strtoull(*p, &ep, 0)); 2696 m->value.q = file_signextend(ms, m, ull); 2697 if (*p == ep) { 2698 file_magwarn(ms, "Unparseable number `%s'", *p); 2699 } else { 2700 size_t ts = typesize(m->type); 2701 uint64_t x; 2702 const char *q; 2703 2704 if (ts == CAST(size_t, ~0)) { 2705 file_magwarn(ms, 2706 "Expected numeric type got `%s'", 2707 type_tbl[m->type].name); 2708 } 2709 for (q = *p; isspace(CAST(unsigned char, *q)); q++) 2710 continue; 2711 if (*q == '-') 2712 ull = -CAST(int64_t, ull); 2713 switch (ts) { 2714 case 1: 2715 x = CAST(uint64_t, ull & ~0xffULL); 2716 break; 2717 case 2: 2718 x = CAST(uint64_t, ull & ~0xffffULL); 2719 break; 2720 case 4: 2721 x = CAST(uint64_t, ull & ~0xffffffffULL); 2722 break; 2723 case 8: 2724 x = 0; 2725 break; 2726 default: 2727 abort(); 2728 } 2729 if (x) { 2730 file_magwarn(ms, "Overflow for numeric" 2731 " type `%s' value %#" PRIx64, 2732 type_tbl[m->type].name, ull); 2733 } 2734 } 2735 if (errno == 0) { 2736 *p = ep; 2737 eatsize(p); 2738 } 2739 return 0; 2740 } 2741 } 2742 2743 /* 2744 * Convert a string containing C character escapes. Stop at an unescaped 2745 * space or tab. 2746 * Copy the converted version to "m->value.s", and the length in m->vallen. 2747 * Return updated scan pointer as function result. Warn if set. 2748 */ 2749 private const char * 2750 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn) 2751 { 2752 const char *origs = s; 2753 char *p = m->value.s; 2754 size_t plen = sizeof(m->value.s); 2755 char *origp = p; 2756 char *pmax = p + plen - 1; 2757 int c; 2758 int val; 2759 2760 while ((c = *s++) != '\0') { 2761 if (isspace(CAST(unsigned char, c))) 2762 break; 2763 if (p >= pmax) { 2764 file_error(ms, 0, "string too long: `%s'", origs); 2765 return NULL; 2766 } 2767 if (c == '\\') { 2768 switch(c = *s++) { 2769 2770 case '\0': 2771 if (warn) 2772 file_magwarn(ms, "incomplete escape"); 2773 s--; 2774 goto out; 2775 2776 case '\t': 2777 if (warn) { 2778 file_magwarn(ms, 2779 "escaped tab found, use \\t instead"); 2780 warn = 0; /* already did */ 2781 } 2782 /*FALLTHROUGH*/ 2783 default: 2784 if (warn) { 2785 if (isprint(CAST(unsigned char, c))) { 2786 /* Allow escaping of 2787 * ``relations'' */ 2788 if (strchr("<>&^=!", c) == NULL 2789 && (m->type != FILE_REGEX || 2790 strchr("[]().*?^$|{}", c) 2791 == NULL)) { 2792 file_magwarn(ms, "no " 2793 "need to escape " 2794 "`%c'", c); 2795 } 2796 } else { 2797 file_magwarn(ms, 2798 "unknown escape sequence: " 2799 "\\%03o", c); 2800 } 2801 } 2802 /*FALLTHROUGH*/ 2803 /* space, perhaps force people to use \040? */ 2804 case ' ': 2805 #if 0 2806 /* 2807 * Other things people escape, but shouldn't need to, 2808 * so we disallow them 2809 */ 2810 case '\'': 2811 case '"': 2812 case '?': 2813 #endif 2814 /* Relations */ 2815 case '>': 2816 case '<': 2817 case '&': 2818 case '^': 2819 case '=': 2820 case '!': 2821 /* and baskslash itself */ 2822 case '\\': 2823 *p++ = CAST(char, c); 2824 break; 2825 2826 case 'a': 2827 *p++ = '\a'; 2828 break; 2829 2830 case 'b': 2831 *p++ = '\b'; 2832 break; 2833 2834 case 'f': 2835 *p++ = '\f'; 2836 break; 2837 2838 case 'n': 2839 *p++ = '\n'; 2840 break; 2841 2842 case 'r': 2843 *p++ = '\r'; 2844 break; 2845 2846 case 't': 2847 *p++ = '\t'; 2848 break; 2849 2850 case 'v': 2851 *p++ = '\v'; 2852 break; 2853 2854 /* \ and up to 3 octal digits */ 2855 case '0': 2856 case '1': 2857 case '2': 2858 case '3': 2859 case '4': 2860 case '5': 2861 case '6': 2862 case '7': 2863 val = c - '0'; 2864 c = *s++; /* try for 2 */ 2865 if (c >= '0' && c <= '7') { 2866 val = (val << 3) | (c - '0'); 2867 c = *s++; /* try for 3 */ 2868 if (c >= '0' && c <= '7') 2869 val = (val << 3) | (c-'0'); 2870 else 2871 --s; 2872 } 2873 else 2874 --s; 2875 *p++ = CAST(char, val); 2876 break; 2877 2878 /* \x and up to 2 hex digits */ 2879 case 'x': 2880 val = 'x'; /* Default if no digits */ 2881 c = hextoint(*s++); /* Get next char */ 2882 if (c >= 0) { 2883 val = c; 2884 c = hextoint(*s++); 2885 if (c >= 0) 2886 val = (val << 4) + c; 2887 else 2888 --s; 2889 } else 2890 --s; 2891 *p++ = CAST(char, val); 2892 break; 2893 } 2894 } else 2895 *p++ = CAST(char, c); 2896 } 2897 --s; 2898 out: 2899 *p = '\0'; 2900 m->vallen = CAST(unsigned char, (p - origp)); 2901 if (m->type == FILE_PSTRING) 2902 m->vallen += CAST(unsigned char, file_pstring_length_size(m)); 2903 return s; 2904 } 2905 2906 2907 /* Single hex char to int; -1 if not a hex char. */ 2908 private int 2909 hextoint(int c) 2910 { 2911 if (!isascii(CAST(unsigned char, c))) 2912 return -1; 2913 if (isdigit(CAST(unsigned char, c))) 2914 return c - '0'; 2915 if ((c >= 'a') && (c <= 'f')) 2916 return c + 10 - 'a'; 2917 if (( c>= 'A') && (c <= 'F')) 2918 return c + 10 - 'A'; 2919 return -1; 2920 } 2921 2922 2923 /* 2924 * Print a string containing C character escapes. 2925 */ 2926 protected void 2927 file_showstr(FILE *fp, const char *s, size_t len) 2928 { 2929 char c; 2930 2931 for (;;) { 2932 if (len == ~0U) { 2933 c = *s++; 2934 if (c == '\0') 2935 break; 2936 } 2937 else { 2938 if (len-- == 0) 2939 break; 2940 c = *s++; 2941 } 2942 if (c >= 040 && c <= 0176) /* TODO isprint && !iscntrl */ 2943 (void) fputc(c, fp); 2944 else { 2945 (void) fputc('\\', fp); 2946 switch (c) { 2947 case '\a': 2948 (void) fputc('a', fp); 2949 break; 2950 2951 case '\b': 2952 (void) fputc('b', fp); 2953 break; 2954 2955 case '\f': 2956 (void) fputc('f', fp); 2957 break; 2958 2959 case '\n': 2960 (void) fputc('n', fp); 2961 break; 2962 2963 case '\r': 2964 (void) fputc('r', fp); 2965 break; 2966 2967 case '\t': 2968 (void) fputc('t', fp); 2969 break; 2970 2971 case '\v': 2972 (void) fputc('v', fp); 2973 break; 2974 2975 default: 2976 (void) fprintf(fp, "%.3o", c & 0377); 2977 break; 2978 } 2979 } 2980 } 2981 } 2982 2983 /* 2984 * eatsize(): Eat the size spec from a number [eg. 10UL] 2985 */ 2986 private void 2987 eatsize(const char **p) 2988 { 2989 const char *l = *p; 2990 2991 if (LOWCASE(*l) == 'u') 2992 l++; 2993 2994 switch (LOWCASE(*l)) { 2995 case 'l': /* long */ 2996 case 's': /* short */ 2997 case 'h': /* short */ 2998 case 'b': /* char/byte */ 2999 case 'c': /* char/byte */ 3000 l++; 3001 /*FALLTHROUGH*/ 3002 default: 3003 break; 3004 } 3005 3006 *p = l; 3007 } 3008 3009 /* 3010 * handle a buffer containing a compiled file. 3011 */ 3012 private struct magic_map * 3013 apprentice_buf(struct magic_set *ms, struct magic *buf, size_t len) 3014 { 3015 struct magic_map *map; 3016 3017 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) { 3018 file_oomem(ms, sizeof(*map)); 3019 return NULL; 3020 } 3021 map->len = len; 3022 map->p = buf; 3023 map->type = MAP_TYPE_USER; 3024 if (check_buffer(ms, map, "buffer") != 0) { 3025 apprentice_unmap(map); 3026 return NULL; 3027 } 3028 return map; 3029 } 3030 3031 /* 3032 * handle a compiled file. 3033 */ 3034 3035 private struct magic_map * 3036 apprentice_map(struct magic_set *ms, const char *fn) 3037 { 3038 int fd; 3039 struct stat st; 3040 char *dbname = NULL; 3041 struct magic_map *map; 3042 struct magic_map *rv = NULL; 3043 3044 fd = -1; 3045 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) { 3046 file_oomem(ms, sizeof(*map)); 3047 goto error; 3048 } 3049 map->type = MAP_TYPE_USER; /* unspecified */ 3050 3051 dbname = mkdbname(ms, fn, 0); 3052 if (dbname == NULL) 3053 goto error; 3054 3055 if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1) 3056 goto error; 3057 3058 if (fstat(fd, &st) == -1) { 3059 file_error(ms, errno, "cannot stat `%s'", dbname); 3060 goto error; 3061 } 3062 if (st.st_size < 8 || st.st_size > maxoff_t()) { 3063 file_error(ms, 0, "file `%s' is too %s", dbname, 3064 st.st_size < 8 ? "small" : "large"); 3065 goto error; 3066 } 3067 3068 map->len = CAST(size_t, st.st_size); 3069 #ifdef QUICK 3070 map->type = MAP_TYPE_MMAP; 3071 if ((map->p = mmap(0, CAST(size_t, st.st_size), PROT_READ|PROT_WRITE, 3072 MAP_PRIVATE|MAP_FILE, fd, CAST(off_t, 0))) == MAP_FAILED) { 3073 file_error(ms, errno, "cannot map `%s'", dbname); 3074 goto error; 3075 } 3076 #else 3077 map->type = MAP_TYPE_MALLOC; 3078 if ((map->p = CAST(void *, malloc(map->len))) == NULL) { 3079 file_oomem(ms, map->len); 3080 goto error; 3081 } 3082 if (read(fd, map->p, map->len) != (ssize_t)map->len) { 3083 file_badread(ms); 3084 goto error; 3085 } 3086 #define RET 1 3087 #endif 3088 (void)close(fd); 3089 fd = -1; 3090 3091 if (check_buffer(ms, map, dbname) != 0) { 3092 rv = RCAST(struct magic_map *, -1); 3093 goto error; 3094 } 3095 #ifdef QUICK 3096 if (mprotect(map->p, CAST(size_t, st.st_size), PROT_READ) == -1) { 3097 file_error(ms, errno, "cannot mprotect `%s'", dbname); 3098 goto error; 3099 } 3100 #endif 3101 3102 free(dbname); 3103 return map; 3104 3105 error: 3106 if (fd != -1) 3107 (void)close(fd); 3108 apprentice_unmap(map); 3109 free(dbname); 3110 return rv; 3111 } 3112 3113 private int 3114 check_buffer(struct magic_set *ms, struct magic_map *map, const char *dbname) 3115 { 3116 uint32_t *ptr; 3117 uint32_t entries, nentries; 3118 uint32_t version; 3119 int i, needsbyteswap; 3120 3121 ptr = CAST(uint32_t *, map->p); 3122 if (*ptr != MAGICNO) { 3123 if (swap4(*ptr) != MAGICNO) { 3124 file_error(ms, 0, "bad magic in `%s'", dbname); 3125 return -1; 3126 } 3127 needsbyteswap = 1; 3128 } else 3129 needsbyteswap = 0; 3130 if (needsbyteswap) 3131 version = swap4(ptr[1]); 3132 else 3133 version = ptr[1]; 3134 if (version != VERSIONNO) { 3135 file_error(ms, 0, "File %s supports only version %d magic " 3136 "files. `%s' is version %d", VERSION, 3137 VERSIONNO, dbname, version); 3138 return -1; 3139 } 3140 entries = CAST(uint32_t, map->len / sizeof(struct magic)); 3141 if ((entries * sizeof(struct magic)) != map->len) { 3142 file_error(ms, 0, "Size of `%s' %" SIZE_T_FORMAT "u is not " 3143 "a multiple of %" SIZE_T_FORMAT "u", 3144 dbname, map->len, sizeof(struct magic)); 3145 return -1; 3146 } 3147 map->magic[0] = CAST(struct magic *, map->p) + 1; 3148 nentries = 0; 3149 for (i = 0; i < MAGIC_SETS; i++) { 3150 if (needsbyteswap) 3151 map->nmagic[i] = swap4(ptr[i + 2]); 3152 else 3153 map->nmagic[i] = ptr[i + 2]; 3154 if (i != MAGIC_SETS - 1) 3155 map->magic[i + 1] = map->magic[i] + map->nmagic[i]; 3156 nentries += map->nmagic[i]; 3157 } 3158 if (entries != nentries + 1) { 3159 file_error(ms, 0, "Inconsistent entries in `%s' %u != %u", 3160 dbname, entries, nentries + 1); 3161 return -1; 3162 } 3163 if (needsbyteswap) 3164 for (i = 0; i < MAGIC_SETS; i++) 3165 byteswap(map->magic[i], map->nmagic[i]); 3166 return 0; 3167 } 3168 3169 /* 3170 * handle an mmaped file. 3171 */ 3172 private int 3173 apprentice_compile(struct magic_set *ms, struct magic_map *map, const char *fn) 3174 { 3175 static const size_t nm = sizeof(*map->nmagic) * MAGIC_SETS; 3176 static const size_t m = sizeof(**map->magic); 3177 int fd = -1; 3178 size_t len; 3179 char *dbname; 3180 int rv = -1; 3181 uint32_t i; 3182 union { 3183 struct magic m; 3184 uint32_t h[2 + MAGIC_SETS]; 3185 } hdr; 3186 3187 dbname = mkdbname(ms, fn, 1); 3188 3189 if (dbname == NULL) 3190 goto out; 3191 3192 if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) 3193 { 3194 file_error(ms, errno, "cannot open `%s'", dbname); 3195 goto out; 3196 } 3197 memset(&hdr, 0, sizeof(hdr)); 3198 hdr.h[0] = MAGICNO; 3199 hdr.h[1] = VERSIONNO; 3200 memcpy(hdr.h + 2, map->nmagic, nm); 3201 3202 if (write(fd, &hdr, sizeof(hdr)) != CAST(ssize_t, sizeof(hdr))) { 3203 file_error(ms, errno, "error writing `%s'", dbname); 3204 goto out2; 3205 } 3206 3207 for (i = 0; i < MAGIC_SETS; i++) { 3208 len = m * map->nmagic[i]; 3209 if (write(fd, map->magic[i], len) != CAST(ssize_t, len)) { 3210 file_error(ms, errno, "error writing `%s'", dbname); 3211 goto out2; 3212 } 3213 } 3214 3215 rv = 0; 3216 out2: 3217 if (fd != -1) 3218 (void)close(fd); 3219 out: 3220 apprentice_unmap(map); 3221 free(dbname); 3222 return rv; 3223 } 3224 3225 private const char ext[] = ".mgc"; 3226 /* 3227 * make a dbname 3228 */ 3229 private char * 3230 mkdbname(struct magic_set *ms, const char *fn, int strip) 3231 { 3232 const char *p, *q; 3233 char *buf; 3234 3235 if (strip) { 3236 if ((p = strrchr(fn, '/')) != NULL) 3237 fn = ++p; 3238 } 3239 3240 for (q = fn; *q; q++) 3241 continue; 3242 /* Look for .mgc */ 3243 for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--) 3244 if (*p != *q) 3245 break; 3246 3247 /* Did not find .mgc, restore q */ 3248 if (p >= ext) 3249 while (*q) 3250 q++; 3251 3252 q++; 3253 /* Compatibility with old code that looked in .mime */ 3254 if (ms->flags & MAGIC_MIME) { 3255 if (asprintf(&buf, "%.*s.mime%s", CAST(int, q - fn), fn, ext) 3256 < 0) 3257 return NULL; 3258 if (access(buf, R_OK) != -1) { 3259 ms->flags &= MAGIC_MIME_TYPE; 3260 return buf; 3261 } 3262 free(buf); 3263 } 3264 if (asprintf(&buf, "%.*s%s", CAST(int, q - fn), fn, ext) < 0) 3265 return NULL; 3266 3267 /* Compatibility with old code that looked in .mime */ 3268 if (strstr(fn, ".mime") != NULL) 3269 ms->flags &= MAGIC_MIME_TYPE; 3270 return buf; 3271 } 3272 3273 /* 3274 * Byteswap an mmap'ed file if needed 3275 */ 3276 private void 3277 byteswap(struct magic *magic, uint32_t nmagic) 3278 { 3279 uint32_t i; 3280 for (i = 0; i < nmagic; i++) 3281 bs1(&magic[i]); 3282 } 3283 3284 /* 3285 * swap a short 3286 */ 3287 private uint16_t 3288 swap2(uint16_t sv) 3289 { 3290 uint16_t rv; 3291 uint8_t *s = RCAST(uint8_t *, RCAST(void *, &sv)); 3292 uint8_t *d = RCAST(uint8_t *, RCAST(void *, &rv)); 3293 d[0] = s[1]; 3294 d[1] = s[0]; 3295 return rv; 3296 } 3297 3298 /* 3299 * swap an int 3300 */ 3301 private uint32_t 3302 swap4(uint32_t sv) 3303 { 3304 uint32_t rv; 3305 uint8_t *s = RCAST(uint8_t *, RCAST(void *, &sv)); 3306 uint8_t *d = RCAST(uint8_t *, RCAST(void *, &rv)); 3307 d[0] = s[3]; 3308 d[1] = s[2]; 3309 d[2] = s[1]; 3310 d[3] = s[0]; 3311 return rv; 3312 } 3313 3314 /* 3315 * swap a quad 3316 */ 3317 private uint64_t 3318 swap8(uint64_t sv) 3319 { 3320 uint64_t rv; 3321 uint8_t *s = RCAST(uint8_t *, RCAST(void *, &sv)); 3322 uint8_t *d = RCAST(uint8_t *, RCAST(void *, &rv)); 3323 #if 0 3324 d[0] = s[3]; 3325 d[1] = s[2]; 3326 d[2] = s[1]; 3327 d[3] = s[0]; 3328 d[4] = s[7]; 3329 d[5] = s[6]; 3330 d[6] = s[5]; 3331 d[7] = s[4]; 3332 #else 3333 d[0] = s[7]; 3334 d[1] = s[6]; 3335 d[2] = s[5]; 3336 d[3] = s[4]; 3337 d[4] = s[3]; 3338 d[5] = s[2]; 3339 d[6] = s[1]; 3340 d[7] = s[0]; 3341 #endif 3342 return rv; 3343 } 3344 3345 /* 3346 * byteswap a single magic entry 3347 */ 3348 private void 3349 bs1(struct magic *m) 3350 { 3351 m->cont_level = swap2(m->cont_level); 3352 m->offset = swap4(CAST(uint32_t, m->offset)); 3353 m->in_offset = swap4(CAST(uint32_t, m->in_offset)); 3354 m->lineno = swap4(CAST(uint32_t, m->lineno)); 3355 if (IS_STRING(m->type)) { 3356 m->str_range = swap4(m->str_range); 3357 m->str_flags = swap4(m->str_flags); 3358 } 3359 else { 3360 m->value.q = swap8(m->value.q); 3361 m->num_mask = swap8(m->num_mask); 3362 } 3363 } 3364 3365 protected size_t 3366 file_pstring_length_size(const struct magic *m) 3367 { 3368 switch (m->str_flags & PSTRING_LEN) { 3369 case PSTRING_1_LE: 3370 return 1; 3371 case PSTRING_2_LE: 3372 case PSTRING_2_BE: 3373 return 2; 3374 case PSTRING_4_LE: 3375 case PSTRING_4_BE: 3376 return 4; 3377 default: 3378 abort(); /* Impossible */ 3379 return 1; 3380 } 3381 } 3382 protected size_t 3383 file_pstring_get_length(const struct magic *m, const char *ss) 3384 { 3385 size_t len = 0; 3386 const unsigned char *s = RCAST(const unsigned char *, ss); 3387 unsigned int s3, s2, s1, s0; 3388 3389 switch (m->str_flags & PSTRING_LEN) { 3390 case PSTRING_1_LE: 3391 len = *s; 3392 break; 3393 case PSTRING_2_LE: 3394 s0 = s[0]; 3395 s1 = s[1]; 3396 len = (s1 << 8) | s0; 3397 break; 3398 case PSTRING_2_BE: 3399 s0 = s[0]; 3400 s1 = s[1]; 3401 len = (s0 << 8) | s1; 3402 break; 3403 case PSTRING_4_LE: 3404 s0 = s[0]; 3405 s1 = s[1]; 3406 s2 = s[2]; 3407 s3 = s[3]; 3408 len = (s3 << 24) | (s2 << 16) | (s1 << 8) | s0; 3409 break; 3410 case PSTRING_4_BE: 3411 s0 = s[0]; 3412 s1 = s[1]; 3413 s2 = s[2]; 3414 s3 = s[3]; 3415 len = (s0 << 24) | (s1 << 16) | (s2 << 8) | s3; 3416 break; 3417 default: 3418 abort(); /* Impossible */ 3419 } 3420 3421 if (m->str_flags & PSTRING_LENGTH_INCLUDES_ITSELF) 3422 len -= file_pstring_length_size(m); 3423 3424 return len; 3425 } 3426 3427 protected int 3428 file_magicfind(struct magic_set *ms, const char *name, struct mlist *v) 3429 { 3430 uint32_t i, j; 3431 struct mlist *mlist, *ml; 3432 3433 mlist = ms->mlist[1]; 3434 3435 for (ml = mlist->next; ml != mlist; ml = ml->next) { 3436 struct magic *ma = ml->magic; 3437 uint32_t nma = ml->nmagic; 3438 for (i = 0; i < nma; i++) { 3439 if (ma[i].type != FILE_NAME) 3440 continue; 3441 if (strcmp(ma[i].value.s, name) == 0) { 3442 v->magic = &ma[i]; 3443 for (j = i + 1; j < nma; j++) 3444 if (ma[j].cont_level == 0) 3445 break; 3446 v->nmagic = j - i; 3447 return 0; 3448 } 3449 } 3450 } 3451 return -1; 3452 } 3453