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.
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
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
29 * softmagic - interpret variable magic from MAGIC
35 FILE_RCSID("@(#)$File: softmagic.c,v 1.135 2009/03/27 22:42:49 christos Exp $")
45 private int match(struct magic_set *, struct magic *, uint32_t,
46 const unsigned char *, size_t, int, int);
47 private int mget(struct magic_set *, const unsigned char *,
48 struct magic *, size_t, unsigned int, int);
49 private int magiccheck(struct magic_set *, struct magic *);
50 private int32_t mprint(struct magic_set *, struct magic *);
51 private int32_t moffset(struct magic_set *, struct magic *);
52 private void mdebug(uint32_t, const char *, size_t);
53 private int mcopy(struct magic_set *, union VALUETYPE *, int, int,
54 const unsigned char *, uint32_t, size_t, size_t);
55 private int mconvert(struct magic_set *, struct magic *);
56 private int print_sep(struct magic_set *, int);
57 private int handle_annotation(struct magic_set *, struct magic *);
58 private void cvt_8(union VALUETYPE *, const struct magic *);
59 private void cvt_16(union VALUETYPE *, const struct magic *);
60 private void cvt_32(union VALUETYPE *, const struct magic *);
61 private void cvt_64(union VALUETYPE *, const struct magic *);
63 #define OFFSET_OOB(n, o, i) ((n) < (o) || (i) > ((n) - (o)))
65 * softmagic - lookup one file in parsed, in-memory copy of database
66 * Passed the name and FILE * of one file to be typed.
68 /*ARGSUSED1*/ /* nbytes passed for regularity, maybe need later */
70 file_softmagic(struct magic_set *ms, const unsigned char *buf, size_t nbytes,
71 size_t level, int mode)
75 for (ml = ms->mlist->next; ml != ms->mlist; ml = ml->next)
76 if ((rv = match(ms, ml->magic, ml->nmagic, buf, nbytes, mode, level)) != 0)
83 * Go through the whole list, stopping if you find a match. Process all
84 * the continuations of that match before returning.
86 * We support multi-level continuations:
88 * At any time when processing a successful top-level match, there is a
89 * current continuation level; it represents the level of the last
90 * successfully matched continuation.
92 * Continuations above that level are skipped as, if we see one, it
93 * means that the continuation that controls them - i.e, the
94 * lower-level continuation preceding them - failed to match.
96 * Continuations below that level are processed as, if we see one,
97 * it means we've finished processing or skipping higher-level
98 * continuations under the control of a successful or unsuccessful
99 * lower-level continuation, and are now seeing the next lower-level
100 * continuation and should process it. The current continuation
101 * level reverts to the level of the one we're seeing.
103 * Continuations at the current level are processed as, if we see
104 * one, there's no lower-level continuation that may have failed.
106 * If a continuation matches, we bump the current continuation level
107 * so that higher-level continuations are processed.
110 match(struct magic_set *ms, struct magic *magic, uint32_t nmagic,
111 const unsigned char *s, size_t nbytes, int mode, int recursion_level)
113 uint32_t magindex = 0;
114 unsigned int cont_level = 0;
115 int need_separator = 0;
116 int returnval = 0, e; /* if a match is found it is set to 1*/
117 int firstline = 1; /* a flag to print X\n X\n- X */
118 int printed_something = 0;
119 int print = (ms->flags & (MAGIC_MIME|MAGIC_APPLE)) == 0;
121 if (file_check_mem(ms, cont_level) == -1)
124 for (magindex = 0; magindex < nmagic; magindex++) {
126 struct magic *m = &magic[magindex];
128 if ((m->flag & BINTEST) != mode) {
130 while (magic[magindex + 1].cont_level != 0 &&
133 continue; /* Skip to next top-level test*/
136 ms->offset = m->offset;
137 ms->line = m->lineno;
139 /* if main entry matches, print it... */
140 switch (mget(ms, s, m, nbytes, cont_level, recursion_level + 1)) {
144 flush = m->reln != '!';
147 if (m->type == FILE_INDIRECT)
150 switch (magiccheck(ms, m)) {
164 * main entry didn't match,
165 * flush its continuations
167 while (magindex < nmagic - 1 &&
168 magic[magindex + 1].cont_level != 0)
174 * If we are going to print something, we'll need to print
175 * a blank before we print something else.
179 printed_something = 1;
180 if ((e = handle_annotation(ms, m)) != 0)
182 if (print_sep(ms, firstline) == -1)
187 if (print && mprint(ms, m) == -1)
190 ms->c.li[cont_level].off = moffset(ms, m);
192 /* and any continuations that match */
193 if (file_check_mem(ms, ++cont_level) == -1)
196 while (magic[magindex+1].cont_level != 0 &&
197 ++magindex < nmagic) {
198 m = &magic[magindex];
199 ms->line = m->lineno; /* for messages */
201 if (cont_level < m->cont_level)
203 if (cont_level > m->cont_level) {
205 * We're at the end of the level
206 * "cont_level" continuations.
208 cont_level = m->cont_level;
210 ms->offset = m->offset;
211 if (m->flag & OFFADD) {
213 ms->c.li[cont_level - 1].off;
216 #ifdef ENABLE_CONDITIONALS
217 if (m->cond == COND_ELSE ||
218 m->cond == COND_ELIF) {
219 if (ms->c.li[cont_level].last_match == 1)
223 switch (mget(ms, s, m, nbytes, cont_level, recursion_level + 1)) {
232 if (m->type == FILE_INDIRECT)
238 switch (flush ? 1 : magiccheck(ms, m)) {
242 #ifdef ENABLE_CONDITIONALS
243 ms->c.li[cont_level].last_match = 0;
247 #ifdef ENABLE_CONDITIONALS
248 ms->c.li[cont_level].last_match = 1;
250 if (m->type != FILE_DEFAULT)
251 ms->c.li[cont_level].got_match = 1;
252 else if (ms->c.li[cont_level].got_match) {
253 ms->c.li[cont_level].got_match = 0;
257 * If we are going to print something,
258 * make sure that we have a separator first.
261 if ((e = handle_annotation(ms, m)) != 0)
263 if (!printed_something) {
264 printed_something = 1;
265 if (print_sep(ms, firstline)
271 * This continuation matched. Print
272 * its message, with a blank before it
273 * if the previous item printed and
274 * this item isn't empty.
276 /* space if previous printed */
278 && ((m->flag & NOSPACE) == 0)
281 file_printf(ms, " ") == -1)
285 if (print && mprint(ms, m) == -1)
288 ms->c.li[cont_level].off = moffset(ms, m);
294 * If we see any continuations
298 if (file_check_mem(ms, ++cont_level) == -1)
303 if (printed_something) {
308 if ((ms->flags & MAGIC_CONTINUE) == 0 && printed_something) {
309 return returnval; /* don't keep searching */
312 return returnval; /* This is hit if -k is set or there is no match */
316 check_fmt(struct magic_set *ms, struct magic *m)
321 if (strchr(m->desc, '%') == NULL)
324 rc = regcomp(&rx, "%[-0-9\\.]*s", REG_EXTENDED|REG_NOSUB);
327 (void)regerror(rc, &rx, errmsg, sizeof(errmsg));
328 file_magerror(ms, "regex error %d, (%s)", rc, errmsg);
331 rc = regexec(&rx, m->desc, 0, 0, 0);
338 char * strndup(const char *, size_t);
341 strndup(const char *str, size_t n)
346 for (len = 0; len < n && str[len]; len++)
348 if ((copy = malloc(len + 1)) == NULL)
350 (void)memcpy(copy, str, len);
354 #endif /* HAVE_STRNDUP */
357 mprint(struct magic_set *ms, struct magic *m)
364 union VALUETYPE *p = &ms->ms_value;
368 v = file_signextend(ms, m, (uint64_t)p->b);
369 switch (check_fmt(ms, m)) {
373 (void)snprintf(buf, sizeof(buf), "%c",
375 if (file_printf(ms, m->desc, buf) == -1)
379 if (file_printf(ms, m->desc, (unsigned char) v) == -1)
383 t = ms->offset + sizeof(char);
389 v = file_signextend(ms, m, (uint64_t)p->h);
390 switch (check_fmt(ms, m)) {
394 (void)snprintf(buf, sizeof(buf), "%hu",
396 if (file_printf(ms, m->desc, buf) == -1)
401 file_printf(ms, m->desc, (unsigned short) v) == -1)
405 t = ms->offset + sizeof(short);
412 v = file_signextend(ms, m, (uint64_t)p->l);
413 switch (check_fmt(ms, m)) {
417 (void)snprintf(buf, sizeof(buf), "%u", (uint32_t)v);
418 if (file_printf(ms, m->desc, buf) == -1)
422 if (file_printf(ms, m->desc, (uint32_t) v) == -1)
426 t = ms->offset + sizeof(int32_t);
432 v = file_signextend(ms, m, p->q);
433 if (file_printf(ms, m->desc, (uint64_t) v) == -1)
435 t = ms->offset + sizeof(int64_t);
440 case FILE_BESTRING16:
441 case FILE_LESTRING16:
442 if (m->reln == '=' || m->reln == '!') {
443 if (file_printf(ms, m->desc, m->value.s) == -1)
445 t = ms->offset + m->vallen;
448 if (*m->value.s == '\0')
449 p->s[strcspn(p->s, "\n")] = '\0';
450 if (file_printf(ms, m->desc, p->s) == -1)
452 t = ms->offset + strlen(p->s);
453 if (m->type == FILE_PSTRING)
462 if (file_printf(ms, m->desc, file_fmttime(p->l, 1)) == -1)
464 t = ms->offset + sizeof(time_t);
471 if (file_printf(ms, m->desc, file_fmttime(p->l, 0)) == -1)
473 t = ms->offset + sizeof(time_t);
479 if (file_printf(ms, m->desc, file_fmttime((uint32_t)p->q,
482 t = ms->offset + sizeof(uint64_t);
488 if (file_printf(ms, m->desc, file_fmttime((uint32_t)p->q,
491 t = ms->offset + sizeof(uint64_t);
498 switch (check_fmt(ms, m)) {
502 (void)snprintf(buf, sizeof(buf), "%g", vf);
503 if (file_printf(ms, m->desc, buf) == -1)
507 if (file_printf(ms, m->desc, vf) == -1)
511 t = ms->offset + sizeof(float);
518 switch (check_fmt(ms, m)) {
522 (void)snprintf(buf, sizeof(buf), "%g", vd);
523 if (file_printf(ms, m->desc, buf) == -1)
527 if (file_printf(ms, m->desc, vd) == -1)
531 t = ms->offset + sizeof(double);
538 cp = strndup((const char *)ms->search.s, ms->search.rm_len);
540 file_oomem(ms, ms->search.rm_len);
543 rval = file_printf(ms, m->desc, cp);
549 if ((m->str_flags & REGEX_OFFSET_START))
550 t = ms->search.offset;
552 t = ms->search.offset + ms->search.rm_len;
557 if (file_printf(ms, m->desc, m->value.s) == -1)
559 if ((m->str_flags & REGEX_OFFSET_START))
560 t = ms->search.offset;
562 t = ms->search.offset + m->vallen;
566 if (file_printf(ms, m->desc, m->value.s) == -1)
576 file_magerror(ms, "invalid m->type (%d) in mprint()", m->type);
583 moffset(struct magic_set *ms, struct magic *m)
587 return ms->offset + sizeof(char);
592 return ms->offset + sizeof(short);
598 return ms->offset + sizeof(int32_t);
603 return ms->offset + sizeof(int64_t);
607 case FILE_BESTRING16:
608 case FILE_LESTRING16:
609 if (m->reln == '=' || m->reln == '!')
610 return ms->offset + m->vallen;
612 union VALUETYPE *p = &ms->ms_value;
615 if (*m->value.s == '\0')
616 p->s[strcspn(p->s, "\n")] = '\0';
617 t = ms->offset + strlen(p->s);
618 if (m->type == FILE_PSTRING)
627 return ms->offset + sizeof(time_t);
633 return ms->offset + sizeof(time_t);
638 return ms->offset + sizeof(uint64_t);
643 return ms->offset + sizeof(uint64_t);
648 return ms->offset + sizeof(float);
653 return ms->offset + sizeof(double);
657 if ((m->str_flags & REGEX_OFFSET_START) != 0)
658 return ms->search.offset;
660 return ms->search.offset + ms->search.rm_len;
663 if ((m->str_flags & REGEX_OFFSET_START) != 0)
664 return ms->search.offset;
666 return ms->search.offset + m->vallen;
679 #define DO_CVT(fld, cast) \
681 switch (m->mask_op & FILE_OPS_MASK) { \
683 p->fld &= cast m->num_mask; \
686 p->fld |= cast m->num_mask; \
689 p->fld ^= cast m->num_mask; \
692 p->fld += cast m->num_mask; \
695 p->fld -= cast m->num_mask; \
697 case FILE_OPMULTIPLY: \
698 p->fld *= cast m->num_mask; \
700 case FILE_OPDIVIDE: \
701 p->fld /= cast m->num_mask; \
703 case FILE_OPMODULO: \
704 p->fld %= cast m->num_mask; \
707 if (m->mask_op & FILE_OPINVERSE) \
711 cvt_8(union VALUETYPE *p, const struct magic *m)
713 DO_CVT(b, (uint8_t));
717 cvt_16(union VALUETYPE *p, const struct magic *m)
719 DO_CVT(h, (uint16_t));
723 cvt_32(union VALUETYPE *p, const struct magic *m)
725 DO_CVT(l, (uint32_t));
729 cvt_64(union VALUETYPE *p, const struct magic *m)
731 DO_CVT(q, (uint64_t));
734 #define DO_CVT2(fld, cast) \
736 switch (m->mask_op & FILE_OPS_MASK) { \
738 p->fld += cast m->num_mask; \
741 p->fld -= cast m->num_mask; \
743 case FILE_OPMULTIPLY: \
744 p->fld *= cast m->num_mask; \
746 case FILE_OPDIVIDE: \
747 p->fld /= cast m->num_mask; \
752 cvt_float(union VALUETYPE *p, const struct magic *m)
758 cvt_double(union VALUETYPE *p, const struct magic *m)
760 DO_CVT2(d, (double));
764 * Convert the byte order of the data we are looking at
765 * While we're here, let's apply the mask operation
766 * (unless you have a better idea)
769 mconvert(struct magic_set *ms, struct magic *m)
771 union VALUETYPE *p = &ms->ms_value;
791 case FILE_BESTRING16:
792 case FILE_LESTRING16: {
793 /* Null terminate and eat *trailing* return */
794 p->s[sizeof(p->s) - 1] = '\0';
796 /* Why? breaks magic numbers that end with \xa */
798 if (len-- && p->s[len] == '\n')
804 char *ptr1 = p->s, *ptr2 = ptr1 + 1;
806 if (len >= sizeof(p->s))
807 len = sizeof(p->s) - 1;
812 /* Why? breaks magic numbers that end with \xa */
814 if (len-- && p->s[len] == '\n')
820 p->h = (short)((p->hs[0]<<8)|(p->hs[1]));
827 ((p->hl[0]<<24)|(p->hl[1]<<16)|(p->hl[2]<<8)|(p->hl[3]));
834 (((uint64_t)p->hq[0]<<56)|((uint64_t)p->hq[1]<<48)|
835 ((uint64_t)p->hq[2]<<40)|((uint64_t)p->hq[3]<<32)|
836 ((uint64_t)p->hq[4]<<24)|((uint64_t)p->hq[5]<<16)|
837 ((uint64_t)p->hq[6]<<8)|((uint64_t)p->hq[7]));
841 p->h = (short)((p->hs[1]<<8)|(p->hs[0]));
848 ((p->hl[3]<<24)|(p->hl[2]<<16)|(p->hl[1]<<8)|(p->hl[0]));
855 (((uint64_t)p->hq[7]<<56)|((uint64_t)p->hq[6]<<48)|
856 ((uint64_t)p->hq[5]<<40)|((uint64_t)p->hq[4]<<32)|
857 ((uint64_t)p->hq[3]<<24)|((uint64_t)p->hq[2]<<16)|
858 ((uint64_t)p->hq[1]<<8)|((uint64_t)p->hq[0]));
865 ((p->hl[1]<<24)|(p->hl[0]<<16)|(p->hl[3]<<8)|(p->hl[2]));
872 p->l = ((uint32_t)p->hl[0]<<24)|((uint32_t)p->hl[1]<<16)|
873 ((uint32_t)p->hl[2]<<8) |((uint32_t)p->hl[3]);
877 p->l = ((uint32_t)p->hl[3]<<24)|((uint32_t)p->hl[2]<<16)|
878 ((uint32_t)p->hl[1]<<8) |((uint32_t)p->hl[0]);
885 p->q = ((uint64_t)p->hq[0]<<56)|((uint64_t)p->hq[1]<<48)|
886 ((uint64_t)p->hq[2]<<40)|((uint64_t)p->hq[3]<<32)|
887 ((uint64_t)p->hq[4]<<24)|((uint64_t)p->hq[5]<<16)|
888 ((uint64_t)p->hq[6]<<8) |((uint64_t)p->hq[7]);
892 p->q = ((uint64_t)p->hq[7]<<56)|((uint64_t)p->hq[6]<<48)|
893 ((uint64_t)p->hq[5]<<40)|((uint64_t)p->hq[4]<<32)|
894 ((uint64_t)p->hq[3]<<24)|((uint64_t)p->hq[2]<<16)|
895 ((uint64_t)p->hq[1]<<8) |((uint64_t)p->hq[0]);
903 file_magerror(ms, "invalid type %d in mconvert()", m->type);
910 mdebug(uint32_t offset, const char *str, size_t len)
912 (void) fprintf(stderr, "mget @%d: ", offset);
913 file_showstr(stderr, str, len);
914 (void) fputc('\n', stderr);
915 (void) fputc('\n', stderr);
919 mcopy(struct magic_set *ms, union VALUETYPE *p, int type, int indir,
920 const unsigned char *s, uint32_t offset, size_t nbytes, size_t linecnt)
923 * Note: FILE_SEARCH and FILE_REGEX do not actually copy
924 * anything, but setup pointers into the source
929 ms->search.s = (const char *)s + offset;
930 ms->search.s_len = nbytes - offset;
931 ms->search.offset = offset;
937 const char *last; /* end of search region */
938 const char *buf; /* start of search region */
943 ms->search.s_len = 0;
947 buf = (const char *)s + offset;
948 end = last = (const char *)s + nbytes;
949 /* mget() guarantees buf <= last */
950 for (lines = linecnt, b = buf;
951 lines && ((b = memchr(c = b, '\n', end - b)) || (b = memchr(c, '\r', end - c)));
954 if (b[0] == '\r' && b[1] == '\n')
958 last = (const char *)s + nbytes;
961 ms->search.s_len = last - buf;
962 ms->search.offset = offset;
963 ms->search.rm_len = 0;
966 case FILE_BESTRING16:
967 case FILE_LESTRING16: {
968 const unsigned char *src = s + offset;
969 const unsigned char *esrc = s + nbytes;
971 char *edst = &p->s[sizeof(p->s) - 1];
973 if (type == FILE_BESTRING16)
976 /* check for pointer overflow */
978 file_magerror(ms, "invalid offset %u in mcopy()",
982 for (/*EMPTY*/; src < esrc; src += 2, dst++) {
988 if (type == FILE_BESTRING16 ?
997 case FILE_STRING: /* XXX - these two should not need */
998 case FILE_PSTRING: /* to copy anything, but do anyway. */
1004 if (offset >= nbytes) {
1005 (void)memset(p, '\0', sizeof(*p));
1008 if (nbytes - offset < sizeof(*p))
1009 nbytes = nbytes - offset;
1011 nbytes = sizeof(*p);
1013 (void)memcpy(p, s + offset, nbytes);
1016 * the usefulness of padding with zeroes eludes me, it
1017 * might even cause problems
1019 if (nbytes < sizeof(*p))
1020 (void)memset(((char *)(void *)p) + nbytes, '\0',
1021 sizeof(*p) - nbytes);
1026 mget(struct magic_set *ms, const unsigned char *s,
1027 struct magic *m, size_t nbytes, unsigned int cont_level, int recursion_level)
1029 uint32_t offset = ms->offset;
1030 uint32_t count = m->str_range;
1031 union VALUETYPE *p = &ms->ms_value;
1033 if (recursion_level >= 20) {
1034 file_error(ms, 0, "recursion nesting exceeded");
1038 if (mcopy(ms, p, m->type, m->flag & INDIR, s, offset, nbytes, count) == -1)
1041 if ((ms->flags & MAGIC_DEBUG) != 0) {
1042 mdebug(offset, (char *)(void *)p, sizeof(union VALUETYPE));
1043 #ifndef COMPILE_ONLY
1048 if (m->flag & INDIR) {
1049 int off = m->in_offset;
1050 if (m->in_op & FILE_OPINDIRECT) {
1051 const union VALUETYPE *q = CAST(const union VALUETYPE *,
1052 ((const void *)(s + offset + off)));
1053 switch (m->in_type) {
1061 off = (short)((q->hs[0]<<8)|(q->hs[1]));
1064 off = (short)((q->hs[1]<<8)|(q->hs[0]));
1071 off = (int32_t)((q->hl[0]<<24)|(q->hl[1]<<16)|
1072 (q->hl[2]<<8)|(q->hl[3]));
1076 off = (int32_t)((q->hl[3]<<24)|(q->hl[2]<<16)|
1077 (q->hl[1]<<8)|(q->hl[0]));
1080 off = (int32_t)((q->hl[1]<<24)|(q->hl[0]<<16)|
1081 (q->hl[3]<<8)|(q->hl[2]));
1085 switch (m->in_type) {
1087 if (OFFSET_OOB(nbytes, offset, 1))
1090 switch (m->in_op & FILE_OPS_MASK) {
1092 offset = p->b & off;
1095 offset = p->b | off;
1098 offset = p->b ^ off;
1101 offset = p->b + off;
1104 offset = p->b - off;
1106 case FILE_OPMULTIPLY:
1107 offset = p->b * off;
1110 offset = p->b / off;
1113 offset = p->b % off;
1118 if (m->in_op & FILE_OPINVERSE)
1122 if (OFFSET_OOB(nbytes, offset, 2))
1125 switch (m->in_op & FILE_OPS_MASK) {
1127 offset = (short)((p->hs[0]<<8)|
1132 offset = (short)((p->hs[0]<<8)|
1137 offset = (short)((p->hs[0]<<8)|
1142 offset = (short)((p->hs[0]<<8)|
1147 offset = (short)((p->hs[0]<<8)|
1151 case FILE_OPMULTIPLY:
1152 offset = (short)((p->hs[0]<<8)|
1157 offset = (short)((p->hs[0]<<8)|
1162 offset = (short)((p->hs[0]<<8)|
1168 offset = (short)((p->hs[0]<<8)|
1170 if (m->in_op & FILE_OPINVERSE)
1174 if (OFFSET_OOB(nbytes, offset, 2))
1177 switch (m->in_op & FILE_OPS_MASK) {
1179 offset = (short)((p->hs[1]<<8)|
1184 offset = (short)((p->hs[1]<<8)|
1189 offset = (short)((p->hs[1]<<8)|
1194 offset = (short)((p->hs[1]<<8)|
1199 offset = (short)((p->hs[1]<<8)|
1203 case FILE_OPMULTIPLY:
1204 offset = (short)((p->hs[1]<<8)|
1209 offset = (short)((p->hs[1]<<8)|
1214 offset = (short)((p->hs[1]<<8)|
1220 offset = (short)((p->hs[1]<<8)|
1222 if (m->in_op & FILE_OPINVERSE)
1226 if (OFFSET_OOB(nbytes, offset, 2))
1229 switch (m->in_op & FILE_OPS_MASK) {
1231 offset = p->h & off;
1234 offset = p->h | off;
1237 offset = p->h ^ off;
1240 offset = p->h + off;
1243 offset = p->h - off;
1245 case FILE_OPMULTIPLY:
1246 offset = p->h * off;
1249 offset = p->h / off;
1252 offset = p->h % off;
1258 if (m->in_op & FILE_OPINVERSE)
1263 if (OFFSET_OOB(nbytes, offset, 4))
1266 switch (m->in_op & FILE_OPS_MASK) {
1268 offset = (int32_t)((p->hl[0]<<24)|
1275 offset = (int32_t)((p->hl[0]<<24)|
1282 offset = (int32_t)((p->hl[0]<<24)|
1289 offset = (int32_t)((p->hl[0]<<24)|
1296 offset = (int32_t)((p->hl[0]<<24)|
1302 case FILE_OPMULTIPLY:
1303 offset = (int32_t)((p->hl[0]<<24)|
1310 offset = (int32_t)((p->hl[0]<<24)|
1317 offset = (int32_t)((p->hl[0]<<24)|
1325 offset = (int32_t)((p->hl[0]<<24)|
1329 if (m->in_op & FILE_OPINVERSE)
1334 if (OFFSET_OOB(nbytes, offset, 4))
1337 switch (m->in_op & FILE_OPS_MASK) {
1339 offset = (int32_t)((p->hl[3]<<24)|
1346 offset = (int32_t)((p->hl[3]<<24)|
1353 offset = (int32_t)((p->hl[3]<<24)|
1360 offset = (int32_t)((p->hl[3]<<24)|
1367 offset = (int32_t)((p->hl[3]<<24)|
1373 case FILE_OPMULTIPLY:
1374 offset = (int32_t)((p->hl[3]<<24)|
1381 offset = (int32_t)((p->hl[3]<<24)|
1388 offset = (int32_t)((p->hl[3]<<24)|
1396 offset = (int32_t)((p->hl[3]<<24)|
1400 if (m->in_op & FILE_OPINVERSE)
1404 if (OFFSET_OOB(nbytes, offset, 4))
1407 switch (m->in_op & FILE_OPS_MASK) {
1409 offset = (int32_t)((p->hl[1]<<24)|
1416 offset = (int32_t)((p->hl[1]<<24)|
1423 offset = (int32_t)((p->hl[1]<<24)|
1430 offset = (int32_t)((p->hl[1]<<24)|
1437 offset = (int32_t)((p->hl[1]<<24)|
1443 case FILE_OPMULTIPLY:
1444 offset = (int32_t)((p->hl[1]<<24)|
1451 offset = (int32_t)((p->hl[1]<<24)|
1458 offset = (int32_t)((p->hl[1]<<24)|
1466 offset = (int32_t)((p->hl[1]<<24)|
1470 if (m->in_op & FILE_OPINVERSE)
1474 if (OFFSET_OOB(nbytes, offset, 4))
1477 switch (m->in_op & FILE_OPS_MASK) {
1479 offset = p->l & off;
1482 offset = p->l | off;
1485 offset = p->l ^ off;
1488 offset = p->l + off;
1491 offset = p->l - off;
1493 case FILE_OPMULTIPLY:
1494 offset = p->l * off;
1497 offset = p->l / off;
1500 offset = p->l % off;
1505 if (m->in_op & FILE_OPINVERSE)
1510 switch (m->in_type) {
1513 offset = ((((offset >> 0) & 0x7f) << 0) |
1514 (((offset >> 8) & 0x7f) << 7) |
1515 (((offset >> 16) & 0x7f) << 14) |
1516 (((offset >> 24) & 0x7f) << 21)) + 10;
1522 if (m->flag & INDIROFFADD) {
1523 offset += ms->c.li[cont_level-1].off;
1525 if (mcopy(ms, p, m->type, 0, s, offset, nbytes, count) == -1)
1527 ms->offset = offset;
1529 if ((ms->flags & MAGIC_DEBUG) != 0) {
1530 mdebug(offset, (char *)(void *)p,
1531 sizeof(union VALUETYPE));
1532 #ifndef COMPILE_ONLY
1538 /* Verify we have enough data to match magic type */
1541 if (OFFSET_OOB(nbytes, offset, 1))
1548 if (OFFSET_OOB(nbytes, offset, 2))
1567 if (OFFSET_OOB(nbytes, offset, 4))
1574 if (OFFSET_OOB(nbytes, offset, 8))
1581 if (OFFSET_OOB(nbytes, offset, m->vallen))
1586 if (OFFSET_OOB(nbytes, offset, 0))
1593 if ((ms->flags & (MAGIC_MIME|MAGIC_APPLE)) == 0 &&
1594 file_printf(ms, m->desc) == -1)
1596 if (OFFSET_OOB(nbytes, offset, 0))
1598 return file_softmagic(ms, s + offset, nbytes - offset,
1599 recursion_level, BINTEST);
1601 case FILE_DEFAULT: /* nothing to check */
1605 if (!mconvert(ms, m))
1611 file_strncmp(const char *s1, const char *s2, size_t len, uint32_t flags)
1614 * Convert the source args to unsigned here so that (1) the
1615 * compare will be unsigned as it is in strncmp() and (2) so
1616 * the ctype functions will work correctly without extra
1619 const unsigned char *a = (const unsigned char *)s1;
1620 const unsigned char *b = (const unsigned char *)s2;
1624 * What we want here is v = strncmp(s1, s2, len),
1625 * but ignoring any nulls.
1628 if (0L == flags) { /* normal string: do it fast */
1630 if ((v = *b++ - *a++) != '\0')
1633 else { /* combine the others */
1635 if ((flags & STRING_IGNORE_LOWERCASE) &&
1637 if ((v = tolower(*b++) - *a++) != '\0')
1640 else if ((flags & STRING_IGNORE_UPPERCASE) &&
1642 if ((v = toupper(*b++) - *a++) != '\0')
1645 else if ((flags & STRING_COMPACT_BLANK) &&
1648 if (isspace(*b++)) {
1657 else if ((flags & STRING_COMPACT_OPTIONAL_BLANK) &&
1664 if ((v = *b++ - *a++) != '\0')
1673 file_strncmp16(const char *a, const char *b, size_t len, uint32_t flags)
1676 * XXX - The 16-bit string compare probably needs to be done
1677 * differently, especially if the flags are to be supported.
1678 * At the moment, I am unsure.
1681 return file_strncmp(a, b, len, flags);
1685 magiccheck(struct magic_set *ms, struct magic *m)
1687 uint64_t l = m->value.q;
1692 union VALUETYPE *p = &ms->ms_value;
1760 file_magerror(ms, "cannot happen with float: invalid relation `%c'",
1794 file_magerror(ms, "cannot happen with double: invalid relation `%c'", m->reln);
1807 v = file_strncmp(m->value.s, p->s, (size_t)m->vallen, m->str_flags);
1810 case FILE_BESTRING16:
1811 case FILE_LESTRING16:
1813 v = file_strncmp16(m->value.s, p->s, (size_t)m->vallen, m->str_flags);
1816 case FILE_SEARCH: { /* search ms->search.s for the string m->value.s */
1820 if (ms->search.s == NULL)
1823 slen = MIN(m->vallen, sizeof(m->value.s));
1827 for (idx = 0; m->str_range == 0 || idx < m->str_range; idx++) {
1828 if (slen + idx > ms->search.s_len)
1831 v = file_strncmp(m->value.s, ms->search.s + idx, slen, m->str_flags);
1832 if (v == 0) { /* found match */
1833 ms->search.offset += idx;
1844 if (ms->search.s == NULL)
1848 rc = regcomp(&rx, m->value.s,
1849 REG_EXTENDED|REG_NEWLINE|
1850 ((m->str_flags & STRING_IGNORE_CASE) ? REG_ICASE : 0));
1852 (void)regerror(rc, &rx, errmsg, sizeof(errmsg));
1853 file_magerror(ms, "regex error %d, (%s)",
1858 regmatch_t pmatch[1];
1859 #ifndef REG_STARTEND
1860 #define REG_STARTEND 0
1861 size_t l = ms->search.s_len - 1;
1862 char c = ms->search.s[l];
1863 ((char *)(intptr_t)ms->search.s)[l] = '\0';
1865 pmatch[0].rm_so = 0;
1866 pmatch[0].rm_eo = ms->search.s_len;
1868 rc = regexec(&rx, (const char *)ms->search.s,
1869 1, pmatch, REG_STARTEND);
1870 #if REG_STARTEND == 0
1871 ((char *)(intptr_t)ms->search.s)[l] = c;
1875 ms->search.s += (int)pmatch[0].rm_so;
1876 ms->search.offset += (size_t)pmatch[0].rm_so;
1878 (size_t)(pmatch[0].rm_eo - pmatch[0].rm_so);
1887 (void)regerror(rc, &rx, errmsg, sizeof(errmsg));
1888 file_magerror(ms, "regexec error %d, (%s)",
1895 if (v == (uint64_t)-1)
1902 file_magerror(ms, "invalid type %d in magiccheck()", m->type);
1906 v = file_signextend(ms, m, v);
1910 if ((ms->flags & MAGIC_DEBUG) != 0)
1911 (void) fprintf(stderr, "%llu == *any* = 1\n",
1912 (unsigned long long)v);
1918 if ((ms->flags & MAGIC_DEBUG) != 0)
1919 (void) fprintf(stderr, "%llu != %llu = %d\n",
1920 (unsigned long long)v, (unsigned long long)l,
1926 if ((ms->flags & MAGIC_DEBUG) != 0)
1927 (void) fprintf(stderr, "%llu == %llu = %d\n",
1928 (unsigned long long)v, (unsigned long long)l,
1933 if (m->flag & UNSIGNED) {
1935 if ((ms->flags & MAGIC_DEBUG) != 0)
1936 (void) fprintf(stderr, "%llu > %llu = %d\n",
1937 (unsigned long long)v,
1938 (unsigned long long)l, matched);
1941 matched = (int64_t) v > (int64_t) l;
1942 if ((ms->flags & MAGIC_DEBUG) != 0)
1943 (void) fprintf(stderr, "%lld > %lld = %d\n",
1944 (long long)v, (long long)l, matched);
1949 if (m->flag & UNSIGNED) {
1951 if ((ms->flags & MAGIC_DEBUG) != 0)
1952 (void) fprintf(stderr, "%llu < %llu = %d\n",
1953 (unsigned long long)v,
1954 (unsigned long long)l, matched);
1957 matched = (int64_t) v < (int64_t) l;
1958 if ((ms->flags & MAGIC_DEBUG) != 0)
1959 (void) fprintf(stderr, "%lld < %lld = %d\n",
1960 (long long)v, (long long)l, matched);
1965 matched = (v & l) == l;
1966 if ((ms->flags & MAGIC_DEBUG) != 0)
1967 (void) fprintf(stderr, "((%llx & %llx) == %llx) = %d\n",
1968 (unsigned long long)v, (unsigned long long)l,
1969 (unsigned long long)l, matched);
1973 matched = (v & l) != l;
1974 if ((ms->flags & MAGIC_DEBUG) != 0)
1975 (void) fprintf(stderr, "((%llx & %llx) != %llx) = %d\n",
1976 (unsigned long long)v, (unsigned long long)l,
1977 (unsigned long long)l, matched);
1982 file_magerror(ms, "cannot happen: invalid relation `%c'",
1991 handle_annotation(struct magic_set *ms, struct magic *m)
1993 if (ms->flags & MAGIC_APPLE) {
1994 if (file_printf(ms, "%.8s", m->apple) == -1)
1998 if ((ms->flags & MAGIC_MIME_TYPE) && m->mimetype[0]) {
1999 if (file_printf(ms, "%s", m->mimetype) == -1)
2007 print_sep(struct magic_set *ms, int firstline)
2009 if (ms->flags & MAGIC_MIME)
2014 * we found another match
2015 * put a newline and '-' to do some simple formatting
2017 return file_printf(ms, "\n- ");