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 1.147 2011/11/05 15:44:22 rrt Exp $")
36 #endif /* lint */
37
38 #include "magic.h"
39 #include <string.h>
40 #include <ctype.h>
41 #include <stdlib.h>
42 #include <time.h>
43
44 #ifndef PREG_OFFSET_CAPTURE
45 # define PREG_OFFSET_CAPTURE (1<<8)
46 #endif
47
48
49
50 private int match(struct magic_set *, struct magic *, uint32_t,
51 const unsigned char *, size_t, int, int);
52 private int mget(struct magic_set *, const unsigned char *,
53 struct magic *, size_t, unsigned int, int);
54 private int magiccheck(struct magic_set *, struct magic *);
55 private int32_t mprint(struct magic_set *, struct magic *);
56 private int32_t moffset(struct magic_set *, struct magic *);
57 private void mdebug(uint32_t, const char *, size_t);
58 private int mcopy(struct magic_set *, union VALUETYPE *, int, int,
59 const unsigned char *, uint32_t, size_t, size_t);
60 private int mconvert(struct magic_set *, struct magic *);
61 private int print_sep(struct magic_set *, int);
62 private int handle_annotation(struct magic_set *, struct magic *);
63 private void cvt_8(union VALUETYPE *, const struct magic *);
64 private void cvt_16(union VALUETYPE *, const struct magic *);
65 private void cvt_32(union VALUETYPE *, const struct magic *);
66 private void cvt_64(union VALUETYPE *, const struct magic *);
67
68 /*
69 * softmagic - lookup one file in parsed, in-memory copy of database
70 * Passed the name and FILE * of one file to be typed.
71 */
72 /*ARGSUSED1*/ /* nbytes passed for regularity, maybe need later */
73 protected int
file_softmagic(struct magic_set * ms,const unsigned char * buf,size_t nbytes,int mode,int text)74 file_softmagic(struct magic_set *ms, const unsigned char *buf, size_t nbytes,
75 int mode, int text)
76 {
77 struct mlist *ml;
78 int rv;
79 for (ml = ms->mlist->next; ml != ms->mlist; ml = ml->next)
80 if ((rv = match(ms, ml->magic, ml->nmagic, buf, nbytes, mode,
81 text)) != 0)
82 return rv;
83
84 return 0;
85 }
86
87 /*
88 * Go through the whole list, stopping if you find a match. Process all
89 * the continuations of that match before returning.
90 *
91 * We support multi-level continuations:
92 *
93 * At any time when processing a successful top-level match, there is a
94 * current continuation level; it represents the level of the last
95 * successfully matched continuation.
96 *
97 * Continuations above that level are skipped as, if we see one, it
98 * means that the continuation that controls them - i.e, the
99 * lower-level continuation preceding them - failed to match.
100 *
101 * Continuations below that level are processed as, if we see one,
102 * it means we've finished processing or skipping higher-level
103 * continuations under the control of a successful or unsuccessful
104 * lower-level continuation, and are now seeing the next lower-level
105 * continuation and should process it. The current continuation
106 * level reverts to the level of the one we're seeing.
107 *
108 * Continuations at the current level are processed as, if we see
109 * one, there's no lower-level continuation that may have failed.
110 *
111 * If a continuation matches, we bump the current continuation level
112 * so that higher-level continuations are processed.
113 */
114 private int
match(struct magic_set * ms,struct magic * magic,uint32_t nmagic,const unsigned char * s,size_t nbytes,int mode,int text)115 match(struct magic_set *ms, struct magic *magic, uint32_t nmagic,
116 const unsigned char *s, size_t nbytes, int mode, int text)
117 {
118 uint32_t magindex = 0;
119 unsigned int cont_level = 0;
120 int need_separator = 0;
121 int returnval = 0, e; /* if a match is found it is set to 1*/
122 int firstline = 1; /* a flag to print X\n X\n- X */
123 int printed_something = 0;
124 int print = (ms->flags & (MAGIC_MIME|MAGIC_APPLE)) == 0;
125
126 if (file_check_mem(ms, cont_level) == -1)
127 return -1;
128
129 for (magindex = 0; magindex < nmagic; magindex++) {
130 int flush = 0;
131 struct magic *m = &magic[magindex];
132
133 if ((IS_LIBMAGIC_STRING(m->type) &&
134 ((text && (m->str_flags & (STRING_BINTEST | STRING_TEXTTEST)) == STRING_BINTEST) ||
135 (!text && (m->str_flags & (STRING_TEXTTEST | STRING_BINTEST)) == STRING_TEXTTEST))) ||
136 (m->flag & mode) != mode) {
137 /* Skip sub-tests */
138 while (magindex < nmagic - 1 && magic[magindex + 1].cont_level != 0) {
139 magindex++;
140 }
141 continue; /* Skip to next top-level test*/
142 }
143
144 ms->offset = m->offset;
145 ms->line = m->lineno;
146
147 /* if main entry matches, print it... */
148 switch (mget(ms, s, m, nbytes, cont_level, text)) {
149 case -1:
150 return -1;
151 case 0:
152 flush = m->reln != '!';
153 break;
154 default:
155 if (m->type == FILE_INDIRECT)
156 returnval = 1;
157
158 switch (magiccheck(ms, m)) {
159 case -1:
160 return -1;
161 case 0:
162 flush++;
163 break;
164 default:
165 flush = 0;
166 break;
167 }
168 break;
169 }
170 if (flush) {
171 /*
172 * main entry didn't match,
173 * flush its continuations
174 */
175 while (magindex < nmagic - 1 && magic[magindex + 1].cont_level != 0) {
176 magindex++;
177 }
178 continue;
179 }
180
181 if ((e = handle_annotation(ms, m)) != 0)
182 return e;
183 /*
184 * If we are going to print something, we'll need to print
185 * a blank before we print something else.
186 */
187 if (*m->desc) {
188 need_separator = 1;
189 printed_something = 1;
190 if (print_sep(ms, firstline) == -1)
191 return -1;
192 }
193
194
195 if (print && mprint(ms, m) == -1)
196 return -1;
197
198 ms->c.li[cont_level].off = moffset(ms, m);
199
200 /* and any continuations that match */
201 if (file_check_mem(ms, ++cont_level) == -1)
202 return -1;
203
204 while (magindex < nmagic - 1 && magic[magindex + 1].cont_level != 0) {
205 magindex++;
206 m = &magic[magindex];
207 ms->line = m->lineno; /* for messages */
208
209 if (cont_level < m->cont_level)
210 continue;
211 if (cont_level > m->cont_level) {
212 /*
213 * We're at the end of the level
214 * "cont_level" continuations.
215 */
216 cont_level = m->cont_level;
217 }
218 ms->offset = m->offset;
219 if (m->flag & OFFADD) {
220 ms->offset += ms->c.li[cont_level - 1].off;
221 }
222
223 #ifdef ENABLE_CONDITIONALS
224 if (m->cond == COND_ELSE ||
225 m->cond == COND_ELIF) {
226 if (ms->c.li[cont_level].last_match == 1)
227 continue;
228 }
229 #endif
230 switch (mget(ms, s, m, nbytes, cont_level, text)) {
231 case -1:
232 return -1;
233 case 0:
234 if (m->reln != '!')
235 continue;
236 flush = 1;
237 break;
238 default:
239 if (m->type == FILE_INDIRECT)
240 returnval = 1;
241 flush = 0;
242 break;
243 }
244
245 switch (flush ? 1 : magiccheck(ms, m)) {
246 case -1:
247 return -1;
248 case 0:
249 #ifdef ENABLE_CONDITIONALS
250 ms->c.li[cont_level].last_match = 0;
251 #endif
252 break;
253 default:
254 #ifdef ENABLE_CONDITIONALS
255 ms->c.li[cont_level].last_match = 1;
256 #endif
257 if (m->type != FILE_DEFAULT)
258 ms->c.li[cont_level].got_match = 1;
259 else if (ms->c.li[cont_level].got_match) {
260 ms->c.li[cont_level].got_match = 0;
261 break;
262 }
263 if ((e = handle_annotation(ms, m)) != 0)
264 return e;
265 /*
266 * If we are going to print something,
267 * make sure that we have a separator first.
268 */
269 if (*m->desc) {
270 if (!printed_something) {
271 printed_something = 1;
272 if (print_sep(ms, firstline)
273 == -1)
274 return -1;
275 }
276 }
277 /*
278 * This continuation matched. Print
279 * its message, with a blank before it
280 * if the previous item printed and
281 * this item isn't empty.
282 */
283 /* space if previous printed */
284 if (need_separator
285 && ((m->flag & NOSPACE) == 0)
286 && *m->desc) {
287 if (print &&
288 file_printf(ms, " ") == -1)
289 return -1;
290 need_separator = 0;
291 }
292 if (print && mprint(ms, m) == -1)
293 return -1;
294
295 ms->c.li[cont_level].off = moffset(ms, m);
296
297 if (*m->desc)
298 need_separator = 1;
299
300 /*
301 * If we see any continuations
302 * at a higher level,
303 * process them.
304 */
305 if (file_check_mem(ms, ++cont_level) == -1)
306 return -1;
307 break;
308 }
309 }
310 if (printed_something) {
311 firstline = 0;
312 if (print)
313 returnval = 1;
314 }
315 if ((ms->flags & MAGIC_CONTINUE) == 0 && printed_something) {
316 return returnval; /* don't keep searching */
317 }
318 }
319 return returnval; /* This is hit if -k is set or there is no match */
320 }
321
322 private int
check_fmt(struct magic_set * ms,struct magic * m)323 check_fmt(struct magic_set *ms, struct magic *m)
324 {
325 pcre *pce;
326 int re_options;
327 pcre_extra *re_extra;
328 TSRMLS_FETCH();
329
330 if (strchr(m->desc, '%') == NULL) {
331 return 0;
332 }
333
334 if ((pce = pcre_get_compiled_regex("~%[-0-9.]*s~", &re_extra, &re_options TSRMLS_CC)) == NULL) {
335 return -1;
336 } else {
337 return !pcre_exec(pce, re_extra, m->desc, strlen(m->desc), 0, re_options, NULL, 0);
338 }
339 }
340
341 private int32_t
mprint(struct magic_set * ms,struct magic * m)342 mprint(struct magic_set *ms, struct magic *m)
343 {
344 uint64_t v;
345 float vf;
346 double vd;
347 int64_t t = 0;
348 char buf[128];
349 union VALUETYPE *p = &ms->ms_value;
350
351 switch (m->type) {
352 case FILE_BYTE:
353 v = file_signextend(ms, m, (uint64_t)p->b);
354 switch (check_fmt(ms, m)) {
355 case -1:
356 return -1;
357 case 1:
358 (void)snprintf(buf, sizeof(buf), "%c",
359 (unsigned char)v);
360 if (file_printf(ms, m->desc, buf) == -1)
361 return -1;
362 break;
363 default:
364 if (file_printf(ms, m->desc, (unsigned char) v) == -1)
365 return -1;
366 break;
367 }
368 t = ms->offset + sizeof(char);
369 break;
370
371 case FILE_SHORT:
372 case FILE_BESHORT:
373 case FILE_LESHORT:
374 v = file_signextend(ms, m, (uint64_t)p->h);
375 switch (check_fmt(ms, m)) {
376 case -1:
377 return -1;
378 case 1:
379 (void)snprintf(buf, sizeof(buf), "%hu",
380 (unsigned short)v);
381 if (file_printf(ms, m->desc, buf) == -1)
382 return -1;
383 break;
384 default:
385 if (
386 file_printf(ms, m->desc, (unsigned short) v) == -1)
387 return -1;
388 break;
389 }
390 t = ms->offset + sizeof(short);
391 break;
392
393 case FILE_LONG:
394 case FILE_BELONG:
395 case FILE_LELONG:
396 case FILE_MELONG:
397 v = file_signextend(ms, m, (uint64_t)p->l);
398 switch (check_fmt(ms, m)) {
399 case -1:
400 return -1;
401 case 1:
402 (void)snprintf(buf, sizeof(buf), "%u", (uint32_t)v);
403 if (file_printf(ms, m->desc, buf) == -1)
404 return -1;
405 break;
406 default:
407 if (file_printf(ms, m->desc, (uint32_t) v) == -1)
408 return -1;
409 break;
410 }
411 t = ms->offset + sizeof(int32_t);
412 break;
413
414 case FILE_QUAD:
415 case FILE_BEQUAD:
416 case FILE_LEQUAD:
417 v = file_signextend(ms, m, p->q);
418 if (file_printf(ms, m->desc, (uint64_t) v) == -1)
419 return -1;
420 t = ms->offset + sizeof(int64_t);
421 break;
422
423 case FILE_STRING:
424 case FILE_PSTRING:
425 case FILE_BESTRING16:
426 case FILE_LESTRING16:
427 if (m->reln == '=' || m->reln == '!') {
428 if (file_printf(ms, m->desc, m->value.s) == -1)
429 return -1;
430 t = ms->offset + m->vallen;
431 }
432 else {
433 if (*m->value.s == '\0')
434 p->s[strcspn(p->s, "\n")] = '\0';
435 if (file_printf(ms, m->desc, p->s) == -1)
436 return -1;
437 t = ms->offset + strlen(p->s);
438 if (m->type == FILE_PSTRING)
439 t += file_pstring_length_size(m);
440 }
441 break;
442
443 case FILE_DATE:
444 case FILE_BEDATE:
445 case FILE_LEDATE:
446 case FILE_MEDATE:
447 if (file_printf(ms, m->desc, file_fmttime(p->l, 1)) == -1)
448 return -1;
449 t = ms->offset + sizeof(time_t);
450 break;
451
452 case FILE_LDATE:
453 case FILE_BELDATE:
454 case FILE_LELDATE:
455 case FILE_MELDATE:
456 if (file_printf(ms, m->desc, file_fmttime(p->l, 0)) == -1)
457 return -1;
458 t = ms->offset + sizeof(time_t);
459 break;
460
461 case FILE_QDATE:
462 case FILE_BEQDATE:
463 case FILE_LEQDATE:
464 if (file_printf(ms, m->desc, file_fmttime((uint32_t)p->q,
465 1)) == -1)
466 return -1;
467 t = ms->offset + sizeof(uint64_t);
468 break;
469
470 case FILE_QLDATE:
471 case FILE_BEQLDATE:
472 case FILE_LEQLDATE:
473 if (file_printf(ms, m->desc, file_fmttime((uint32_t)p->q,
474 0)) == -1)
475 return -1;
476 t = ms->offset + sizeof(uint64_t);
477 break;
478
479 case FILE_FLOAT:
480 case FILE_BEFLOAT:
481 case FILE_LEFLOAT:
482 vf = p->f;
483 switch (check_fmt(ms, m)) {
484 case -1:
485 return -1;
486 case 1:
487 (void)snprintf(buf, sizeof(buf), "%g", vf);
488 if (file_printf(ms, m->desc, buf) == -1)
489 return -1;
490 break;
491 default:
492 if (file_printf(ms, m->desc, vf) == -1)
493 return -1;
494 break;
495 }
496 t = ms->offset + sizeof(float);
497 break;
498
499 case FILE_DOUBLE:
500 case FILE_BEDOUBLE:
501 case FILE_LEDOUBLE:
502 vd = p->d;
503 switch (check_fmt(ms, m)) {
504 case -1:
505 return -1;
506 case 1:
507 (void)snprintf(buf, sizeof(buf), "%g", vd);
508 if (file_printf(ms, m->desc, buf) == -1)
509 return -1;
510 break;
511 default:
512 if (file_printf(ms, m->desc, vd) == -1)
513 return -1;
514 break;
515 }
516 t = ms->offset + sizeof(double);
517 break;
518
519 case FILE_REGEX: {
520 char *cp;
521 int rval;
522
523 cp = estrndup((const char *)ms->search.s, ms->search.rm_len);
524
525 rval = file_printf(ms, m->desc, cp);
526 efree(cp);
527
528 if (rval == -1)
529 return -1;
530
531 if ((m->str_flags & REGEX_OFFSET_START))
532 t = ms->search.offset;
533 else
534 t = ms->search.offset + ms->search.rm_len;
535 break;
536 }
537
538 case FILE_SEARCH:
539 if (file_printf(ms, m->desc, m->value.s) == -1)
540 return -1;
541 if ((m->str_flags & REGEX_OFFSET_START))
542 t = ms->search.offset;
543 else
544 t = ms->search.offset + m->vallen;
545 break;
546
547 case FILE_DEFAULT:
548 if (file_printf(ms, m->desc, m->value.s) == -1)
549 return -1;
550 t = ms->offset;
551 break;
552
553 case FILE_INDIRECT:
554 t = ms->offset;
555 break;
556
557 default:
558 file_magerror(ms, "invalid m->type (%d) in mprint()", m->type);
559 return -1;
560 }
561 return (int32_t)t;
562 }
563
564 private int32_t
moffset(struct magic_set * ms,struct magic * m)565 moffset(struct magic_set *ms, struct magic *m)
566 {
567 switch (m->type) {
568 case FILE_BYTE:
569 return CAST(int32_t, (ms->offset + sizeof(char)));
570
571 case FILE_SHORT:
572 case FILE_BESHORT:
573 case FILE_LESHORT:
574 return CAST(int32_t, (ms->offset + sizeof(short)));
575
576 case FILE_LONG:
577 case FILE_BELONG:
578 case FILE_LELONG:
579 case FILE_MELONG:
580 return CAST(int32_t, (ms->offset + sizeof(int32_t)));
581
582 case FILE_QUAD:
583 case FILE_BEQUAD:
584 case FILE_LEQUAD:
585 return CAST(int32_t, (ms->offset + sizeof(int64_t)));
586
587 case FILE_STRING:
588 case FILE_PSTRING:
589 case FILE_BESTRING16:
590 case FILE_LESTRING16:
591 if (m->reln == '=' || m->reln == '!')
592 return ms->offset + m->vallen;
593 else {
594 union VALUETYPE *p = &ms->ms_value;
595 uint32_t t;
596
597 if (*m->value.s == '\0')
598 p->s[strcspn(p->s, "\n")] = '\0';
599 t = CAST(uint32_t, (ms->offset + strlen(p->s)));
600 if (m->type == FILE_PSTRING)
601 t += file_pstring_length_size(m);
602 return t;
603 }
604
605 case FILE_DATE:
606 case FILE_BEDATE:
607 case FILE_LEDATE:
608 case FILE_MEDATE:
609 return CAST(int32_t, (ms->offset + sizeof(time_t)));
610
611 case FILE_LDATE:
612 case FILE_BELDATE:
613 case FILE_LELDATE:
614 case FILE_MELDATE:
615 return CAST(int32_t, (ms->offset + sizeof(time_t)));
616
617 case FILE_QDATE:
618 case FILE_BEQDATE:
619 case FILE_LEQDATE:
620 return CAST(int32_t, (ms->offset + sizeof(uint64_t)));
621
622 case FILE_QLDATE:
623 case FILE_BEQLDATE:
624 case FILE_LEQLDATE:
625 return CAST(int32_t, (ms->offset + sizeof(uint64_t)));
626
627 case FILE_FLOAT:
628 case FILE_BEFLOAT:
629 case FILE_LEFLOAT:
630 return CAST(int32_t, (ms->offset + sizeof(float)));
631
632 case FILE_DOUBLE:
633 case FILE_BEDOUBLE:
634 case FILE_LEDOUBLE:
635 return CAST(int32_t, (ms->offset + sizeof(double)));
636
637 case FILE_REGEX:
638 if ((m->str_flags & REGEX_OFFSET_START) != 0)
639 return CAST(int32_t, ms->search.offset);
640 else
641 return CAST(int32_t, (ms->search.offset +
642 ms->search.rm_len));
643
644 case FILE_SEARCH:
645 if ((m->str_flags & REGEX_OFFSET_START) != 0)
646 return CAST(int32_t, ms->search.offset);
647 else
648 return CAST(int32_t, (ms->search.offset + m->vallen));
649
650 case FILE_DEFAULT:
651 return ms->offset;
652
653 case FILE_INDIRECT:
654 return ms->offset;
655
656 default:
657 return 0;
658 }
659 }
660
661 #define DO_CVT(fld, cast) \
662 if (m->num_mask) \
663 switch (m->mask_op & FILE_OPS_MASK) { \
664 case FILE_OPAND: \
665 p->fld &= cast m->num_mask; \
666 break; \
667 case FILE_OPOR: \
668 p->fld |= cast m->num_mask; \
669 break; \
670 case FILE_OPXOR: \
671 p->fld ^= cast m->num_mask; \
672 break; \
673 case FILE_OPADD: \
674 p->fld += cast m->num_mask; \
675 break; \
676 case FILE_OPMINUS: \
677 p->fld -= cast m->num_mask; \
678 break; \
679 case FILE_OPMULTIPLY: \
680 p->fld *= cast m->num_mask; \
681 break; \
682 case FILE_OPDIVIDE: \
683 p->fld /= cast m->num_mask; \
684 break; \
685 case FILE_OPMODULO: \
686 p->fld %= cast m->num_mask; \
687 break; \
688 } \
689 if (m->mask_op & FILE_OPINVERSE) \
690 p->fld = ~p->fld \
691
692 private void
cvt_8(union VALUETYPE * p,const struct magic * m)693 cvt_8(union VALUETYPE *p, const struct magic *m)
694 {
695 DO_CVT(b, (uint8_t));
696 }
697
698 private void
cvt_16(union VALUETYPE * p,const struct magic * m)699 cvt_16(union VALUETYPE *p, const struct magic *m)
700 {
701 DO_CVT(h, (uint16_t));
702 }
703
704 private void
cvt_32(union VALUETYPE * p,const struct magic * m)705 cvt_32(union VALUETYPE *p, const struct magic *m)
706 {
707 DO_CVT(l, (uint32_t));
708 }
709
710 private void
cvt_64(union VALUETYPE * p,const struct magic * m)711 cvt_64(union VALUETYPE *p, const struct magic *m)
712 {
713 DO_CVT(q, (uint64_t));
714 }
715
716 #define DO_CVT2(fld, cast) \
717 if (m->num_mask) \
718 switch (m->mask_op & FILE_OPS_MASK) { \
719 case FILE_OPADD: \
720 p->fld += cast (int64_t)m->num_mask; \
721 break; \
722 case FILE_OPMINUS: \
723 p->fld -= cast (int64_t)m->num_mask; \
724 break; \
725 case FILE_OPMULTIPLY: \
726 p->fld *= cast (int64_t)m->num_mask; \
727 break; \
728 case FILE_OPDIVIDE: \
729 p->fld /= cast (int64_t)m->num_mask; \
730 break; \
731 } \
732
733 private void
cvt_float(union VALUETYPE * p,const struct magic * m)734 cvt_float(union VALUETYPE *p, const struct magic *m)
735 {
736 DO_CVT2(f, (float));
737 }
738
739 private void
cvt_double(union VALUETYPE * p,const struct magic * m)740 cvt_double(union VALUETYPE *p, const struct magic *m)
741 {
742 DO_CVT2(d, (double));
743 }
744
745 /*
746 * Convert the byte order of the data we are looking at
747 * While we're here, let's apply the mask operation
748 * (unless you have a better idea)
749 */
750 private int
mconvert(struct magic_set * ms,struct magic * m)751 mconvert(struct magic_set *ms, struct magic *m)
752 {
753 union VALUETYPE *p = &ms->ms_value;
754
755 switch (m->type) {
756 case FILE_BYTE:
757 cvt_8(p, m);
758 return 1;
759 case FILE_SHORT:
760 cvt_16(p, m);
761 return 1;
762 case FILE_LONG:
763 case FILE_DATE:
764 case FILE_LDATE:
765 cvt_32(p, m);
766 return 1;
767 case FILE_QUAD:
768 case FILE_QDATE:
769 case FILE_QLDATE:
770 cvt_64(p, m);
771 return 1;
772 case FILE_STRING:
773 case FILE_BESTRING16:
774 case FILE_LESTRING16: {
775 /* Null terminate and eat *trailing* return */
776 p->s[sizeof(p->s) - 1] = '\0';
777 return 1;
778 }
779 case FILE_PSTRING: {
780 size_t sz = file_pstring_length_size(m);
781 char *ptr1 = p->s, *ptr2 = ptr1 + sz;
782 size_t len = file_pstring_get_length(m, ptr1);
783 if (len >= sizeof(p->s)) {
784 /*
785 * The size of the pascal string length (sz)
786 * is 1, 2, or 4. We need at least 1 byte for NUL
787 * termination, but we've already truncated the
788 * string by p->s, so we need to deduct sz.
789 */
790 len = sizeof(p->s) - sz;
791 }
792 while (len--)
793 *ptr1++ = *ptr2++;
794 *ptr1 = '\0';
795 return 1;
796 }
797 case FILE_BESHORT:
798 p->h = (short)((p->hs[0]<<8)|(p->hs[1]));
799 cvt_16(p, m);
800 return 1;
801 case FILE_BELONG:
802 case FILE_BEDATE:
803 case FILE_BELDATE:
804 p->l = (int32_t)
805 ((p->hl[0]<<24)|(p->hl[1]<<16)|(p->hl[2]<<8)|(p->hl[3]));
806 cvt_32(p, m);
807 return 1;
808 case FILE_BEQUAD:
809 case FILE_BEQDATE:
810 case FILE_BEQLDATE:
811 p->q = (uint64_t)
812 (((uint64_t)p->hq[0]<<56)|((uint64_t)p->hq[1]<<48)|
813 ((uint64_t)p->hq[2]<<40)|((uint64_t)p->hq[3]<<32)|
814 ((uint64_t)p->hq[4]<<24)|((uint64_t)p->hq[5]<<16)|
815 ((uint64_t)p->hq[6]<<8)|((uint64_t)p->hq[7]));
816 cvt_64(p, m);
817 return 1;
818 case FILE_LESHORT:
819 p->h = (short)((p->hs[1]<<8)|(p->hs[0]));
820 cvt_16(p, m);
821 return 1;
822 case FILE_LELONG:
823 case FILE_LEDATE:
824 case FILE_LELDATE:
825 p->l = (int32_t)
826 ((p->hl[3]<<24)|(p->hl[2]<<16)|(p->hl[1]<<8)|(p->hl[0]));
827 cvt_32(p, m);
828 return 1;
829 case FILE_LEQUAD:
830 case FILE_LEQDATE:
831 case FILE_LEQLDATE:
832 p->q = (uint64_t)
833 (((uint64_t)p->hq[7]<<56)|((uint64_t)p->hq[6]<<48)|
834 ((uint64_t)p->hq[5]<<40)|((uint64_t)p->hq[4]<<32)|
835 ((uint64_t)p->hq[3]<<24)|((uint64_t)p->hq[2]<<16)|
836 ((uint64_t)p->hq[1]<<8)|((uint64_t)p->hq[0]));
837 cvt_64(p, m);
838 return 1;
839 case FILE_MELONG:
840 case FILE_MEDATE:
841 case FILE_MELDATE:
842 p->l = (int32_t)
843 ((p->hl[1]<<24)|(p->hl[0]<<16)|(p->hl[3]<<8)|(p->hl[2]));
844 cvt_32(p, m);
845 return 1;
846 case FILE_FLOAT:
847 cvt_float(p, m);
848 return 1;
849 case FILE_BEFLOAT:
850 p->l = ((uint32_t)p->hl[0]<<24)|((uint32_t)p->hl[1]<<16)|
851 ((uint32_t)p->hl[2]<<8) |((uint32_t)p->hl[3]);
852 cvt_float(p, m);
853 return 1;
854 case FILE_LEFLOAT:
855 p->l = ((uint32_t)p->hl[3]<<24)|((uint32_t)p->hl[2]<<16)|
856 ((uint32_t)p->hl[1]<<8) |((uint32_t)p->hl[0]);
857 cvt_float(p, m);
858 return 1;
859 case FILE_DOUBLE:
860 cvt_double(p, m);
861 return 1;
862 case FILE_BEDOUBLE:
863 p->q = ((uint64_t)p->hq[0]<<56)|((uint64_t)p->hq[1]<<48)|
864 ((uint64_t)p->hq[2]<<40)|((uint64_t)p->hq[3]<<32)|
865 ((uint64_t)p->hq[4]<<24)|((uint64_t)p->hq[5]<<16)|
866 ((uint64_t)p->hq[6]<<8) |((uint64_t)p->hq[7]);
867 cvt_double(p, m);
868 return 1;
869 case FILE_LEDOUBLE:
870 p->q = ((uint64_t)p->hq[7]<<56)|((uint64_t)p->hq[6]<<48)|
871 ((uint64_t)p->hq[5]<<40)|((uint64_t)p->hq[4]<<32)|
872 ((uint64_t)p->hq[3]<<24)|((uint64_t)p->hq[2]<<16)|
873 ((uint64_t)p->hq[1]<<8) |((uint64_t)p->hq[0]);
874 cvt_double(p, m);
875 return 1;
876 case FILE_REGEX:
877 case FILE_SEARCH:
878 case FILE_DEFAULT:
879 return 1;
880 default:
881 file_magerror(ms, "invalid type %d in mconvert()", m->type);
882 return 0;
883 }
884 }
885
886
887 private void
mdebug(uint32_t offset,const char * str,size_t len)888 mdebug(uint32_t offset, const char *str, size_t len)
889 {
890 (void) fprintf(stderr, "mget @%d: ", offset);
891 file_showstr(stderr, str, len);
892 (void) fputc('\n', stderr);
893 (void) fputc('\n', stderr);
894 }
895
896 private int
mcopy(struct magic_set * ms,union VALUETYPE * p,int type,int indir,const unsigned char * s,uint32_t offset,size_t nbytes,size_t linecnt)897 mcopy(struct magic_set *ms, union VALUETYPE *p, int type, int indir,
898 const unsigned char *s, uint32_t offset, size_t nbytes, size_t linecnt)
899 {
900 /*
901 * Note: FILE_SEARCH and FILE_REGEX do not actually copy
902 * anything, but setup pointers into the source
903 */
904 if (indir == 0) {
905 switch (type) {
906 case FILE_SEARCH:
907 ms->search.s = RCAST(const char *, s) + offset;
908 ms->search.s_len = nbytes - offset;
909 ms->search.offset = offset;
910 return 0;
911
912 case FILE_REGEX: {
913 const char *b;
914 const char *c;
915 const char *last; /* end of search region */
916 const char *buf; /* start of search region */
917 const char *end;
918 size_t lines;
919
920 if (s == NULL) {
921 ms->search.s_len = 0;
922 ms->search.s = NULL;
923 return 0;
924 }
925 buf = RCAST(const char *, s) + offset;
926 end = last = RCAST(const char *, s) + nbytes;
927 /* mget() guarantees buf <= last */
928 for (lines = linecnt, b = buf; lines && b < end &&
929 ((b = CAST(const char *,
930 memchr(c = b, '\n', CAST(size_t, (end - b)))))
931 || (b = CAST(const char *,
932 memchr(c, '\r', CAST(size_t, (end - c))))));
933 lines--, b++) {
934 last = b;
935 if (b[0] == '\r' && b[1] == '\n')
936 b++;
937 }
938 if (lines)
939 last = RCAST(const char *, s) + nbytes;
940
941 ms->search.s = buf;
942 ms->search.s_len = last - buf;
943 ms->search.offset = offset;
944 ms->search.rm_len = 0;
945 return 0;
946 }
947 case FILE_BESTRING16:
948 case FILE_LESTRING16: {
949 const unsigned char *src = s + offset;
950 const unsigned char *esrc = s + nbytes;
951 char *dst = p->s;
952 char *edst = &p->s[sizeof(p->s) - 1];
953
954 if (type == FILE_BESTRING16)
955 src++;
956
957 /* check for pointer overflow */
958 if (src < s) {
959 file_magerror(ms, "invalid offset %u in mcopy()",
960 offset);
961 return -1;
962 }
963 for (/*EMPTY*/; src < esrc; src += 2, dst++) {
964 if (dst < edst)
965 *dst = *src;
966 else
967 break;
968 if (*dst == '\0') {
969 if (type == FILE_BESTRING16 ?
970 *(src - 1) != '\0' :
971 *(src + 1) != '\0')
972 *dst = ' ';
973 }
974 }
975 *edst = '\0';
976 return 0;
977 }
978 case FILE_STRING: /* XXX - these two should not need */
979 case FILE_PSTRING: /* to copy anything, but do anyway. */
980 default:
981 break;
982 }
983 }
984
985 if (offset >= nbytes) {
986 (void)memset(p, '\0', sizeof(*p));
987 return 0;
988 }
989 if (nbytes - offset < sizeof(*p))
990 nbytes = nbytes - offset;
991 else
992 nbytes = sizeof(*p);
993
994 (void)memcpy(p, s + offset, nbytes);
995
996 /*
997 * the usefulness of padding with zeroes eludes me, it
998 * might even cause problems
999 */
1000 if (nbytes < sizeof(*p))
1001 (void)memset(((char *)(void *)p) + nbytes, '\0',
1002 sizeof(*p) - nbytes);
1003 return 0;
1004 }
1005
1006 private int
mget(struct magic_set * ms,const unsigned char * s,struct magic * m,size_t nbytes,unsigned int cont_level,int text)1007 mget(struct magic_set *ms, const unsigned char *s,
1008 struct magic *m, size_t nbytes, unsigned int cont_level, int text)
1009 {
1010 uint32_t offset = ms->offset;
1011 uint32_t count = m->str_range;
1012 union VALUETYPE *p = &ms->ms_value;
1013
1014 if (mcopy(ms, p, m->type, m->flag & INDIR, s, offset, nbytes, count) == -1)
1015 return -1;
1016
1017 if ((ms->flags & MAGIC_DEBUG) != 0) {
1018 mdebug(offset, (char *)(void *)p, sizeof(union VALUETYPE));
1019 }
1020
1021 if (m->flag & INDIR) {
1022 int off = m->in_offset;
1023 if (m->in_op & FILE_OPINDIRECT) {
1024 const union VALUETYPE *q =
1025 ((const void *)(s + offset + off));
1026 switch (m->in_type) {
1027 case FILE_BYTE:
1028 off = q->b;
1029 break;
1030 case FILE_SHORT:
1031 off = q->h;
1032 break;
1033 case FILE_BESHORT:
1034 off = (short)((q->hs[0]<<8)|(q->hs[1]));
1035 break;
1036 case FILE_LESHORT:
1037 off = (short)((q->hs[1]<<8)|(q->hs[0]));
1038 break;
1039 case FILE_LONG:
1040 off = q->l;
1041 break;
1042 case FILE_BELONG:
1043 case FILE_BEID3:
1044 off = (int32_t)((q->hl[0]<<24)|(q->hl[1]<<16)|
1045 (q->hl[2]<<8)|(q->hl[3]));
1046 break;
1047 case FILE_LEID3:
1048 case FILE_LELONG:
1049 off = (int32_t)((q->hl[3]<<24)|(q->hl[2]<<16)|
1050 (q->hl[1]<<8)|(q->hl[0]));
1051 break;
1052 case FILE_MELONG:
1053 off = (int32_t)((q->hl[1]<<24)|(q->hl[0]<<16)|
1054 (q->hl[3]<<8)|(q->hl[2]));
1055 break;
1056 }
1057 }
1058 switch (m->in_type) {
1059 case FILE_BYTE:
1060 if (nbytes < (offset + 1))
1061 return 0;
1062 if (off) {
1063 switch (m->in_op & FILE_OPS_MASK) {
1064 case FILE_OPAND:
1065 offset = p->b & off;
1066 break;
1067 case FILE_OPOR:
1068 offset = p->b | off;
1069 break;
1070 case FILE_OPXOR:
1071 offset = p->b ^ off;
1072 break;
1073 case FILE_OPADD:
1074 offset = p->b + off;
1075 break;
1076 case FILE_OPMINUS:
1077 offset = p->b - off;
1078 break;
1079 case FILE_OPMULTIPLY:
1080 offset = p->b * off;
1081 break;
1082 case FILE_OPDIVIDE:
1083 offset = p->b / off;
1084 break;
1085 case FILE_OPMODULO:
1086 offset = p->b % off;
1087 break;
1088 }
1089 } else
1090 offset = p->b;
1091 if (m->in_op & FILE_OPINVERSE)
1092 offset = ~offset;
1093 break;
1094 case FILE_BESHORT:
1095 if (nbytes < (offset + 2))
1096 return 0;
1097 if (off) {
1098 switch (m->in_op & FILE_OPS_MASK) {
1099 case FILE_OPAND:
1100 offset = (short)((p->hs[0]<<8)|
1101 (p->hs[1])) &
1102 off;
1103 break;
1104 case FILE_OPOR:
1105 offset = (short)((p->hs[0]<<8)|
1106 (p->hs[1])) |
1107 off;
1108 break;
1109 case FILE_OPXOR:
1110 offset = (short)((p->hs[0]<<8)|
1111 (p->hs[1])) ^
1112 off;
1113 break;
1114 case FILE_OPADD:
1115 offset = (short)((p->hs[0]<<8)|
1116 (p->hs[1])) +
1117 off;
1118 break;
1119 case FILE_OPMINUS:
1120 offset = (short)((p->hs[0]<<8)|
1121 (p->hs[1])) -
1122 off;
1123 break;
1124 case FILE_OPMULTIPLY:
1125 offset = (short)((p->hs[0]<<8)|
1126 (p->hs[1])) *
1127 off;
1128 break;
1129 case FILE_OPDIVIDE:
1130 offset = (short)((p->hs[0]<<8)|
1131 (p->hs[1])) /
1132 off;
1133 break;
1134 case FILE_OPMODULO:
1135 offset = (short)((p->hs[0]<<8)|
1136 (p->hs[1])) %
1137 off;
1138 break;
1139 }
1140 } else
1141 offset = (short)((p->hs[0]<<8)|
1142 (p->hs[1]));
1143 if (m->in_op & FILE_OPINVERSE)
1144 offset = ~offset;
1145 break;
1146 case FILE_LESHORT:
1147 if (nbytes < (offset + 2))
1148 return 0;
1149 if (off) {
1150 switch (m->in_op & FILE_OPS_MASK) {
1151 case FILE_OPAND:
1152 offset = (short)((p->hs[1]<<8)|
1153 (p->hs[0])) &
1154 off;
1155 break;
1156 case FILE_OPOR:
1157 offset = (short)((p->hs[1]<<8)|
1158 (p->hs[0])) |
1159 off;
1160 break;
1161 case FILE_OPXOR:
1162 offset = (short)((p->hs[1]<<8)|
1163 (p->hs[0])) ^
1164 off;
1165 break;
1166 case FILE_OPADD:
1167 offset = (short)((p->hs[1]<<8)|
1168 (p->hs[0])) +
1169 off;
1170 break;
1171 case FILE_OPMINUS:
1172 offset = (short)((p->hs[1]<<8)|
1173 (p->hs[0])) -
1174 off;
1175 break;
1176 case FILE_OPMULTIPLY:
1177 offset = (short)((p->hs[1]<<8)|
1178 (p->hs[0])) *
1179 off;
1180 break;
1181 case FILE_OPDIVIDE:
1182 offset = (short)((p->hs[1]<<8)|
1183 (p->hs[0])) /
1184 off;
1185 break;
1186 case FILE_OPMODULO:
1187 offset = (short)((p->hs[1]<<8)|
1188 (p->hs[0])) %
1189 off;
1190 break;
1191 }
1192 } else
1193 offset = (short)((p->hs[1]<<8)|
1194 (p->hs[0]));
1195 if (m->in_op & FILE_OPINVERSE)
1196 offset = ~offset;
1197 break;
1198 case FILE_SHORT:
1199 if (nbytes < (offset + 2))
1200 return 0;
1201 if (off) {
1202 switch (m->in_op & FILE_OPS_MASK) {
1203 case FILE_OPAND:
1204 offset = p->h & off;
1205 break;
1206 case FILE_OPOR:
1207 offset = p->h | off;
1208 break;
1209 case FILE_OPXOR:
1210 offset = p->h ^ off;
1211 break;
1212 case FILE_OPADD:
1213 offset = p->h + off;
1214 break;
1215 case FILE_OPMINUS:
1216 offset = p->h - off;
1217 break;
1218 case FILE_OPMULTIPLY:
1219 offset = p->h * off;
1220 break;
1221 case FILE_OPDIVIDE:
1222 offset = p->h / off;
1223 break;
1224 case FILE_OPMODULO:
1225 offset = p->h % off;
1226 break;
1227 }
1228 }
1229 else
1230 offset = p->h;
1231 if (m->in_op & FILE_OPINVERSE)
1232 offset = ~offset;
1233 break;
1234 case FILE_BELONG:
1235 case FILE_BEID3:
1236 if (nbytes < (offset + 4))
1237 return 0;
1238 if (off) {
1239 switch (m->in_op & FILE_OPS_MASK) {
1240 case FILE_OPAND:
1241 offset = (int32_t)((p->hl[0]<<24)|
1242 (p->hl[1]<<16)|
1243 (p->hl[2]<<8)|
1244 (p->hl[3])) &
1245 off;
1246 break;
1247 case FILE_OPOR:
1248 offset = (int32_t)((p->hl[0]<<24)|
1249 (p->hl[1]<<16)|
1250 (p->hl[2]<<8)|
1251 (p->hl[3])) |
1252 off;
1253 break;
1254 case FILE_OPXOR:
1255 offset = (int32_t)((p->hl[0]<<24)|
1256 (p->hl[1]<<16)|
1257 (p->hl[2]<<8)|
1258 (p->hl[3])) ^
1259 off;
1260 break;
1261 case FILE_OPADD:
1262 offset = (int32_t)((p->hl[0]<<24)|
1263 (p->hl[1]<<16)|
1264 (p->hl[2]<<8)|
1265 (p->hl[3])) +
1266 off;
1267 break;
1268 case FILE_OPMINUS:
1269 offset = (int32_t)((p->hl[0]<<24)|
1270 (p->hl[1]<<16)|
1271 (p->hl[2]<<8)|
1272 (p->hl[3])) -
1273 off;
1274 break;
1275 case FILE_OPMULTIPLY:
1276 offset = (int32_t)((p->hl[0]<<24)|
1277 (p->hl[1]<<16)|
1278 (p->hl[2]<<8)|
1279 (p->hl[3])) *
1280 off;
1281 break;
1282 case FILE_OPDIVIDE:
1283 offset = (int32_t)((p->hl[0]<<24)|
1284 (p->hl[1]<<16)|
1285 (p->hl[2]<<8)|
1286 (p->hl[3])) /
1287 off;
1288 break;
1289 case FILE_OPMODULO:
1290 offset = (int32_t)((p->hl[0]<<24)|
1291 (p->hl[1]<<16)|
1292 (p->hl[2]<<8)|
1293 (p->hl[3])) %
1294 off;
1295 break;
1296 }
1297 } else
1298 offset = (int32_t)((p->hl[0]<<24)|
1299 (p->hl[1]<<16)|
1300 (p->hl[2]<<8)|
1301 (p->hl[3]));
1302 if (m->in_op & FILE_OPINVERSE)
1303 offset = ~offset;
1304 break;
1305 case FILE_LELONG:
1306 case FILE_LEID3:
1307 if (nbytes < (offset + 4))
1308 return 0;
1309 if (off) {
1310 switch (m->in_op & FILE_OPS_MASK) {
1311 case FILE_OPAND:
1312 offset = (int32_t)((p->hl[3]<<24)|
1313 (p->hl[2]<<16)|
1314 (p->hl[1]<<8)|
1315 (p->hl[0])) &
1316 off;
1317 break;
1318 case FILE_OPOR:
1319 offset = (int32_t)((p->hl[3]<<24)|
1320 (p->hl[2]<<16)|
1321 (p->hl[1]<<8)|
1322 (p->hl[0])) |
1323 off;
1324 break;
1325 case FILE_OPXOR:
1326 offset = (int32_t)((p->hl[3]<<24)|
1327 (p->hl[2]<<16)|
1328 (p->hl[1]<<8)|
1329 (p->hl[0])) ^
1330 off;
1331 break;
1332 case FILE_OPADD:
1333 offset = (int32_t)((p->hl[3]<<24)|
1334 (p->hl[2]<<16)|
1335 (p->hl[1]<<8)|
1336 (p->hl[0])) +
1337 off;
1338 break;
1339 case FILE_OPMINUS:
1340 offset = (int32_t)((p->hl[3]<<24)|
1341 (p->hl[2]<<16)|
1342 (p->hl[1]<<8)|
1343 (p->hl[0])) -
1344 off;
1345 break;
1346 case FILE_OPMULTIPLY:
1347 offset = (int32_t)((p->hl[3]<<24)|
1348 (p->hl[2]<<16)|
1349 (p->hl[1]<<8)|
1350 (p->hl[0])) *
1351 off;
1352 break;
1353 case FILE_OPDIVIDE:
1354 offset = (int32_t)((p->hl[3]<<24)|
1355 (p->hl[2]<<16)|
1356 (p->hl[1]<<8)|
1357 (p->hl[0])) /
1358 off;
1359 break;
1360 case FILE_OPMODULO:
1361 offset = (int32_t)((p->hl[3]<<24)|
1362 (p->hl[2]<<16)|
1363 (p->hl[1]<<8)|
1364 (p->hl[0])) %
1365 off;
1366 break;
1367 }
1368 } else
1369 offset = (int32_t)((p->hl[3]<<24)|
1370 (p->hl[2]<<16)|
1371 (p->hl[1]<<8)|
1372 (p->hl[0]));
1373 if (m->in_op & FILE_OPINVERSE)
1374 offset = ~offset;
1375 break;
1376 case FILE_MELONG:
1377 if (nbytes < (offset + 4))
1378 return 0;
1379 if (off) {
1380 switch (m->in_op & FILE_OPS_MASK) {
1381 case FILE_OPAND:
1382 offset = (int32_t)((p->hl[1]<<24)|
1383 (p->hl[0]<<16)|
1384 (p->hl[3]<<8)|
1385 (p->hl[2])) &
1386 off;
1387 break;
1388 case FILE_OPOR:
1389 offset = (int32_t)((p->hl[1]<<24)|
1390 (p->hl[0]<<16)|
1391 (p->hl[3]<<8)|
1392 (p->hl[2])) |
1393 off;
1394 break;
1395 case FILE_OPXOR:
1396 offset = (int32_t)((p->hl[1]<<24)|
1397 (p->hl[0]<<16)|
1398 (p->hl[3]<<8)|
1399 (p->hl[2])) ^
1400 off;
1401 break;
1402 case FILE_OPADD:
1403 offset = (int32_t)((p->hl[1]<<24)|
1404 (p->hl[0]<<16)|
1405 (p->hl[3]<<8)|
1406 (p->hl[2])) +
1407 off;
1408 break;
1409 case FILE_OPMINUS:
1410 offset = (int32_t)((p->hl[1]<<24)|
1411 (p->hl[0]<<16)|
1412 (p->hl[3]<<8)|
1413 (p->hl[2])) -
1414 off;
1415 break;
1416 case FILE_OPMULTIPLY:
1417 offset = (int32_t)((p->hl[1]<<24)|
1418 (p->hl[0]<<16)|
1419 (p->hl[3]<<8)|
1420 (p->hl[2])) *
1421 off;
1422 break;
1423 case FILE_OPDIVIDE:
1424 offset = (int32_t)((p->hl[1]<<24)|
1425 (p->hl[0]<<16)|
1426 (p->hl[3]<<8)|
1427 (p->hl[2])) /
1428 off;
1429 break;
1430 case FILE_OPMODULO:
1431 offset = (int32_t)((p->hl[1]<<24)|
1432 (p->hl[0]<<16)|
1433 (p->hl[3]<<8)|
1434 (p->hl[2])) %
1435 off;
1436 break;
1437 }
1438 } else
1439 offset = (int32_t)((p->hl[1]<<24)|
1440 (p->hl[0]<<16)|
1441 (p->hl[3]<<8)|
1442 (p->hl[2]));
1443 if (m->in_op & FILE_OPINVERSE)
1444 offset = ~offset;
1445 break;
1446 case FILE_LONG:
1447 if (nbytes < (offset + 4))
1448 return 0;
1449 if (off) {
1450 switch (m->in_op & FILE_OPS_MASK) {
1451 case FILE_OPAND:
1452 offset = p->l & off;
1453 break;
1454 case FILE_OPOR:
1455 offset = p->l | off;
1456 break;
1457 case FILE_OPXOR:
1458 offset = p->l ^ off;
1459 break;
1460 case FILE_OPADD:
1461 offset = p->l + off;
1462 break;
1463 case FILE_OPMINUS:
1464 offset = p->l - off;
1465 break;
1466 case FILE_OPMULTIPLY:
1467 offset = p->l * off;
1468 break;
1469 case FILE_OPDIVIDE:
1470 offset = p->l / off;
1471 break;
1472 case FILE_OPMODULO:
1473 offset = p->l % off;
1474 break;
1475 }
1476 } else
1477 offset = p->l;
1478 if (m->in_op & FILE_OPINVERSE)
1479 offset = ~offset;
1480 break;
1481 }
1482
1483 switch (m->in_type) {
1484 case FILE_LEID3:
1485 case FILE_BEID3:
1486 offset = ((((offset >> 0) & 0x7f) << 0) |
1487 (((offset >> 8) & 0x7f) << 7) |
1488 (((offset >> 16) & 0x7f) << 14) |
1489 (((offset >> 24) & 0x7f) << 21)) + 10;
1490 break;
1491 default:
1492 break;
1493 }
1494
1495 if (m->flag & INDIROFFADD) {
1496 offset += ms->c.li[cont_level-1].off;
1497 }
1498 if (mcopy(ms, p, m->type, 0, s, offset, nbytes, count) == -1)
1499 return -1;
1500 ms->offset = offset;
1501
1502 if ((ms->flags & MAGIC_DEBUG) != 0) {
1503 mdebug(offset, (char *)(void *)p,
1504 sizeof(union VALUETYPE));
1505 }
1506 }
1507
1508 /* Verify we have enough data to match magic type */
1509 switch (m->type) {
1510 case FILE_BYTE:
1511 if (nbytes < (offset + 1)) /* should alway be true */
1512 return 0;
1513 break;
1514
1515 case FILE_SHORT:
1516 case FILE_BESHORT:
1517 case FILE_LESHORT:
1518 if (nbytes < (offset + 2))
1519 return 0;
1520 break;
1521
1522 case FILE_LONG:
1523 case FILE_BELONG:
1524 case FILE_LELONG:
1525 case FILE_MELONG:
1526 case FILE_DATE:
1527 case FILE_BEDATE:
1528 case FILE_LEDATE:
1529 case FILE_MEDATE:
1530 case FILE_LDATE:
1531 case FILE_BELDATE:
1532 case FILE_LELDATE:
1533 case FILE_MELDATE:
1534 case FILE_FLOAT:
1535 case FILE_BEFLOAT:
1536 case FILE_LEFLOAT:
1537 if (nbytes < (offset + 4))
1538 return 0;
1539 break;
1540
1541 case FILE_DOUBLE:
1542 case FILE_BEDOUBLE:
1543 case FILE_LEDOUBLE:
1544 if (nbytes < (offset + 8))
1545 return 0;
1546 break;
1547
1548 case FILE_STRING:
1549 case FILE_PSTRING:
1550 case FILE_SEARCH:
1551 if (nbytes < (offset + m->vallen))
1552 return 0;
1553 break;
1554
1555 case FILE_REGEX:
1556 if (nbytes < offset)
1557 return 0;
1558 break;
1559
1560 case FILE_INDIRECT:
1561 if ((ms->flags & (MAGIC_MIME|MAGIC_APPLE)) == 0 &&
1562 file_printf(ms, "%s", m->desc) == -1)
1563 return -1;
1564 if (nbytes < offset)
1565 return 0;
1566 return file_softmagic(ms, s + offset, nbytes - offset,
1567 BINTEST, text);
1568
1569 case FILE_DEFAULT: /* nothing to check */
1570 default:
1571 break;
1572 }
1573 if (!mconvert(ms, m))
1574 return 0;
1575 return 1;
1576 }
1577
1578 private uint64_t
file_strncmp(const char * s1,const char * s2,size_t len,uint32_t flags)1579 file_strncmp(const char *s1, const char *s2, size_t len, uint32_t flags)
1580 {
1581 /*
1582 * Convert the source args to unsigned here so that (1) the
1583 * compare will be unsigned as it is in strncmp() and (2) so
1584 * the ctype functions will work correctly without extra
1585 * casting.
1586 */
1587 const unsigned char *a = (const unsigned char *)s1;
1588 const unsigned char *b = (const unsigned char *)s2;
1589 uint64_t v;
1590
1591 /*
1592 * What we want here is v = strncmp(s1, s2, len),
1593 * but ignoring any nulls.
1594 */
1595 v = 0;
1596 if (0L == flags) { /* normal string: do it fast */
1597 while (len-- > 0)
1598 if ((v = *b++ - *a++) != '\0')
1599 break;
1600 }
1601 else { /* combine the others */
1602 while (len-- > 0) {
1603 if ((flags & STRING_IGNORE_LOWERCASE) &&
1604 islower(*a)) {
1605 if ((v = tolower(*b++) - *a++) != '\0')
1606 break;
1607 }
1608 else if ((flags & STRING_IGNORE_UPPERCASE) &&
1609 isupper(*a)) {
1610 if ((v = toupper(*b++) - *a++) != '\0')
1611 break;
1612 }
1613 else if ((flags & STRING_COMPACT_WHITESPACE) &&
1614 isspace(*a)) {
1615 a++;
1616 if (isspace(*b++)) {
1617 if (!isspace(*a))
1618 while (isspace(*b))
1619 b++;
1620 }
1621 else {
1622 v = 1;
1623 break;
1624 }
1625 }
1626 else if ((flags & STRING_COMPACT_OPTIONAL_WHITESPACE) &&
1627 isspace(*a)) {
1628 a++;
1629 while (isspace(*b))
1630 b++;
1631 }
1632 else {
1633 if ((v = *b++ - *a++) != '\0')
1634 break;
1635 }
1636 }
1637 }
1638 return v;
1639 }
1640
1641 private uint64_t
file_strncmp16(const char * a,const char * b,size_t len,uint32_t flags)1642 file_strncmp16(const char *a, const char *b, size_t len, uint32_t flags)
1643 {
1644 /*
1645 * XXX - The 16-bit string compare probably needs to be done
1646 * differently, especially if the flags are to be supported.
1647 * At the moment, I am unsure.
1648 */
1649 flags = 0;
1650 return file_strncmp(a, b, len, flags);
1651 }
1652
1653 public void
convert_libmagic_pattern(zval * pattern,int options)1654 convert_libmagic_pattern(zval *pattern, int options)
1655 {
1656 int i, j=0;
1657 char *t;
1658
1659 t = (char *) safe_emalloc(Z_STRLEN_P(pattern), 2, 5);
1660
1661 t[j++] = '~';
1662
1663 for (i=0; i<Z_STRLEN_P(pattern); i++, j++) {
1664 switch (Z_STRVAL_P(pattern)[i]) {
1665 case '~':
1666 t[j++] = '\\';
1667 t[j] = '~';
1668 break;
1669 default:
1670 t[j] = Z_STRVAL_P(pattern)[i];
1671 break;
1672 }
1673 }
1674 t[j++] = '~';
1675
1676 if (options & PCRE_CASELESS)
1677 t[j++] = 'i';
1678
1679 if (options & PCRE_MULTILINE)
1680 t[j++] = 'm';
1681
1682 t[j]='\0';
1683
1684 Z_STRVAL_P(pattern) = t;
1685 Z_STRLEN_P(pattern) = j;
1686
1687 }
1688
1689 private int
magiccheck(struct magic_set * ms,struct magic * m)1690 magiccheck(struct magic_set *ms, struct magic *m)
1691 {
1692 uint64_t l = m->value.q;
1693 uint64_t v;
1694 float fl, fv;
1695 double dl, dv;
1696 int matched;
1697 union VALUETYPE *p = &ms->ms_value;
1698
1699 switch (m->type) {
1700 case FILE_BYTE:
1701 v = p->b;
1702 break;
1703
1704 case FILE_SHORT:
1705 case FILE_BESHORT:
1706 case FILE_LESHORT:
1707 v = p->h;
1708 break;
1709
1710 case FILE_LONG:
1711 case FILE_BELONG:
1712 case FILE_LELONG:
1713 case FILE_MELONG:
1714 case FILE_DATE:
1715 case FILE_BEDATE:
1716 case FILE_LEDATE:
1717 case FILE_MEDATE:
1718 case FILE_LDATE:
1719 case FILE_BELDATE:
1720 case FILE_LELDATE:
1721 case FILE_MELDATE:
1722 v = p->l;
1723 break;
1724
1725 case FILE_QUAD:
1726 case FILE_LEQUAD:
1727 case FILE_BEQUAD:
1728 case FILE_QDATE:
1729 case FILE_BEQDATE:
1730 case FILE_LEQDATE:
1731 case FILE_QLDATE:
1732 case FILE_BEQLDATE:
1733 case FILE_LEQLDATE:
1734 v = p->q;
1735 break;
1736
1737 case FILE_FLOAT:
1738 case FILE_BEFLOAT:
1739 case FILE_LEFLOAT:
1740 fl = m->value.f;
1741 fv = p->f;
1742 switch (m->reln) {
1743 case 'x':
1744 matched = 1;
1745 break;
1746
1747 case '!':
1748 matched = fv != fl;
1749 break;
1750
1751 case '=':
1752 matched = fv == fl;
1753 break;
1754
1755 case '>':
1756 matched = fv > fl;
1757 break;
1758
1759 case '<':
1760 matched = fv < fl;
1761 break;
1762
1763 default:
1764 matched = 0;
1765 file_magerror(ms, "cannot happen with float: invalid relation `%c'",
1766 m->reln);
1767 return -1;
1768 }
1769 return matched;
1770
1771 case FILE_DOUBLE:
1772 case FILE_BEDOUBLE:
1773 case FILE_LEDOUBLE:
1774 dl = m->value.d;
1775 dv = p->d;
1776 switch (m->reln) {
1777 case 'x':
1778 matched = 1;
1779 break;
1780
1781 case '!':
1782 matched = dv != dl;
1783 break;
1784
1785 case '=':
1786 matched = dv == dl;
1787 break;
1788
1789 case '>':
1790 matched = dv > dl;
1791 break;
1792
1793 case '<':
1794 matched = dv < dl;
1795 break;
1796
1797 default:
1798 matched = 0;
1799 file_magerror(ms, "cannot happen with double: invalid relation `%c'", m->reln);
1800 return -1;
1801 }
1802 return matched;
1803
1804 case FILE_DEFAULT:
1805 l = 0;
1806 v = 0;
1807 break;
1808
1809 case FILE_STRING:
1810 case FILE_PSTRING:
1811 l = 0;
1812 v = file_strncmp(m->value.s, p->s, (size_t)m->vallen, m->str_flags);
1813 break;
1814
1815 case FILE_BESTRING16:
1816 case FILE_LESTRING16:
1817 l = 0;
1818 v = file_strncmp16(m->value.s, p->s, (size_t)m->vallen, m->str_flags);
1819 break;
1820
1821 case FILE_SEARCH: { /* search ms->search.s for the string m->value.s */
1822 size_t slen;
1823 size_t idx;
1824
1825 if (ms->search.s == NULL)
1826 return 0;
1827
1828 slen = MIN(m->vallen, sizeof(m->value.s));
1829 l = 0;
1830 v = 0;
1831
1832 for (idx = 0; m->str_range == 0 || idx < m->str_range; idx++) {
1833 if (slen + idx > ms->search.s_len)
1834 break;
1835
1836 v = file_strncmp(m->value.s, ms->search.s + idx, slen, m->str_flags);
1837 if (v == 0) { /* found match */
1838 ms->search.offset += idx;
1839 break;
1840 }
1841 }
1842 break;
1843 }
1844 case FILE_REGEX: {
1845 zval *pattern;
1846 int options = 0;
1847 pcre_cache_entry *pce;
1848 TSRMLS_FETCH();
1849
1850 MAKE_STD_ZVAL(pattern);
1851 ZVAL_STRINGL(pattern, (char *)m->value.s, m->vallen, 0);
1852
1853 options |= PCRE_MULTILINE;
1854
1855 if (m->str_flags & STRING_IGNORE_CASE) {
1856 options |= PCRE_CASELESS;
1857 }
1858
1859 convert_libmagic_pattern(pattern, options);
1860
1861 l = v = 0;
1862 #if (PHP_MAJOR_VERSION < 6)
1863 if ((pce = pcre_get_compiled_regex_cache(Z_STRVAL_P(pattern), Z_STRLEN_P(pattern) TSRMLS_CC)) == NULL) {
1864 #else
1865 if ((pce = pcre_get_compiled_regex_cache(IS_STRING, Z_STRVAL_P(pattern), Z_STRLEN_P(pattern) TSRMLS_CC)) == NULL) {
1866 #endif
1867 zval_dtor(pattern);
1868 FREE_ZVAL(pattern);
1869 return -1;
1870 } else {
1871 /* pce now contains the compiled regex */
1872 zval *retval;
1873 zval *subpats;
1874 char *haystack;
1875
1876 MAKE_STD_ZVAL(retval);
1877 ALLOC_INIT_ZVAL(subpats);
1878
1879 /* Cut the search len from haystack, equals to REG_STARTEND */
1880 haystack = estrndup(ms->search.s, ms->search.s_len);
1881
1882 /* match v = 0, no match v = 1 */
1883 #if (PHP_MAJOR_VERSION < 6)
1884 php_pcre_match_impl(pce, haystack, ms->search.s_len, retval, subpats, 1, 1, PREG_OFFSET_CAPTURE, 0 TSRMLS_CC);
1885 #else
1886 php_pcre_match_impl(pce, IS_STRING, haystack, ms->search.s_len, retval, subpats, 1, 1, PREG_OFFSET_CAPTURE, 0 TSRMLS_CC);
1887 #endif
1888 /* Free haystack */
1889 efree(haystack);
1890
1891 if (Z_LVAL_P(retval) < 0) {
1892 zval_ptr_dtor(&subpats);
1893 FREE_ZVAL(retval);
1894 zval_dtor(pattern);
1895 FREE_ZVAL(pattern);
1896 return -1;
1897 } else if ((Z_LVAL_P(retval) > 0) && (Z_TYPE_P(subpats) == IS_ARRAY)) {
1898
1899 /* Need to fetch global match which equals pmatch[0] */
1900 HashTable *ht = Z_ARRVAL_P(subpats);
1901 HashPosition outer_pos;
1902 zval *pattern_match = NULL, *pattern_offset = NULL;
1903
1904 zend_hash_internal_pointer_reset_ex(ht, &outer_pos);
1905
1906 if (zend_hash_has_more_elements_ex(ht, &outer_pos) == SUCCESS &&
1907 zend_hash_move_forward_ex(ht, &outer_pos)) {
1908
1909 zval **ppzval;
1910
1911 /* The first element (should be) is the global match
1912 Need to move to the inner array to get the global match */
1913
1914 if (zend_hash_get_current_data_ex(ht, (void**)&ppzval, &outer_pos) != FAILURE) {
1915
1916 HashTable *inner_ht;
1917 HashPosition inner_pos;
1918 zval **match, **offset;
1919 zval tmpcopy = **ppzval, matchcopy, offsetcopy;
1920
1921 zval_copy_ctor(&tmpcopy);
1922 INIT_PZVAL(&tmpcopy);
1923
1924 inner_ht = Z_ARRVAL(tmpcopy);
1925
1926 /* If everything goes according to the master plan
1927 tmpcopy now contains two elements:
1928 0 = the match
1929 1 = starting position of the match */
1930 zend_hash_internal_pointer_reset_ex(inner_ht, &inner_pos);
1931
1932 if (zend_hash_has_more_elements_ex(inner_ht, &inner_pos) == SUCCESS &&
1933 zend_hash_move_forward_ex(inner_ht, &inner_pos)) {
1934
1935 if (zend_hash_get_current_data_ex(inner_ht, (void**)&match, &inner_pos) != FAILURE) {
1936
1937 matchcopy = **match;
1938 zval_copy_ctor(&matchcopy);
1939 INIT_PZVAL(&matchcopy);
1940 convert_to_string(&matchcopy);
1941
1942 MAKE_STD_ZVAL(pattern_match);
1943 Z_STRVAL_P(pattern_match) = (char *)Z_STRVAL(matchcopy);
1944 Z_STRLEN_P(pattern_match) = Z_STRLEN(matchcopy);
1945 Z_TYPE_P(pattern_match) = IS_STRING;
1946
1947 zval_dtor(&matchcopy);
1948 }
1949 }
1950
1951 if (zend_hash_has_more_elements_ex(inner_ht, &inner_pos) == SUCCESS &&
1952 zend_hash_move_forward_ex(inner_ht, &inner_pos)) {
1953
1954 if (zend_hash_get_current_data_ex(inner_ht, (void**)&offset, &inner_pos) != FAILURE) {
1955
1956 offsetcopy = **offset;
1957 zval_copy_ctor(&offsetcopy);
1958 INIT_PZVAL(&offsetcopy);
1959 convert_to_long(&offsetcopy);
1960
1961 MAKE_STD_ZVAL(pattern_offset);
1962 Z_LVAL_P(pattern_offset) = Z_LVAL(offsetcopy);
1963 Z_TYPE_P(pattern_offset) = IS_LONG;
1964
1965 zval_dtor(&offsetcopy);
1966 }
1967 }
1968 zval_dtor(&tmpcopy);
1969 }
1970
1971 if ((pattern_match != NULL) && (pattern_offset != NULL)) {
1972 ms->search.s += (int)Z_LVAL_P(pattern_offset); /* this is where the match starts */
1973 ms->search.offset += (size_t)Z_LVAL_P(pattern_offset); /* this is where the match starts as size_t */
1974 ms->search.rm_len = Z_STRLEN_P(pattern_match) /* This is the length of the matched pattern */;
1975 v = 0;
1976
1977 efree(pattern_match);
1978 efree(pattern_offset);
1979
1980 } else {
1981 zval_ptr_dtor(&subpats);
1982 FREE_ZVAL(retval);
1983 zval_dtor(pattern);
1984 FREE_ZVAL(pattern);
1985 return -1;
1986 }
1987 }
1988
1989
1990 } else {
1991 v = 1;
1992 }
1993 zval_ptr_dtor(&subpats);
1994 FREE_ZVAL(retval);
1995 }
1996 zval_dtor(pattern);
1997 FREE_ZVAL(pattern);
1998 break;
1999 }
2000 case FILE_INDIRECT:
2001 return 1;
2002 default:
2003 file_magerror(ms, "invalid type %d in magiccheck()", m->type);
2004 return -1;
2005 }
2006
2007 v = file_signextend(ms, m, v);
2008
2009 switch (m->reln) {
2010 case 'x':
2011 if ((ms->flags & MAGIC_DEBUG) != 0)
2012 (void) fprintf(stderr, "%" INT64_T_FORMAT
2013 "u == *any* = 1\n", (unsigned long long)v);
2014 matched = 1;
2015 break;
2016
2017 case '!':
2018 matched = v != l;
2019 if ((ms->flags & MAGIC_DEBUG) != 0)
2020 (void) fprintf(stderr, "%" INT64_T_FORMAT "u != %"
2021 INT64_T_FORMAT "u = %d\n", (unsigned long long)v,
2022 (unsigned long long)l, matched);
2023 break;
2024
2025 case '=':
2026 matched = v == l;
2027 if ((ms->flags & MAGIC_DEBUG) != 0)
2028 (void) fprintf(stderr, "%" INT64_T_FORMAT "u == %"
2029 INT64_T_FORMAT "u = %d\n", (unsigned long long)v,
2030 (unsigned long long)l, matched);
2031 break;
2032
2033 case '>':
2034 if (m->flag & UNSIGNED) {
2035 matched = v > l;
2036 if ((ms->flags & MAGIC_DEBUG) != 0)
2037 (void) fprintf(stderr, "%" INT64_T_FORMAT
2038 "u > %" INT64_T_FORMAT "u = %d\n",
2039 (unsigned long long)v,
2040 (unsigned long long)l, matched);
2041 }
2042 else {
2043 matched = (int64_t) v > (int64_t) l;
2044 if ((ms->flags & MAGIC_DEBUG) != 0)
2045 (void) fprintf(stderr, "%" INT64_T_FORMAT
2046 "d > %" INT64_T_FORMAT "d = %d\n",
2047 (long long)v, (long long)l, matched);
2048 }
2049 break;
2050
2051 case '<':
2052 if (m->flag & UNSIGNED) {
2053 matched = v < l;
2054 if ((ms->flags & MAGIC_DEBUG) != 0)
2055 (void) fprintf(stderr, "%" INT64_T_FORMAT
2056 "u < %" INT64_T_FORMAT "u = %d\n",
2057 (unsigned long long)v,
2058 (unsigned long long)l, matched);
2059 }
2060 else {
2061 matched = (int64_t) v < (int64_t) l;
2062 if ((ms->flags & MAGIC_DEBUG) != 0)
2063 (void) fprintf(stderr, "%" INT64_T_FORMAT
2064 "d < %" INT64_T_FORMAT "d = %d\n",
2065 (long long)v, (long long)l, matched);
2066 }
2067 break;
2068
2069 case '&':
2070 matched = (v & l) == l;
2071 if ((ms->flags & MAGIC_DEBUG) != 0)
2072 (void) fprintf(stderr, "((%" INT64_T_FORMAT "x & %"
2073 INT64_T_FORMAT "x) == %" INT64_T_FORMAT
2074 "x) = %d\n", (unsigned long long)v,
2075 (unsigned long long)l, (unsigned long long)l,
2076 matched);
2077 break;
2078
2079 case '^':
2080 matched = (v & l) != l;
2081 if ((ms->flags & MAGIC_DEBUG) != 0)
2082 (void) fprintf(stderr, "((%" INT64_T_FORMAT "x & %"
2083 INT64_T_FORMAT "x) != %" INT64_T_FORMAT
2084 "x) = %d\n", (unsigned long long)v,
2085 (unsigned long long)l, (unsigned long long)l,
2086 matched);
2087 break;
2088
2089 default:
2090 matched = 0;
2091 file_magerror(ms, "cannot happen: invalid relation `%c'",
2092 m->reln);
2093 return -1;
2094 }
2095
2096 return matched;
2097 }
2098
2099 private int
2100 handle_annotation(struct magic_set *ms, struct magic *m)
2101 {
2102 if (ms->flags & MAGIC_APPLE) {
2103 if (file_printf(ms, "%.8s", m->apple) == -1)
2104 return -1;
2105 return 1;
2106 }
2107 if ((ms->flags & MAGIC_MIME_TYPE) && m->mimetype[0]) {
2108 if (file_printf(ms, "%s", m->mimetype) == -1)
2109 return -1;
2110 return 1;
2111 }
2112 return 0;
2113 }
2114
2115 private int
2116 print_sep(struct magic_set *ms, int firstline)
2117 {
2118 if (ms->flags & MAGIC_MIME)
2119 return 0;
2120 if (firstline)
2121 return 0;
2122 /*
2123 * we found another match
2124 * put a newline and '-' to do some simple formatting
2125 */
2126 return file_printf(ms, "\n- ");
2127 }
2128
