1 /* $NetBSD: softmagic.c,v 1.12 2014/10/10 20:15:02 christos Exp $ */ 2 /* 3 * Copyright (c) Ian F. Darwin 1986-1995. 4 * Software written by Ian F. Darwin and others; 5 * maintained 1995-present by Christos Zoulas and others. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice immediately at the beginning of the file, without modification, 12 * this list of conditions, and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR 21 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 */ 29 /* 30 * softmagic - interpret variable magic from MAGIC 31 */ 32 33 #include "file.h" 34 35 #ifndef lint 36 #if 0 37 FILE_RCSID("@(#)$File: softmagic.c,v 1.195 2014/09/24 19:49:07 christos Exp $") 38 #else 39 __RCSID("$NetBSD: softmagic.c,v 1.12 2014/10/10 20:15:02 christos Exp $"); 40 #endif 41 #endif /* lint */ 42 43 #include "magic.h" 44 #include <assert.h> 45 #include <string.h> 46 #include <ctype.h> 47 #include <stdlib.h> 48 #include <time.h> 49 50 private int match(struct magic_set *, struct magic *, uint32_t, 51 const unsigned char *, size_t, size_t, int, int, int, int, int *, int *, 52 int *); 53 private int mget(struct magic_set *, const unsigned char *, 54 struct magic *, size_t, size_t, unsigned int, int, int, int, int, int *, 55 int *, int *); 56 private int magiccheck(struct magic_set *, struct magic *); 57 private int32_t mprint(struct magic_set *, struct magic *); 58 private int32_t moffset(struct magic_set *, struct magic *); 59 private void mdebug(uint32_t, const char *, size_t); 60 private int mcopy(struct magic_set *, union VALUETYPE *, int, int, 61 const unsigned char *, uint32_t, size_t, struct magic *); 62 private int mconvert(struct magic_set *, struct magic *, int); 63 private int print_sep(struct magic_set *, int); 64 private int handle_annotation(struct magic_set *, struct magic *); 65 private void cvt_8(union VALUETYPE *, const struct magic *); 66 private void cvt_16(union VALUETYPE *, const struct magic *); 67 private void cvt_32(union VALUETYPE *, const struct magic *); 68 private void cvt_64(union VALUETYPE *, const struct magic *); 69 70 #define OFFSET_OOB(n, o, i) ((n) < (o) || (i) > ((n) - (o))) 71 /* 72 * softmagic - lookup one file in parsed, in-memory copy of database 73 * Passed the name and FILE * of one file to be typed. 74 */ 75 /*ARGSUSED1*/ /* nbytes passed for regularity, maybe need later */ 76 protected int 77 file_softmagic(struct magic_set *ms, const unsigned char *buf, size_t nbytes, 78 size_t level, int mode, int text) 79 { 80 struct mlist *ml; 81 int rv, printed_something = 0, need_separator = 0; 82 83 for (ml = ms->mlist[0]->next; ml != ms->mlist[0]; ml = ml->next) 84 if ((rv = match(ms, ml->magic, ml->nmagic, buf, nbytes, 0, mode, 85 text, 0, level, &printed_something, &need_separator, 86 NULL)) != 0) 87 return rv; 88 89 return 0; 90 } 91 92 #define FILE_FMTDEBUG 93 #ifdef FILE_FMTDEBUG 94 #define F(a, b, c) file_fmtcheck((a), (b), (c), __FILE__, __LINE__) 95 96 private const char * __attribute__((__format_arg__(3))) 97 file_fmtcheck(struct magic_set *ms, const struct magic *m, const char *def, 98 const char *file, size_t line) 99 { 100 const char *ptr = fmtcheck(m->desc, def); 101 if (ptr == def) 102 file_magerror(ms, 103 "%s, %" SIZE_T_FORMAT "u: format `%s' does not match" 104 " with `%s'", file, line, m->desc, def); 105 return ptr; 106 } 107 #else 108 #define F(a, b, c) fmtcheck((b)->desc, (c)) 109 #endif 110 111 /* 112 * Go through the whole list, stopping if you find a match. Process all 113 * the continuations of that match before returning. 114 * 115 * We support multi-level continuations: 116 * 117 * At any time when processing a successful top-level match, there is a 118 * current continuation level; it represents the level of the last 119 * successfully matched continuation. 120 * 121 * Continuations above that level are skipped as, if we see one, it 122 * means that the continuation that controls them - i.e, the 123 * lower-level continuation preceding them - failed to match. 124 * 125 * Continuations below that level are processed as, if we see one, 126 * it means we've finished processing or skipping higher-level 127 * continuations under the control of a successful or unsuccessful 128 * lower-level continuation, and are now seeing the next lower-level 129 * continuation and should process it. The current continuation 130 * level reverts to the level of the one we're seeing. 131 * 132 * Continuations at the current level are processed as, if we see 133 * one, there's no lower-level continuation that may have failed. 134 * 135 * If a continuation matches, we bump the current continuation level 136 * so that higher-level continuations are processed. 137 */ 138 private int 139 match(struct magic_set *ms, struct magic *magic, uint32_t nmagic, 140 const unsigned char *s, size_t nbytes, size_t offset, int mode, int text, 141 int flip, int recursion_level, int *printed_something, int *need_separator, 142 int *returnval) 143 { 144 uint32_t magindex = 0; 145 unsigned int cont_level = 0; 146 int returnvalv = 0, e; /* if a match is found it is set to 1*/ 147 int firstline = 1; /* a flag to print X\n X\n- X */ 148 int print = (ms->flags & (MAGIC_MIME|MAGIC_APPLE)) == 0; 149 150 if (returnval == NULL) 151 returnval = &returnvalv; 152 153 if (file_check_mem(ms, cont_level) == -1) 154 return -1; 155 156 for (magindex = 0; magindex < nmagic; magindex++) { 157 int flush = 0; 158 struct magic *m = &magic[magindex]; 159 160 if (m->type != FILE_NAME) 161 if ((IS_STRING(m->type) && 162 #define FLT (STRING_BINTEST | STRING_TEXTTEST) 163 ((text && (m->str_flags & FLT) == STRING_BINTEST) || 164 (!text && (m->str_flags & FLT) == STRING_TEXTTEST))) || 165 (m->flag & mode) != mode) { 166 /* Skip sub-tests */ 167 while (magindex + 1 < nmagic && 168 magic[magindex + 1].cont_level != 0 && 169 ++magindex) 170 continue; 171 continue; /* Skip to next top-level test*/ 172 } 173 174 ms->offset = m->offset; 175 ms->line = m->lineno; 176 177 /* if main entry matches, print it... */ 178 switch (mget(ms, s, m, nbytes, offset, cont_level, mode, text, 179 flip, recursion_level + 1, printed_something, 180 need_separator, returnval)) { 181 case -1: 182 return -1; 183 case 0: 184 flush = m->reln != '!'; 185 break; 186 default: 187 if (m->type == FILE_INDIRECT) 188 *returnval = 1; 189 190 switch (magiccheck(ms, m)) { 191 case -1: 192 return -1; 193 case 0: 194 flush++; 195 break; 196 default: 197 flush = 0; 198 break; 199 } 200 break; 201 } 202 if (flush) { 203 /* 204 * main entry didn't match, 205 * flush its continuations 206 */ 207 while (magindex < nmagic - 1 && 208 magic[magindex + 1].cont_level != 0) 209 magindex++; 210 continue; 211 } 212 213 if ((e = handle_annotation(ms, m)) != 0) { 214 *need_separator = 1; 215 *printed_something = 1; 216 *returnval = 1; 217 return e; 218 } 219 /* 220 * If we are going to print something, we'll need to print 221 * a blank before we print something else. 222 */ 223 if (*m->desc) { 224 *need_separator = 1; 225 *printed_something = 1; 226 if (print_sep(ms, firstline) == -1) 227 return -1; 228 } 229 230 231 if (print && mprint(ms, m) == -1) 232 return -1; 233 234 ms->c.li[cont_level].off = moffset(ms, m); 235 236 /* and any continuations that match */ 237 if (file_check_mem(ms, ++cont_level) == -1) 238 return -1; 239 240 while (magindex + 1 < nmagic && 241 magic[magindex + 1].cont_level != 0) { 242 m = &magic[++magindex]; 243 ms->line = m->lineno; /* for messages */ 244 245 if (cont_level < m->cont_level) 246 continue; 247 if (cont_level > m->cont_level) { 248 /* 249 * We're at the end of the level 250 * "cont_level" continuations. 251 */ 252 cont_level = m->cont_level; 253 } 254 ms->offset = m->offset; 255 if (m->flag & OFFADD) { 256 ms->offset += 257 ms->c.li[cont_level - 1].off; 258 } 259 260 #ifdef ENABLE_CONDITIONALS 261 if (m->cond == COND_ELSE || 262 m->cond == COND_ELIF) { 263 if (ms->c.li[cont_level].last_match == 1) 264 continue; 265 } 266 #endif 267 switch (mget(ms, s, m, nbytes, offset, cont_level, mode, 268 text, flip, recursion_level + 1, printed_something, 269 need_separator, returnval)) { 270 case -1: 271 return -1; 272 case 0: 273 if (m->reln != '!') 274 continue; 275 flush = 1; 276 break; 277 default: 278 if (m->type == FILE_INDIRECT) 279 *returnval = 1; 280 flush = 0; 281 break; 282 } 283 284 switch (flush ? 1 : magiccheck(ms, m)) { 285 case -1: 286 return -1; 287 case 0: 288 #ifdef ENABLE_CONDITIONALS 289 ms->c.li[cont_level].last_match = 0; 290 #endif 291 break; 292 default: 293 #ifdef ENABLE_CONDITIONALS 294 ms->c.li[cont_level].last_match = 1; 295 #endif 296 if (m->type == FILE_CLEAR) 297 ms->c.li[cont_level].got_match = 0; 298 else if (ms->c.li[cont_level].got_match) { 299 if (m->type == FILE_DEFAULT) 300 break; 301 } else 302 ms->c.li[cont_level].got_match = 1; 303 if ((e = handle_annotation(ms, m)) != 0) { 304 *need_separator = 1; 305 *printed_something = 1; 306 *returnval = 1; 307 return e; 308 } 309 /* 310 * If we are going to print something, 311 * make sure that we have a separator first. 312 */ 313 if (*m->desc) { 314 if (!*printed_something) { 315 *printed_something = 1; 316 if (print_sep(ms, firstline) 317 == -1) 318 return -1; 319 } 320 } 321 /* 322 * This continuation matched. Print 323 * its message, with a blank before it 324 * if the previous item printed and 325 * this item isn't empty. 326 */ 327 /* space if previous printed */ 328 if (*need_separator 329 && ((m->flag & NOSPACE) == 0) 330 && *m->desc) { 331 if (print && 332 file_printf(ms, " ") == -1) 333 return -1; 334 *need_separator = 0; 335 } 336 if (print && mprint(ms, m) == -1) 337 return -1; 338 339 ms->c.li[cont_level].off = moffset(ms, m); 340 341 if (*m->desc) 342 *need_separator = 1; 343 344 /* 345 * If we see any continuations 346 * at a higher level, 347 * process them. 348 */ 349 if (file_check_mem(ms, ++cont_level) == -1) 350 return -1; 351 break; 352 } 353 } 354 if (*printed_something) { 355 firstline = 0; 356 if (print) 357 *returnval = 1; 358 } 359 if ((ms->flags & MAGIC_CONTINUE) == 0 && *printed_something) { 360 return *returnval; /* don't keep searching */ 361 } 362 } 363 return *returnval; /* This is hit if -k is set or there is no match */ 364 } 365 366 private int 367 check_fmt(struct magic_set *ms, struct magic *m) 368 { 369 file_regex_t rx; 370 int rc, rv = -1; 371 372 if (strchr(m->desc, '%') == NULL) 373 return 0; 374 375 rc = file_regcomp(&rx, "%[-0-9\\.]*s", REG_EXTENDED|REG_NOSUB); 376 if (rc) { 377 file_regerror(&rx, rc, ms); 378 } else { 379 rc = file_regexec(&rx, m->desc, 0, 0, 0); 380 rv = !rc; 381 } 382 file_regfree(&rx); 383 return rv; 384 } 385 386 #ifndef HAVE_STRNDUP 387 char * strndup(const char *, size_t); 388 389 char * 390 strndup(const char *str, size_t n) 391 { 392 size_t len; 393 char *copy; 394 395 for (len = 0; len < n && str[len]; len++) 396 continue; 397 if ((copy = malloc(len + 1)) == NULL) 398 return NULL; 399 (void)memcpy(copy, str, len); 400 copy[len] = '\0'; 401 return copy; 402 } 403 #endif /* HAVE_STRNDUP */ 404 405 private int32_t 406 mprint(struct magic_set *ms, struct magic *m) 407 { 408 uint64_t v; 409 float vf; 410 double vd; 411 int64_t t = 0; 412 char buf[128], tbuf[26]; 413 union VALUETYPE *p = &ms->ms_value; 414 415 switch (m->type) { 416 case FILE_BYTE: 417 v = file_signextend(ms, m, (uint64_t)p->b); 418 switch (check_fmt(ms, m)) { 419 case -1: 420 return -1; 421 case 1: 422 (void)snprintf(buf, sizeof(buf), "%d", 423 (unsigned char)v); 424 if (file_printf(ms, F(ms, m, "%s"), buf) == -1) 425 return -1; 426 break; 427 default: 428 if (file_printf(ms, F(ms, m, "%d"), 429 (unsigned char) v) == -1) 430 return -1; 431 break; 432 } 433 t = ms->offset + sizeof(char); 434 break; 435 436 case FILE_SHORT: 437 case FILE_BESHORT: 438 case FILE_LESHORT: 439 v = file_signextend(ms, m, (uint64_t)p->h); 440 switch (check_fmt(ms, m)) { 441 case -1: 442 return -1; 443 case 1: 444 (void)snprintf(buf, sizeof(buf), "%u", 445 (unsigned short)v); 446 if (file_printf(ms, F(ms, m, "%s"), buf) == -1) 447 return -1; 448 break; 449 default: 450 if (file_printf(ms, F(ms, m, "%u"), 451 (unsigned short) v) == -1) 452 return -1; 453 break; 454 } 455 t = ms->offset + sizeof(short); 456 break; 457 458 case FILE_LONG: 459 case FILE_BELONG: 460 case FILE_LELONG: 461 case FILE_MELONG: 462 v = file_signextend(ms, m, (uint64_t)p->l); 463 switch (check_fmt(ms, m)) { 464 case -1: 465 return -1; 466 case 1: 467 (void)snprintf(buf, sizeof(buf), "%u", (uint32_t) v); 468 if (file_printf(ms, F(ms, m, "%s"), buf) == -1) 469 return -1; 470 break; 471 default: 472 if (file_printf(ms, F(ms, m, "%u"), (uint32_t) v) == -1) 473 return -1; 474 break; 475 } 476 t = ms->offset + sizeof(int32_t); 477 break; 478 479 case FILE_QUAD: 480 case FILE_BEQUAD: 481 case FILE_LEQUAD: 482 v = file_signextend(ms, m, p->q); 483 switch (check_fmt(ms, m)) { 484 case -1: 485 return -1; 486 case 1: 487 (void)snprintf(buf, sizeof(buf), "%" INT64_T_FORMAT "u", 488 (unsigned long long)v); 489 if (file_printf(ms, F(ms, m, "%s"), buf) == -1) 490 return -1; 491 break; 492 default: 493 if (file_printf(ms, F(ms, m, "%" INT64_T_FORMAT "u"), 494 (unsigned long long) v) == -1) 495 return -1; 496 break; 497 } 498 t = ms->offset + sizeof(int64_t); 499 break; 500 501 case FILE_STRING: 502 case FILE_PSTRING: 503 case FILE_BESTRING16: 504 case FILE_LESTRING16: 505 if (m->reln == '=' || m->reln == '!') { 506 if (file_printf(ms, F(ms, m, "%s"), m->value.s) == -1) 507 return -1; 508 t = ms->offset + m->vallen; 509 } 510 else { 511 char *str = p->s; 512 513 /* compute t before we mangle the string? */ 514 t = ms->offset + strlen(str); 515 516 if (*m->value.s == '\0') 517 str[strcspn(str, "\n")] = '\0'; 518 519 if (m->str_flags & STRING_TRIM) { 520 char *last; 521 while (isspace((unsigned char)*str)) 522 str++; 523 last = str; 524 while (*last) 525 last++; 526 --last; 527 while (isspace((unsigned char)*last)) 528 last--; 529 *++last = '\0'; 530 } 531 532 if (file_printf(ms, F(ms, m, "%s"), str) == -1) 533 return -1; 534 535 if (m->type == FILE_PSTRING) 536 t += file_pstring_length_size(m); 537 } 538 break; 539 540 case FILE_DATE: 541 case FILE_BEDATE: 542 case FILE_LEDATE: 543 case FILE_MEDATE: 544 if (file_printf(ms, F(ms, m, "%s"), 545 file_fmttime(p->l + m->num_mask, FILE_T_LOCAL, tbuf)) == -1) 546 return -1; 547 t = ms->offset + sizeof(uint32_t); 548 break; 549 550 case FILE_LDATE: 551 case FILE_BELDATE: 552 case FILE_LELDATE: 553 case FILE_MELDATE: 554 if (file_printf(ms, F(ms, m, "%s"), 555 file_fmttime(p->l + m->num_mask, 0, tbuf)) == -1) 556 return -1; 557 t = ms->offset + sizeof(uint32_t); 558 break; 559 560 case FILE_QDATE: 561 case FILE_BEQDATE: 562 case FILE_LEQDATE: 563 if (file_printf(ms, F(ms, m, "%s"), 564 file_fmttime(p->q + m->num_mask, FILE_T_LOCAL, tbuf)) == -1) 565 return -1; 566 t = ms->offset + sizeof(uint64_t); 567 break; 568 569 case FILE_QLDATE: 570 case FILE_BEQLDATE: 571 case FILE_LEQLDATE: 572 if (file_printf(ms, F(ms, m, "%s"), 573 file_fmttime(p->q + m->num_mask, 0, tbuf)) == -1) 574 return -1; 575 t = ms->offset + sizeof(uint64_t); 576 break; 577 578 case FILE_QWDATE: 579 case FILE_BEQWDATE: 580 case FILE_LEQWDATE: 581 if (file_printf(ms, F(ms, m, "%s"), 582 file_fmttime(p->q + m->num_mask, FILE_T_WINDOWS, tbuf)) == -1) 583 return -1; 584 t = ms->offset + sizeof(uint64_t); 585 break; 586 587 case FILE_FLOAT: 588 case FILE_BEFLOAT: 589 case FILE_LEFLOAT: 590 vf = p->f; 591 switch (check_fmt(ms, m)) { 592 case -1: 593 return -1; 594 case 1: 595 (void)snprintf(buf, sizeof(buf), "%g", vf); 596 if (file_printf(ms, F(ms, m, "%s"), buf) == -1) 597 return -1; 598 break; 599 default: 600 if (file_printf(ms, F(ms, m, "%g"), vf) == -1) 601 return -1; 602 break; 603 } 604 t = ms->offset + sizeof(float); 605 break; 606 607 case FILE_DOUBLE: 608 case FILE_BEDOUBLE: 609 case FILE_LEDOUBLE: 610 vd = p->d; 611 switch (check_fmt(ms, m)) { 612 case -1: 613 return -1; 614 case 1: 615 (void)snprintf(buf, sizeof(buf), "%g", vd); 616 if (file_printf(ms, F(ms, m, "%s"), buf) == -1) 617 return -1; 618 break; 619 default: 620 if (file_printf(ms, F(ms, m, "%g"), vd) == -1) 621 return -1; 622 break; 623 } 624 t = ms->offset + sizeof(double); 625 break; 626 627 case FILE_REGEX: { 628 char *cp; 629 int rval; 630 631 cp = strndup((const char *)ms->search.s, ms->search.rm_len); 632 if (cp == NULL) { 633 file_oomem(ms, ms->search.rm_len); 634 return -1; 635 } 636 rval = file_printf(ms, F(ms, m, "%s"), cp); 637 free(cp); 638 639 if (rval == -1) 640 return -1; 641 642 if ((m->str_flags & REGEX_OFFSET_START)) 643 t = ms->search.offset; 644 else 645 t = ms->search.offset + ms->search.rm_len; 646 break; 647 } 648 649 case FILE_SEARCH: 650 if (file_printf(ms, F(ms, m, "%s"), m->value.s) == -1) 651 return -1; 652 if ((m->str_flags & REGEX_OFFSET_START)) 653 t = ms->search.offset; 654 else 655 t = ms->search.offset + m->vallen; 656 break; 657 658 case FILE_DEFAULT: 659 case FILE_CLEAR: 660 if (file_printf(ms, "%s", m->desc) == -1) 661 return -1; 662 t = ms->offset; 663 break; 664 665 case FILE_INDIRECT: 666 case FILE_USE: 667 case FILE_NAME: 668 t = ms->offset; 669 break; 670 671 default: 672 file_magerror(ms, "invalid m->type (%d) in mprint()", m->type); 673 return -1; 674 } 675 return (int32_t)t; 676 } 677 678 private int32_t 679 moffset(struct magic_set *ms, struct magic *m) 680 { 681 switch (m->type) { 682 case FILE_BYTE: 683 return CAST(int32_t, (ms->offset + sizeof(char))); 684 685 case FILE_SHORT: 686 case FILE_BESHORT: 687 case FILE_LESHORT: 688 return CAST(int32_t, (ms->offset + sizeof(short))); 689 690 case FILE_LONG: 691 case FILE_BELONG: 692 case FILE_LELONG: 693 case FILE_MELONG: 694 return CAST(int32_t, (ms->offset + sizeof(int32_t))); 695 696 case FILE_QUAD: 697 case FILE_BEQUAD: 698 case FILE_LEQUAD: 699 return CAST(int32_t, (ms->offset + sizeof(int64_t))); 700 701 case FILE_STRING: 702 case FILE_PSTRING: 703 case FILE_BESTRING16: 704 case FILE_LESTRING16: 705 if (m->reln == '=' || m->reln == '!') 706 return ms->offset + m->vallen; 707 else { 708 union VALUETYPE *p = &ms->ms_value; 709 uint32_t t; 710 711 if (*m->value.s == '\0') 712 p->s[strcspn(p->s, "\n")] = '\0'; 713 t = CAST(uint32_t, (ms->offset + strlen(p->s))); 714 if (m->type == FILE_PSTRING) 715 t += (uint32_t)file_pstring_length_size(m); 716 return t; 717 } 718 719 case FILE_DATE: 720 case FILE_BEDATE: 721 case FILE_LEDATE: 722 case FILE_MEDATE: 723 return CAST(int32_t, (ms->offset + sizeof(uint32_t))); 724 725 case FILE_LDATE: 726 case FILE_BELDATE: 727 case FILE_LELDATE: 728 case FILE_MELDATE: 729 return CAST(int32_t, (ms->offset + sizeof(uint32_t))); 730 731 case FILE_QDATE: 732 case FILE_BEQDATE: 733 case FILE_LEQDATE: 734 return CAST(int32_t, (ms->offset + sizeof(uint64_t))); 735 736 case FILE_QLDATE: 737 case FILE_BEQLDATE: 738 case FILE_LEQLDATE: 739 return CAST(int32_t, (ms->offset + sizeof(uint64_t))); 740 741 case FILE_FLOAT: 742 case FILE_BEFLOAT: 743 case FILE_LEFLOAT: 744 return CAST(int32_t, (ms->offset + sizeof(float))); 745 746 case FILE_DOUBLE: 747 case FILE_BEDOUBLE: 748 case FILE_LEDOUBLE: 749 return CAST(int32_t, (ms->offset + sizeof(double))); 750 751 case FILE_REGEX: 752 if ((m->str_flags & REGEX_OFFSET_START) != 0) 753 return CAST(int32_t, ms->search.offset); 754 else 755 return CAST(int32_t, (ms->search.offset + 756 ms->search.rm_len)); 757 758 case FILE_SEARCH: 759 if ((m->str_flags & REGEX_OFFSET_START) != 0) 760 return CAST(int32_t, ms->search.offset); 761 else 762 return CAST(int32_t, (ms->search.offset + m->vallen)); 763 764 case FILE_CLEAR: 765 case FILE_DEFAULT: 766 case FILE_INDIRECT: 767 return ms->offset; 768 769 default: 770 return 0; 771 } 772 } 773 774 private int 775 cvt_flip(int type, int flip) 776 { 777 if (flip == 0) 778 return type; 779 switch (type) { 780 case FILE_BESHORT: 781 return FILE_LESHORT; 782 case FILE_BELONG: 783 return FILE_LELONG; 784 case FILE_BEDATE: 785 return FILE_LEDATE; 786 case FILE_BELDATE: 787 return FILE_LELDATE; 788 case FILE_BEQUAD: 789 return FILE_LEQUAD; 790 case FILE_BEQDATE: 791 return FILE_LEQDATE; 792 case FILE_BEQLDATE: 793 return FILE_LEQLDATE; 794 case FILE_BEQWDATE: 795 return FILE_LEQWDATE; 796 case FILE_LESHORT: 797 return FILE_BESHORT; 798 case FILE_LELONG: 799 return FILE_BELONG; 800 case FILE_LEDATE: 801 return FILE_BEDATE; 802 case FILE_LELDATE: 803 return FILE_BELDATE; 804 case FILE_LEQUAD: 805 return FILE_BEQUAD; 806 case FILE_LEQDATE: 807 return FILE_BEQDATE; 808 case FILE_LEQLDATE: 809 return FILE_BEQLDATE; 810 case FILE_LEQWDATE: 811 return FILE_BEQWDATE; 812 case FILE_BEFLOAT: 813 return FILE_LEFLOAT; 814 case FILE_LEFLOAT: 815 return FILE_BEFLOAT; 816 case FILE_BEDOUBLE: 817 return FILE_LEDOUBLE; 818 case FILE_LEDOUBLE: 819 return FILE_BEDOUBLE; 820 default: 821 return type; 822 } 823 } 824 #define DO_CVT(fld, cast) \ 825 if (m->num_mask) \ 826 switch (m->mask_op & FILE_OPS_MASK) { \ 827 case FILE_OPAND: \ 828 p->fld &= cast m->num_mask; \ 829 break; \ 830 case FILE_OPOR: \ 831 p->fld |= cast m->num_mask; \ 832 break; \ 833 case FILE_OPXOR: \ 834 p->fld ^= cast m->num_mask; \ 835 break; \ 836 case FILE_OPADD: \ 837 p->fld += cast m->num_mask; \ 838 break; \ 839 case FILE_OPMINUS: \ 840 p->fld -= cast m->num_mask; \ 841 break; \ 842 case FILE_OPMULTIPLY: \ 843 p->fld *= cast m->num_mask; \ 844 break; \ 845 case FILE_OPDIVIDE: \ 846 p->fld /= cast m->num_mask; \ 847 break; \ 848 case FILE_OPMODULO: \ 849 p->fld %= cast m->num_mask; \ 850 break; \ 851 } \ 852 if (m->mask_op & FILE_OPINVERSE) \ 853 p->fld = ~p->fld \ 854 855 private void 856 cvt_8(union VALUETYPE *p, const struct magic *m) 857 { 858 DO_CVT(b, (uint8_t)); 859 } 860 861 private void 862 cvt_16(union VALUETYPE *p, const struct magic *m) 863 { 864 DO_CVT(h, (uint16_t)); 865 } 866 867 private void 868 cvt_32(union VALUETYPE *p, const struct magic *m) 869 { 870 DO_CVT(l, (uint32_t)); 871 } 872 873 private void 874 cvt_64(union VALUETYPE *p, const struct magic *m) 875 { 876 DO_CVT(q, (uint64_t)); 877 } 878 879 #define DO_CVT2(fld, cast) \ 880 if (m->num_mask) \ 881 switch (m->mask_op & FILE_OPS_MASK) { \ 882 case FILE_OPADD: \ 883 p->fld += cast m->num_mask; \ 884 break; \ 885 case FILE_OPMINUS: \ 886 p->fld -= cast m->num_mask; \ 887 break; \ 888 case FILE_OPMULTIPLY: \ 889 p->fld *= cast m->num_mask; \ 890 break; \ 891 case FILE_OPDIVIDE: \ 892 p->fld /= cast m->num_mask; \ 893 break; \ 894 } \ 895 896 private void 897 cvt_float(union VALUETYPE *p, const struct magic *m) 898 { 899 DO_CVT2(f, (float)); 900 } 901 902 private void 903 cvt_double(union VALUETYPE *p, const struct magic *m) 904 { 905 DO_CVT2(d, (double)); 906 } 907 908 /* 909 * Convert the byte order of the data we are looking at 910 * While we're here, let's apply the mask operation 911 * (unless you have a better idea) 912 */ 913 private int 914 mconvert(struct magic_set *ms, struct magic *m, int flip) 915 { 916 union VALUETYPE *p = &ms->ms_value; 917 uint8_t type; 918 919 switch (type = cvt_flip(m->type, flip)) { 920 case FILE_BYTE: 921 cvt_8(p, m); 922 return 1; 923 case FILE_SHORT: 924 cvt_16(p, m); 925 return 1; 926 case FILE_LONG: 927 case FILE_DATE: 928 case FILE_LDATE: 929 cvt_32(p, m); 930 return 1; 931 case FILE_QUAD: 932 case FILE_QDATE: 933 case FILE_QLDATE: 934 case FILE_QWDATE: 935 cvt_64(p, m); 936 return 1; 937 case FILE_STRING: 938 case FILE_BESTRING16: 939 case FILE_LESTRING16: { 940 /* Null terminate and eat *trailing* return */ 941 p->s[sizeof(p->s) - 1] = '\0'; 942 return 1; 943 } 944 case FILE_PSTRING: { 945 size_t sz = file_pstring_length_size(m); 946 char *ptr1 = p->s, *ptr2 = ptr1 + sz; 947 size_t len = file_pstring_get_length(m, ptr1); 948 if (len >= sizeof(p->s)) { 949 /* 950 * The size of the pascal string length (sz) 951 * is 1, 2, or 4. We need at least 1 byte for NUL 952 * termination, but we've already truncated the 953 * string by p->s, so we need to deduct sz. 954 */ 955 len = sizeof(p->s) - sz; 956 } 957 while (len--) 958 *ptr1++ = *ptr2++; 959 *ptr1 = '\0'; 960 return 1; 961 } 962 case FILE_BESHORT: 963 p->h = (short)((p->hs[0]<<8)|(p->hs[1])); 964 cvt_16(p, m); 965 return 1; 966 case FILE_BELONG: 967 case FILE_BEDATE: 968 case FILE_BELDATE: 969 p->l = (int32_t) 970 ((p->hl[0]<<24)|(p->hl[1]<<16)|(p->hl[2]<<8)|(p->hl[3])); 971 if (type == FILE_BELONG) 972 cvt_32(p, m); 973 return 1; 974 case FILE_BEQUAD: 975 case FILE_BEQDATE: 976 case FILE_BEQLDATE: 977 case FILE_BEQWDATE: 978 p->q = (uint64_t) 979 (((uint64_t)p->hq[0]<<56)|((uint64_t)p->hq[1]<<48)| 980 ((uint64_t)p->hq[2]<<40)|((uint64_t)p->hq[3]<<32)| 981 ((uint64_t)p->hq[4]<<24)|((uint64_t)p->hq[5]<<16)| 982 ((uint64_t)p->hq[6]<<8)|((uint64_t)p->hq[7])); 983 if (type == FILE_BEQUAD) 984 cvt_64(p, m); 985 return 1; 986 case FILE_LESHORT: 987 p->h = (short)((p->hs[1]<<8)|(p->hs[0])); 988 cvt_16(p, m); 989 return 1; 990 case FILE_LELONG: 991 case FILE_LEDATE: 992 case FILE_LELDATE: 993 p->l = (int32_t) 994 ((p->hl[3]<<24)|(p->hl[2]<<16)|(p->hl[1]<<8)|(p->hl[0])); 995 if (type == FILE_LELONG) 996 cvt_32(p, m); 997 return 1; 998 case FILE_LEQUAD: 999 case FILE_LEQDATE: 1000 case FILE_LEQLDATE: 1001 case FILE_LEQWDATE: 1002 p->q = (uint64_t) 1003 (((uint64_t)p->hq[7]<<56)|((uint64_t)p->hq[6]<<48)| 1004 ((uint64_t)p->hq[5]<<40)|((uint64_t)p->hq[4]<<32)| 1005 ((uint64_t)p->hq[3]<<24)|((uint64_t)p->hq[2]<<16)| 1006 ((uint64_t)p->hq[1]<<8)|((uint64_t)p->hq[0])); 1007 if (type == FILE_LEQUAD) 1008 cvt_64(p, m); 1009 return 1; 1010 case FILE_MELONG: 1011 case FILE_MEDATE: 1012 case FILE_MELDATE: 1013 p->l = (int32_t) 1014 ((p->hl[1]<<24)|(p->hl[0]<<16)|(p->hl[3]<<8)|(p->hl[2])); 1015 if (type == FILE_MELONG) 1016 cvt_32(p, m); 1017 return 1; 1018 case FILE_FLOAT: 1019 cvt_float(p, m); 1020 return 1; 1021 case FILE_BEFLOAT: 1022 p->l = ((uint32_t)p->hl[0]<<24)|((uint32_t)p->hl[1]<<16)| 1023 ((uint32_t)p->hl[2]<<8) |((uint32_t)p->hl[3]); 1024 cvt_float(p, m); 1025 return 1; 1026 case FILE_LEFLOAT: 1027 p->l = ((uint32_t)p->hl[3]<<24)|((uint32_t)p->hl[2]<<16)| 1028 ((uint32_t)p->hl[1]<<8) |((uint32_t)p->hl[0]); 1029 cvt_float(p, m); 1030 return 1; 1031 case FILE_DOUBLE: 1032 cvt_double(p, m); 1033 return 1; 1034 case FILE_BEDOUBLE: 1035 p->q = ((uint64_t)p->hq[0]<<56)|((uint64_t)p->hq[1]<<48)| 1036 ((uint64_t)p->hq[2]<<40)|((uint64_t)p->hq[3]<<32)| 1037 ((uint64_t)p->hq[4]<<24)|((uint64_t)p->hq[5]<<16)| 1038 ((uint64_t)p->hq[6]<<8) |((uint64_t)p->hq[7]); 1039 cvt_double(p, m); 1040 return 1; 1041 case FILE_LEDOUBLE: 1042 p->q = ((uint64_t)p->hq[7]<<56)|((uint64_t)p->hq[6]<<48)| 1043 ((uint64_t)p->hq[5]<<40)|((uint64_t)p->hq[4]<<32)| 1044 ((uint64_t)p->hq[3]<<24)|((uint64_t)p->hq[2]<<16)| 1045 ((uint64_t)p->hq[1]<<8) |((uint64_t)p->hq[0]); 1046 cvt_double(p, m); 1047 return 1; 1048 case FILE_REGEX: 1049 case FILE_SEARCH: 1050 case FILE_DEFAULT: 1051 case FILE_CLEAR: 1052 case FILE_NAME: 1053 case FILE_USE: 1054 return 1; 1055 default: 1056 file_magerror(ms, "invalid type %d in mconvert()", m->type); 1057 return 0; 1058 } 1059 } 1060 1061 1062 private void 1063 mdebug(uint32_t offset, const char *str, size_t len) 1064 { 1065 (void) fprintf(stderr, "mget/%" SIZE_T_FORMAT "u @%d: ", len, offset); 1066 file_showstr(stderr, str, len); 1067 (void) fputc('\n', stderr); 1068 (void) fputc('\n', stderr); 1069 } 1070 1071 private int 1072 mcopy(struct magic_set *ms, union VALUETYPE *p, int type, int indir, 1073 const unsigned char *s, uint32_t offset, size_t nbytes, struct magic *m) 1074 { 1075 /* 1076 * Note: FILE_SEARCH and FILE_REGEX do not actually copy 1077 * anything, but setup pointers into the source 1078 */ 1079 if (indir == 0) { 1080 switch (type) { 1081 case FILE_SEARCH: 1082 ms->search.s = RCAST(const char *, s) + offset; 1083 ms->search.s_len = nbytes - offset; 1084 ms->search.offset = offset; 1085 return 0; 1086 1087 case FILE_REGEX: { 1088 const char *b; 1089 const char *c; 1090 const char *last; /* end of search region */ 1091 const char *buf; /* start of search region */ 1092 const char *end; 1093 size_t lines, linecnt, bytecnt; 1094 1095 if (s == NULL) { 1096 ms->search.s_len = 0; 1097 ms->search.s = NULL; 1098 return 0; 1099 } 1100 1101 if (m->str_flags & REGEX_LINE_COUNT) { 1102 linecnt = m->str_range; 1103 bytecnt = linecnt * 80; 1104 } else { 1105 linecnt = 0; 1106 bytecnt = m->str_range; 1107 } 1108 1109 if (bytecnt == 0) 1110 bytecnt = 8192; 1111 if (bytecnt > nbytes) 1112 bytecnt = nbytes; 1113 1114 buf = RCAST(const char *, s) + offset; 1115 end = last = RCAST(const char *, s) + bytecnt; 1116 /* mget() guarantees buf <= last */ 1117 for (lines = linecnt, b = buf; lines && b < end && 1118 ((b = CAST(const char *, 1119 memchr(c = b, '\n', CAST(size_t, (end - b))))) 1120 || (b = CAST(const char *, 1121 memchr(c, '\r', CAST(size_t, (end - c)))))); 1122 lines--, b++) { 1123 last = b; 1124 if (b[0] == '\r' && b[1] == '\n') 1125 b++; 1126 } 1127 if (lines) 1128 last = RCAST(const char *, s) + bytecnt; 1129 1130 ms->search.s = buf; 1131 ms->search.s_len = last - buf; 1132 ms->search.offset = offset; 1133 ms->search.rm_len = 0; 1134 return 0; 1135 } 1136 case FILE_BESTRING16: 1137 case FILE_LESTRING16: { 1138 const unsigned char *src = s + offset; 1139 const unsigned char *esrc = s + nbytes; 1140 char *dst = p->s; 1141 char *edst = &p->s[sizeof(p->s) - 1]; 1142 1143 if (type == FILE_BESTRING16) 1144 src++; 1145 1146 /* check that offset is within range */ 1147 if (offset >= nbytes) 1148 break; 1149 for (/*EMPTY*/; src < esrc; src += 2, dst++) { 1150 if (dst < edst) 1151 *dst = *src; 1152 else 1153 break; 1154 if (*dst == '\0') { 1155 if (type == FILE_BESTRING16 ? 1156 *(src - 1) != '\0' : 1157 *(src + 1) != '\0') 1158 *dst = ' '; 1159 } 1160 } 1161 *edst = '\0'; 1162 return 0; 1163 } 1164 case FILE_STRING: /* XXX - these two should not need */ 1165 case FILE_PSTRING: /* to copy anything, but do anyway. */ 1166 default: 1167 break; 1168 } 1169 } 1170 1171 if (offset >= nbytes) { 1172 (void)memset(p, '\0', sizeof(*p)); 1173 return 0; 1174 } 1175 if (nbytes - offset < sizeof(*p)) 1176 nbytes = nbytes - offset; 1177 else 1178 nbytes = sizeof(*p); 1179 1180 (void)memcpy(p, s + offset, nbytes); 1181 1182 /* 1183 * the usefulness of padding with zeroes eludes me, it 1184 * might even cause problems 1185 */ 1186 if (nbytes < sizeof(*p)) 1187 (void)memset(((char *)(void *)p) + nbytes, '\0', 1188 sizeof(*p) - nbytes); 1189 return 0; 1190 } 1191 1192 private int 1193 mget(struct magic_set *ms, const unsigned char *s, struct magic *m, 1194 size_t nbytes, size_t o, unsigned int cont_level, int mode, int text, 1195 int flip, int recursion_level, int *printed_something, 1196 int *need_separator, int *returnval) 1197 { 1198 uint32_t soffset, offset = ms->offset; 1199 uint32_t lhs; 1200 int rv, oneed_separator, in_type; 1201 char *sbuf, *rbuf; 1202 union VALUETYPE *p = &ms->ms_value; 1203 struct mlist ml; 1204 1205 if (recursion_level >= 20) { 1206 file_error(ms, 0, "recursion nesting exceeded"); 1207 return -1; 1208 } 1209 1210 if (mcopy(ms, p, m->type, m->flag & INDIR, s, (uint32_t)(offset + o), 1211 (uint32_t)nbytes, m) == -1) 1212 return -1; 1213 1214 if ((ms->flags & MAGIC_DEBUG) != 0) { 1215 fprintf(stderr, "mget(type=%d, flag=%x, offset=%u, o=%" 1216 SIZE_T_FORMAT "u, " "nbytes=%" SIZE_T_FORMAT "u)\n", 1217 m->type, m->flag, offset, o, nbytes); 1218 mdebug(offset, (char *)(void *)p, sizeof(union VALUETYPE)); 1219 #ifndef COMPILE_ONLY 1220 file_mdump(m); 1221 #endif 1222 } 1223 1224 if (m->flag & INDIR) { 1225 int off = m->in_offset; 1226 if (m->in_op & FILE_OPINDIRECT) { 1227 const union VALUETYPE *q = CAST(const union VALUETYPE *, 1228 ((const void *)(s + offset + off))); 1229 switch (cvt_flip(m->in_type, flip)) { 1230 case FILE_BYTE: 1231 off = q->b; 1232 break; 1233 case FILE_SHORT: 1234 off = q->h; 1235 break; 1236 case FILE_BESHORT: 1237 off = (short)((q->hs[0]<<8)|(q->hs[1])); 1238 break; 1239 case FILE_LESHORT: 1240 off = (short)((q->hs[1]<<8)|(q->hs[0])); 1241 break; 1242 case FILE_LONG: 1243 off = q->l; 1244 break; 1245 case FILE_BELONG: 1246 case FILE_BEID3: 1247 off = (int32_t)((q->hl[0]<<24)|(q->hl[1]<<16)| 1248 (q->hl[2]<<8)|(q->hl[3])); 1249 break; 1250 case FILE_LEID3: 1251 case FILE_LELONG: 1252 off = (int32_t)((q->hl[3]<<24)|(q->hl[2]<<16)| 1253 (q->hl[1]<<8)|(q->hl[0])); 1254 break; 1255 case FILE_MELONG: 1256 off = (int32_t)((q->hl[1]<<24)|(q->hl[0]<<16)| 1257 (q->hl[3]<<8)|(q->hl[2])); 1258 break; 1259 } 1260 if ((ms->flags & MAGIC_DEBUG) != 0) 1261 fprintf(stderr, "indirect offs=%u\n", off); 1262 } 1263 switch (in_type = cvt_flip(m->in_type, flip)) { 1264 case FILE_BYTE: 1265 if (OFFSET_OOB(nbytes, offset, 1)) 1266 return 0; 1267 if (off) { 1268 switch (m->in_op & FILE_OPS_MASK) { 1269 case FILE_OPAND: 1270 offset = p->b & off; 1271 break; 1272 case FILE_OPOR: 1273 offset = p->b | off; 1274 break; 1275 case FILE_OPXOR: 1276 offset = p->b ^ off; 1277 break; 1278 case FILE_OPADD: 1279 offset = p->b + off; 1280 break; 1281 case FILE_OPMINUS: 1282 offset = p->b - off; 1283 break; 1284 case FILE_OPMULTIPLY: 1285 offset = p->b * off; 1286 break; 1287 case FILE_OPDIVIDE: 1288 offset = p->b / off; 1289 break; 1290 case FILE_OPMODULO: 1291 offset = p->b % off; 1292 break; 1293 } 1294 } else 1295 offset = p->b; 1296 if (m->in_op & FILE_OPINVERSE) 1297 offset = ~offset; 1298 break; 1299 case FILE_BESHORT: 1300 if (OFFSET_OOB(nbytes, offset, 2)) 1301 return 0; 1302 lhs = (p->hs[0] << 8) | p->hs[1]; 1303 if (off) { 1304 switch (m->in_op & FILE_OPS_MASK) { 1305 case FILE_OPAND: 1306 offset = lhs & off; 1307 break; 1308 case FILE_OPOR: 1309 offset = lhs | off; 1310 break; 1311 case FILE_OPXOR: 1312 offset = lhs ^ off; 1313 break; 1314 case FILE_OPADD: 1315 offset = lhs + off; 1316 break; 1317 case FILE_OPMINUS: 1318 offset = lhs - off; 1319 break; 1320 case FILE_OPMULTIPLY: 1321 offset = lhs * off; 1322 break; 1323 case FILE_OPDIVIDE: 1324 offset = lhs / off; 1325 break; 1326 case FILE_OPMODULO: 1327 offset = lhs % off; 1328 break; 1329 } 1330 } else 1331 offset = lhs; 1332 if (m->in_op & FILE_OPINVERSE) 1333 offset = ~offset; 1334 break; 1335 case FILE_LESHORT: 1336 if (OFFSET_OOB(nbytes, offset, 2)) 1337 return 0; 1338 lhs = (p->hs[1] << 8) | p->hs[0]; 1339 if (off) { 1340 switch (m->in_op & FILE_OPS_MASK) { 1341 case FILE_OPAND: 1342 offset = lhs & off; 1343 break; 1344 case FILE_OPOR: 1345 offset = lhs | off; 1346 break; 1347 case FILE_OPXOR: 1348 offset = lhs ^ off; 1349 break; 1350 case FILE_OPADD: 1351 offset = lhs + off; 1352 break; 1353 case FILE_OPMINUS: 1354 offset = lhs - off; 1355 break; 1356 case FILE_OPMULTIPLY: 1357 offset = lhs * off; 1358 break; 1359 case FILE_OPDIVIDE: 1360 offset = lhs / off; 1361 break; 1362 case FILE_OPMODULO: 1363 offset = lhs % off; 1364 break; 1365 } 1366 } else 1367 offset = lhs; 1368 if (m->in_op & FILE_OPINVERSE) 1369 offset = ~offset; 1370 break; 1371 case FILE_SHORT: 1372 if (OFFSET_OOB(nbytes, offset, 2)) 1373 return 0; 1374 if (off) { 1375 switch (m->in_op & FILE_OPS_MASK) { 1376 case FILE_OPAND: 1377 offset = p->h & off; 1378 break; 1379 case FILE_OPOR: 1380 offset = p->h | off; 1381 break; 1382 case FILE_OPXOR: 1383 offset = p->h ^ off; 1384 break; 1385 case FILE_OPADD: 1386 offset = p->h + off; 1387 break; 1388 case FILE_OPMINUS: 1389 offset = p->h - off; 1390 break; 1391 case FILE_OPMULTIPLY: 1392 offset = p->h * off; 1393 break; 1394 case FILE_OPDIVIDE: 1395 offset = p->h / off; 1396 break; 1397 case FILE_OPMODULO: 1398 offset = p->h % off; 1399 break; 1400 } 1401 } 1402 else 1403 offset = p->h; 1404 if (m->in_op & FILE_OPINVERSE) 1405 offset = ~offset; 1406 break; 1407 case FILE_BELONG: 1408 case FILE_BEID3: 1409 if (OFFSET_OOB(nbytes, offset, 4)) 1410 return 0; 1411 lhs = (p->hl[0] << 24) | (p->hl[1] << 16) | 1412 (p->hl[2] << 8) | p->hl[3]; 1413 if (off) { 1414 switch (m->in_op & FILE_OPS_MASK) { 1415 case FILE_OPAND: 1416 offset = lhs & off; 1417 break; 1418 case FILE_OPOR: 1419 offset = lhs | off; 1420 break; 1421 case FILE_OPXOR: 1422 offset = lhs ^ off; 1423 break; 1424 case FILE_OPADD: 1425 offset = lhs + off; 1426 break; 1427 case FILE_OPMINUS: 1428 offset = lhs - off; 1429 break; 1430 case FILE_OPMULTIPLY: 1431 offset = lhs * off; 1432 break; 1433 case FILE_OPDIVIDE: 1434 offset = lhs / off; 1435 break; 1436 case FILE_OPMODULO: 1437 offset = lhs % off; 1438 break; 1439 } 1440 } else 1441 offset = lhs; 1442 if (m->in_op & FILE_OPINVERSE) 1443 offset = ~offset; 1444 break; 1445 case FILE_LELONG: 1446 case FILE_LEID3: 1447 if (OFFSET_OOB(nbytes, offset, 4)) 1448 return 0; 1449 lhs = (p->hl[3] << 24) | (p->hl[2] << 16) | 1450 (p->hl[1] << 8) | p->hl[0]; 1451 if (off) { 1452 switch (m->in_op & FILE_OPS_MASK) { 1453 case FILE_OPAND: 1454 offset = lhs & off; 1455 break; 1456 case FILE_OPOR: 1457 offset = lhs | off; 1458 break; 1459 case FILE_OPXOR: 1460 offset = lhs ^ off; 1461 break; 1462 case FILE_OPADD: 1463 offset = lhs + off; 1464 break; 1465 case FILE_OPMINUS: 1466 offset = lhs - off; 1467 break; 1468 case FILE_OPMULTIPLY: 1469 offset = lhs * off; 1470 break; 1471 case FILE_OPDIVIDE: 1472 offset = lhs / off; 1473 break; 1474 case FILE_OPMODULO: 1475 offset = lhs % off; 1476 break; 1477 } 1478 } else 1479 offset = lhs; 1480 if (m->in_op & FILE_OPINVERSE) 1481 offset = ~offset; 1482 break; 1483 case FILE_MELONG: 1484 if (OFFSET_OOB(nbytes, offset, 4)) 1485 return 0; 1486 lhs = (p->hl[1] << 24) | (p->hl[0] << 16) | 1487 (p->hl[3] << 8) | p->hl[2]; 1488 if (off) { 1489 switch (m->in_op & FILE_OPS_MASK) { 1490 case FILE_OPAND: 1491 offset = lhs & off; 1492 break; 1493 case FILE_OPOR: 1494 offset = lhs | off; 1495 break; 1496 case FILE_OPXOR: 1497 offset = lhs ^ off; 1498 break; 1499 case FILE_OPADD: 1500 offset = lhs + off; 1501 break; 1502 case FILE_OPMINUS: 1503 offset = lhs - off; 1504 break; 1505 case FILE_OPMULTIPLY: 1506 offset = lhs * off; 1507 break; 1508 case FILE_OPDIVIDE: 1509 offset = lhs / off; 1510 break; 1511 case FILE_OPMODULO: 1512 offset = lhs % off; 1513 break; 1514 } 1515 } else 1516 offset = lhs; 1517 if (m->in_op & FILE_OPINVERSE) 1518 offset = ~offset; 1519 break; 1520 case FILE_LONG: 1521 if (OFFSET_OOB(nbytes, offset, 4)) 1522 return 0; 1523 if (off) { 1524 switch (m->in_op & FILE_OPS_MASK) { 1525 case FILE_OPAND: 1526 offset = p->l & off; 1527 break; 1528 case FILE_OPOR: 1529 offset = p->l | off; 1530 break; 1531 case FILE_OPXOR: 1532 offset = p->l ^ off; 1533 break; 1534 case FILE_OPADD: 1535 offset = p->l + off; 1536 break; 1537 case FILE_OPMINUS: 1538 offset = p->l - off; 1539 break; 1540 case FILE_OPMULTIPLY: 1541 offset = p->l * off; 1542 break; 1543 case FILE_OPDIVIDE: 1544 offset = p->l / off; 1545 break; 1546 case FILE_OPMODULO: 1547 offset = p->l % off; 1548 break; 1549 } 1550 } else 1551 offset = p->l; 1552 if (m->in_op & FILE_OPINVERSE) 1553 offset = ~offset; 1554 break; 1555 default: 1556 break; 1557 } 1558 1559 switch (in_type) { 1560 case FILE_LEID3: 1561 case FILE_BEID3: 1562 offset = ((((offset >> 0) & 0x7f) << 0) | 1563 (((offset >> 8) & 0x7f) << 7) | 1564 (((offset >> 16) & 0x7f) << 14) | 1565 (((offset >> 24) & 0x7f) << 21)) + 10; 1566 break; 1567 default: 1568 break; 1569 } 1570 1571 if (m->flag & INDIROFFADD) { 1572 offset += ms->c.li[cont_level-1].off; 1573 if (offset == 0) { 1574 if ((ms->flags & MAGIC_DEBUG) != 0) 1575 fprintf(stderr, 1576 "indirect *zero* offset\n"); 1577 return 0; 1578 } 1579 if ((ms->flags & MAGIC_DEBUG) != 0) 1580 fprintf(stderr, "indirect +offs=%u\n", offset); 1581 } 1582 if (mcopy(ms, p, m->type, 0, s, offset, nbytes, m) == -1) 1583 return -1; 1584 ms->offset = offset; 1585 1586 if ((ms->flags & MAGIC_DEBUG) != 0) { 1587 mdebug(offset, (char *)(void *)p, 1588 sizeof(union VALUETYPE)); 1589 #ifndef COMPILE_ONLY 1590 file_mdump(m); 1591 #endif 1592 } 1593 } 1594 1595 /* Verify we have enough data to match magic type */ 1596 switch (m->type) { 1597 case FILE_BYTE: 1598 if (OFFSET_OOB(nbytes, offset, 1)) 1599 return 0; 1600 break; 1601 1602 case FILE_SHORT: 1603 case FILE_BESHORT: 1604 case FILE_LESHORT: 1605 if (OFFSET_OOB(nbytes, offset, 2)) 1606 return 0; 1607 break; 1608 1609 case FILE_LONG: 1610 case FILE_BELONG: 1611 case FILE_LELONG: 1612 case FILE_MELONG: 1613 case FILE_DATE: 1614 case FILE_BEDATE: 1615 case FILE_LEDATE: 1616 case FILE_MEDATE: 1617 case FILE_LDATE: 1618 case FILE_BELDATE: 1619 case FILE_LELDATE: 1620 case FILE_MELDATE: 1621 case FILE_FLOAT: 1622 case FILE_BEFLOAT: 1623 case FILE_LEFLOAT: 1624 if (OFFSET_OOB(nbytes, offset, 4)) 1625 return 0; 1626 break; 1627 1628 case FILE_DOUBLE: 1629 case FILE_BEDOUBLE: 1630 case FILE_LEDOUBLE: 1631 if (OFFSET_OOB(nbytes, offset, 8)) 1632 return 0; 1633 break; 1634 1635 case FILE_STRING: 1636 case FILE_PSTRING: 1637 case FILE_SEARCH: 1638 if (OFFSET_OOB(nbytes, offset, m->vallen)) 1639 return 0; 1640 break; 1641 1642 case FILE_REGEX: 1643 if (nbytes < offset) 1644 return 0; 1645 break; 1646 1647 case FILE_INDIRECT: 1648 if (offset == 0) 1649 return 0; 1650 if (nbytes < offset) 1651 return 0; 1652 sbuf = ms->o.buf; 1653 soffset = ms->offset; 1654 ms->o.buf = NULL; 1655 ms->offset = 0; 1656 rv = file_softmagic(ms, s + offset, nbytes - offset, 1657 recursion_level, BINTEST, text); 1658 if ((ms->flags & MAGIC_DEBUG) != 0) 1659 fprintf(stderr, "indirect @offs=%u[%d]\n", offset, rv); 1660 rbuf = ms->o.buf; 1661 ms->o.buf = sbuf; 1662 ms->offset = soffset; 1663 if (rv == 1) { 1664 if ((ms->flags & (MAGIC_MIME|MAGIC_APPLE)) == 0 && 1665 file_printf(ms, F(ms, m, "%u"), offset) == -1) { 1666 free(rbuf); 1667 return -1; 1668 } 1669 if (file_printf(ms, "%s", rbuf) == -1) { 1670 free(rbuf); 1671 return -1; 1672 } 1673 } 1674 free(rbuf); 1675 return rv; 1676 1677 case FILE_USE: 1678 if (nbytes < offset) 1679 return 0; 1680 sbuf = m->value.s; 1681 if (*sbuf == '^') { 1682 sbuf++; 1683 flip = !flip; 1684 } 1685 if (file_magicfind(ms, sbuf, &ml) == -1) { 1686 file_error(ms, 0, "cannot find entry `%s'", sbuf); 1687 return -1; 1688 } 1689 1690 oneed_separator = *need_separator; 1691 if (m->flag & NOSPACE) 1692 *need_separator = 0; 1693 rv = match(ms, ml.magic, ml.nmagic, s, nbytes, offset + o, 1694 mode, text, flip, recursion_level, printed_something, 1695 need_separator, returnval); 1696 if (rv != 1) 1697 *need_separator = oneed_separator; 1698 return rv; 1699 1700 case FILE_NAME: 1701 if (file_printf(ms, "%s", m->desc) == -1) 1702 return -1; 1703 return 1; 1704 case FILE_DEFAULT: /* nothing to check */ 1705 case FILE_CLEAR: 1706 default: 1707 break; 1708 } 1709 if (!mconvert(ms, m, flip)) 1710 return 0; 1711 return 1; 1712 } 1713 1714 private uint64_t 1715 file_strncmp(const char *s1, const char *s2, size_t len, uint32_t flags) 1716 { 1717 /* 1718 * Convert the source args to unsigned here so that (1) the 1719 * compare will be unsigned as it is in strncmp() and (2) so 1720 * the ctype functions will work correctly without extra 1721 * casting. 1722 */ 1723 const unsigned char *a = (const unsigned char *)s1; 1724 const unsigned char *b = (const unsigned char *)s2; 1725 uint64_t v; 1726 1727 /* 1728 * What we want here is v = strncmp(s1, s2, len), 1729 * but ignoring any nulls. 1730 */ 1731 v = 0; 1732 if (0L == flags) { /* normal string: do it fast */ 1733 while (len-- > 0) 1734 if ((v = *b++ - *a++) != '\0') 1735 break; 1736 } 1737 else { /* combine the others */ 1738 while (len-- > 0) { 1739 if ((flags & STRING_IGNORE_LOWERCASE) && 1740 islower(*a)) { 1741 if ((v = tolower(*b++) - *a++) != '\0') 1742 break; 1743 } 1744 else if ((flags & STRING_IGNORE_UPPERCASE) && 1745 isupper(*a)) { 1746 if ((v = toupper(*b++) - *a++) != '\0') 1747 break; 1748 } 1749 else if ((flags & STRING_COMPACT_WHITESPACE) && 1750 isspace(*a)) { 1751 a++; 1752 if (isspace(*b++)) { 1753 if (!isspace(*a)) 1754 while (isspace(*b)) 1755 b++; 1756 } 1757 else { 1758 v = 1; 1759 break; 1760 } 1761 } 1762 else if ((flags & STRING_COMPACT_OPTIONAL_WHITESPACE) && 1763 isspace(*a)) { 1764 a++; 1765 while (isspace(*b)) 1766 b++; 1767 } 1768 else { 1769 if ((v = *b++ - *a++) != '\0') 1770 break; 1771 } 1772 } 1773 } 1774 return v; 1775 } 1776 1777 private uint64_t 1778 file_strncmp16(const char *a, const char *b, size_t len, uint32_t flags) 1779 { 1780 /* 1781 * XXX - The 16-bit string compare probably needs to be done 1782 * differently, especially if the flags are to be supported. 1783 * At the moment, I am unsure. 1784 */ 1785 flags = 0; 1786 return file_strncmp(a, b, len, flags); 1787 } 1788 1789 private int 1790 magiccheck(struct magic_set *ms, struct magic *m) 1791 { 1792 uint64_t l = m->value.q; 1793 uint64_t v; 1794 float fl, fv; 1795 double dl, dv; 1796 int matched; 1797 union VALUETYPE *p = &ms->ms_value; 1798 1799 switch (m->type) { 1800 case FILE_BYTE: 1801 v = p->b; 1802 break; 1803 1804 case FILE_SHORT: 1805 case FILE_BESHORT: 1806 case FILE_LESHORT: 1807 v = p->h; 1808 break; 1809 1810 case FILE_LONG: 1811 case FILE_BELONG: 1812 case FILE_LELONG: 1813 case FILE_MELONG: 1814 case FILE_DATE: 1815 case FILE_BEDATE: 1816 case FILE_LEDATE: 1817 case FILE_MEDATE: 1818 case FILE_LDATE: 1819 case FILE_BELDATE: 1820 case FILE_LELDATE: 1821 case FILE_MELDATE: 1822 v = p->l; 1823 break; 1824 1825 case FILE_QUAD: 1826 case FILE_LEQUAD: 1827 case FILE_BEQUAD: 1828 case FILE_QDATE: 1829 case FILE_BEQDATE: 1830 case FILE_LEQDATE: 1831 case FILE_QLDATE: 1832 case FILE_BEQLDATE: 1833 case FILE_LEQLDATE: 1834 case FILE_QWDATE: 1835 case FILE_BEQWDATE: 1836 case FILE_LEQWDATE: 1837 v = p->q; 1838 break; 1839 1840 case FILE_FLOAT: 1841 case FILE_BEFLOAT: 1842 case FILE_LEFLOAT: 1843 fl = m->value.f; 1844 fv = p->f; 1845 switch (m->reln) { 1846 case 'x': 1847 matched = 1; 1848 break; 1849 1850 case '!': 1851 matched = fv != fl; 1852 break; 1853 1854 case '=': 1855 matched = fv == fl; 1856 break; 1857 1858 case '>': 1859 matched = fv > fl; 1860 break; 1861 1862 case '<': 1863 matched = fv < fl; 1864 break; 1865 1866 default: 1867 file_magerror(ms, "cannot happen with float: invalid relation `%c'", 1868 m->reln); 1869 return -1; 1870 } 1871 return matched; 1872 1873 case FILE_DOUBLE: 1874 case FILE_BEDOUBLE: 1875 case FILE_LEDOUBLE: 1876 dl = m->value.d; 1877 dv = p->d; 1878 switch (m->reln) { 1879 case 'x': 1880 matched = 1; 1881 break; 1882 1883 case '!': 1884 matched = dv != dl; 1885 break; 1886 1887 case '=': 1888 matched = dv == dl; 1889 break; 1890 1891 case '>': 1892 matched = dv > dl; 1893 break; 1894 1895 case '<': 1896 matched = dv < dl; 1897 break; 1898 1899 default: 1900 file_magerror(ms, "cannot happen with double: invalid relation `%c'", m->reln); 1901 return -1; 1902 } 1903 return matched; 1904 1905 case FILE_DEFAULT: 1906 case FILE_CLEAR: 1907 l = 0; 1908 v = 0; 1909 break; 1910 1911 case FILE_STRING: 1912 case FILE_PSTRING: 1913 l = 0; 1914 v = file_strncmp(m->value.s, p->s, (size_t)m->vallen, m->str_flags); 1915 break; 1916 1917 case FILE_BESTRING16: 1918 case FILE_LESTRING16: 1919 l = 0; 1920 v = file_strncmp16(m->value.s, p->s, (size_t)m->vallen, m->str_flags); 1921 break; 1922 1923 case FILE_SEARCH: { /* search ms->search.s for the string m->value.s */ 1924 size_t slen; 1925 size_t idx; 1926 1927 if (ms->search.s == NULL) 1928 return 0; 1929 1930 slen = MIN(m->vallen, sizeof(m->value.s)); 1931 l = 0; 1932 v = 0; 1933 1934 for (idx = 0; m->str_range == 0 || idx < m->str_range; idx++) { 1935 if (slen + idx > ms->search.s_len) 1936 break; 1937 1938 v = file_strncmp(m->value.s, ms->search.s + idx, slen, 1939 m->str_flags); 1940 if (v == 0) { /* found match */ 1941 ms->search.offset += idx; 1942 break; 1943 } 1944 } 1945 break; 1946 } 1947 case FILE_REGEX: { 1948 int rc; 1949 file_regex_t rx; 1950 const char *search; 1951 1952 if (ms->search.s == NULL) 1953 return 0; 1954 1955 l = 0; 1956 rc = file_regcomp(&rx, m->value.s, 1957 REG_EXTENDED|REG_NEWLINE| 1958 ((m->str_flags & STRING_IGNORE_CASE) ? REG_ICASE : 0)); 1959 if (rc) { 1960 file_regerror(&rx, rc, ms); 1961 v = (uint64_t)-1; 1962 } else { 1963 regmatch_t pmatch[1]; 1964 size_t slen = ms->search.s_len; 1965 #ifndef REG_STARTEND 1966 #define REG_STARTEND 0 1967 char *copy; 1968 if (slen != 0) { 1969 copy = malloc(slen); 1970 if (copy == NULL) { 1971 file_error(ms, errno, 1972 "can't allocate %" SIZE_T_FORMAT "u bytes", 1973 slen); 1974 return -1; 1975 } 1976 memcpy(copy, ms->search.s, slen); 1977 copy[--slen] = '\0'; 1978 search = copy; 1979 } else { 1980 search = ms->search.s; 1981 copy = NULL; 1982 } 1983 #else 1984 search = ms->search.s; 1985 pmatch[0].rm_so = 0; 1986 pmatch[0].rm_eo = slen; 1987 #endif 1988 rc = file_regexec(&rx, (const char *)search, 1989 1, pmatch, REG_STARTEND); 1990 #if REG_STARTEND == 0 1991 free(copy); 1992 #endif 1993 switch (rc) { 1994 case 0: 1995 ms->search.s += (int)pmatch[0].rm_so; 1996 ms->search.offset += (size_t)pmatch[0].rm_so; 1997 ms->search.rm_len = 1998 (size_t)(pmatch[0].rm_eo - pmatch[0].rm_so); 1999 v = 0; 2000 break; 2001 2002 case REG_NOMATCH: 2003 v = 1; 2004 break; 2005 2006 default: 2007 file_regerror(&rx, rc, ms); 2008 v = (uint64_t)-1; 2009 break; 2010 } 2011 } 2012 file_regfree(&rx); 2013 if (v == (uint64_t)-1) 2014 return -1; 2015 break; 2016 } 2017 case FILE_INDIRECT: 2018 case FILE_USE: 2019 case FILE_NAME: 2020 return 1; 2021 default: 2022 file_magerror(ms, "invalid type %d in magiccheck()", m->type); 2023 return -1; 2024 } 2025 2026 v = file_signextend(ms, m, v); 2027 2028 switch (m->reln) { 2029 case 'x': 2030 if ((ms->flags & MAGIC_DEBUG) != 0) 2031 (void) fprintf(stderr, "%" INT64_T_FORMAT 2032 "u == *any* = 1\n", (unsigned long long)v); 2033 matched = 1; 2034 break; 2035 2036 case '!': 2037 matched = v != l; 2038 if ((ms->flags & MAGIC_DEBUG) != 0) 2039 (void) fprintf(stderr, "%" INT64_T_FORMAT "u != %" 2040 INT64_T_FORMAT "u = %d\n", (unsigned long long)v, 2041 (unsigned long long)l, matched); 2042 break; 2043 2044 case '=': 2045 matched = v == l; 2046 if ((ms->flags & MAGIC_DEBUG) != 0) 2047 (void) fprintf(stderr, "%" INT64_T_FORMAT "u == %" 2048 INT64_T_FORMAT "u = %d\n", (unsigned long long)v, 2049 (unsigned long long)l, matched); 2050 break; 2051 2052 case '>': 2053 if (m->flag & UNSIGNED) { 2054 matched = v > l; 2055 if ((ms->flags & MAGIC_DEBUG) != 0) 2056 (void) fprintf(stderr, "%" INT64_T_FORMAT 2057 "u > %" INT64_T_FORMAT "u = %d\n", 2058 (unsigned long long)v, 2059 (unsigned long long)l, matched); 2060 } 2061 else { 2062 matched = (int64_t) v > (int64_t) l; 2063 if ((ms->flags & MAGIC_DEBUG) != 0) 2064 (void) fprintf(stderr, "%" INT64_T_FORMAT 2065 "d > %" INT64_T_FORMAT "d = %d\n", 2066 (long long)v, (long long)l, matched); 2067 } 2068 break; 2069 2070 case '<': 2071 if (m->flag & UNSIGNED) { 2072 matched = v < l; 2073 if ((ms->flags & MAGIC_DEBUG) != 0) 2074 (void) fprintf(stderr, "%" INT64_T_FORMAT 2075 "u < %" INT64_T_FORMAT "u = %d\n", 2076 (unsigned long long)v, 2077 (unsigned long long)l, matched); 2078 } 2079 else { 2080 matched = (int64_t) v < (int64_t) l; 2081 if ((ms->flags & MAGIC_DEBUG) != 0) 2082 (void) fprintf(stderr, "%" INT64_T_FORMAT 2083 "d < %" INT64_T_FORMAT "d = %d\n", 2084 (long long)v, (long long)l, matched); 2085 } 2086 break; 2087 2088 case '&': 2089 matched = (v & l) == l; 2090 if ((ms->flags & MAGIC_DEBUG) != 0) 2091 (void) fprintf(stderr, "((%" INT64_T_FORMAT "x & %" 2092 INT64_T_FORMAT "x) == %" INT64_T_FORMAT 2093 "x) = %d\n", (unsigned long long)v, 2094 (unsigned long long)l, (unsigned long long)l, 2095 matched); 2096 break; 2097 2098 case '^': 2099 matched = (v & l) != l; 2100 if ((ms->flags & MAGIC_DEBUG) != 0) 2101 (void) fprintf(stderr, "((%" INT64_T_FORMAT "x & %" 2102 INT64_T_FORMAT "x) != %" INT64_T_FORMAT 2103 "x) = %d\n", (unsigned long long)v, 2104 (unsigned long long)l, (unsigned long long)l, 2105 matched); 2106 break; 2107 2108 default: 2109 file_magerror(ms, "cannot happen: invalid relation `%c'", 2110 m->reln); 2111 return -1; 2112 } 2113 2114 return matched; 2115 } 2116 2117 private int 2118 handle_annotation(struct magic_set *ms, struct magic *m) 2119 { 2120 if (ms->flags & MAGIC_APPLE) { 2121 if (file_printf(ms, "%.8s", m->apple) == -1) 2122 return -1; 2123 return 1; 2124 } 2125 if ((ms->flags & MAGIC_MIME_TYPE) && m->mimetype[0]) { 2126 if (file_printf(ms, "%s", m->mimetype) == -1) 2127 return -1; 2128 return 1; 2129 } 2130 return 0; 2131 } 2132 2133 private int 2134 print_sep(struct magic_set *ms, int firstline) 2135 { 2136 if (ms->flags & MAGIC_MIME) 2137 return 0; 2138 if (firstline) 2139 return 0; 2140 /* 2141 * we found another match 2142 * put a newline and '-' to do some simple formatting 2143 */ 2144 return file_printf(ms, "\n- "); 2145 } 2146