Update to file-4.19.
[dragonfly.git] / contrib / file-4 / src / softmagic.c
blob77250487bdc63f23bfe193bcf7265366ab02d895
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.
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.
29 * softmagic - interpret variable magic from MAGIC
32 #include "file.h"
33 #include "magic.h"
34 #include <string.h>
35 #include <ctype.h>
36 #include <stdlib.h>
37 #include <time.h>
38 #include <regex.h>
41 #ifndef lint
42 FILE_RCSID("@(#)$Id: softmagic.c,v 1.87 2006/12/11 21:48:49 christos Exp $")
43 #endif /* lint */
45 private int match(struct magic_set *, struct magic *, uint32_t,
46 const unsigned char *, size_t);
47 private int mget(struct magic_set *, union VALUETYPE *, const unsigned char *,
48 struct magic *, size_t, unsigned int);
49 private int magiccheck(struct magic_set *, union VALUETYPE *, struct magic *);
50 private int32_t mprint(struct magic_set *, union VALUETYPE *, struct magic *);
51 private void mdebug(uint32_t, const char *, size_t);
52 private int mcopy(struct magic_set *, union VALUETYPE *, int, int,
53 const unsigned char *, uint32_t, size_t);
54 private int mconvert(struct magic_set *, union VALUETYPE *, struct magic *);
55 private int check_mem(struct magic_set *, unsigned int);
56 private int print_sep(struct magic_set *, int);
57 private void cvt_8(union VALUETYPE *, const struct magic *);
58 private void cvt_16(union VALUETYPE *, const struct magic *);
59 private void cvt_32(union VALUETYPE *, const struct magic *);
60 private void cvt_64(union VALUETYPE *, const struct magic *);
63 * softmagic - lookup one file in parsed, in-memory copy of database
64 * Passed the name and FILE * of one file to be typed.
66 /*ARGSUSED1*/ /* nbytes passed for regularity, maybe need later */
67 protected int
68 file_softmagic(struct magic_set *ms, const unsigned char *buf, size_t nbytes)
70 struct mlist *ml;
71 int rv;
72 for (ml = ms->mlist->next; ml != ms->mlist; ml = ml->next)
73 if ((rv = match(ms, ml->magic, ml->nmagic, buf, nbytes)) != 0)
74 return rv;
76 return 0;
80 * Go through the whole list, stopping if you find a match. Process all
81 * the continuations of that match before returning.
83 * We support multi-level continuations:
85 * At any time when processing a successful top-level match, there is a
86 * current continuation level; it represents the level of the last
87 * successfully matched continuation.
89 * Continuations above that level are skipped as, if we see one, it
90 * means that the continuation that controls them - i.e, the
91 * lower-level continuation preceding them - failed to match.
93 * Continuations below that level are processed as, if we see one,
94 * it means we've finished processing or skipping higher-level
95 * continuations under the control of a successful or unsuccessful
96 * lower-level continuation, and are now seeing the next lower-level
97 * continuation and should process it. The current continuation
98 * level reverts to the level of the one we're seeing.
100 * Continuations at the current level are processed as, if we see
101 * one, there's no lower-level continuation that may have failed.
103 * If a continuation matches, we bump the current continuation level
104 * so that higher-level continuations are processed.
106 private int
107 match(struct magic_set *ms, struct magic *magic, uint32_t nmagic,
108 const unsigned char *s, size_t nbytes)
110 uint32_t magindex = 0;
111 unsigned int cont_level = 0;
112 int need_separator = 0;
113 union VALUETYPE p;
114 int32_t oldoff = 0;
115 int returnval = 0; /* if a match is found it is set to 1*/
116 int firstline = 1; /* a flag to print X\n X\n- X */
117 int printed_something = 0;
119 if (check_mem(ms, cont_level) == -1)
120 return -1;
122 for (magindex = 0; magindex < nmagic; magindex++) {
123 /* if main entry matches, print it... */
124 ms->offset = magic[magindex].offset;
125 int flush = !mget(ms, &p, s, &magic[magindex], nbytes,
126 cont_level);
127 if (flush) {
128 if (magic[magindex].reln == '!')
129 flush = 0;
130 } else {
131 switch (magiccheck(ms, &p, &magic[magindex])) {
132 case -1:
133 return -1;
134 case 0:
135 flush++;
136 break;
137 default:
138 break;
141 if (flush) {
143 * main entry didn't match,
144 * flush its continuations
146 while (magindex < nmagic - 1 &&
147 magic[magindex + 1].cont_level != 0)
148 magindex++;
149 continue;
153 * If we are going to print something, we'll need to print
154 * a blank before we print something else.
156 if (magic[magindex].desc[0]) {
157 need_separator = 1;
158 printed_something = 1;
159 if (print_sep(ms, firstline) == -1)
160 return -1;
163 if ((ms->c.off[cont_level] = mprint(ms, &p, &magic[magindex]))
164 == -1)
165 return -1;
167 /* and any continuations that match */
168 if (check_mem(ms, ++cont_level) == -1)
169 return -1;
171 while (magic[magindex+1].cont_level != 0 &&
172 ++magindex < nmagic) {
173 if (cont_level < magic[magindex].cont_level)
174 continue;
175 if (cont_level > magic[magindex].cont_level) {
177 * We're at the end of the level
178 * "cont_level" continuations.
180 cont_level = magic[magindex].cont_level;
182 ms->offset = magic[magindex].offset;
183 if (magic[magindex].flag & OFFADD) {
184 ms->offset +=
185 ms->c.off[cont_level - 1];
188 flush = !mget(ms, &p, s, &magic[magindex], nbytes,
189 cont_level);
190 if (flush && magic[magindex].reln != '!')
191 continue;
193 switch (flush ? 1 : magiccheck(ms, &p, &magic[magindex])) {
194 case -1:
195 return -1;
196 case 0:
197 break;
198 default:
200 * If we are going to print something,
201 * make sure that we have a separator first.
203 if (magic[magindex].desc[0]) {
204 printed_something = 1;
205 if (print_sep(ms, firstline) == -1)
206 return -1;
209 * This continuation matched.
210 * Print its message, with
211 * a blank before it if
212 * the previous item printed
213 * and this item isn't empty.
215 /* space if previous printed */
216 if (need_separator
217 && (magic[magindex].nospflag == 0)
218 && (magic[magindex].desc[0] != '\0')) {
219 if (file_printf(ms, " ") == -1)
220 return -1;
221 need_separator = 0;
223 if ((ms->c.off[cont_level] = mprint(ms, &p,
224 &magic[magindex])) == -1)
225 return -1;
226 if (magic[magindex].desc[0])
227 need_separator = 1;
230 * If we see any continuations
231 * at a higher level,
232 * process them.
234 if (check_mem(ms, ++cont_level) == -1)
235 return -1;
238 firstline = 0;
239 if (printed_something)
240 returnval = 1;
241 if ((ms->flags & MAGIC_CONTINUE) == 0 && printed_something) {
242 return 1; /* don't keep searching */
245 return returnval; /* This is hit if -k is set or there is no match */
248 private int
249 check_mem(struct magic_set *ms, unsigned int level)
251 size_t len;
253 if (level < ms->c.len)
254 return 0;
256 len = (ms->c.len += 20) * sizeof(*ms->c.off);
257 ms->c.off = (ms->c.off == NULL) ? malloc(len) : realloc(ms->c.off, len);
258 if (ms->c.off != NULL)
259 return 0;
260 file_oomem(ms, len);
261 return -1;
264 private int
265 check_fmt(struct magic_set *ms, struct magic *m)
267 regex_t rx;
268 int rc;
270 if (strchr(m->desc, '%') == NULL)
271 return 0;
273 rc = regcomp(&rx, "%[-0-9\\.]*s", REG_EXTENDED|REG_NOSUB);
274 if (rc) {
275 char errmsg[512];
276 regerror(rc, &rx, errmsg, sizeof(errmsg));
277 file_error(ms, 0, "regex error %d, (%s)", rc, errmsg);
278 return -1;
279 } else {
280 rc = regexec(&rx, m->desc, 0, 0, 0);
281 regfree(&rx);
282 return !rc;
286 private int32_t
287 mprint(struct magic_set *ms, union VALUETYPE *p, struct magic *m)
289 uint64_t v;
290 int32_t t = 0;
291 char buf[512];
294 switch (m->type) {
295 case FILE_BYTE:
296 v = file_signextend(ms, m, (uint64_t)p->b);
297 switch (check_fmt(ms, m)) {
298 case -1:
299 return -1;
300 case 1:
301 if (snprintf(buf, sizeof(buf), "%c",
302 (unsigned char)v) < 0)
303 return -1;
304 if (file_printf(ms, m->desc, buf) == -1)
305 return -1;
306 break;
307 default:
308 if (file_printf(ms, m->desc, (unsigned char) v) == -1)
309 return -1;
310 break;
312 t = ms->offset + sizeof(char);
313 break;
315 case FILE_SHORT:
316 case FILE_BESHORT:
317 case FILE_LESHORT:
318 v = file_signextend(ms, m, (uint64_t)p->h);
319 switch (check_fmt(ms, m)) {
320 case -1:
321 return -1;
322 case 1:
323 if (snprintf(buf, sizeof(buf), "%hu",
324 (unsigned short)v) < 0)
325 return -1;
326 if (file_printf(ms, m->desc, buf) == -1)
327 return -1;
328 break;
329 default:
330 if (file_printf(ms, m->desc, (unsigned short) v) == -1)
331 return -1;
332 break;
334 t = ms->offset + sizeof(short);
335 break;
337 case FILE_LONG:
338 case FILE_BELONG:
339 case FILE_LELONG:
340 case FILE_MELONG:
341 v = file_signextend(ms, m, (uint64_t)p->l);
342 switch (check_fmt(ms, m)) {
343 case -1:
344 return -1;
345 case 1:
346 if (snprintf(buf, sizeof(buf), "%u", (uint32_t)v) < 0)
347 return -1;
348 if (file_printf(ms, m->desc, buf) == -1)
349 return -1;
350 break;
351 default:
352 if (file_printf(ms, m->desc, (uint32_t) v) == -1)
353 return -1;
354 break;
356 t = ms->offset + sizeof(int32_t);
357 break;
359 case FILE_QUAD:
360 case FILE_BEQUAD:
361 case FILE_LEQUAD:
362 v = file_signextend(ms, m, p->q);
363 if (file_printf(ms, m->desc, (uint64_t) v) == -1)
364 return -1;
365 t = ms->offset + sizeof(int64_t);
366 break;
367 case FILE_STRING:
368 case FILE_PSTRING:
369 case FILE_BESTRING16:
370 case FILE_LESTRING16:
371 if (m->reln == '=' || m->reln == '!') {
372 if (file_printf(ms, m->desc, m->value.s) == -1)
373 return -1;
374 t = ms->offset + m->vallen;
376 else {
377 if (*m->value.s == '\0') {
378 char *cp = strchr(p->s,'\n');
379 if (cp)
380 *cp = '\0';
382 if (file_printf(ms, m->desc, p->s) == -1)
383 return -1;
384 t = ms->offset + strlen(p->s);
386 break;
388 case FILE_DATE:
389 case FILE_BEDATE:
390 case FILE_LEDATE:
391 case FILE_MEDATE:
392 if (file_printf(ms, m->desc, file_fmttime(p->l, 1)) == -1)
393 return -1;
394 t = ms->offset + sizeof(time_t);
395 break;
397 case FILE_LDATE:
398 case FILE_BELDATE:
399 case FILE_LELDATE:
400 case FILE_MELDATE:
401 if (file_printf(ms, m->desc, file_fmttime(p->l, 0)) == -1)
402 return -1;
403 t = ms->offset + sizeof(time_t);
404 break;
406 case FILE_QDATE:
407 case FILE_BEQDATE:
408 case FILE_LEQDATE:
409 if (file_printf(ms, m->desc, file_fmttime((uint32_t)p->q, 1))
410 == -1)
411 return -1;
412 t = ms->offset + sizeof(uint64_t);
413 break;
415 case FILE_QLDATE:
416 case FILE_BEQLDATE:
417 case FILE_LEQLDATE:
418 if (file_printf(ms, m->desc, file_fmttime((uint32_t)p->q, 0))
419 == -1)
420 return -1;
421 t = ms->offset + sizeof(uint64_t);
422 break;
424 case FILE_REGEX:
425 if (file_printf(ms, m->desc, p->s) == -1)
426 return -1;
427 t = ms->offset + strlen(p->s);
428 break;
430 case FILE_SEARCH:
431 if (file_printf(ms, m->desc, m->value.s) == -1)
432 return -1;
433 t = ms->offset + m->vallen;
434 break;
436 default:
437 file_error(ms, 0, "invalid m->type (%d) in mprint()", m->type);
438 return -1;
440 return(t);
444 #define DO_CVT(fld, cast) \
445 if (m->mask) \
446 switch (m->mask_op & 0x7F) { \
447 case FILE_OPAND: \
448 p->fld &= cast m->mask; \
449 break; \
450 case FILE_OPOR: \
451 p->fld |= cast m->mask; \
452 break; \
453 case FILE_OPXOR: \
454 p->fld ^= cast m->mask; \
455 break; \
456 case FILE_OPADD: \
457 p->fld += cast m->mask; \
458 break; \
459 case FILE_OPMINUS: \
460 p->fld -= cast m->mask; \
461 break; \
462 case FILE_OPMULTIPLY: \
463 p->fld *= cast m->mask; \
464 break; \
465 case FILE_OPDIVIDE: \
466 p->fld /= cast m->mask; \
467 break; \
468 case FILE_OPMODULO: \
469 p->fld %= cast m->mask; \
470 break; \
472 if (m->mask_op & FILE_OPINVERSE) \
473 p->fld = ~p->fld \
475 private void
476 cvt_8(union VALUETYPE *p, const struct magic *m)
478 DO_CVT(b, (uint8_t));
481 private void
482 cvt_16(union VALUETYPE *p, const struct magic *m)
484 DO_CVT(h, (uint16_t));
487 private void
488 cvt_32(union VALUETYPE *p, const struct magic *m)
490 DO_CVT(l, (uint32_t));
493 private void
494 cvt_64(union VALUETYPE *p, const struct magic *m)
496 DO_CVT(q, (uint64_t));
500 * Convert the byte order of the data we are looking at
501 * While we're here, let's apply the mask operation
502 * (unless you have a better idea)
504 private int
505 mconvert(struct magic_set *ms, union VALUETYPE *p, struct magic *m)
507 switch (m->type) {
508 case FILE_BYTE:
509 cvt_8(p, m);
510 return 1;
511 case FILE_SHORT:
512 cvt_16(p, m);
513 return 1;
514 case FILE_LONG:
515 case FILE_DATE:
516 case FILE_LDATE:
517 cvt_32(p, m);
518 return 1;
519 case FILE_QUAD:
520 case FILE_QDATE:
521 case FILE_QLDATE:
522 cvt_64(p, m);
523 return 1;
524 case FILE_STRING:
525 case FILE_BESTRING16:
526 case FILE_LESTRING16:
528 size_t len;
530 /* Null terminate and eat *trailing* return */
531 p->s[sizeof(p->s) - 1] = '\0';
532 len = strlen(p->s);
533 if (len-- && p->s[len] == '\n')
534 p->s[len] = '\0';
535 return 1;
537 case FILE_PSTRING:
539 char *ptr1 = p->s, *ptr2 = ptr1 + 1;
540 size_t len = *p->s;
541 if (len >= sizeof(p->s))
542 len = sizeof(p->s) - 1;
543 while (len--)
544 *ptr1++ = *ptr2++;
545 *ptr1 = '\0';
546 len = strlen(p->s);
547 if (len-- && p->s[len] == '\n')
548 p->s[len] = '\0';
549 return 1;
551 case FILE_BESHORT:
552 p->h = (short)((p->hs[0]<<8)|(p->hs[1]));
553 cvt_16(p, m);
554 return 1;
555 case FILE_BELONG:
556 case FILE_BEDATE:
557 case FILE_BELDATE:
558 p->l = (int32_t)
559 ((p->hl[0]<<24)|(p->hl[1]<<16)|(p->hl[2]<<8)|(p->hl[3]));
560 cvt_32(p, m);
561 return 1;
562 case FILE_BEQUAD:
563 case FILE_BEQDATE:
564 case FILE_BEQLDATE:
565 p->q = (int64_t)
566 (((int64_t)p->hq[0]<<56)|((int64_t)p->hq[1]<<48)|
567 ((int64_t)p->hq[2]<<40)|((int64_t)p->hq[3]<<32)|
568 (p->hq[4]<<24)|(p->hq[5]<<16)|(p->hq[6]<<8)|(p->hq[7]));
569 cvt_64(p, m);
570 return 1;
571 case FILE_LESHORT:
572 p->h = (short)((p->hs[1]<<8)|(p->hs[0]));
573 cvt_16(p, m);
574 return 1;
575 case FILE_LELONG:
576 case FILE_LEDATE:
577 case FILE_LELDATE:
578 p->l = (int32_t)
579 ((p->hl[3]<<24)|(p->hl[2]<<16)|(p->hl[1]<<8)|(p->hl[0]));
580 cvt_32(p, m);
581 return 1;
582 case FILE_LEQUAD:
583 case FILE_LEQDATE:
584 case FILE_LEQLDATE:
585 p->q = (int64_t)
586 (((int64_t)p->hq[7]<<56)|((int64_t)p->hq[6]<<48)|
587 ((int64_t)p->hq[5]<<40)|((int64_t)p->hq[4]<<32)|
588 (p->hq[3]<<24)|(p->hq[2]<<16)|(p->hq[1]<<8)|(p->hq[0]));
589 cvt_64(p, m);
590 return 1;
591 case FILE_MELONG:
592 case FILE_MEDATE:
593 case FILE_MELDATE:
594 p->l = (int32_t)
595 ((p->hl[1]<<24)|(p->hl[0]<<16)|(p->hl[3]<<8)|(p->hl[2]));
596 cvt_32(p, m);
597 return 1;
598 case FILE_REGEX:
599 case FILE_SEARCH:
600 return 1;
601 default:
602 file_error(ms, 0, "invalid type %d in mconvert()", m->type);
603 return 0;
608 private void
609 mdebug(uint32_t offset, const char *str, size_t len)
611 (void) fprintf(stderr, "mget @%d: ", offset);
612 file_showstr(stderr, str, len);
613 (void) fputc('\n', stderr);
614 (void) fputc('\n', stderr);
617 private int
618 mcopy(struct magic_set *ms, union VALUETYPE *p, int type, int indir,
619 const unsigned char *s, uint32_t offset, size_t nbytes)
621 if (type == FILE_REGEX && indir == 0) {
623 * offset is interpreted as last line to search,
624 * (starting at 1), not as bytes-from start-of-file
626 char *b, *c, *last = NULL;
627 if (s == NULL) {
628 p->search.buflen = 0;
629 p->search.buf = NULL;
630 return 0;
632 if ((p->search.buf = strdup((const char *)s)) == NULL) {
633 file_oomem(ms, strlen((const char *)s));
634 return -1;
636 for (b = p->search.buf; offset &&
637 ((b = strchr(c = b, '\n')) || (b = strchr(c, '\r')));
638 offset--, b++) {
639 last = b;
640 if (b[0] == '\r' && b[1] == '\n') b++;
642 if (last != NULL)
643 *last = '\0';
644 p->search.buflen = last - p->search.buf;
645 return 0;
648 if (indir == 0 && (type == FILE_BESTRING16 || type == FILE_LESTRING16))
650 const unsigned char *src = s + offset;
651 const unsigned char *esrc = s + nbytes;
652 char *dst = p->s, *edst = &p->s[sizeof(p->s) - 1];
654 if (type == FILE_BESTRING16)
655 src++;
657 /* check for pointer overflow */
658 if (src < s) {
659 file_error(ms, 0, "invalid offset %zu in mcopy()",
660 offset);
661 return -1;
664 for (;src < esrc; src++, dst++) {
665 if (dst < edst)
666 *dst = *src++;
667 else
668 break;
669 if (*dst == '\0')
670 *dst = ' ';
672 *edst = '\0';
673 return 0;
676 if (offset >= nbytes) {
677 (void)memset(p, '\0', sizeof(*p));
678 return 0;
680 if (nbytes - offset < sizeof(*p))
681 nbytes = nbytes - offset;
682 else
683 nbytes = sizeof(*p);
685 (void)memcpy(p, s + offset, nbytes);
688 * the usefulness of padding with zeroes eludes me, it
689 * might even cause problems
691 if (nbytes < sizeof(*p))
692 (void)memset(((char *)(void *)p) + nbytes, '\0',
693 sizeof(*p) - nbytes);
694 return 0;
697 private int
698 mget(struct magic_set *ms, union VALUETYPE *p, const unsigned char *s,
699 struct magic *m, size_t nbytes, unsigned int cont_level)
701 uint32_t offset = ms->offset;
703 if (mcopy(ms, p, m->type, m->flag & INDIR, s, offset, nbytes) == -1)
704 return -1;
706 if ((ms->flags & MAGIC_DEBUG) != 0) {
707 mdebug(offset, (char *)(void *)p, sizeof(union VALUETYPE));
708 file_mdump(m);
711 if (m->flag & INDIR) {
712 int off = m->in_offset;
713 if (m->in_op & FILE_OPINDIRECT) {
714 const union VALUETYPE *q =
715 ((const void *)(s + offset + off));
716 switch (m->in_type) {
717 case FILE_BYTE:
718 off = q->b;
719 break;
720 case FILE_SHORT:
721 off = q->h;
722 break;
723 case FILE_BESHORT:
724 off = (short)((q->hs[0]<<8)|(q->hs[1]));
725 break;
726 case FILE_LESHORT:
727 off = (short)((q->hs[1]<<8)|(q->hs[0]));
728 break;
729 case FILE_LONG:
730 off = q->l;
731 break;
732 case FILE_BELONG:
733 off = (int32_t)((q->hl[0]<<24)|(q->hl[1]<<16)|
734 (q->hl[2]<<8)|(q->hl[3]));
735 break;
736 case FILE_LELONG:
737 off = (int32_t)((q->hl[3]<<24)|(q->hl[2]<<16)|
738 (q->hl[1]<<8)|(q->hl[0]));
739 break;
740 case FILE_MELONG:
741 off = (int32_t)((q->hl[1]<<24)|(q->hl[0]<<16)|
742 (q->hl[3]<<8)|(q->hl[2]));
743 break;
746 switch (m->in_type) {
747 case FILE_BYTE:
748 if (nbytes < (offset + 1)) return 0;
749 if (off) {
750 switch (m->in_op & 0x3F) {
751 case FILE_OPAND:
752 offset = p->b & off;
753 break;
754 case FILE_OPOR:
755 offset = p->b | off;
756 break;
757 case FILE_OPXOR:
758 offset = p->b ^ off;
759 break;
760 case FILE_OPADD:
761 offset = p->b + off;
762 break;
763 case FILE_OPMINUS:
764 offset = p->b - off;
765 break;
766 case FILE_OPMULTIPLY:
767 offset = p->b * off;
768 break;
769 case FILE_OPDIVIDE:
770 offset = p->b / off;
771 break;
772 case FILE_OPMODULO:
773 offset = p->b % off;
774 break;
776 } else
777 offset = p->b;
778 if (m->in_op & FILE_OPINVERSE)
779 offset = ~offset;
780 break;
781 case FILE_BESHORT:
782 if (nbytes < (offset + 2))
783 return 0;
784 if (off) {
785 switch (m->in_op & 0x7F) {
786 case FILE_OPAND:
787 offset = (short)((p->hs[0]<<8)|
788 (p->hs[1])) &
789 off;
790 break;
791 case FILE_OPOR:
792 offset = (short)((p->hs[0]<<8)|
793 (p->hs[1])) |
794 off;
795 break;
796 case FILE_OPXOR:
797 offset = (short)((p->hs[0]<<8)|
798 (p->hs[1])) ^
799 off;
800 break;
801 case FILE_OPADD:
802 offset = (short)((p->hs[0]<<8)|
803 (p->hs[1])) +
804 off;
805 break;
806 case FILE_OPMINUS:
807 offset = (short)((p->hs[0]<<8)|
808 (p->hs[1])) -
809 off;
810 break;
811 case FILE_OPMULTIPLY:
812 offset = (short)((p->hs[0]<<8)|
813 (p->hs[1])) *
814 off;
815 break;
816 case FILE_OPDIVIDE:
817 offset = (short)((p->hs[0]<<8)|
818 (p->hs[1])) /
819 off;
820 break;
821 case FILE_OPMODULO:
822 offset = (short)((p->hs[0]<<8)|
823 (p->hs[1])) %
824 off;
825 break;
827 } else
828 offset = (short)((p->hs[0]<<8)|
829 (p->hs[1]));
830 if (m->in_op & FILE_OPINVERSE)
831 offset = ~offset;
832 break;
833 case FILE_LESHORT:
834 if (nbytes < (offset + 2))
835 return 0;
836 if (off) {
837 switch (m->in_op & 0x7F) {
838 case FILE_OPAND:
839 offset = (short)((p->hs[1]<<8)|
840 (p->hs[0])) &
841 off;
842 break;
843 case FILE_OPOR:
844 offset = (short)((p->hs[1]<<8)|
845 (p->hs[0])) |
846 off;
847 break;
848 case FILE_OPXOR:
849 offset = (short)((p->hs[1]<<8)|
850 (p->hs[0])) ^
851 off;
852 break;
853 case FILE_OPADD:
854 offset = (short)((p->hs[1]<<8)|
855 (p->hs[0])) +
856 off;
857 break;
858 case FILE_OPMINUS:
859 offset = (short)((p->hs[1]<<8)|
860 (p->hs[0])) -
861 off;
862 break;
863 case FILE_OPMULTIPLY:
864 offset = (short)((p->hs[1]<<8)|
865 (p->hs[0])) *
866 off;
867 break;
868 case FILE_OPDIVIDE:
869 offset = (short)((p->hs[1]<<8)|
870 (p->hs[0])) /
871 off;
872 break;
873 case FILE_OPMODULO:
874 offset = (short)((p->hs[1]<<8)|
875 (p->hs[0])) %
876 off;
877 break;
879 } else
880 offset = (short)((p->hs[1]<<8)|
881 (p->hs[0]));
882 if (m->in_op & FILE_OPINVERSE)
883 offset = ~offset;
884 break;
885 case FILE_SHORT:
886 if (nbytes < (offset + 2))
887 return 0;
888 if (off) {
889 switch (m->in_op & 0x7F) {
890 case FILE_OPAND:
891 offset = p->h & off;
892 break;
893 case FILE_OPOR:
894 offset = p->h | off;
895 break;
896 case FILE_OPXOR:
897 offset = p->h ^ off;
898 break;
899 case FILE_OPADD:
900 offset = p->h + off;
901 break;
902 case FILE_OPMINUS:
903 offset = p->h - off;
904 break;
905 case FILE_OPMULTIPLY:
906 offset = p->h * off;
907 break;
908 case FILE_OPDIVIDE:
909 offset = p->h / off;
910 break;
911 case FILE_OPMODULO:
912 offset = p->h % off;
913 break;
916 else
917 offset = p->h;
918 if (m->in_op & FILE_OPINVERSE)
919 offset = ~offset;
920 break;
921 case FILE_BELONG:
922 if (nbytes < (offset + 4))
923 return 0;
924 if (off) {
925 switch (m->in_op & 0x7F) {
926 case FILE_OPAND:
927 offset = (int32_t)((p->hl[0]<<24)|
928 (p->hl[1]<<16)|
929 (p->hl[2]<<8)|
930 (p->hl[3])) &
931 off;
932 break;
933 case FILE_OPOR:
934 offset = (int32_t)((p->hl[0]<<24)|
935 (p->hl[1]<<16)|
936 (p->hl[2]<<8)|
937 (p->hl[3])) |
938 off;
939 break;
940 case FILE_OPXOR:
941 offset = (int32_t)((p->hl[0]<<24)|
942 (p->hl[1]<<16)|
943 (p->hl[2]<<8)|
944 (p->hl[3])) ^
945 off;
946 break;
947 case FILE_OPADD:
948 offset = (int32_t)((p->hl[0]<<24)|
949 (p->hl[1]<<16)|
950 (p->hl[2]<<8)|
951 (p->hl[3])) +
952 off;
953 break;
954 case FILE_OPMINUS:
955 offset = (int32_t)((p->hl[0]<<24)|
956 (p->hl[1]<<16)|
957 (p->hl[2]<<8)|
958 (p->hl[3])) -
959 off;
960 break;
961 case FILE_OPMULTIPLY:
962 offset = (int32_t)((p->hl[0]<<24)|
963 (p->hl[1]<<16)|
964 (p->hl[2]<<8)|
965 (p->hl[3])) *
966 off;
967 break;
968 case FILE_OPDIVIDE:
969 offset = (int32_t)((p->hl[0]<<24)|
970 (p->hl[1]<<16)|
971 (p->hl[2]<<8)|
972 (p->hl[3])) /
973 off;
974 break;
975 case FILE_OPMODULO:
976 offset = (int32_t)((p->hl[0]<<24)|
977 (p->hl[1]<<16)|
978 (p->hl[2]<<8)|
979 (p->hl[3])) %
980 off;
981 break;
983 } else
984 offset = (int32_t)((p->hl[0]<<24)|
985 (p->hl[1]<<16)|
986 (p->hl[2]<<8)|
987 (p->hl[3]));
988 if (m->in_op & FILE_OPINVERSE)
989 offset = ~offset;
990 break;
991 case FILE_LELONG:
992 if (nbytes < (offset + 4))
993 return 0;
994 if (off) {
995 switch (m->in_op & 0x7F) {
996 case FILE_OPAND:
997 offset = (int32_t)((p->hl[3]<<24)|
998 (p->hl[2]<<16)|
999 (p->hl[1]<<8)|
1000 (p->hl[0])) &
1001 off;
1002 break;
1003 case FILE_OPOR:
1004 offset = (int32_t)((p->hl[3]<<24)|
1005 (p->hl[2]<<16)|
1006 (p->hl[1]<<8)|
1007 (p->hl[0])) |
1008 off;
1009 break;
1010 case FILE_OPXOR:
1011 offset = (int32_t)((p->hl[3]<<24)|
1012 (p->hl[2]<<16)|
1013 (p->hl[1]<<8)|
1014 (p->hl[0])) ^
1015 off;
1016 break;
1017 case FILE_OPADD:
1018 offset = (int32_t)((p->hl[3]<<24)|
1019 (p->hl[2]<<16)|
1020 (p->hl[1]<<8)|
1021 (p->hl[0])) +
1022 off;
1023 break;
1024 case FILE_OPMINUS:
1025 offset = (int32_t)((p->hl[3]<<24)|
1026 (p->hl[2]<<16)|
1027 (p->hl[1]<<8)|
1028 (p->hl[0])) -
1029 off;
1030 break;
1031 case FILE_OPMULTIPLY:
1032 offset = (int32_t)((p->hl[3]<<24)|
1033 (p->hl[2]<<16)|
1034 (p->hl[1]<<8)|
1035 (p->hl[0])) *
1036 off;
1037 break;
1038 case FILE_OPDIVIDE:
1039 offset = (int32_t)((p->hl[3]<<24)|
1040 (p->hl[2]<<16)|
1041 (p->hl[1]<<8)|
1042 (p->hl[0])) /
1043 off;
1044 break;
1045 case FILE_OPMODULO:
1046 offset = (int32_t)((p->hl[3]<<24)|
1047 (p->hl[2]<<16)|
1048 (p->hl[1]<<8)|
1049 (p->hl[0])) %
1050 off;
1051 break;
1053 } else
1054 offset = (int32_t)((p->hl[3]<<24)|
1055 (p->hl[2]<<16)|
1056 (p->hl[1]<<8)|
1057 (p->hl[0]));
1058 if (m->in_op & FILE_OPINVERSE)
1059 offset = ~offset;
1060 break;
1061 case FILE_MELONG:
1062 if (nbytes < (offset + 4))
1063 return 0;
1064 if (off) {
1065 switch (m->in_op & 0x7F) {
1066 case FILE_OPAND:
1067 offset = (int32_t)((p->hl[1]<<24)|
1068 (p->hl[0]<<16)|
1069 (p->hl[3]<<8)|
1070 (p->hl[2])) &
1071 off;
1072 break;
1073 case FILE_OPOR:
1074 offset = (int32_t)((p->hl[1]<<24)|
1075 (p->hl[0]<<16)|
1076 (p->hl[3]<<8)|
1077 (p->hl[2])) |
1078 off;
1079 break;
1080 case FILE_OPXOR:
1081 offset = (int32_t)((p->hl[1]<<24)|
1082 (p->hl[0]<<16)|
1083 (p->hl[3]<<8)|
1084 (p->hl[2])) ^
1085 off;
1086 break;
1087 case FILE_OPADD:
1088 offset = (int32_t)((p->hl[1]<<24)|
1089 (p->hl[0]<<16)|
1090 (p->hl[3]<<8)|
1091 (p->hl[2])) +
1092 off;
1093 break;
1094 case FILE_OPMINUS:
1095 offset = (int32_t)((p->hl[1]<<24)|
1096 (p->hl[0]<<16)|
1097 (p->hl[3]<<8)|
1098 (p->hl[2])) -
1099 off;
1100 break;
1101 case FILE_OPMULTIPLY:
1102 offset = (int32_t)((p->hl[1]<<24)|
1103 (p->hl[0]<<16)|
1104 (p->hl[3]<<8)|
1105 (p->hl[2])) *
1106 off;
1107 break;
1108 case FILE_OPDIVIDE:
1109 offset = (int32_t)((p->hl[1]<<24)|
1110 (p->hl[0]<<16)|
1111 (p->hl[3]<<8)|
1112 (p->hl[2])) /
1113 off;
1114 break;
1115 case FILE_OPMODULO:
1116 offset = (int32_t)((p->hl[1]<<24)|
1117 (p->hl[0]<<16)|
1118 (p->hl[3]<<8)|
1119 (p->hl[2])) %
1120 off;
1121 break;
1123 } else
1124 offset = (int32_t)((p->hl[1]<<24)|
1125 (p->hl[0]<<16)|
1126 (p->hl[3]<<8)|
1127 (p->hl[2]));
1128 if (m->in_op & FILE_OPINVERSE)
1129 offset = ~offset;
1130 break;
1131 case FILE_LONG:
1132 if (nbytes < (offset + 4))
1133 return 0;
1134 if (off) {
1135 switch (m->in_op & 0x7F) {
1136 case FILE_OPAND:
1137 offset = p->l & off;
1138 break;
1139 case FILE_OPOR:
1140 offset = p->l | off;
1141 break;
1142 case FILE_OPXOR:
1143 offset = p->l ^ off;
1144 break;
1145 case FILE_OPADD:
1146 offset = p->l + off;
1147 break;
1148 case FILE_OPMINUS:
1149 offset = p->l - off;
1150 break;
1151 case FILE_OPMULTIPLY:
1152 offset = p->l * off;
1153 break;
1154 case FILE_OPDIVIDE:
1155 offset = p->l / off;
1156 break;
1157 case FILE_OPMODULO:
1158 offset = p->l % off;
1159 break;
1160 /* case TOOMANYSWITCHBLOCKS:
1161 * ugh = p->eye % m->strain;
1162 * rub;
1163 * case BEER:
1164 * off = p->tab & m->in_gest;
1165 * sleep;
1168 } else
1169 offset = p->l;
1170 if (m->in_op & FILE_OPINVERSE)
1171 offset = ~offset;
1172 break;
1175 if (m->flag & INDIROFFADD) offset += ms->c.off[cont_level-1];
1176 if (mcopy(ms, p, m->type, 0, s, offset, nbytes) == -1)
1177 return -1;
1178 ms->offset = offset;
1180 if ((ms->flags & MAGIC_DEBUG) != 0) {
1181 mdebug(offset, (char *)(void *)p,
1182 sizeof(union VALUETYPE));
1183 file_mdump(m);
1187 /* Verify we have enough data to match magic type */
1188 switch (m->type) {
1189 case FILE_BYTE:
1190 if (nbytes < (offset + 1)) /* should alway be true */
1191 return 0;
1192 break;
1194 case FILE_SHORT:
1195 case FILE_BESHORT:
1196 case FILE_LESHORT:
1197 if (nbytes < (offset + 2))
1198 return 0;
1199 break;
1201 case FILE_LONG:
1202 case FILE_BELONG:
1203 case FILE_LELONG:
1204 case FILE_MELONG:
1205 case FILE_DATE:
1206 case FILE_BEDATE:
1207 case FILE_LEDATE:
1208 case FILE_MEDATE:
1209 case FILE_LDATE:
1210 case FILE_BELDATE:
1211 case FILE_LELDATE:
1212 case FILE_MELDATE:
1213 if (nbytes < (offset + 4))
1214 return 0;
1215 break;
1217 case FILE_STRING:
1218 case FILE_PSTRING:
1219 case FILE_SEARCH:
1220 if (nbytes < (offset + m->vallen))
1221 return 0;
1222 break;
1223 default: break;
1226 if (m->type == FILE_SEARCH) {
1227 size_t mlen = (size_t)(m->mask + m->vallen);
1228 size_t flen = nbytes - offset;
1229 if (flen < mlen)
1230 mlen = flen;
1231 p->search.buflen = mlen;
1232 p->search.buf = malloc(mlen + 1);
1233 if (p->search.buf == NULL) {
1234 file_error(ms, errno, "Cannot allocate search buffer");
1235 return 0;
1237 (void)memcpy(p->search.buf, s + offset, mlen);
1238 p->search.buf[mlen] = '\0';
1240 if (!mconvert(ms, p, m))
1241 return 0;
1242 return 1;
1245 private int
1246 magiccheck(struct magic_set *ms, union VALUETYPE *p, struct magic *m)
1248 uint64_t l = m->value.q;
1249 uint64_t v;
1250 int matched;
1252 if ( (m->value.s[0] == 'x') && (m->value.s[1] == '\0') ) {
1253 return 1;
1257 switch (m->type) {
1258 case FILE_BYTE:
1259 v = p->b;
1260 break;
1262 case FILE_SHORT:
1263 case FILE_BESHORT:
1264 case FILE_LESHORT:
1265 v = p->h;
1266 break;
1268 case FILE_LONG:
1269 case FILE_BELONG:
1270 case FILE_LELONG:
1271 case FILE_MELONG:
1272 case FILE_DATE:
1273 case FILE_BEDATE:
1274 case FILE_LEDATE:
1275 case FILE_MEDATE:
1276 case FILE_LDATE:
1277 case FILE_BELDATE:
1278 case FILE_LELDATE:
1279 case FILE_MELDATE:
1280 v = p->l;
1281 break;
1283 case FILE_QUAD:
1284 case FILE_LEQUAD:
1285 case FILE_BEQUAD:
1286 case FILE_QDATE:
1287 case FILE_BEQDATE:
1288 case FILE_LEQDATE:
1289 case FILE_QLDATE:
1290 case FILE_BEQLDATE:
1291 case FILE_LEQLDATE:
1292 v = p->q;
1293 break;
1295 case FILE_STRING:
1296 case FILE_BESTRING16:
1297 case FILE_LESTRING16:
1298 case FILE_PSTRING:
1301 * What we want here is:
1302 * v = strncmp(m->value.s, p->s, m->vallen);
1303 * but ignoring any nulls. bcmp doesn't give -/+/0
1304 * and isn't universally available anyway.
1306 unsigned char *a = (unsigned char*)m->value.s;
1307 unsigned char *b = (unsigned char*)p->s;
1308 int len = m->vallen;
1309 l = 0;
1310 v = 0;
1311 if (0L == m->mask) { /* normal string: do it fast */
1312 while (--len >= 0)
1313 if ((v = *b++ - *a++) != '\0')
1314 break;
1315 } else { /* combine the others */
1316 while (--len >= 0) {
1317 if ((m->mask & STRING_IGNORE_LOWERCASE) &&
1318 islower(*a)) {
1319 if ((v = tolower(*b++) - *a++) != '\0')
1320 break;
1321 } else if ((m->mask & STRING_COMPACT_BLANK) &&
1322 isspace(*a)) {
1323 a++;
1324 if (isspace(*b++)) {
1325 while (isspace(*b))
1326 b++;
1327 } else {
1328 v = 1;
1329 break;
1331 } else if (isspace(*a) &&
1332 (m->mask & STRING_COMPACT_OPTIONAL_BLANK)) {
1333 a++;
1334 while (isspace(*b))
1335 b++;
1336 } else {
1337 if ((v = *b++ - *a++) != '\0')
1338 break;
1342 break;
1344 case FILE_REGEX:
1346 int rc;
1347 regex_t rx;
1348 char errmsg[512];
1350 if (p->search.buf == NULL)
1351 return 0;
1353 rc = regcomp(&rx, m->value.s,
1354 REG_EXTENDED|REG_NOSUB|REG_NEWLINE|
1355 ((m->mask & STRING_IGNORE_LOWERCASE) ? REG_ICASE : 0));
1356 if (rc) {
1357 free(p->search.buf);
1358 p->search.buf = NULL;
1359 regerror(rc, &rx, errmsg, sizeof(errmsg));
1360 file_error(ms, 0, "regex error %d, (%s)", rc, errmsg);
1361 return -1;
1362 } else {
1363 rc = regexec(&rx, p->search.buf, 0, 0, 0);
1364 regfree(&rx);
1365 free(p->search.buf);
1366 p->search.buf = NULL;
1367 return !rc;
1370 case FILE_SEARCH:
1373 * search for a string in a certain range
1375 unsigned char *a = (unsigned char*)m->value.s;
1376 unsigned char *b = (unsigned char*)p->search.buf;
1377 size_t len, slen = m->vallen;
1378 size_t range = 0;
1379 if (slen > sizeof(m->value.s))
1380 slen = sizeof(m->value.s);
1381 l = 0;
1382 v = 0;
1383 if (b == NULL)
1384 return 0;
1385 len = slen;
1386 while (++range <= m->mask) {
1387 while (len-- > 0 && (v = *b++ - *a++) == 0)
1388 continue;
1389 if (!v) {
1390 ms->offset += range - 1;
1391 break;
1393 if (range + slen >= p->search.buflen) {
1394 v = 1;
1395 break;
1397 len = slen;
1398 a = (unsigned char*)m->value.s;
1399 b = (unsigned char*)p->search.buf + range;
1401 free(p->search.buf);
1402 p->search.buf = NULL;
1403 if (v)
1404 return 0;
1405 break;
1407 default:
1408 file_error(ms, 0, "invalid type %d in magiccheck()", m->type);
1409 return -1;
1412 if (m->type != FILE_STRING && m->type != FILE_PSTRING)
1413 v = file_signextend(ms, m, v);
1415 switch (m->reln) {
1416 case 'x':
1417 if ((ms->flags & MAGIC_DEBUG) != 0)
1418 (void) fprintf(stderr, "%llu == *any* = 1\n",
1419 (unsigned long long)v);
1420 matched = 1;
1421 break;
1423 case '!':
1424 matched = v != l;
1425 if ((ms->flags & MAGIC_DEBUG) != 0)
1426 (void) fprintf(stderr, "%llu != %llu = %d\n",
1427 (unsigned long long)v, (unsigned long long)l,
1428 matched);
1429 break;
1431 case '=':
1432 matched = v == l;
1433 if ((ms->flags & MAGIC_DEBUG) != 0)
1434 (void) fprintf(stderr, "%llu == %llu = %d\n",
1435 (unsigned long long)v, (unsigned long long)l,
1436 matched);
1437 break;
1439 case '>':
1440 if (m->flag & UNSIGNED) {
1441 matched = v > l;
1442 if ((ms->flags & MAGIC_DEBUG) != 0)
1443 (void) fprintf(stderr, "%llu > %llu = %d\n",
1444 (unsigned long long)v,
1445 (unsigned long long)l, matched);
1447 else {
1448 matched = (int32_t) v > (int32_t) l;
1449 if ((ms->flags & MAGIC_DEBUG) != 0)
1450 (void) fprintf(stderr, "%lld > %lld = %d\n",
1451 (long long)v, (long long)l, matched);
1453 break;
1455 case '<':
1456 if (m->flag & UNSIGNED) {
1457 matched = v < l;
1458 if ((ms->flags & MAGIC_DEBUG) != 0)
1459 (void) fprintf(stderr, "%llu < %llu = %d\n",
1460 (unsigned long long)v,
1461 (unsigned long long)l, matched);
1463 else {
1464 matched = (int32_t) v < (int32_t) l;
1465 if ((ms->flags & MAGIC_DEBUG) != 0)
1466 (void) fprintf(stderr, "%lld < %lld = %d\n",
1467 (long long)v, (long long)l, matched);
1469 break;
1471 case '&':
1472 matched = (v & l) == l;
1473 if ((ms->flags & MAGIC_DEBUG) != 0)
1474 (void) fprintf(stderr, "((%llx & %llx) == %llx) = %d\n",
1475 (unsigned long long)v, (unsigned long long)l,
1476 (unsigned long long)l, matched);
1477 break;
1479 case '^':
1480 matched = (v & l) != l;
1481 if ((ms->flags & MAGIC_DEBUG) != 0)
1482 (void) fprintf(stderr, "((%llx & %llx) != %llx) = %d\n",
1483 (unsigned long long)v, (unsigned long long)l,
1484 (unsigned long long)l, matched);
1485 break;
1487 default:
1488 matched = 0;
1489 file_error(ms, 0, "cannot happen: invalid relation `%c'",
1490 m->reln);
1491 return -1;
1494 return matched;
1497 private int
1498 print_sep(struct magic_set *ms, int firstline)
1500 if (firstline)
1501 return 0;
1503 * we found another match
1504 * put a newline and '-' to do some simple formatting
1506 return file_printf(ms, "\n- ");