?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
sbin/hammer: Have consistent naming for buffer variables
[dragonfly.git] / contrib / file / src / softmagic.c
blob5e277e31fdda21da7d8c0ec73872bad884d07697
1 /*
2 * Copyright (c) Ian F. Darwin 1986-1995.
3 * Software written by Ian F. Darwin and others;
4 * maintained 1995-present by Christos Zoulas and others.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 * notice immediately at the beginning of the file, without modification,
11 * this list of conditions, and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 *
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
20 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26 * SUCH DAMAGE.
27 */
28 /*
29 * softmagic - interpret variable magic from MAGIC
30 */
31
32 #include "file.h"
33
34 #ifndef lint
35 FILE_RCSID("@(#)$File: softmagic.c,v 1.206 2015/01/01 17:07:34 christos Exp $")
36 #endif /* lint */
37
38 #include "magic.h"
39 #include <assert.h>
40 #include <string.h>
41 #include <ctype.h>
42 #include <stdlib.h>
43 #include <time.h>
44
45 private int match(struct magic_set *, struct magic *, uint32_t,
46 const unsigned char *, size_t, size_t, int, int, int, uint16_t,
47 uint16_t *, int *, int *, int *);
48 private int mget(struct magic_set *, const unsigned char *,
49 struct magic *, size_t, size_t, unsigned int, int, int, int, uint16_t,
50 uint16_t *, int *, int *, int *);
51 private int magiccheck(struct magic_set *, struct magic *);
52 private int32_t mprint(struct magic_set *, struct magic *);
53 private int32_t moffset(struct magic_set *, struct magic *);
54 private void mdebug(uint32_t, const char *, size_t);
55 private int mcopy(struct magic_set *, union VALUETYPE *, int, int,
56 const unsigned char *, uint32_t, size_t, struct magic *);
57 private int mconvert(struct magic_set *, struct magic *, int);
58 private int print_sep(struct magic_set *, int);
59 private int handle_annotation(struct magic_set *, struct magic *);
60 private void cvt_8(union VALUETYPE *, const struct magic *);
61 private void cvt_16(union VALUETYPE *, const struct magic *);
62 private void cvt_32(union VALUETYPE *, const struct magic *);
63 private void cvt_64(union VALUETYPE *, const struct magic *);
64
65 #define OFFSET_OOB(n, o, i) ((n) < (o) || (i) > ((n) - (o)))
66
67 /*
68 * softmagic - lookup one file in parsed, in-memory copy of database
69 * Passed the name and FILE * of one file to be typed.
70 */
71 /*ARGSUSED1*/ /* nbytes passed for regularity, maybe need later */
72 protected int
73 file_softmagic(struct magic_set *ms, const unsigned char *buf, size_t nbytes,
74 uint16_t indir_level, uint16_t *name_count, int mode, int text)
75 {
76 struct mlist *ml;
77 int rv, printed_something = 0, need_separator = 0;
78 uint16_t nc;
79
80 if (name_count == NULL) {
81 nc = 0;
82 name_count = &nc;
83 }
84
85 for (ml = ms->mlist[0]->next; ml != ms->mlist[0]; ml = ml->next)
86 if ((rv = match(ms, ml->magic, ml->nmagic, buf, nbytes, 0, mode,
87 text, 0, indir_level, name_count,
88 &printed_something, &need_separator, NULL)) != 0)
89 return rv;
90
91 return 0;
92 }
93
94 #define FILE_FMTDEBUG
95 #ifdef FILE_FMTDEBUG
96 #define F(a, b, c) file_fmtcheck((a), (b), (c), __FILE__, __LINE__)
97
98 private const char * __attribute__((__format_arg__(3)))
99 file_fmtcheck(struct magic_set *ms, const struct magic *m, const char *def,
100 const char *file, size_t line)
101 {
102 const char *ptr = fmtcheck(m->desc, def);
103 if (ptr == def)
104 file_magerror(ms,
105 "%s, %" SIZE_T_FORMAT "u: format `%s' does not match"
106 " with `%s'", file, line, m->desc, def);
107 return ptr;
108 }
109 #else
110 #define F(a, b, c) fmtcheck((b)->desc, (c))
111 #endif
112
113 /*
114 * Go through the whole list, stopping if you find a match. Process all
115 * the continuations of that match before returning.
116 *
117 * We support multi-level continuations:
118 *
119 * At any time when processing a successful top-level match, there is a
120 * current continuation level; it represents the level of the last
121 * successfully matched continuation.
122 *
123 * Continuations above that level are skipped as, if we see one, it
124 * means that the continuation that controls them - i.e, the
125 * lower-level continuation preceding them - failed to match.
126 *
127 * Continuations below that level are processed as, if we see one,
128 * it means we've finished processing or skipping higher-level
129 * continuations under the control of a successful or unsuccessful
130 * lower-level continuation, and are now seeing the next lower-level
131 * continuation and should process it. The current continuation
132 * level reverts to the level of the one we're seeing.
133 *
134 * Continuations at the current level are processed as, if we see
135 * one, there's no lower-level continuation that may have failed.
136 *
137 * If a continuation matches, we bump the current continuation level
138 * so that higher-level continuations are processed.
139 */
140 private int
141 match(struct magic_set *ms, struct magic *magic, uint32_t nmagic,
142 const unsigned char *s, size_t nbytes, size_t offset, int mode, int text,
143 int flip, uint16_t indir_level, uint16_t *name_count,
144 int *printed_something, int *need_separator, int *returnval)
145 {
146 uint32_t magindex = 0;
147 unsigned int cont_level = 0;
148 int returnvalv = 0, e; /* if a match is found it is set to 1*/
149 int firstline = 1; /* a flag to print X\n X\n- X */
150 int print = (ms->flags & (MAGIC_MIME|MAGIC_APPLE)) == 0;
151
152 if (returnval == NULL)
153 returnval = &returnvalv;
154
155 if (file_check_mem(ms, cont_level) == -1)
156 return -1;
157
158 for (magindex = 0; magindex < nmagic; magindex++) {
159 int flush = 0;
160 struct magic *m = &magic[magindex];
161
162 if (m->type != FILE_NAME)
163 if ((IS_STRING(m->type) &&
164 #define FLT (STRING_BINTEST | STRING_TEXTTEST)
165 ((text && (m->str_flags & FLT) == STRING_BINTEST) ||
166 (!text && (m->str_flags & FLT) == STRING_TEXTTEST))) ||
167 (m->flag & mode) != mode) {
168 /* Skip sub-tests */
169 while (magindex + 1 < nmagic &&
170 magic[magindex + 1].cont_level != 0 &&
171 ++magindex)
172 continue;
173 continue; /* Skip to next top-level test*/
174 }
175
176 ms->offset = m->offset;
177 ms->line = m->lineno;
178
179 /* if main entry matches, print it... */
180 switch (mget(ms, s, m, nbytes, offset, cont_level, mode, text,
181 flip, indir_level, name_count,
182 printed_something, need_separator, returnval)) {
183 case -1:
184 return -1;
185 case 0:
186 flush = m->reln != '!';
187 break;
188 default:
189 if (m->type == FILE_INDIRECT)
190 *returnval = 1;
191
192 switch (magiccheck(ms, m)) {
193 case -1:
194 return -1;
195 case 0:
196 flush++;
197 break;
198 default:
199 flush = 0;
200 break;
201 }
202 break;
203 }
204 if (flush) {
205 /*
206 * main entry didn't match,
207 * flush its continuations
208 */
209 while (magindex < nmagic - 1 &&
210 magic[magindex + 1].cont_level != 0)
211 magindex++;
212 continue;
213 }
214
215 if ((e = handle_annotation(ms, m)) != 0) {
216 *need_separator = 1;
217 *printed_something = 1;
218 *returnval = 1;
219 return e;
220 }
221 /*
222 * If we are going to print something, we'll need to print
223 * a blank before we print something else.
224 */
225 if (*m->desc) {
226 *need_separator = 1;
227 *printed_something = 1;
228 if (print_sep(ms, firstline) == -1)
229 return -1;
230 }
231
232
233 if (print && mprint(ms, m) == -1)
234 return -1;
235
236 ms->c.li[cont_level].off = moffset(ms, m);
237
238 /* and any continuations that match */
239 if (file_check_mem(ms, ++cont_level) == -1)
240 return -1;
241
242 while (magindex + 1 < nmagic &&
243 magic[magindex + 1].cont_level != 0) {
244 m = &magic[++magindex];
245 ms->line = m->lineno; /* for messages */
246
247 if (cont_level < m->cont_level)
248 continue;
249 if (cont_level > m->cont_level) {
250 /*
251 * We're at the end of the level
252 * "cont_level" continuations.
253 */
254 cont_level = m->cont_level;
255 }
256 ms->offset = m->offset;
257 if (m->flag & OFFADD) {
258 ms->offset +=
259 ms->c.li[cont_level - 1].off;
260 }
261
262 #ifdef ENABLE_CONDITIONALS
263 if (m->cond == COND_ELSE ||
264 m->cond == COND_ELIF) {
265 if (ms->c.li[cont_level].last_match == 1)
266 continue;
267 }
268 #endif
269 switch (mget(ms, s, m, nbytes, offset, cont_level, mode,
270 text, flip, indir_level, name_count,
271 printed_something, need_separator, returnval)) {
272 case -1:
273 return -1;
274 case 0:
275 if (m->reln != '!')
276 continue;
277 flush = 1;
278 break;
279 default:
280 if (m->type == FILE_INDIRECT)
281 *returnval = 1;
282 flush = 0;
283 break;
284 }
285
286 switch (flush ? 1 : magiccheck(ms, m)) {
287 case -1:
288 return -1;
289 case 0:
290 #ifdef ENABLE_CONDITIONALS
291 ms->c.li[cont_level].last_match = 0;
292 #endif
293 break;
294 default:
295 #ifdef ENABLE_CONDITIONALS
296 ms->c.li[cont_level].last_match = 1;
297 #endif
298 if (m->type == FILE_CLEAR)
299 ms->c.li[cont_level].got_match = 0;
300 else if (ms->c.li[cont_level].got_match) {
301 if (m->type == FILE_DEFAULT)
302 break;
303 } else
304 ms->c.li[cont_level].got_match = 1;
305 if ((e = handle_annotation(ms, m)) != 0) {
306 *need_separator = 1;
307 *printed_something = 1;
308 *returnval = 1;
309 return e;
310 }
311 /*
312 * If we are going to print something,
313 * make sure that we have a separator first.
314 */
315 if (*m->desc) {
316 if (!*printed_something) {
317 *printed_something = 1;
318 if (print_sep(ms, firstline)
319 == -1)
320 return -1;
321 }
322 }
323 /*
324 * This continuation matched. Print
325 * its message, with a blank before it
326 * if the previous item printed and
327 * this item isn't empty.
328 */
329 /* space if previous printed */
330 if (*need_separator
331 && ((m->flag & NOSPACE) == 0)
332 && *m->desc) {
333 if (print &&
334 file_printf(ms, " ") == -1)
335 return -1;
336 *need_separator = 0;
337 }
338 if (print && mprint(ms, m) == -1)
339 return -1;
340
341 ms->c.li[cont_level].off = moffset(ms, m);
342
343 if (*m->desc)
344 *need_separator = 1;
345
346 /*
347 * If we see any continuations
348 * at a higher level,
349 * process them.
350 */
351 if (file_check_mem(ms, ++cont_level) == -1)
352 return -1;
353 break;
354 }
355 }
356 if (*printed_something) {
357 firstline = 0;
358 if (print)
359 *returnval = 1;
360 }
361 if ((ms->flags & MAGIC_CONTINUE) == 0 && *printed_something) {
362 return *returnval; /* don't keep searching */
363 }
364 }
365 return *returnval; /* This is hit if -k is set or there is no match */
366 }
367
368 private int
369 check_fmt(struct magic_set *ms, struct magic *m)
370 {
371 file_regex_t rx;
372 int rc, rv = -1;
373
374 if (strchr(m->desc, '%') == NULL)
375 return 0;
376
377 rc = file_regcomp(&rx, "%[-0-9\\.]*s", REG_EXTENDED|REG_NOSUB);
378 if (rc) {
379 file_regerror(&rx, rc, ms);
380 } else {
381 rc = file_regexec(&rx, m->desc, 0, 0, 0);
382 rv = !rc;
383 }
384 file_regfree(&rx);
385 return rv;
386 }
387
388 #ifndef HAVE_STRNDUP
389 char * strndup(const char *, size_t);
390
391 char *
392 strndup(const char *str, size_t n)
393 {
394 size_t len;
395 char *copy;
396
397 for (len = 0; len < n && str[len]; len++)
398 continue;
399 if ((copy = malloc(len + 1)) == NULL)
400 return NULL;
401 (void)memcpy(copy, str, len);
402 copy[len] = '\0';
403 return copy;
404 }
405 #endif /* HAVE_STRNDUP */
406
407 private int32_t
408 mprint(struct magic_set *ms, struct magic *m)
409 {
410 uint64_t v;
411 float vf;
412 double vd;
413 int64_t t = 0;
414 char buf[128], tbuf[26], sbuf[512];
415 union VALUETYPE *p = &ms->ms_value;
416
417 switch (m->type) {
418 case FILE_BYTE:
419 v = file_signextend(ms, m, (uint64_t)p->b);
420 switch (check_fmt(ms, m)) {
421 case -1:
422 return -1;
423 case 1:
424 (void)snprintf(buf, sizeof(buf), "%d",
425 (unsigned char)v);
426 if (file_printf(ms, F(ms, m, "%s"), buf) == -1)
427 return -1;
428 break;
429 default:
430 if (file_printf(ms, F(ms, m, "%d"),
431 (unsigned char) v) == -1)
432 return -1;
433 break;
434 }
435 t = ms->offset + sizeof(char);
436 break;
437
438 case FILE_SHORT:
439 case FILE_BESHORT:
440 case FILE_LESHORT:
441 v = file_signextend(ms, m, (uint64_t)p->h);
442 switch (check_fmt(ms, m)) {
443 case -1:
444 return -1;
445 case 1:
446 (void)snprintf(buf, sizeof(buf), "%u",
447 (unsigned short)v);
448 if (file_printf(ms, F(ms, m, "%s"), buf) == -1)
449 return -1;
450 break;
451 default:
452 if (file_printf(ms, F(ms, m, "%u"),
453 (unsigned short) v) == -1)
454 return -1;
455 break;
456 }
457 t = ms->offset + sizeof(short);
458 break;
459
460 case FILE_LONG:
461 case FILE_BELONG:
462 case FILE_LELONG:
463 case FILE_MELONG:
464 v = file_signextend(ms, m, (uint64_t)p->l);
465 switch (check_fmt(ms, m)) {
466 case -1:
467 return -1;
468 case 1:
469 (void)snprintf(buf, sizeof(buf), "%u", (uint32_t) v);
470 if (file_printf(ms, F(ms, m, "%s"), buf) == -1)
471 return -1;
472 break;
473 default:
474 if (file_printf(ms, F(ms, m, "%u"), (uint32_t) v) == -1)
475 return -1;
476 break;
477 }
478 t = ms->offset + sizeof(int32_t);
479 break;
480
481 case FILE_QUAD:
482 case FILE_BEQUAD:
483 case FILE_LEQUAD:
484 v = file_signextend(ms, m, p->q);
485 switch (check_fmt(ms, m)) {
486 case -1:
487 return -1;
488 case 1:
489 (void)snprintf(buf, sizeof(buf), "%" INT64_T_FORMAT "u",
490 (unsigned long long)v);
491 if (file_printf(ms, F(ms, m, "%s"), buf) == -1)
492 return -1;
493 break;
494 default:
495 if (file_printf(ms, F(ms, m, "%" INT64_T_FORMAT "u"),
496 (unsigned long long) v) == -1)
497 return -1;
498 break;
499 }
500 t = ms->offset + sizeof(int64_t);
501 break;
502
503 case FILE_STRING:
504 case FILE_PSTRING:
505 case FILE_BESTRING16:
506 case FILE_LESTRING16:
507 if (m->reln == '=' || m->reln == '!') {
508 if (file_printf(ms, F(ms, m, "%s"),
509 file_printable(sbuf, sizeof(sbuf), m->value.s))
510 == -1)
511 return -1;
512 t = ms->offset + m->vallen;
513 }
514 else {
515 char *str = p->s;
516
517 /* compute t before we mangle the string? */
518 t = ms->offset + strlen(str);
519
520 if (*m->value.s == '\0')
521 str[strcspn(str, "\n")] = '\0';
522
523 if (m->str_flags & STRING_TRIM) {
524 char *last;
525 while (isspace((unsigned char)*str))
526 str++;
527 last = str;
528 while (*last)
529 last++;
530 --last;
531 while (isspace((unsigned char)*last))
532 last--;
533 *++last = '\0';
534 }
535
536 if (file_printf(ms, F(ms, m, "%s"),
537 file_printable(sbuf, sizeof(sbuf), str)) == -1)
538 return -1;
539
540 if (m->type == FILE_PSTRING)
541 t += file_pstring_length_size(m);
542 }
543 break;
544
545 case FILE_DATE:
546 case FILE_BEDATE:
547 case FILE_LEDATE:
548 case FILE_MEDATE:
549 if (file_printf(ms, F(ms, m, "%s"),
550 file_fmttime(p->l + m->num_mask, FILE_T_LOCAL, tbuf)) == -1)
551 return -1;
552 t = ms->offset + sizeof(uint32_t);
553 break;
554
555 case FILE_LDATE:
556 case FILE_BELDATE:
557 case FILE_LELDATE:
558 case FILE_MELDATE:
559 if (file_printf(ms, F(ms, m, "%s"),
560 file_fmttime(p->l + m->num_mask, 0, tbuf)) == -1)
561 return -1;
562 t = ms->offset + sizeof(uint32_t);
563 break;
564
565 case FILE_QDATE:
566 case FILE_BEQDATE:
567 case FILE_LEQDATE:
568 if (file_printf(ms, F(ms, m, "%s"),
569 file_fmttime(p->q + m->num_mask, FILE_T_LOCAL, tbuf)) == -1)
570 return -1;
571 t = ms->offset + sizeof(uint64_t);
572 break;
573
574 case FILE_QLDATE:
575 case FILE_BEQLDATE:
576 case FILE_LEQLDATE:
577 if (file_printf(ms, F(ms, m, "%s"),
578 file_fmttime(p->q + m->num_mask, 0, tbuf)) == -1)
579 return -1;
580 t = ms->offset + sizeof(uint64_t);
581 break;
582
583 case FILE_QWDATE:
584 case FILE_BEQWDATE:
585 case FILE_LEQWDATE:
586 if (file_printf(ms, F(ms, m, "%s"),
587 file_fmttime(p->q + m->num_mask, FILE_T_WINDOWS, tbuf)) == -1)
588 return -1;
589 t = ms->offset + sizeof(uint64_t);
590 break;
591
592 case FILE_FLOAT:
593 case FILE_BEFLOAT:
594 case FILE_LEFLOAT:
595 vf = p->f;
596 switch (check_fmt(ms, m)) {
597 case -1:
598 return -1;
599 case 1:
600 (void)snprintf(buf, sizeof(buf), "%g", vf);
601 if (file_printf(ms, F(ms, m, "%s"), buf) == -1)
602 return -1;
603 break;
604 default:
605 if (file_printf(ms, F(ms, m, "%g"), vf) == -1)
606 return -1;
607 break;
608 }
609 t = ms->offset + sizeof(float);
610 break;
611
612 case FILE_DOUBLE:
613 case FILE_BEDOUBLE:
614 case FILE_LEDOUBLE:
615 vd = p->d;
616 switch (check_fmt(ms, m)) {
617 case -1:
618 return -1;
619 case 1:
620 (void)snprintf(buf, sizeof(buf), "%g", vd);
621 if (file_printf(ms, F(ms, m, "%s"), buf) == -1)
622 return -1;
623 break;
624 default:
625 if (file_printf(ms, F(ms, m, "%g"), vd) == -1)
626 return -1;
627 break;
628 }
629 t = ms->offset + sizeof(double);
630 break;
631
632 case FILE_REGEX: {
633 char *cp;
634 int rval;
635
636 cp = strndup((const char *)ms->search.s, ms->search.rm_len);
637 if (cp == NULL) {
638 file_oomem(ms, ms->search.rm_len);
639 return -1;
640 }
641 rval = file_printf(ms, F(ms, m, "%s"),
642 file_printable(sbuf, sizeof(sbuf), cp));
643 free(cp);
644
645 if (rval == -1)
646 return -1;
647
648 if ((m->str_flags & REGEX_OFFSET_START))
649 t = ms->search.offset;
650 else
651 t = ms->search.offset + ms->search.rm_len;
652 break;
653 }
654
655 case FILE_SEARCH:
656 if (file_printf(ms, F(ms, m, "%s"),
657 file_printable(sbuf, sizeof(sbuf), m->value.s)) == -1)
658 return -1;
659 if ((m->str_flags & REGEX_OFFSET_START))
660 t = ms->search.offset;
661 else
662 t = ms->search.offset + m->vallen;
663 break;
664
665 case FILE_DEFAULT:
666 case FILE_CLEAR:
667 if (file_printf(ms, "%s", m->desc) == -1)
668 return -1;
669 t = ms->offset;
670 break;
671
672 case FILE_INDIRECT:
673 case FILE_USE:
674 case FILE_NAME:
675 t = ms->offset;
676 break;
677
678 default:
679 file_magerror(ms, "invalid m->type (%d) in mprint()", m->type);
680 return -1;
681 }
682 return (int32_t)t;
683 }
684
685 private int32_t
686 moffset(struct magic_set *ms, struct magic *m)
687 {
688 switch (m->type) {
689 case FILE_BYTE:
690 return CAST(int32_t, (ms->offset + sizeof(char)));
691
692 case FILE_SHORT:
693 case FILE_BESHORT:
694 case FILE_LESHORT:
695 return CAST(int32_t, (ms->offset + sizeof(short)));
696
697 case FILE_LONG:
698 case FILE_BELONG:
699 case FILE_LELONG:
700 case FILE_MELONG:
701 return CAST(int32_t, (ms->offset + sizeof(int32_t)));
702
703 case FILE_QUAD:
704 case FILE_BEQUAD:
705 case FILE_LEQUAD:
706 return CAST(int32_t, (ms->offset + sizeof(int64_t)));
707
708 case FILE_STRING:
709 case FILE_PSTRING:
710 case FILE_BESTRING16:
711 case FILE_LESTRING16:
712 if (m->reln == '=' || m->reln == '!')
713 return ms->offset + m->vallen;
714 else {
715 union VALUETYPE *p = &ms->ms_value;
716 uint32_t t;
717
718 if (*m->value.s == '\0')
719 p->s[strcspn(p->s, "\n")] = '\0';
720 t = CAST(uint32_t, (ms->offset + strlen(p->s)));
721 if (m->type == FILE_PSTRING)
722 t += (uint32_t)file_pstring_length_size(m);
723 return t;
724 }
725
726 case FILE_DATE:
727 case FILE_BEDATE:
728 case FILE_LEDATE:
729 case FILE_MEDATE:
730 return CAST(int32_t, (ms->offset + sizeof(uint32_t)));
731
732 case FILE_LDATE:
733 case FILE_BELDATE:
734 case FILE_LELDATE:
735 case FILE_MELDATE:
736 return CAST(int32_t, (ms->offset + sizeof(uint32_t)));
737
738 case FILE_QDATE:
739 case FILE_BEQDATE:
740 case FILE_LEQDATE:
741 return CAST(int32_t, (ms->offset + sizeof(uint64_t)));
742
743 case FILE_QLDATE:
744 case FILE_BEQLDATE:
745 case FILE_LEQLDATE:
746 return CAST(int32_t, (ms->offset + sizeof(uint64_t)));
747
748 case FILE_FLOAT:
749 case FILE_BEFLOAT:
750 case FILE_LEFLOAT:
751 return CAST(int32_t, (ms->offset + sizeof(float)));
752
753 case FILE_DOUBLE:
754 case FILE_BEDOUBLE:
755 case FILE_LEDOUBLE:
756 return CAST(int32_t, (ms->offset + sizeof(double)));
757
758 case FILE_REGEX:
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 +
763 ms->search.rm_len));
764
765 case FILE_SEARCH:
766 if ((m->str_flags & REGEX_OFFSET_START) != 0)
767 return CAST(int32_t, ms->search.offset);
768 else
769 return CAST(int32_t, (ms->search.offset + m->vallen));
770
771 case FILE_CLEAR:
772 case FILE_DEFAULT:
773 case FILE_INDIRECT:
774 return ms->offset;
775
776 default:
777 return 0;
778 }
779 }
780
781 private int
782 cvt_flip(int type, int flip)
783 {
784 if (flip == 0)
785 return type;
786 switch (type) {
787 case FILE_BESHORT:
788 return FILE_LESHORT;
789 case FILE_BELONG:
790 return FILE_LELONG;
791 case FILE_BEDATE:
792 return FILE_LEDATE;
793 case FILE_BELDATE:
794 return FILE_LELDATE;
795 case FILE_BEQUAD:
796 return FILE_LEQUAD;
797 case FILE_BEQDATE:
798 return FILE_LEQDATE;
799 case FILE_BEQLDATE:
800 return FILE_LEQLDATE;
801 case FILE_BEQWDATE:
802 return FILE_LEQWDATE;
803 case FILE_LESHORT:
804 return FILE_BESHORT;
805 case FILE_LELONG:
806 return FILE_BELONG;
807 case FILE_LEDATE:
808 return FILE_BEDATE;
809 case FILE_LELDATE:
810 return FILE_BELDATE;
811 case FILE_LEQUAD:
812 return FILE_BEQUAD;
813 case FILE_LEQDATE:
814 return FILE_BEQDATE;
815 case FILE_LEQLDATE:
816 return FILE_BEQLDATE;
817 case FILE_LEQWDATE:
818 return FILE_BEQWDATE;
819 case FILE_BEFLOAT:
820 return FILE_LEFLOAT;
821 case FILE_LEFLOAT:
822 return FILE_BEFLOAT;
823 case FILE_BEDOUBLE:
824 return FILE_LEDOUBLE;
825 case FILE_LEDOUBLE:
826 return FILE_BEDOUBLE;
827 default:
828 return type;
829 }
830 }
831 #define DO_CVT(fld, cast) \
832 if (m->num_mask) \
833 switch (m->mask_op & FILE_OPS_MASK) { \
834 case FILE_OPAND: \
835 p->fld &= cast m->num_mask; \
836 break; \
837 case FILE_OPOR: \
838 p->fld |= cast m->num_mask; \
839 break; \
840 case FILE_OPXOR: \
841 p->fld ^= cast m->num_mask; \
842 break; \
843 case FILE_OPADD: \
844 p->fld += cast m->num_mask; \
845 break; \
846 case FILE_OPMINUS: \
847 p->fld -= cast m->num_mask; \
848 break; \
849 case FILE_OPMULTIPLY: \
850 p->fld *= cast m->num_mask; \
851 break; \
852 case FILE_OPDIVIDE: \
853 p->fld /= cast m->num_mask; \
854 break; \
855 case FILE_OPMODULO: \
856 p->fld %= cast m->num_mask; \
857 break; \
858 } \
859 if (m->mask_op & FILE_OPINVERSE) \
860 p->fld = ~p->fld \
861
862 private void
863 cvt_8(union VALUETYPE *p, const struct magic *m)
864 {
865 DO_CVT(b, (uint8_t));
866 }
867
868 private void
869 cvt_16(union VALUETYPE *p, const struct magic *m)
870 {
871 DO_CVT(h, (uint16_t));
872 }
873
874 private void
875 cvt_32(union VALUETYPE *p, const struct magic *m)
876 {
877 DO_CVT(l, (uint32_t));
878 }
879
880 private void
881 cvt_64(union VALUETYPE *p, const struct magic *m)
882 {
883 DO_CVT(q, (uint64_t));
884 }
885
886 #define DO_CVT2(fld, cast) \
887 if (m->num_mask) \
888 switch (m->mask_op & FILE_OPS_MASK) { \
889 case FILE_OPADD: \
890 p->fld += cast m->num_mask; \
891 break; \
892 case FILE_OPMINUS: \
893 p->fld -= cast m->num_mask; \
894 break; \
895 case FILE_OPMULTIPLY: \
896 p->fld *= cast m->num_mask; \
897 break; \
898 case FILE_OPDIVIDE: \
899 p->fld /= cast m->num_mask; \
900 break; \
901 } \
902
903 private void
904 cvt_float(union VALUETYPE *p, const struct magic *m)
905 {
906 DO_CVT2(f, (float));
907 }
908
909 private void
910 cvt_double(union VALUETYPE *p, const struct magic *m)
911 {
912 DO_CVT2(d, (double));
913 }
914
915 /*
916 * Convert the byte order of the data we are looking at
917 * While we're here, let's apply the mask operation
918 * (unless you have a better idea)
919 */
920 private int
921 mconvert(struct magic_set *ms, struct magic *m, int flip)
922 {
923 union VALUETYPE *p = &ms->ms_value;
924 uint8_t type;
925
926 switch (type = cvt_flip(m->type, flip)) {
927 case FILE_BYTE:
928 cvt_8(p, m);
929 return 1;
930 case FILE_SHORT:
931 cvt_16(p, m);
932 return 1;
933 case FILE_LONG:
934 case FILE_DATE:
935 case FILE_LDATE:
936 cvt_32(p, m);
937 return 1;
938 case FILE_QUAD:
939 case FILE_QDATE:
940 case FILE_QLDATE:
941 case FILE_QWDATE:
942 cvt_64(p, m);
943 return 1;
944 case FILE_STRING:
945 case FILE_BESTRING16:
946 case FILE_LESTRING16: {
947 /* Null terminate and eat *trailing* return */
948 p->s[sizeof(p->s) - 1] = '\0';
949 return 1;
950 }
951 case FILE_PSTRING: {
952 size_t sz = file_pstring_length_size(m);
953 char *ptr1 = p->s, *ptr2 = ptr1 + sz;
954 size_t len = file_pstring_get_length(m, ptr1);
955 sz = sizeof(p->s) - sz; /* maximum length of string */
956 if (len >= sz) {
957 /*
958 * The size of the pascal string length (sz)
959 * is 1, 2, or 4. We need at least 1 byte for NUL
960 * termination, but we've already truncated the
961 * string by p->s, so we need to deduct sz.
962 * Because we can use one of the bytes of the length
963 * after we shifted as NUL termination.
964 */
965 len = sz;
966 }
967 while (len--)
968 *ptr1++ = *ptr2++;
969 *ptr1 = '\0';
970 return 1;
971 }
972 case FILE_BESHORT:
973 p->h = (short)((p->hs[0]<<8)|(p->hs[1]));
974 cvt_16(p, m);
975 return 1;
976 case FILE_BELONG:
977 case FILE_BEDATE:
978 case FILE_BELDATE:
979 p->l = (int32_t)
980 ((p->hl[0]<<24)|(p->hl[1]<<16)|(p->hl[2]<<8)|(p->hl[3]));
981 if (type == FILE_BELONG)
982 cvt_32(p, m);
983 return 1;
984 case FILE_BEQUAD:
985 case FILE_BEQDATE:
986 case FILE_BEQLDATE:
987 case FILE_BEQWDATE:
988 p->q = (uint64_t)
989 (((uint64_t)p->hq[0]<<56)|((uint64_t)p->hq[1]<<48)|
990 ((uint64_t)p->hq[2]<<40)|((uint64_t)p->hq[3]<<32)|
991 ((uint64_t)p->hq[4]<<24)|((uint64_t)p->hq[5]<<16)|
992 ((uint64_t)p->hq[6]<<8)|((uint64_t)p->hq[7]));
993 if (type == FILE_BEQUAD)
994 cvt_64(p, m);
995 return 1;
996 case FILE_LESHORT:
997 p->h = (short)((p->hs[1]<<8)|(p->hs[0]));
998 cvt_16(p, m);
999 return 1;
1000 case FILE_LELONG:
1001 case FILE_LEDATE:
1002 case FILE_LELDATE:
1003 p->l = (int32_t)
1004 ((p->hl[3]<<24)|(p->hl[2]<<16)|(p->hl[1]<<8)|(p->hl[0]));
1005 if (type == FILE_LELONG)
1006 cvt_32(p, m);
1007 return 1;
1008 case FILE_LEQUAD:
1009 case FILE_LEQDATE:
1010 case FILE_LEQLDATE:
1011 case FILE_LEQWDATE:
1012 p->q = (uint64_t)
1013 (((uint64_t)p->hq[7]<<56)|((uint64_t)p->hq[6]<<48)|
1014 ((uint64_t)p->hq[5]<<40)|((uint64_t)p->hq[4]<<32)|
1015 ((uint64_t)p->hq[3]<<24)|((uint64_t)p->hq[2]<<16)|
1016 ((uint64_t)p->hq[1]<<8)|((uint64_t)p->hq[0]));
1017 if (type == FILE_LEQUAD)
1018 cvt_64(p, m);
1019 return 1;
1020 case FILE_MELONG:
1021 case FILE_MEDATE:
1022 case FILE_MELDATE:
1023 p->l = (int32_t)
1024 ((p->hl[1]<<24)|(p->hl[0]<<16)|(p->hl[3]<<8)|(p->hl[2]));
1025 if (type == FILE_MELONG)
1026 cvt_32(p, m);
1027 return 1;
1028 case FILE_FLOAT:
1029 cvt_float(p, m);
1030 return 1;
1031 case FILE_BEFLOAT:
1032 p->l = ((uint32_t)p->hl[0]<<24)|((uint32_t)p->hl[1]<<16)|
1033 ((uint32_t)p->hl[2]<<8) |((uint32_t)p->hl[3]);
1034 cvt_float(p, m);
1035 return 1;
1036 case FILE_LEFLOAT:
1037 p->l = ((uint32_t)p->hl[3]<<24)|((uint32_t)p->hl[2]<<16)|
1038 ((uint32_t)p->hl[1]<<8) |((uint32_t)p->hl[0]);
1039 cvt_float(p, m);
1040 return 1;
1041 case FILE_DOUBLE:
1042 cvt_double(p, m);
1043 return 1;
1044 case FILE_BEDOUBLE:
1045 p->q = ((uint64_t)p->hq[0]<<56)|((uint64_t)p->hq[1]<<48)|
1046 ((uint64_t)p->hq[2]<<40)|((uint64_t)p->hq[3]<<32)|
1047 ((uint64_t)p->hq[4]<<24)|((uint64_t)p->hq[5]<<16)|
1048 ((uint64_t)p->hq[6]<<8) |((uint64_t)p->hq[7]);
1049 cvt_double(p, m);
1050 return 1;
1051 case FILE_LEDOUBLE:
1052 p->q = ((uint64_t)p->hq[7]<<56)|((uint64_t)p->hq[6]<<48)|
1053 ((uint64_t)p->hq[5]<<40)|((uint64_t)p->hq[4]<<32)|
1054 ((uint64_t)p->hq[3]<<24)|((uint64_t)p->hq[2]<<16)|
1055 ((uint64_t)p->hq[1]<<8) |((uint64_t)p->hq[0]);
1056 cvt_double(p, m);
1057 return 1;
1058 case FILE_REGEX:
1059 case FILE_SEARCH:
1060 case FILE_DEFAULT:
1061 case FILE_CLEAR:
1062 case FILE_NAME:
1063 case FILE_USE:
1064 return 1;
1065 default:
1066 file_magerror(ms, "invalid type %d in mconvert()", m->type);
1067 return 0;
1068 }
1069 }
1070
1071
1072 private void
1073 mdebug(uint32_t offset, const char *str, size_t len)
1074 {
1075 (void) fprintf(stderr, "mget/%" SIZE_T_FORMAT "u @%d: ", len, offset);
1076 file_showstr(stderr, str, len);
1077 (void) fputc('\n', stderr);
1078 (void) fputc('\n', stderr);
1079 }
1080
1081 private int
1082 mcopy(struct magic_set *ms, union VALUETYPE *p, int type, int indir,
1083 const unsigned char *s, uint32_t offset, size_t nbytes, struct magic *m)
1084 {
1085 /*
1086 * Note: FILE_SEARCH and FILE_REGEX do not actually copy
1087 * anything, but setup pointers into the source
1088 */
1089 if (indir == 0) {
1090 switch (type) {
1091 case FILE_SEARCH:
1092 ms->search.s = RCAST(const char *, s) + offset;
1093 ms->search.s_len = nbytes - offset;
1094 ms->search.offset = offset;
1095 return 0;
1096
1097 case FILE_REGEX: {
1098 const char *b;
1099 const char *c;
1100 const char *last; /* end of search region */
1101 const char *buf; /* start of search region */
1102 const char *end;
1103 size_t lines, linecnt, bytecnt;
1104
1105 if (s == NULL) {
1106 ms->search.s_len = 0;
1107 ms->search.s = NULL;
1108 return 0;
1109 }
1110
1111 if (m->str_flags & REGEX_LINE_COUNT) {
1112 linecnt = m->str_range;
1113 bytecnt = linecnt * 80;
1114 } else {
1115 linecnt = 0;
1116 bytecnt = m->str_range;
1117 }
1118
1119 if (bytecnt == 0)
1120 bytecnt = 8192;
1121 if (bytecnt > nbytes)
1122 bytecnt = nbytes;
1123
1124 buf = RCAST(const char *, s) + offset;
1125 end = last = RCAST(const char *, s) + bytecnt;
1126 /* mget() guarantees buf <= last */
1127 for (lines = linecnt, b = buf; lines && b < end &&
1128 ((b = CAST(const char *,
1129 memchr(c = b, '\n', CAST(size_t, (end - b)))))
1130 || (b = CAST(const char *,
1131 memchr(c, '\r', CAST(size_t, (end - c))))));
1132 lines--, b++) {
1133 last = b;
1134 if (b[0] == '\r' && b[1] == '\n')
1135 b++;
1136 }
1137 if (lines)
1138 last = RCAST(const char *, s) + bytecnt;
1139
1140 ms->search.s = buf;
1141 ms->search.s_len = last - buf;
1142 ms->search.offset = offset;
1143 ms->search.rm_len = 0;
1144 return 0;
1145 }
1146 case FILE_BESTRING16:
1147 case FILE_LESTRING16: {
1148 const unsigned char *src = s + offset;
1149 const unsigned char *esrc = s + nbytes;
1150 char *dst = p->s;
1151 char *edst = &p->s[sizeof(p->s) - 1];
1152
1153 if (type == FILE_BESTRING16)
1154 src++;
1155
1156 /* check that offset is within range */
1157 if (offset >= nbytes)
1158 break;
1159 for (/*EMPTY*/; src < esrc; src += 2, dst++) {
1160 if (dst < edst)
1161 *dst = *src;
1162 else
1163 break;
1164 if (*dst == '\0') {
1165 if (type == FILE_BESTRING16 ?
1166 *(src - 1) != '\0' :
1167 *(src + 1) != '\0')
1168 *dst = ' ';
1169 }
1170 }
1171 *edst = '\0';
1172 return 0;
1173 }
1174 case FILE_STRING: /* XXX - these two should not need */
1175 case FILE_PSTRING: /* to copy anything, but do anyway. */
1176 default:
1177 break;
1178 }
1179 }
1180
1181 if (offset >= nbytes) {
1182 (void)memset(p, '\0', sizeof(*p));
1183 return 0;
1184 }
1185 if (nbytes - offset < sizeof(*p))
1186 nbytes = nbytes - offset;
1187 else
1188 nbytes = sizeof(*p);
1189
1190 (void)memcpy(p, s + offset, nbytes);
1191
1192 /*
1193 * the usefulness of padding with zeroes eludes me, it
1194 * might even cause problems
1195 */
1196 if (nbytes < sizeof(*p))
1197 (void)memset(((char *)(void *)p) + nbytes, '\0',
1198 sizeof(*p) - nbytes);
1199 return 0;
1200 }
1201
1202 private int
1203 mget(struct magic_set *ms, const unsigned char *s, struct magic *m,
1204 size_t nbytes, size_t o, unsigned int cont_level, int mode, int text,
1205 int flip, uint16_t indir_level, uint16_t *name_count,
1206 int *printed_something, int *need_separator, int *returnval)
1207 {
1208 uint32_t offset = ms->offset;
1209 uint32_t lhs;
1210 file_pushbuf_t *pb;
1211 int rv, oneed_separator, in_type;
1212 char *rbuf;
1213 union VALUETYPE *p = &ms->ms_value;
1214 struct mlist ml;
1215
1216 if (indir_level >= ms->indir_max) {
1217 file_error(ms, 0, "indirect recursion nesting (%hu) exceeded",
1218 indir_level);
1219 return -1;
1220 }
1221
1222 if (*name_count >= ms->name_max) {
1223 file_error(ms, 0, "name use count (%hu) exceeded",
1224 *name_count);
1225 return -1;
1226 }
1227
1228 if (mcopy(ms, p, m->type, m->flag & INDIR, s, (uint32_t)(offset + o),
1229 (uint32_t)nbytes, m) == -1)
1230 return -1;
1231
1232 if ((ms->flags & MAGIC_DEBUG) != 0) {
1233 fprintf(stderr, "mget(type=%d, flag=%x, offset=%u, o=%"
1234 SIZE_T_FORMAT "u, " "nbytes=%" SIZE_T_FORMAT
1235 "u, il=%hu, nc=%hu)\n",
1236 m->type, m->flag, offset, o, nbytes,
1237 indir_level, *name_count);
1238 mdebug(offset, (char *)(void *)p, sizeof(union VALUETYPE));
1239 #ifndef COMPILE_ONLY
1240 file_mdump(m);
1241 #endif
1242 }
1243
1244 if (m->flag & INDIR) {
1245 int off = m->in_offset;
1246 if (m->in_op & FILE_OPINDIRECT) {
1247 const union VALUETYPE *q = CAST(const union VALUETYPE *,
1248 ((const void *)(s + offset + off)));
1249 switch (cvt_flip(m->in_type, flip)) {
1250 case FILE_BYTE:
1251 off = q->b;
1252 break;
1253 case FILE_SHORT:
1254 off = q->h;
1255 break;
1256 case FILE_BESHORT:
1257 off = (short)((q->hs[0]<<8)|(q->hs[1]));
1258 break;
1259 case FILE_LESHORT:
1260 off = (short)((q->hs[1]<<8)|(q->hs[0]));
1261 break;
1262 case FILE_LONG:
1263 off = q->l;
1264 break;
1265 case FILE_BELONG:
1266 case FILE_BEID3:
1267 off = (int32_t)((q->hl[0]<<24)|(q->hl[1]<<16)|
1268 (q->hl[2]<<8)|(q->hl[3]));
1269 break;
1270 case FILE_LEID3:
1271 case FILE_LELONG:
1272 off = (int32_t)((q->hl[3]<<24)|(q->hl[2]<<16)|
1273 (q->hl[1]<<8)|(q->hl[0]));
1274 break;
1275 case FILE_MELONG:
1276 off = (int32_t)((q->hl[1]<<24)|(q->hl[0]<<16)|
1277 (q->hl[3]<<8)|(q->hl[2]));
1278 break;
1279 }
1280 if ((ms->flags & MAGIC_DEBUG) != 0)
1281 fprintf(stderr, "indirect offs=%u\n", off);
1282 }
1283 switch (in_type = cvt_flip(m->in_type, flip)) {
1284 case FILE_BYTE:
1285 if (OFFSET_OOB(nbytes, offset, 1))
1286 return 0;
1287 if (off) {
1288 switch (m->in_op & FILE_OPS_MASK) {
1289 case FILE_OPAND:
1290 offset = p->b & off;
1291 break;
1292 case FILE_OPOR:
1293 offset = p->b | off;
1294 break;
1295 case FILE_OPXOR:
1296 offset = p->b ^ off;
1297 break;
1298 case FILE_OPADD:
1299 offset = p->b + off;
1300 break;
1301 case FILE_OPMINUS:
1302 offset = p->b - off;
1303 break;
1304 case FILE_OPMULTIPLY:
1305 offset = p->b * off;
1306 break;
1307 case FILE_OPDIVIDE:
1308 offset = p->b / off;
1309 break;
1310 case FILE_OPMODULO:
1311 offset = p->b % off;
1312 break;
1313 }
1314 } else
1315 offset = p->b;
1316 if (m->in_op & FILE_OPINVERSE)
1317 offset = ~offset;
1318 break;
1319 case FILE_BESHORT:
1320 if (OFFSET_OOB(nbytes, offset, 2))
1321 return 0;
1322 lhs = (p->hs[0] << 8) | p->hs[1];
1323 if (off) {
1324 switch (m->in_op & FILE_OPS_MASK) {
1325 case FILE_OPAND:
1326 offset = lhs & off;
1327 break;
1328 case FILE_OPOR:
1329 offset = lhs | off;
1330 break;
1331 case FILE_OPXOR:
1332 offset = lhs ^ off;
1333 break;
1334 case FILE_OPADD:
1335 offset = lhs + off;
1336 break;
1337 case FILE_OPMINUS:
1338 offset = lhs - off;
1339 break;
1340 case FILE_OPMULTIPLY:
1341 offset = lhs * off;
1342 break;
1343 case FILE_OPDIVIDE:
1344 offset = lhs / off;
1345 break;
1346 case FILE_OPMODULO:
1347 offset = lhs % off;
1348 break;
1349 }
1350 } else
1351 offset = lhs;
1352 if (m->in_op & FILE_OPINVERSE)
1353 offset = ~offset;
1354 break;
1355 case FILE_LESHORT:
1356 if (OFFSET_OOB(nbytes, offset, 2))
1357 return 0;
1358 lhs = (p->hs[1] << 8) | p->hs[0];
1359 if (off) {
1360 switch (m->in_op & FILE_OPS_MASK) {
1361 case FILE_OPAND:
1362 offset = lhs & off;
1363 break;
1364 case FILE_OPOR:
1365 offset = lhs | off;
1366 break;
1367 case FILE_OPXOR:
1368 offset = lhs ^ off;
1369 break;
1370 case FILE_OPADD:
1371 offset = lhs + off;
1372 break;
1373 case FILE_OPMINUS:
1374 offset = lhs - off;
1375 break;
1376 case FILE_OPMULTIPLY:
1377 offset = lhs * off;
1378 break;
1379 case FILE_OPDIVIDE:
1380 offset = lhs / off;
1381 break;
1382 case FILE_OPMODULO:
1383 offset = lhs % off;
1384 break;
1385 }
1386 } else
1387 offset = lhs;
1388 if (m->in_op & FILE_OPINVERSE)
1389 offset = ~offset;
1390 break;
1391 case FILE_SHORT:
1392 if (OFFSET_OOB(nbytes, offset, 2))
1393 return 0;
1394 if (off) {
1395 switch (m->in_op & FILE_OPS_MASK) {
1396 case FILE_OPAND:
1397 offset = p->h & off;
1398 break;
1399 case FILE_OPOR:
1400 offset = p->h | off;
1401 break;
1402 case FILE_OPXOR:
1403 offset = p->h ^ off;
1404 break;
1405 case FILE_OPADD:
1406 offset = p->h + off;
1407 break;
1408 case FILE_OPMINUS:
1409 offset = p->h - off;
1410 break;
1411 case FILE_OPMULTIPLY:
1412 offset = p->h * off;
1413 break;
1414 case FILE_OPDIVIDE:
1415 offset = p->h / off;
1416 break;
1417 case FILE_OPMODULO:
1418 offset = p->h % off;
1419 break;
1420 }
1421 }
1422 else
1423 offset = p->h;
1424 if (m->in_op & FILE_OPINVERSE)
1425 offset = ~offset;
1426 break;
1427 case FILE_BELONG:
1428 case FILE_BEID3:
1429 if (OFFSET_OOB(nbytes, offset, 4))
1430 return 0;
1431 lhs = (p->hl[0] << 24) | (p->hl[1] << 16) |
1432 (p->hl[2] << 8) | p->hl[3];
1433 if (off) {
1434 switch (m->in_op & FILE_OPS_MASK) {
1435 case FILE_OPAND:
1436 offset = lhs & off;
1437 break;
1438 case FILE_OPOR:
1439 offset = lhs | off;
1440 break;
1441 case FILE_OPXOR:
1442 offset = lhs ^ off;
1443 break;
1444 case FILE_OPADD:
1445 offset = lhs + off;
1446 break;
1447 case FILE_OPMINUS:
1448 offset = lhs - off;
1449 break;
1450 case FILE_OPMULTIPLY:
1451 offset = lhs * off;
1452 break;
1453 case FILE_OPDIVIDE:
1454 offset = lhs / off;
1455 break;
1456 case FILE_OPMODULO:
1457 offset = lhs % off;
1458 break;
1459 }
1460 } else
1461 offset = lhs;
1462 if (m->in_op & FILE_OPINVERSE)
1463 offset = ~offset;
1464 break;
1465 case FILE_LELONG:
1466 case FILE_LEID3:
1467 if (OFFSET_OOB(nbytes, offset, 4))
1468 return 0;
1469 lhs = (p->hl[3] << 24) | (p->hl[2] << 16) |
1470 (p->hl[1] << 8) | p->hl[0];
1471 if (off) {
1472 switch (m->in_op & FILE_OPS_MASK) {
1473 case FILE_OPAND:
1474 offset = lhs & off;
1475 break;
1476 case FILE_OPOR:
1477 offset = lhs | off;
1478 break;
1479 case FILE_OPXOR:
1480 offset = lhs ^ off;
1481 break;
1482 case FILE_OPADD:
1483 offset = lhs + off;
1484 break;
1485 case FILE_OPMINUS:
1486 offset = lhs - off;
1487 break;
1488 case FILE_OPMULTIPLY:
1489 offset = lhs * off;
1490 break;
1491 case FILE_OPDIVIDE:
1492 offset = lhs / off;
1493 break;
1494 case FILE_OPMODULO:
1495 offset = lhs % off;
1496 break;
1497 }
1498 } else
1499 offset = lhs;
1500 if (m->in_op & FILE_OPINVERSE)
1501 offset = ~offset;
1502 break;
1503 case FILE_MELONG:
1504 if (OFFSET_OOB(nbytes, offset, 4))
1505 return 0;
1506 lhs = (p->hl[1] << 24) | (p->hl[0] << 16) |
1507 (p->hl[3] << 8) | p->hl[2];
1508 if (off) {
1509 switch (m->in_op & FILE_OPS_MASK) {
1510 case FILE_OPAND:
1511 offset = lhs & off;
1512 break;
1513 case FILE_OPOR:
1514 offset = lhs | off;
1515 break;
1516 case FILE_OPXOR:
1517 offset = lhs ^ off;
1518 break;
1519 case FILE_OPADD:
1520 offset = lhs + off;
1521 break;
1522 case FILE_OPMINUS:
1523 offset = lhs - off;
1524 break;
1525 case FILE_OPMULTIPLY:
1526 offset = lhs * off;
1527 break;
1528 case FILE_OPDIVIDE:
1529 offset = lhs / off;
1530 break;
1531 case FILE_OPMODULO:
1532 offset = lhs % off;
1533 break;
1534 }
1535 } else
1536 offset = lhs;
1537 if (m->in_op & FILE_OPINVERSE)
1538 offset = ~offset;
1539 break;
1540 case FILE_LONG:
1541 if (OFFSET_OOB(nbytes, offset, 4))
1542 return 0;
1543 if (off) {
1544 switch (m->in_op & FILE_OPS_MASK) {
1545 case FILE_OPAND:
1546 offset = p->l & off;
1547 break;
1548 case FILE_OPOR:
1549 offset = p->l | off;
1550 break;
1551 case FILE_OPXOR:
1552 offset = p->l ^ off;
1553 break;
1554 case FILE_OPADD:
1555 offset = p->l + off;
1556 break;
1557 case FILE_OPMINUS:
1558 offset = p->l - off;
1559 break;
1560 case FILE_OPMULTIPLY:
1561 offset = p->l * off;
1562 break;
1563 case FILE_OPDIVIDE:
1564 offset = p->l / off;
1565 break;
1566 case FILE_OPMODULO:
1567 offset = p->l % off;
1568 break;
1569 }
1570 } else
1571 offset = p->l;
1572 if (m->in_op & FILE_OPINVERSE)
1573 offset = ~offset;
1574 break;
1575 default:
1576 break;
1577 }
1578
1579 switch (in_type) {
1580 case FILE_LEID3:
1581 case FILE_BEID3:
1582 offset = ((((offset >> 0) & 0x7f) << 0) |
1583 (((offset >> 8) & 0x7f) << 7) |
1584 (((offset >> 16) & 0x7f) << 14) |
1585 (((offset >> 24) & 0x7f) << 21)) + 10;
1586 break;
1587 default:
1588 break;
1589 }
1590
1591 if (m->flag & INDIROFFADD) {
1592 offset += ms->c.li[cont_level-1].off;
1593 if (offset == 0) {
1594 if ((ms->flags & MAGIC_DEBUG) != 0)
1595 fprintf(stderr,
1596 "indirect *zero* offset\n");
1597 return 0;
1598 }
1599 if ((ms->flags & MAGIC_DEBUG) != 0)
1600 fprintf(stderr, "indirect +offs=%u\n", offset);
1601 }
1602 if (mcopy(ms, p, m->type, 0, s, offset, nbytes, m) == -1)
1603 return -1;
1604 ms->offset = offset;
1605
1606 if ((ms->flags & MAGIC_DEBUG) != 0) {
1607 mdebug(offset, (char *)(void *)p,
1608 sizeof(union VALUETYPE));
1609 #ifndef COMPILE_ONLY
1610 file_mdump(m);
1611 #endif
1612 }
1613 }
1614
1615 /* Verify we have enough data to match magic type */
1616 switch (m->type) {
1617 case FILE_BYTE:
1618 if (OFFSET_OOB(nbytes, offset, 1))
1619 return 0;
1620 break;
1621
1622 case FILE_SHORT:
1623 case FILE_BESHORT:
1624 case FILE_LESHORT:
1625 if (OFFSET_OOB(nbytes, offset, 2))
1626 return 0;
1627 break;
1628
1629 case FILE_LONG:
1630 case FILE_BELONG:
1631 case FILE_LELONG:
1632 case FILE_MELONG:
1633 case FILE_DATE:
1634 case FILE_BEDATE:
1635 case FILE_LEDATE:
1636 case FILE_MEDATE:
1637 case FILE_LDATE:
1638 case FILE_BELDATE:
1639 case FILE_LELDATE:
1640 case FILE_MELDATE:
1641 case FILE_FLOAT:
1642 case FILE_BEFLOAT:
1643 case FILE_LEFLOAT:
1644 if (OFFSET_OOB(nbytes, offset, 4))
1645 return 0;
1646 break;
1647
1648 case FILE_DOUBLE:
1649 case FILE_BEDOUBLE:
1650 case FILE_LEDOUBLE:
1651 if (OFFSET_OOB(nbytes, offset, 8))
1652 return 0;
1653 break;
1654
1655 case FILE_STRING:
1656 case FILE_PSTRING:
1657 case FILE_SEARCH:
1658 if (OFFSET_OOB(nbytes, offset, m->vallen))
1659 return 0;
1660 break;
1661
1662 case FILE_REGEX:
1663 if (nbytes < offset)
1664 return 0;
1665 break;
1666
1667 case FILE_INDIRECT:
1668 if (m->str_flags & INDIRECT_RELATIVE)
1669 offset += o;
1670 if (offset == 0)
1671 return 0;
1672
1673 if (nbytes < offset)
1674 return 0;
1675
1676 if ((pb = file_push_buffer(ms)) == NULL)
1677 return -1;
1678
1679 rv = file_softmagic(ms, s + offset, nbytes - offset,
1680 indir_level + 1, name_count, BINTEST, text);
1681
1682 if ((ms->flags & MAGIC_DEBUG) != 0)
1683 fprintf(stderr, "indirect @offs=%u[%d]\n", offset, rv);
1684
1685 rbuf = file_pop_buffer(ms, pb);
1686 if (rbuf == NULL && ms->event_flags & EVENT_HAD_ERR)
1687 return -1;
1688
1689 if (rv == 1) {
1690 if ((ms->flags & (MAGIC_MIME|MAGIC_APPLE)) == 0 &&
1691 file_printf(ms, F(ms, m, "%u"), offset) == -1) {
1692 free(rbuf);
1693 return -1;
1694 }
1695 if (file_printf(ms, "%s", rbuf) == -1) {
1696 free(rbuf);
1697 return -1;
1698 }
1699 }
1700 free(rbuf);
1701 return rv;
1702
1703 case FILE_USE:
1704 if (nbytes < offset)
1705 return 0;
1706 rbuf = m->value.s;
1707 if (*rbuf == '^') {
1708 rbuf++;
1709 flip = !flip;
1710 }
1711 if (file_magicfind(ms, rbuf, &ml) == -1) {
1712 file_error(ms, 0, "cannot find entry `%s'", rbuf);
1713 return -1;
1714 }
1715 (*name_count)++;
1716 oneed_separator = *need_separator;
1717 if (m->flag & NOSPACE)
1718 *need_separator = 0;
1719 rv = match(ms, ml.magic, ml.nmagic, s, nbytes, offset + o,
1720 mode, text, flip, indir_level, name_count,
1721 printed_something, need_separator, returnval);
1722 if (rv != 1)
1723 *need_separator = oneed_separator;
1724 return rv;
1725
1726 case FILE_NAME:
1727 if (file_printf(ms, "%s", m->desc) == -1)
1728 return -1;
1729 return 1;
1730 case FILE_DEFAULT: /* nothing to check */
1731 case FILE_CLEAR:
1732 default:
1733 break;
1734 }
1735 if (!mconvert(ms, m, flip))
1736 return 0;
1737 return 1;
1738 }
1739
1740 private uint64_t
1741 file_strncmp(const char *s1, const char *s2, size_t len, uint32_t flags)
1742 {
1743 /*
1744 * Convert the source args to unsigned here so that (1) the
1745 * compare will be unsigned as it is in strncmp() and (2) so
1746 * the ctype functions will work correctly without extra
1747 * casting.
1748 */
1749 const unsigned char *a = (const unsigned char *)s1;
1750 const unsigned char *b = (const unsigned char *)s2;
1751 uint64_t v;
1752
1753 /*
1754 * What we want here is v = strncmp(s1, s2, len),
1755 * but ignoring any nulls.
1756 */
1757 v = 0;
1758 if (0L == flags) { /* normal string: do it fast */
1759 while (len-- > 0)
1760 if ((v = *b++ - *a++) != '\0')
1761 break;
1762 }
1763 else { /* combine the others */
1764 while (len-- > 0) {
1765 if ((flags & STRING_IGNORE_LOWERCASE) &&
1766 islower(*a)) {
1767 if ((v = tolower(*b++) - *a++) != '\0')
1768 break;
1769 }
1770 else if ((flags & STRING_IGNORE_UPPERCASE) &&
1771 isupper(*a)) {
1772 if ((v = toupper(*b++) - *a++) != '\0')
1773 break;
1774 }
1775 else if ((flags & STRING_COMPACT_WHITESPACE) &&
1776 isspace(*a)) {
1777 a++;
1778 if (isspace(*b++)) {
1779 if (!isspace(*a))
1780 while (isspace(*b))
1781 b++;
1782 }
1783 else {
1784 v = 1;
1785 break;
1786 }
1787 }
1788 else if ((flags & STRING_COMPACT_OPTIONAL_WHITESPACE) &&
1789 isspace(*a)) {
1790 a++;
1791 while (isspace(*b))
1792 b++;
1793 }
1794 else {
1795 if ((v = *b++ - *a++) != '\0')
1796 break;
1797 }
1798 }
1799 }
1800 return v;
1801 }
1802
1803 private uint64_t
1804 file_strncmp16(const char *a, const char *b, size_t len, uint32_t flags)
1805 {
1806 /*
1807 * XXX - The 16-bit string compare probably needs to be done
1808 * differently, especially if the flags are to be supported.
1809 * At the moment, I am unsure.
1810 */
1811 flags = 0;
1812 return file_strncmp(a, b, len, flags);
1813 }
1814
1815 private int
1816 magiccheck(struct magic_set *ms, struct magic *m)
1817 {
1818 uint64_t l = m->value.q;
1819 uint64_t v;
1820 float fl, fv;
1821 double dl, dv;
1822 int matched;
1823 union VALUETYPE *p = &ms->ms_value;
1824
1825 switch (m->type) {
1826 case FILE_BYTE:
1827 v = p->b;
1828 break;
1829
1830 case FILE_SHORT:
1831 case FILE_BESHORT:
1832 case FILE_LESHORT:
1833 v = p->h;
1834 break;
1835
1836 case FILE_LONG:
1837 case FILE_BELONG:
1838 case FILE_LELONG:
1839 case FILE_MELONG:
1840 case FILE_DATE:
1841 case FILE_BEDATE:
1842 case FILE_LEDATE:
1843 case FILE_MEDATE:
1844 case FILE_LDATE:
1845 case FILE_BELDATE:
1846 case FILE_LELDATE:
1847 case FILE_MELDATE:
1848 v = p->l;
1849 break;
1850
1851 case FILE_QUAD:
1852 case FILE_LEQUAD:
1853 case FILE_BEQUAD:
1854 case FILE_QDATE:
1855 case FILE_BEQDATE:
1856 case FILE_LEQDATE:
1857 case FILE_QLDATE:
1858 case FILE_BEQLDATE:
1859 case FILE_LEQLDATE:
1860 case FILE_QWDATE:
1861 case FILE_BEQWDATE:
1862 case FILE_LEQWDATE:
1863 v = p->q;
1864 break;
1865
1866 case FILE_FLOAT:
1867 case FILE_BEFLOAT:
1868 case FILE_LEFLOAT:
1869 fl = m->value.f;
1870 fv = p->f;
1871 switch (m->reln) {
1872 case 'x':
1873 matched = 1;
1874 break;
1875
1876 case '!':
1877 matched = fv != fl;
1878 break;
1879
1880 case '=':
1881 matched = fv == fl;
1882 break;
1883
1884 case '>':
1885 matched = fv > fl;
1886 break;
1887
1888 case '<':
1889 matched = fv < fl;
1890 break;
1891
1892 default:
1893 file_magerror(ms, "cannot happen with float: invalid relation `%c'",
1894 m->reln);
1895 return -1;
1896 }
1897 return matched;
1898
1899 case FILE_DOUBLE:
1900 case FILE_BEDOUBLE:
1901 case FILE_LEDOUBLE:
1902 dl = m->value.d;
1903 dv = p->d;
1904 switch (m->reln) {
1905 case 'x':
1906 matched = 1;
1907 break;
1908
1909 case '!':
1910 matched = dv != dl;
1911 break;
1912
1913 case '=':
1914 matched = dv == dl;
1915 break;
1916
1917 case '>':
1918 matched = dv > dl;
1919 break;
1920
1921 case '<':
1922 matched = dv < dl;
1923 break;
1924
1925 default:
1926 file_magerror(ms, "cannot happen with double: invalid relation `%c'", m->reln);
1927 return -1;
1928 }
1929 return matched;
1930
1931 case FILE_DEFAULT:
1932 case FILE_CLEAR:
1933 l = 0;
1934 v = 0;
1935 break;
1936
1937 case FILE_STRING:
1938 case FILE_PSTRING:
1939 l = 0;
1940 v = file_strncmp(m->value.s, p->s, (size_t)m->vallen, m->str_flags);
1941 break;
1942
1943 case FILE_BESTRING16:
1944 case FILE_LESTRING16:
1945 l = 0;
1946 v = file_strncmp16(m->value.s, p->s, (size_t)m->vallen, m->str_flags);
1947 break;
1948
1949 case FILE_SEARCH: { /* search ms->search.s for the string m->value.s */
1950 size_t slen;
1951 size_t idx;
1952
1953 if (ms->search.s == NULL)
1954 return 0;
1955
1956 slen = MIN(m->vallen, sizeof(m->value.s));
1957 l = 0;
1958 v = 0;
1959
1960 for (idx = 0; m->str_range == 0 || idx < m->str_range; idx++) {
1961 if (slen + idx > ms->search.s_len)
1962 break;
1963
1964 v = file_strncmp(m->value.s, ms->search.s + idx, slen,
1965 m->str_flags);
1966 if (v == 0) { /* found match */
1967 ms->search.offset += idx;
1968 break;
1969 }
1970 }
1971 break;
1972 }
1973 case FILE_REGEX: {
1974 int rc;
1975 file_regex_t rx;
1976 const char *search;
1977
1978 if (ms->search.s == NULL)
1979 return 0;
1980
1981 l = 0;
1982 rc = file_regcomp(&rx, m->value.s,
1983 REG_EXTENDED|REG_NEWLINE|
1984 ((m->str_flags & STRING_IGNORE_CASE) ? REG_ICASE : 0));
1985 if (rc) {
1986 file_regerror(&rx, rc, ms);
1987 v = (uint64_t)-1;
1988 } else {
1989 regmatch_t pmatch[1];
1990 size_t slen = ms->search.s_len;
1991 #ifndef REG_STARTEND
1992 #define REG_STARTEND 0
1993 char *copy;
1994 if (slen != 0) {
1995 copy = malloc(slen);
1996 if (copy == NULL) {
1997 file_error(ms, errno,
1998 "can't allocate %" SIZE_T_FORMAT "u bytes",
1999 slen);
2000 return -1;
2001 }
2002 memcpy(copy, ms->search.s, slen);
2003 copy[--slen] = '\0';
2004 search = copy;
2005 } else {
2006 search = ms->search.s;
2007 copy = NULL;
2008 }
2009 #else
2010 search = ms->search.s;
2011 pmatch[0].rm_so = 0;
2012 pmatch[0].rm_eo = slen;
2013 #endif
2014 rc = file_regexec(&rx, (const char *)search,
2015 1, pmatch, REG_STARTEND);
2016 #if REG_STARTEND == 0
2017 free(copy);
2018 #endif
2019 switch (rc) {
2020 case 0:
2021 ms->search.s += (int)pmatch[0].rm_so;
2022 ms->search.offset += (size_t)pmatch[0].rm_so;
2023 ms->search.rm_len =
2024 (size_t)(pmatch[0].rm_eo - pmatch[0].rm_so);
2025 v = 0;
2026 break;
2027
2028 case REG_NOMATCH:
2029 v = 1;
2030 break;
2031
2032 default:
2033 file_regerror(&rx, rc, ms);
2034 v = (uint64_t)-1;
2035 break;
2036 }
2037 }
2038 file_regfree(&rx);
2039 if (v == (uint64_t)-1)
2040 return -1;
2041 break;
2042 }
2043 case FILE_INDIRECT:
2044 case FILE_USE:
2045 case FILE_NAME:
2046 return 1;
2047 default:
2048 file_magerror(ms, "invalid type %d in magiccheck()", m->type);
2049 return -1;
2050 }
2051
2052 v = file_signextend(ms, m, v);
2053
2054 switch (m->reln) {
2055 case 'x':
2056 if ((ms->flags & MAGIC_DEBUG) != 0)
2057 (void) fprintf(stderr, "%" INT64_T_FORMAT
2058 "u == *any* = 1\n", (unsigned long long)v);
2059 matched = 1;
2060 break;
2061
2062 case '!':
2063 matched = v != l;
2064 if ((ms->flags & MAGIC_DEBUG) != 0)
2065 (void) fprintf(stderr, "%" INT64_T_FORMAT "u != %"
2066 INT64_T_FORMAT "u = %d\n", (unsigned long long)v,
2067 (unsigned long long)l, matched);
2068 break;
2069
2070 case '=':
2071 matched = v == l;
2072 if ((ms->flags & MAGIC_DEBUG) != 0)
2073 (void) fprintf(stderr, "%" INT64_T_FORMAT "u == %"
2074 INT64_T_FORMAT "u = %d\n", (unsigned long long)v,
2075 (unsigned long long)l, matched);
2076 break;
2077
2078 case '>':
2079 if (m->flag & UNSIGNED) {
2080 matched = v > l;
2081 if ((ms->flags & MAGIC_DEBUG) != 0)
2082 (void) fprintf(stderr, "%" INT64_T_FORMAT
2083 "u > %" INT64_T_FORMAT "u = %d\n",
2084 (unsigned long long)v,
2085 (unsigned long long)l, matched);
2086 }
2087 else {
2088 matched = (int64_t) v > (int64_t) l;
2089 if ((ms->flags & MAGIC_DEBUG) != 0)
2090 (void) fprintf(stderr, "%" INT64_T_FORMAT
2091 "d > %" INT64_T_FORMAT "d = %d\n",
2092 (long long)v, (long long)l, matched);
2093 }
2094 break;
2095
2096 case '<':
2097 if (m->flag & UNSIGNED) {
2098 matched = v < l;
2099 if ((ms->flags & MAGIC_DEBUG) != 0)
2100 (void) fprintf(stderr, "%" INT64_T_FORMAT
2101 "u < %" INT64_T_FORMAT "u = %d\n",
2102 (unsigned long long)v,
2103 (unsigned long long)l, matched);
2104 }
2105 else {
2106 matched = (int64_t) v < (int64_t) l;
2107 if ((ms->flags & MAGIC_DEBUG) != 0)
2108 (void) fprintf(stderr, "%" INT64_T_FORMAT
2109 "d < %" INT64_T_FORMAT "d = %d\n",
2110 (long long)v, (long long)l, matched);
2111 }
2112 break;
2113
2114 case '&':
2115 matched = (v & l) == l;
2116 if ((ms->flags & MAGIC_DEBUG) != 0)
2117 (void) fprintf(stderr, "((%" INT64_T_FORMAT "x & %"
2118 INT64_T_FORMAT "x) == %" INT64_T_FORMAT
2119 "x) = %d\n", (unsigned long long)v,
2120 (unsigned long long)l, (unsigned long long)l,
2121 matched);
2122 break;
2123
2124 case '^':
2125 matched = (v & l) != l;
2126 if ((ms->flags & MAGIC_DEBUG) != 0)
2127 (void) fprintf(stderr, "((%" INT64_T_FORMAT "x & %"
2128 INT64_T_FORMAT "x) != %" INT64_T_FORMAT
2129 "x) = %d\n", (unsigned long long)v,
2130 (unsigned long long)l, (unsigned long long)l,
2131 matched);
2132 break;
2133
2134 default:
2135 file_magerror(ms, "cannot happen: invalid relation `%c'",
2136 m->reln);
2137 return -1;
2138 }
2139
2140 return matched;
2141 }
2142
2143 private int
2144 handle_annotation(struct magic_set *ms, struct magic *m)
2145 {
2146 if (ms->flags & MAGIC_APPLE) {
2147 if (file_printf(ms, "%.8s", m->apple) == -1)
2148 return -1;
2149 return 1;
2150 }
2151 if ((ms->flags & MAGIC_MIME_TYPE) && m->mimetype[0]) {
2152 if (file_printf(ms, "%s", m->mimetype) == -1)
2153 return -1;
2154 return 1;
2155 }
2156 return 0;
2157 }
2158
2159 private int
2160 print_sep(struct magic_set *ms, int firstline)
2161 {
2162 if (ms->flags & MAGIC_MIME)
2163 return 0;
2164 if (firstline)
2165 return 0;
2166 /*
2167 * we found another match
2168 * put a newline and '-' to do some simple formatting
2169 */
2170 return file_printf(ms, "\n- ");
2171 }
DragonFly BSD