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 * apprentice - make one pass through /etc/magic, learning its secrets.
30 */
31
32 #include "file.h"
33
34 #ifndef lint
35 FILE_RCSID("@(#)$File: apprentice.c,v 1.356 2024/11/27 15:37:00 christos Exp $")
36 #endif /* lint */
37
38 #include "magic.h"
39 #include <stdlib.h>
40 #ifdef HAVE_UNISTD_H
41 #include <unistd.h>
42 #endif
43 #include <stddef.h>
44 #include <string.h>
45 #include <assert.h>
46 #include <ctype.h>
47 #include <fcntl.h>
48 #ifdef QUICK
49 #include <sys/mman.h>
50 #endif
51 #include <dirent.h>
52 #include <limits.h>
53 #ifdef HAVE_BYTESWAP_H
54 #include <byteswap.h>
55 #endif
56 #ifdef HAVE_SYS_BSWAP_H
57 #include <sys/bswap.h>
58 #endif
59
60
61 #define EATAB {while (isascii(CAST(unsigned char, *l)) && \
62 isspace(CAST(unsigned char, *l))) ++l;}
63 #define LOWCASE(l) (isupper(CAST(unsigned char, l)) ? \
64 tolower(CAST(unsigned char, l)) : (l))
65 /*
66 * Work around a bug in headers on Digital Unix.
67 * At least confirmed for: OSF1 V4.0 878
68 */
69 #if defined(__osf__) && defined(__DECC)
70 #ifdef MAP_FAILED
71 #undef MAP_FAILED
72 #endif
73 #endif
74
75 #ifndef MAP_FAILED
76 #define MAP_FAILED (void *) -1
77 #endif
78
79 #ifndef MAP_FILE
80 #define MAP_FILE 0
81 #endif
82
83 #define ALLOC_CHUNK CAST(size_t, 10)
84 #define ALLOC_INCR CAST(size_t, 200)
85
86 #define MAP_TYPE_USER 0
87 #define MAP_TYPE_MALLOC 1
88 #define MAP_TYPE_MMAP 2
89
90 struct magic_entry {
91 struct magic *mp;
92 uint32_t cont_count;
93 uint32_t max_count;
94 };
95
96 struct magic_entry_set {
97 struct magic_entry *me;
98 uint32_t count;
99 uint32_t max;
100 };
101
102 struct magic_map {
103 void *p;
104 size_t len;
105 int type;
106 struct magic *magic[MAGIC_SETS];
107 uint32_t nmagic[MAGIC_SETS];
108 };
109
110 file_private int file_formats[FILE_NAMES_SIZE];
111 file_private const size_t file_nformats = FILE_NAMES_SIZE;
112 file_protected const char *file_names[FILE_NAMES_SIZE];
113 file_protected const size_t file_nnames = FILE_NAMES_SIZE;
114
115 file_private int getvalue(struct magic_set *ms, struct magic *, const char **, int);
116 file_private int hextoint(int);
117 file_private const char *getstr(struct magic_set *, struct magic *, const char *,
118 int);
119 file_private int parse(struct magic_set *, struct magic_entry *, const char *,
120 const char *, size_t, int);
121 file_private void eatsize(const char **);
122 file_private int apprentice_1(struct magic_set *, const char *, int);
123 file_private ssize_t apprentice_magic_strength_1(const struct magic *);
124 file_private int apprentice_sort(const void *, const void *);
125 file_private void apprentice_list(struct mlist *, int );
126 file_private struct magic_map *apprentice_load(struct magic_set *,
127 const char *, int);
128 file_private struct mlist *mlist_alloc(void);
129 file_private void mlist_free_all(struct magic_set *);
130 file_private void mlist_free(struct mlist *);
131 file_private void byteswap(struct magic *, uint32_t);
132 file_private void bs1(struct magic *);
133
134 #if defined(HAVE_BYTESWAP_H)
135 #define swap2(x) bswap_16(x)
136 #define swap4(x) bswap_32(x)
137 #define swap8(x) bswap_64(x)
138 #elif defined(HAVE_SYS_BSWAP_H)
139 #define swap2(x) bswap16(x)
140 #define swap4(x) bswap32(x)
141 #define swap8(x) bswap64(x)
142 #else
143 file_private uint16_t swap2(uint16_t);
144 file_private uint32_t swap4(uint32_t);
145 file_private uint64_t swap8(uint64_t);
146 #endif
147
148 file_private char *mkdbname(struct magic_set *, const char *, int);
149 file_private struct magic_map *apprentice_buf(struct magic_set *, struct magic *,
150 size_t);
151 file_private struct magic_map *apprentice_map(struct magic_set *, const char *);
152 file_private int check_buffer(struct magic_set *, struct magic_map *, const char *);
153 file_private void apprentice_unmap(struct magic_map *);
154 file_private int apprentice_compile(struct magic_set *, struct magic_map *,
155 const char *);
156 file_private int check_format_type(const char *, int, const char **);
157 file_private int check_format(struct magic_set *, struct magic *);
158 file_private int get_op(char);
159 file_private int parse_mime(struct magic_set *, struct magic_entry *, const char *,
160 size_t);
161 file_private int parse_strength(struct magic_set *, struct magic_entry *,
162 const char *, size_t);
163 file_private int parse_apple(struct magic_set *, struct magic_entry *, const char *,
164 size_t);
165 file_private int parse_ext(struct magic_set *, struct magic_entry *, const char *,
166 size_t);
167
168
169 file_private size_t magicsize = sizeof(struct magic);
170
171 file_private const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc";
172
173 file_private struct {
174 const char *name;
175 size_t len;
176 int (*fun)(struct magic_set *, struct magic_entry *, const char *,
177 size_t);
178 } bang[] = {
179 #define DECLARE_FIELD(name) { # name, sizeof(# name) - 1, parse_ ## name }
180 DECLARE_FIELD(mime),
181 DECLARE_FIELD(apple),
182 DECLARE_FIELD(ext),
183 DECLARE_FIELD(strength),
184 #undef DECLARE_FIELD
185 { NULL, 0, NULL }
186 };
187
188 #ifdef COMPILE_ONLY
189
190 int main(int, char *[]);
191
192 int
main(int argc,char * argv[])193 main(int argc, char *argv[])
194 {
195 int ret;
196 struct magic_set *ms;
197 char *progname;
198
199 if ((progname = strrchr(argv[0], '/')) != NULL)
200 progname++;
201 else
202 progname = argv[0];
203
204 if (argc != 2) {
205 (void)fprintf(stderr, "Usage: %s file\n", progname);
206 return 1;
207 }
208
209 if ((ms = magic_open(MAGIC_CHECK)) == NULL) {
210 (void)fprintf(stderr, "%s: %s\n", progname, strerror(errno));
211 return 1;
212 }
213 ret = magic_compile(ms, argv[1]) == -1 ? 1 : 0;
214 if (ret == 1)
215 (void)fprintf(stderr, "%s: %s\n", progname, magic_error(ms));
216 magic_close(ms);
217 return ret;
218 }
219 #endif /* COMPILE_ONLY */
220
221 struct type_tbl_s {
222 const char name[16];
223 const size_t len;
224 const int type;
225 const int format;
226 };
227
228 /*
229 * XXX - the actual Single UNIX Specification says that "long" means "long",
230 * as in the C data type, but we treat it as meaning "4-byte integer".
231 * Given that the OS X version of file 5.04 did the same, I guess that passes
232 * the actual test; having "long" be dependent on how big a "long" is on
233 * the machine running "file" is silly.
234 */
235 static const struct type_tbl_s type_tbl[] = {
236 # define XX(s) s, (sizeof(s) - 1)
237 # define XX_NULL "", 0
238 { XX("invalid"), FILE_INVALID, FILE_FMT_NONE },
239 { XX("byte"), FILE_BYTE, FILE_FMT_NUM },
240 { XX("short"), FILE_SHORT, FILE_FMT_NUM },
241 { XX("default"), FILE_DEFAULT, FILE_FMT_NONE },
242 { XX("long"), FILE_LONG, FILE_FMT_NUM },
243 { XX("string"), FILE_STRING, FILE_FMT_STR },
244 { XX("date"), FILE_DATE, FILE_FMT_STR },
245 { XX("beshort"), FILE_BESHORT, FILE_FMT_NUM },
246 { XX("belong"), FILE_BELONG, FILE_FMT_NUM },
247 { XX("bedate"), FILE_BEDATE, FILE_FMT_STR },
248 { XX("leshort"), FILE_LESHORT, FILE_FMT_NUM },
249 { XX("lelong"), FILE_LELONG, FILE_FMT_NUM },
250 { XX("ledate"), FILE_LEDATE, FILE_FMT_STR },
251 { XX("pstring"), FILE_PSTRING, FILE_FMT_STR },
252 { XX("ldate"), FILE_LDATE, FILE_FMT_STR },
253 { XX("beldate"), FILE_BELDATE, FILE_FMT_STR },
254 { XX("leldate"), FILE_LELDATE, FILE_FMT_STR },
255 { XX("regex"), FILE_REGEX, FILE_FMT_STR },
256 { XX("bestring16"), FILE_BESTRING16, FILE_FMT_STR },
257 { XX("lestring16"), FILE_LESTRING16, FILE_FMT_STR },
258 { XX("search"), FILE_SEARCH, FILE_FMT_STR },
259 { XX("medate"), FILE_MEDATE, FILE_FMT_STR },
260 { XX("meldate"), FILE_MELDATE, FILE_FMT_STR },
261 { XX("melong"), FILE_MELONG, FILE_FMT_NUM },
262 { XX("quad"), FILE_QUAD, FILE_FMT_QUAD },
263 { XX("lequad"), FILE_LEQUAD, FILE_FMT_QUAD },
264 { XX("bequad"), FILE_BEQUAD, FILE_FMT_QUAD },
265 { XX("qdate"), FILE_QDATE, FILE_FMT_STR },
266 { XX("leqdate"), FILE_LEQDATE, FILE_FMT_STR },
267 { XX("beqdate"), FILE_BEQDATE, FILE_FMT_STR },
268 { XX("qldate"), FILE_QLDATE, FILE_FMT_STR },
269 { XX("leqldate"), FILE_LEQLDATE, FILE_FMT_STR },
270 { XX("beqldate"), FILE_BEQLDATE, FILE_FMT_STR },
271 { XX("float"), FILE_FLOAT, FILE_FMT_FLOAT },
272 { XX("befloat"), FILE_BEFLOAT, FILE_FMT_FLOAT },
273 { XX("lefloat"), FILE_LEFLOAT, FILE_FMT_FLOAT },
274 { XX("double"), FILE_DOUBLE, FILE_FMT_DOUBLE },
275 { XX("bedouble"), FILE_BEDOUBLE, FILE_FMT_DOUBLE },
276 { XX("ledouble"), FILE_LEDOUBLE, FILE_FMT_DOUBLE },
277 { XX("leid3"), FILE_LEID3, FILE_FMT_NUM },
278 { XX("beid3"), FILE_BEID3, FILE_FMT_NUM },
279 { XX("indirect"), FILE_INDIRECT, FILE_FMT_NUM },
280 { XX("qwdate"), FILE_QWDATE, FILE_FMT_STR },
281 { XX("leqwdate"), FILE_LEQWDATE, FILE_FMT_STR },
282 { XX("beqwdate"), FILE_BEQWDATE, FILE_FMT_STR },
283 { XX("name"), FILE_NAME, FILE_FMT_NONE },
284 { XX("use"), FILE_USE, FILE_FMT_NONE },
285 { XX("clear"), FILE_CLEAR, FILE_FMT_NONE },
286 { XX("der"), FILE_DER, FILE_FMT_STR },
287 { XX("guid"), FILE_GUID, FILE_FMT_STR },
288 { XX("offset"), FILE_OFFSET, FILE_FMT_QUAD },
289 { XX("bevarint"), FILE_BEVARINT, FILE_FMT_STR },
290 { XX("levarint"), FILE_LEVARINT, FILE_FMT_STR },
291 { XX("msdosdate"), FILE_MSDOSDATE, FILE_FMT_STR },
292 { XX("lemsdosdate"), FILE_LEMSDOSDATE, FILE_FMT_STR },
293 { XX("bemsdosdate"), FILE_BEMSDOSDATE, FILE_FMT_STR },
294 { XX("msdostime"), FILE_MSDOSTIME, FILE_FMT_STR },
295 { XX("lemsdostime"), FILE_LEMSDOSTIME, FILE_FMT_STR },
296 { XX("bemsdostime"), FILE_BEMSDOSTIME, FILE_FMT_STR },
297 { XX("octal"), FILE_OCTAL, FILE_FMT_STR },
298 { XX_NULL, FILE_INVALID, FILE_FMT_NONE },
299 };
300
301 /*
302 * These are not types, and cannot be preceded by "u" to make them
303 * unsigned.
304 */
305 static const struct type_tbl_s special_tbl[] = {
306 { XX("der"), FILE_DER, FILE_FMT_STR },
307 { XX("name"), FILE_NAME, FILE_FMT_STR },
308 { XX("use"), FILE_USE, FILE_FMT_STR },
309 { XX("octal"), FILE_OCTAL, FILE_FMT_STR },
310 { XX_NULL, FILE_INVALID, FILE_FMT_NONE },
311 };
312 # undef XX
313 # undef XX_NULL
314
315 file_private int
get_type(const struct type_tbl_s * tbl,const char * l,const char ** t)316 get_type(const struct type_tbl_s *tbl, const char *l, const char **t)
317 {
318 const struct type_tbl_s *p;
319
320 for (p = tbl; p->len; p++) {
321 if (strncmp(l, p->name, p->len) == 0) {
322 if (t)
323 *t = l + p->len;
324 break;
325 }
326 }
327 return p->type;
328 }
329
330 file_private off_t
maxoff_t(void)331 maxoff_t(void) {
332 if (/*CONSTCOND*/sizeof(off_t) == sizeof(int))
333 return CAST(off_t, INT_MAX);
334 if (/*CONSTCOND*/sizeof(off_t) == sizeof(long))
335 return CAST(off_t, LONG_MAX);
336 return 0x7fffffff;
337 }
338
339 file_private int
get_standard_integer_type(const char * l,const char ** t)340 get_standard_integer_type(const char *l, const char **t)
341 {
342 int type;
343
344 if (isalpha(CAST(unsigned char, l[1]))) {
345 switch (l[1]) {
346 case 'C':
347 /* "dC" and "uC" */
348 type = FILE_BYTE;
349 break;
350 case 'S':
351 /* "dS" and "uS" */
352 type = FILE_SHORT;
353 break;
354 case 'I':
355 case 'L':
356 /*
357 * "dI", "dL", "uI", and "uL".
358 *
359 * XXX - the actual Single UNIX Specification says
360 * that "L" means "long", as in the C data type,
361 * but we treat it as meaning "4-byte integer".
362 * Given that the OS X version of file 5.04 did
363 * the same, I guess that passes the actual SUS
364 * validation suite; having "dL" be dependent on
365 * how big a "long" is on the machine running
366 * "file" is silly.
367 */
368 type = FILE_LONG;
369 break;
370 case 'Q':
371 /* "dQ" and "uQ" */
372 type = FILE_QUAD;
373 break;
374 default:
375 /* "d{anything else}", "u{anything else}" */
376 return FILE_INVALID;
377 }
378 l += 2;
379 } else if (isdigit(CAST(unsigned char, l[1]))) {
380 /*
381 * "d{num}" and "u{num}"; we only support {num} values
382 * of 1, 2, 4, and 8 - the Single UNIX Specification
383 * doesn't say anything about whether arbitrary
384 * values should be supported, but both the Solaris 10
385 * and OS X Mountain Lion versions of file passed the
386 * Single UNIX Specification validation suite, and
387 * neither of them support values bigger than 8 or
388 * non-power-of-2 values.
389 */
390 if (isdigit(CAST(unsigned char, l[2]))) {
391 /* Multi-digit, so > 9 */
392 return FILE_INVALID;
393 }
394 switch (l[1]) {
395 case '1':
396 type = FILE_BYTE;
397 break;
398 case '2':
399 type = FILE_SHORT;
400 break;
401 case '4':
402 type = FILE_LONG;
403 break;
404 case '8':
405 type = FILE_QUAD;
406 break;
407 default:
408 /* XXX - what about 3, 5, 6, or 7? */
409 return FILE_INVALID;
410 }
411 l += 2;
412 } else {
413 /*
414 * "d" or "u" by itself.
415 */
416 type = FILE_LONG;
417 ++l;
418 }
419 if (t)
420 *t = l;
421 return type;
422 }
423
424 file_private void
init_file_tables(void)425 init_file_tables(void)
426 {
427 static int done = 0;
428 const struct type_tbl_s *p;
429
430 if (done)
431 return;
432 done++;
433
434 for (p = type_tbl; p->len; p++) {
435 assert(p->type < FILE_NAMES_SIZE);
436 file_names[p->type] = p->name;
437 file_formats[p->type] = p->format;
438 }
439 assert(p - type_tbl == FILE_NAMES_SIZE);
440 }
441
442 file_private int
add_mlist(struct mlist * mlp,struct magic_map * map,size_t idx)443 add_mlist(struct mlist *mlp, struct magic_map *map, size_t idx)
444 {
445 struct mlist *ml;
446
447 mlp->map = NULL;
448 if ((ml = CAST(struct mlist *, malloc(sizeof(*ml)))) == NULL)
449 return -1;
450
451 ml->map = idx == 0 ? map : NULL;
452 ml->magic = map->magic[idx];
453 ml->nmagic = map->nmagic[idx];
454 if (ml->nmagic) {
455 ml->magic_rxcomp = CAST(file_regex_t **,
456 calloc(ml->nmagic, sizeof(*ml->magic_rxcomp)));
457 if (ml->magic_rxcomp == NULL) {
458 free(ml);
459 return -1;
460 }
461 } else
462 ml->magic_rxcomp = NULL;
463 mlp->prev->next = ml;
464 ml->prev = mlp->prev;
465 ml->next = mlp;
466 mlp->prev = ml;
467 return 0;
468 }
469
470 /*
471 * Handle one file or directory.
472 */
473 file_private int
apprentice_1(struct magic_set * ms,const char * fn,int action)474 apprentice_1(struct magic_set *ms, const char *fn, int action)
475 {
476 struct magic_map *map;
477 #ifndef COMPILE_ONLY
478 size_t i;
479 #endif
480
481 if (magicsize != FILE_MAGICSIZE) {
482 file_error(ms, 0, "magic element size %lu != %lu",
483 CAST(unsigned long, sizeof(*map->magic[0])),
484 CAST(unsigned long, FILE_MAGICSIZE));
485 return -1;
486 }
487
488 if (action == FILE_COMPILE) {
489 map = apprentice_load(ms, fn, action);
490 if (map == NULL)
491 return -1;
492 return apprentice_compile(ms, map, fn);
493 }
494
495 #ifndef COMPILE_ONLY
496 map = apprentice_map(ms, fn);
497 if (map == NULL) {
498 if (ms->flags & MAGIC_CHECK)
499 file_magwarn(ms, "using regular magic file `%s'", fn);
500 map = apprentice_load(ms, fn, action);
501 if (map == NULL)
502 return -1;
503 }
504
505 for (i = 0; i < MAGIC_SETS; i++) {
506 if (add_mlist(ms->mlist[i], map, i) == -1) {
507 /* failed to add to any list, free explicitly */
508 if (i == 0)
509 apprentice_unmap(map);
510 else
511 mlist_free_all(ms);
512 file_oomem(ms, sizeof(*ms->mlist[0]));
513 return -1;
514 }
515 }
516
517 if (action == FILE_LIST) {
518 for (i = 0; i < MAGIC_SETS; i++) {
519 printf("Set %" SIZE_T_FORMAT "u:\nBinary patterns:\n",
520 i);
521 apprentice_list(ms->mlist[i], BINTEST);
522 printf("Text patterns:\n");
523 apprentice_list(ms->mlist[i], TEXTTEST);
524 }
525 }
526 return 0;
527 #else
528 return 0;
529 #endif /* COMPILE_ONLY */
530 }
531
532 file_protected void
file_ms_free(struct magic_set * ms)533 file_ms_free(struct magic_set *ms)
534 {
535 size_t i;
536 if (ms == NULL)
537 return;
538 for (i = 0; i < MAGIC_SETS; i++)
539 mlist_free(ms->mlist[i]);
540 free(ms->o.pbuf);
541 free(ms->o.buf);
542 free(ms->c.li);
543 free(ms->fnamebuf);
544 #ifdef USE_C_LOCALE
545 freelocale(ms->c_lc_ctype);
546 #endif
547 free(ms);
548 }
549
550 file_protected struct magic_set *
file_ms_alloc(int flags)551 file_ms_alloc(int flags)
552 {
553 struct magic_set *ms;
554 size_t i, len;
555
556 if ((ms = CAST(struct magic_set *, calloc(CAST(size_t, 1u),
557 sizeof(*ms)))) == NULL)
558 return NULL;
559
560 if (magic_setflags(ms, flags) == -1) {
561 errno = EINVAL;
562 goto free;
563 }
564
565 ms->o.buf = ms->o.pbuf = NULL;
566 ms->o.blen = 0;
567 len = (ms->c.len = 10) * sizeof(*ms->c.li);
568
569 if ((ms->c.li = CAST(struct level_info *, malloc(len))) == NULL)
570 goto free;
571
572 ms->event_flags = 0;
573 ms->error = -1;
574 for (i = 0; i < MAGIC_SETS; i++)
575 ms->mlist[i] = NULL;
576 ms->fnamebuf = NULL;
577 ms->file = "unknown";
578 ms->line = 0;
579 ms->magwarn = 0;
580 ms->indir_max = FILE_INDIR_MAX;
581 ms->name_max = FILE_NAME_MAX;
582 ms->elf_shnum_max = FILE_ELF_SHNUM_MAX;
583 ms->elf_shsize_max = FILE_ELF_SHSIZE_MAX;
584 ms->elf_phnum_max = FILE_ELF_PHNUM_MAX;
585 ms->elf_notes_max = FILE_ELF_NOTES_MAX;
586 ms->regex_max = FILE_REGEX_MAX;
587 ms->bytes_max = FILE_BYTES_MAX;
588 ms->encoding_max = FILE_ENCODING_MAX;
589 ms->magwarn_max = FILE_MAGWARN_MAX;
590 #ifdef USE_C_LOCALE
591 ms->c_lc_ctype = newlocale(LC_CTYPE_MASK, "C", 0);
592 assert(ms->c_lc_ctype != NULL);
593 #endif
594 return ms;
595 free:
596 free(ms);
597 return NULL;
598 }
599
600 file_private void
apprentice_unmap(struct magic_map * map)601 apprentice_unmap(struct magic_map *map)
602 {
603 size_t i;
604 char *p;
605 if (map == NULL)
606 return;
607
608 switch (map->type) {
609 case MAP_TYPE_USER:
610 break;
611 case MAP_TYPE_MALLOC:
612 p = CAST(char *, map->p);
613 for (i = 0; i < MAGIC_SETS; i++) {
614 char *b = RCAST(char *, map->magic[i]);
615 if (p != NULL && b >= p && b <= p + map->len)
616 continue;
617 free(b);
618 }
619 free(p);
620 break;
621 #ifdef QUICK
622 case MAP_TYPE_MMAP:
623 if (map->p && map->p != MAP_FAILED)
624 (void)munmap(map->p, map->len);
625 break;
626 #endif
627 default:
628 fprintf(stderr, "Bad map type %d", map->type);
629 abort();
630 }
631 free(map);
632 }
633
634 file_private struct mlist *
mlist_alloc(void)635 mlist_alloc(void)
636 {
637 struct mlist *mlist;
638 if ((mlist = CAST(struct mlist *, calloc(1, sizeof(*mlist)))) == NULL) {
639 return NULL;
640 }
641 mlist->next = mlist->prev = mlist;
642 return mlist;
643 }
644
645 file_private void
mlist_free_all(struct magic_set * ms)646 mlist_free_all(struct magic_set *ms)
647 {
648 size_t i;
649
650 for (i = 0; i < MAGIC_SETS; i++) {
651 mlist_free(ms->mlist[i]);
652 ms->mlist[i] = NULL;
653 }
654 }
655
656 file_private void
mlist_free_one(struct mlist * ml)657 mlist_free_one(struct mlist *ml)
658 {
659 size_t i;
660
661 if (ml->map)
662 apprentice_unmap(CAST(struct magic_map *, ml->map));
663
664 for (i = 0; i < ml->nmagic; ++i) {
665 if (ml->magic_rxcomp[i]) {
666 file_regfree(ml->magic_rxcomp[i]);
667 free(ml->magic_rxcomp[i]);
668 ml->magic_rxcomp[i] = NULL;
669 }
670 }
671 free(ml->magic_rxcomp);
672 ml->magic_rxcomp = NULL;
673 free(ml);
674 }
675
676 file_private void
mlist_free(struct mlist * mlist)677 mlist_free(struct mlist *mlist)
678 {
679 struct mlist *ml, *next;
680
681 if (mlist == NULL)
682 return;
683
684 for (ml = mlist->next; ml != mlist;) {
685 next = ml->next;
686 mlist_free_one(ml);
687 ml = next;
688 }
689 mlist_free_one(mlist);
690 }
691
692 #ifndef COMPILE_ONLY
693 /* void **bufs: an array of compiled magic files */
694 file_protected int
buffer_apprentice(struct magic_set * ms,struct magic ** bufs,size_t * sizes,size_t nbufs)695 buffer_apprentice(struct magic_set *ms, struct magic **bufs,
696 size_t *sizes, size_t nbufs)
697 {
698 size_t i, j;
699 struct magic_map *map;
700
701 if (nbufs == 0)
702 return -1;
703
704 (void)file_reset(ms, 0);
705
706 init_file_tables();
707
708 for (i = 0; i < MAGIC_SETS; i++) {
709 mlist_free(ms->mlist[i]);
710 if ((ms->mlist[i] = mlist_alloc()) == NULL) {
711 file_oomem(ms, sizeof(*ms->mlist[0]));
712 goto fail;
713 }
714 }
715
716 for (i = 0; i < nbufs; i++) {
717 map = apprentice_buf(ms, bufs[i], sizes[i]);
718 if (map == NULL)
719 goto fail;
720
721 for (j = 0; j < MAGIC_SETS; j++) {
722 if (add_mlist(ms->mlist[j], map, j) == -1) {
723 file_oomem(ms, sizeof(*ms->mlist[0]));
724 goto fail;
725 }
726 }
727 }
728
729 return 0;
730 fail:
731 mlist_free_all(ms);
732 return -1;
733 }
734 #endif
735
736 /* const char *fn: list of magic files and directories */
737 file_protected int
file_apprentice(struct magic_set * ms,const char * fn,int action)738 file_apprentice(struct magic_set *ms, const char *fn, int action)
739 {
740 char *p;
741 int fileerr, errs = -1;
742 size_t i, j;
743
744 (void)file_reset(ms, 0);
745
746 if ((fn = magic_getpath(fn, action)) == NULL)
747 return -1;
748
749 init_file_tables();
750
751 free(ms->fnamebuf);
752 if ((ms->fnamebuf = strdup(fn)) == NULL) {
753 file_oomem(ms, strlen(fn));
754 return -1;
755 }
756
757 for (i = 0; i < MAGIC_SETS; i++) {
758 mlist_free(ms->mlist[i]);
759 if ((ms->mlist[i] = mlist_alloc()) == NULL) {
760 file_oomem(ms, sizeof(*ms->mlist[0]));
761 for (j = 0; j < i; j++) {
762 mlist_free(ms->mlist[j]);
763 ms->mlist[j] = NULL;
764 }
765 return -1;
766 }
767 }
768 fn = ms->fnamebuf;
769
770 while (fn) {
771 p = CCAST(char *, strchr(fn, PATHSEP));
772 if (p)
773 *p++ = '\0';
774 if (*fn == '\0')
775 break;
776 fileerr = apprentice_1(ms, fn, action);
777 errs = MAX(errs, fileerr);
778 fn = p;
779 }
780
781 if (errs == -1) {
782 for (i = 0; i < MAGIC_SETS; i++) {
783 mlist_free(ms->mlist[i]);
784 ms->mlist[i] = NULL;
785 }
786 file_error(ms, 0, "could not find any valid magic files!");
787 return -1;
788 }
789
790 #if 0
791 /*
792 * Always leave the database loaded
793 */
794 if (action == FILE_LOAD)
795 return 0;
796
797 for (i = 0; i < MAGIC_SETS; i++) {
798 mlist_free(ms->mlist[i]);
799 ms->mlist[i] = NULL;
800 }
801 #endif
802
803 switch (action) {
804 case FILE_LOAD:
805 case FILE_COMPILE:
806 case FILE_CHECK:
807 case FILE_LIST:
808 return 0;
809 default:
810 file_error(ms, 0, "Invalid action %d", action);
811 return -1;
812 }
813 }
814
815 /*
816 * Compute the real length of a magic expression, for the purposes
817 * of determining how "strong" a magic expression is (approximating
818 * how specific its matches are):
819 * - magic characters count 0 unless escaped.
820 * - [] expressions count 1
821 * - {} expressions count 0
822 * - regular characters or escaped magic characters count 1
823 * - 0 length expressions count as one
824 */
825 file_private size_t
nonmagic(const char * str)826 nonmagic(const char *str)
827 {
828 const char *p;
829 size_t rv = 0;
830
831 for (p = str; *p; p++)
832 switch (*p) {
833 case '\\': /* Escaped anything counts 1 */
834 if (!*++p)
835 p--;
836 rv++;
837 continue;
838 case '?': /* Magic characters count 0 */
839 case '*':
840 case '.':
841 case '+':
842 case '^':
843 case '$':
844 continue;
845 case '[': /* Bracketed expressions count 1 the ']' */
846 while (*p && *p != ']')
847 p++;
848 p--;
849 continue;
850 case '{': /* Braced expressions count 0 */
851 while (*p && *p != '}')
852 p++;
853 if (!*p)
854 p--;
855 continue;
856 default: /* Anything else counts 1 */
857 rv++;
858 continue;
859 }
860
861 return rv == 0 ? 1 : rv; /* Return at least 1 */
862 }
863
864
865 file_private size_t
typesize(int type)866 typesize(int type)
867 {
868 switch (type) {
869 case FILE_BYTE:
870 return 1;
871
872 case FILE_SHORT:
873 case FILE_LESHORT:
874 case FILE_BESHORT:
875 case FILE_MSDOSDATE:
876 case FILE_BEMSDOSDATE:
877 case FILE_LEMSDOSDATE:
878 case FILE_MSDOSTIME:
879 case FILE_BEMSDOSTIME:
880 case FILE_LEMSDOSTIME:
881 return 2;
882
883 case FILE_LONG:
884 case FILE_LELONG:
885 case FILE_BELONG:
886 case FILE_MELONG:
887 return 4;
888
889 case FILE_DATE:
890 case FILE_LEDATE:
891 case FILE_BEDATE:
892 case FILE_MEDATE:
893 case FILE_LDATE:
894 case FILE_LELDATE:
895 case FILE_BELDATE:
896 case FILE_MELDATE:
897 case FILE_FLOAT:
898 case FILE_BEFLOAT:
899 case FILE_LEFLOAT:
900 case FILE_BEID3:
901 case FILE_LEID3:
902 return 4;
903
904 case FILE_QUAD:
905 case FILE_BEQUAD:
906 case FILE_LEQUAD:
907 case FILE_QDATE:
908 case FILE_LEQDATE:
909 case FILE_BEQDATE:
910 case FILE_QLDATE:
911 case FILE_LEQLDATE:
912 case FILE_BEQLDATE:
913 case FILE_QWDATE:
914 case FILE_LEQWDATE:
915 case FILE_BEQWDATE:
916 case FILE_DOUBLE:
917 case FILE_BEDOUBLE:
918 case FILE_LEDOUBLE:
919 case FILE_OFFSET:
920 case FILE_BEVARINT:
921 case FILE_LEVARINT:
922 return 8;
923
924 case FILE_GUID:
925 return 16;
926
927 default:
928 return FILE_BADSIZE;
929 }
930 }
931
932 /*
933 * Get weight of this magic entry, for sorting purposes.
934 */
935 file_private ssize_t
apprentice_magic_strength_1(const struct magic * m)936 apprentice_magic_strength_1(const struct magic *m)
937 {
938 #define MULT 10U
939 size_t ts, v;
940 ssize_t val = 2 * MULT; /* baseline strength */
941
942 switch (m->type) {
943 case FILE_DEFAULT: /* make sure this sorts last */
944 if (m->factor_op != FILE_FACTOR_OP_NONE) {
945 file_magwarn1("Unsupported factor_op in default %d",
946 m->factor_op);
947 }
948 return 0;
949
950 case FILE_BYTE:
951 case FILE_SHORT:
952 case FILE_LESHORT:
953 case FILE_BESHORT:
954 case FILE_LONG:
955 case FILE_LELONG:
956 case FILE_BELONG:
957 case FILE_MELONG:
958 case FILE_DATE:
959 case FILE_LEDATE:
960 case FILE_BEDATE:
961 case FILE_MEDATE:
962 case FILE_LDATE:
963 case FILE_LELDATE:
964 case FILE_BELDATE:
965 case FILE_MELDATE:
966 case FILE_FLOAT:
967 case FILE_BEFLOAT:
968 case FILE_LEFLOAT:
969 case FILE_QUAD:
970 case FILE_BEQUAD:
971 case FILE_LEQUAD:
972 case FILE_QDATE:
973 case FILE_LEQDATE:
974 case FILE_BEQDATE:
975 case FILE_QLDATE:
976 case FILE_LEQLDATE:
977 case FILE_BEQLDATE:
978 case FILE_QWDATE:
979 case FILE_LEQWDATE:
980 case FILE_BEQWDATE:
981 case FILE_DOUBLE:
982 case FILE_BEDOUBLE:
983 case FILE_LEDOUBLE:
984 case FILE_BEVARINT:
985 case FILE_LEVARINT:
986 case FILE_GUID:
987 case FILE_BEID3:
988 case FILE_LEID3:
989 case FILE_OFFSET:
990 case FILE_MSDOSDATE:
991 case FILE_BEMSDOSDATE:
992 case FILE_LEMSDOSDATE:
993 case FILE_MSDOSTIME:
994 case FILE_BEMSDOSTIME:
995 case FILE_LEMSDOSTIME:
996 ts = typesize(m->type);
997 if (ts == FILE_BADSIZE) {
998 (void)fprintf(stderr, "Bad size for type %d\n",
999 m->type);
1000 abort();
1001 }
1002 val += ts * MULT;
1003 break;
1004
1005 case FILE_PSTRING:
1006 case FILE_STRING:
1007 case FILE_OCTAL:
1008 val += m->vallen * MULT;
1009 break;
1010
1011 case FILE_BESTRING16:
1012 case FILE_LESTRING16:
1013 val += m->vallen * MULT / 2;
1014 break;
1015
1016 case FILE_SEARCH:
1017 if (m->vallen == 0)
1018 break;
1019 val += m->vallen * MAX(MULT / m->vallen, 1);
1020 break;
1021
1022 case FILE_REGEX:
1023 v = nonmagic(m->value.s);
1024 val += v * MAX(MULT / v, 1);
1025 break;
1026
1027 case FILE_INDIRECT:
1028 case FILE_NAME:
1029 case FILE_USE:
1030 case FILE_CLEAR:
1031 break;
1032
1033 case FILE_DER:
1034 val += MULT;
1035 break;
1036
1037 default:
1038 (void)fprintf(stderr, "Bad type %d\n", m->type);
1039 abort();
1040 }
1041
1042 switch (m->reln) {
1043 case 'x': /* matches anything penalize */
1044 case '!': /* matches almost anything penalize */
1045 val = 0;
1046 break;
1047
1048 case '=': /* Exact match, prefer */
1049 val += MULT;
1050 break;
1051
1052 case '>':
1053 case '<': /* comparison match reduce strength */
1054 val -= 2 * MULT;
1055 break;
1056
1057 case '^':
1058 case '&': /* masking bits, we could count them too */
1059 val -= MULT;
1060 break;
1061
1062 default:
1063 (void)fprintf(stderr, "Bad relation %c\n", m->reln);
1064 abort();
1065 }
1066
1067 return val;
1068 }
1069
1070
1071 /*ARGSUSED*/
1072 file_protected size_t
file_magic_strength(const struct magic * m,size_t nmagic)1073 file_magic_strength(const struct magic *m,
1074 size_t nmagic __attribute__((__unused__)))
1075 {
1076 ssize_t val = apprentice_magic_strength_1(m);
1077
1078 #ifdef notyet
1079 if (m->desc[0] == '\0') {
1080 size_t i;
1081 /*
1082 * Magic entries with no description get their continuations
1083 * added
1084 */
1085 for (i = 1; m[i].cont_level != 0 && i < MIN(nmagic, 3); i++) {
1086 ssize_t v = apprentice_magic_strength_1(&m[i]) >>
1087 (i + 1);
1088 val += v;
1089 if (m[i].desc[0] != '\0')
1090 break;
1091 }
1092 }
1093 #endif
1094
1095 switch (m->factor_op) {
1096 case FILE_FACTOR_OP_NONE:
1097 break;
1098 case FILE_FACTOR_OP_PLUS:
1099 val += m->factor;
1100 break;
1101 case FILE_FACTOR_OP_MINUS:
1102 val -= m->factor;
1103 break;
1104 case FILE_FACTOR_OP_TIMES:
1105 val *= m->factor;
1106 break;
1107 case FILE_FACTOR_OP_DIV:
1108 val /= m->factor;
1109 break;
1110 default:
1111 (void)fprintf(stderr, "Bad factor_op %u\n", m->factor_op);
1112 abort();
1113 }
1114
1115 if (val <= 0) /* ensure we only return 0 for FILE_DEFAULT */
1116 val = 1;
1117
1118 #ifndef notyet
1119 /*
1120 * Magic entries with no description get a bonus because they depend
1121 * on subsequent magic entries to print something.
1122 */
1123 if (m->desc[0] == '\0')
1124 val++;
1125 #endif
1126
1127 return val;
1128 }
1129
1130 /*
1131 * Sort callback for sorting entries by "strength" (basically length)
1132 */
1133 file_private int
apprentice_sort(const void * a,const void * b)1134 apprentice_sort(const void *a, const void *b)
1135 {
1136 const struct magic_entry *ma = CAST(const struct magic_entry *, a);
1137 const struct magic_entry *mb = CAST(const struct magic_entry *, b);
1138 size_t sa = file_magic_strength(ma->mp, ma->cont_count);
1139 size_t sb = file_magic_strength(mb->mp, mb->cont_count);
1140 if (sa == sb) {
1141 struct magic mpa = *ma->mp;
1142 struct magic mpb = *mb->mp;
1143 mpa.lineno = mpb.lineno = 0;
1144 int x = memcmp(&mpa, &mpb, sizeof(mpa));
1145 if (x == 0) {
1146 // Don't warn for DER
1147 if (mpa.type == FILE_DER)
1148 return 0;
1149 file_magwarn1("Duplicate magic entry `%s'",
1150 ma->mp->desc);
1151 #ifndef COMPILE_ONLY
1152 file_mdump(ma->mp);
1153 file_mdump(mb->mp);
1154 #endif
1155 return 0;
1156 }
1157 return x > 0 ? -1 : 1;
1158 }
1159 return sa > sb ? -1 : 1;
1160 }
1161
1162 /*
1163 * Shows sorted patterns list in the order which is used for the matching
1164 */
1165 file_private void
apprentice_list(struct mlist * mlist,int mode)1166 apprentice_list(struct mlist *mlist, int mode)
1167 {
1168 uint32_t magindex, descindex, mimeindex, lineindex;
1169 struct mlist *ml;
1170 for (ml = mlist->next; ml != mlist; ml = ml->next) {
1171 for (magindex = 0; magindex < ml->nmagic; magindex++) {
1172 struct magic *m = &ml->magic[magindex];
1173 if ((m->flag & mode) != mode) {
1174 /* Skip sub-tests */
1175 while (magindex + 1 < ml->nmagic &&
1176 ml->magic[magindex + 1].cont_level != 0)
1177 ++magindex;
1178 continue; /* Skip to next top-level test*/
1179 }
1180
1181 /*
1182 * Try to iterate over the tree until we find item with
1183 * description/mimetype.
1184 */
1185 lineindex = descindex = mimeindex = magindex;
1186 for (; magindex + 1 < ml->nmagic &&
1187 ml->magic[magindex + 1].cont_level != 0;
1188 magindex++) {
1189 uint32_t mi = magindex + 1;
1190 if (*ml->magic[descindex].desc == '\0'
1191 && *ml->magic[mi].desc)
1192 descindex = mi;
1193 if (*ml->magic[mimeindex].mimetype == '\0'
1194 && *ml->magic[mi].mimetype)
1195 mimeindex = mi;
1196 }
1197
1198 printf("Strength = %3" SIZE_T_FORMAT "u@%u: %s [%s]\n",
1199 file_magic_strength(m, ml->nmagic - magindex),
1200 ml->magic[lineindex].lineno,
1201 ml->magic[descindex].desc,
1202 ml->magic[mimeindex].mimetype);
1203 }
1204 }
1205 }
1206
1207 file_private void
set_test_type(struct magic * mstart,struct magic * m)1208 set_test_type(struct magic *mstart, struct magic *m)
1209 {
1210 switch (m->type) {
1211 case FILE_BYTE:
1212 case FILE_SHORT:
1213 case FILE_LONG:
1214 case FILE_DATE:
1215 case FILE_BESHORT:
1216 case FILE_BELONG:
1217 case FILE_BEDATE:
1218 case FILE_LESHORT:
1219 case FILE_LELONG:
1220 case FILE_LEDATE:
1221 case FILE_LDATE:
1222 case FILE_BELDATE:
1223 case FILE_LELDATE:
1224 case FILE_MEDATE:
1225 case FILE_MELDATE:
1226 case FILE_MELONG:
1227 case FILE_QUAD:
1228 case FILE_LEQUAD:
1229 case FILE_BEQUAD:
1230 case FILE_QDATE:
1231 case FILE_LEQDATE:
1232 case FILE_BEQDATE:
1233 case FILE_QLDATE:
1234 case FILE_LEQLDATE:
1235 case FILE_BEQLDATE:
1236 case FILE_QWDATE:
1237 case FILE_LEQWDATE:
1238 case FILE_BEQWDATE:
1239 case FILE_FLOAT:
1240 case FILE_BEFLOAT:
1241 case FILE_LEFLOAT:
1242 case FILE_DOUBLE:
1243 case FILE_BEDOUBLE:
1244 case FILE_LEDOUBLE:
1245 case FILE_BEVARINT:
1246 case FILE_LEVARINT:
1247 case FILE_DER:
1248 case FILE_GUID:
1249 case FILE_OFFSET:
1250 case FILE_MSDOSDATE:
1251 case FILE_BEMSDOSDATE:
1252 case FILE_LEMSDOSDATE:
1253 case FILE_MSDOSTIME:
1254 case FILE_BEMSDOSTIME:
1255 case FILE_LEMSDOSTIME:
1256 case FILE_OCTAL:
1257 mstart->flag |= BINTEST;
1258 break;
1259 case FILE_STRING:
1260 case FILE_PSTRING:
1261 case FILE_BESTRING16:
1262 case FILE_LESTRING16:
1263 /* Allow text overrides */
1264 if (mstart->str_flags & STRING_TEXTTEST)
1265 mstart->flag |= TEXTTEST;
1266 else
1267 mstart->flag |= BINTEST;
1268 break;
1269 case FILE_REGEX:
1270 case FILE_SEARCH:
1271 /* Check for override */
1272 if (mstart->str_flags & STRING_BINTEST)
1273 mstart->flag |= BINTEST;
1274 if (mstart->str_flags & STRING_TEXTTEST)
1275 mstart->flag |= TEXTTEST;
1276
1277 if (mstart->flag & (TEXTTEST|BINTEST))
1278 break;
1279
1280 /* binary test if pattern is not text */
1281 if (file_looks_utf8(m->value.us, CAST(size_t, m->vallen), NULL,
1282 NULL) <= 0)
1283 mstart->flag |= BINTEST;
1284 else
1285 mstart->flag |= TEXTTEST;
1286 break;
1287 case FILE_DEFAULT:
1288 /* can't deduce anything; we shouldn't see this at the
1289 top level anyway */
1290 break;
1291 case FILE_INVALID:
1292 default:
1293 /* invalid search type, but no need to complain here */
1294 break;
1295 }
1296 }
1297
1298 file_private int
addentry(struct magic_set * ms,struct magic_entry * me,struct magic_entry_set * mset)1299 addentry(struct magic_set *ms, struct magic_entry *me,
1300 struct magic_entry_set *mset)
1301 {
1302 size_t i = me->mp->type == FILE_NAME ? 1 : 0;
1303 if (mset[i].me == NULL || mset[i].count == mset[i].max) {
1304 struct magic_entry *mp;
1305
1306 size_t incr = mset[i].max + ALLOC_INCR;
1307 if ((mp = CAST(struct magic_entry *,
1308 realloc(mset[i].me, sizeof(*mp) * incr))) ==
1309 NULL) {
1310 file_oomem(ms, sizeof(*mp) * incr);
1311 return -1;
1312 }
1313 (void)memset(&mp[mset[i].count], 0, sizeof(*mp) *
1314 ALLOC_INCR);
1315 mset[i].me = mp;
1316 mset[i].max = CAST(uint32_t, incr);
1317 assert(mset[i].max == incr);
1318 }
1319 mset[i].me[mset[i].count++] = *me;
1320 memset(me, 0, sizeof(*me));
1321 return 0;
1322 }
1323
1324 /*
1325 * Load and parse one file.
1326 */
1327 file_private void
load_1(struct magic_set * ms,int action,const char * fn,int * errs,struct magic_entry_set * mset)1328 load_1(struct magic_set *ms, int action, const char *fn, int *errs,
1329 struct magic_entry_set *mset)
1330 {
1331 size_t lineno = 0, llen = 0;
1332 char *line = NULL;
1333 ssize_t len;
1334 struct magic_entry me;
1335
1336 FILE *f = fopen(ms->file = fn, "r");
1337 if (f == NULL) {
1338 if (errno != ENOENT)
1339 file_error(ms, errno, "cannot read magic file `%s'",
1340 fn);
1341 (*errs)++;
1342 return;
1343 }
1344
1345 memset(&me, 0, sizeof(me));
1346 /* read and parse this file */
1347 for (ms->line = 1; (len = getline(&line, &llen, f)) != -1;
1348 ms->line++) {
1349 if (ms->magwarn >= ms->magwarn_max)
1350 break;
1351 if (len == 0) /* null line, garbage, etc */
1352 continue;
1353 if (line[len - 1] == '\n') {
1354 lineno++;
1355 line[len - 1] = '\0'; /* delete newline */
1356 }
1357 switch (line[0]) {
1358 case '\0': /* empty, do not parse */
1359 case '#': /* comment, do not parse */
1360 continue;
1361 case '!':
1362 if (line[1] == ':') {
1363 size_t i;
1364
1365 for (i = 0; bang[i].name != NULL; i++) {
1366 if (CAST(size_t, len - 2) > bang[i].len &&
1367 memcmp(bang[i].name, line + 2,
1368 bang[i].len) == 0)
1369 break;
1370 }
1371 if (bang[i].name == NULL) {
1372 file_error(ms, 0,
1373 "Unknown !: entry `%s'", line);
1374 (*errs)++;
1375 continue;
1376 }
1377 if (me.mp == NULL) {
1378 file_error(ms, 0,
1379 "No current entry for :!%s type",
1380 bang[i].name);
1381 (*errs)++;
1382 continue;
1383 }
1384 if ((*bang[i].fun)(ms, &me,
1385 line + bang[i].len + 2,
1386 len - bang[i].len - 2) != 0) {
1387 (*errs)++;
1388 continue;
1389 }
1390 continue;
1391 }
1392 /*FALLTHROUGH*/
1393 default:
1394 again:
1395 switch (parse(ms, &me, fn, line, lineno, action)) {
1396 case 0:
1397 continue;
1398 case 1:
1399 (void)addentry(ms, &me, mset);
1400 goto again;
1401 default:
1402 (*errs)++;
1403 break;
1404 }
1405 }
1406 }
1407 if (me.mp)
1408 (void)addentry(ms, &me, mset);
1409 free(line);
1410 (void)fclose(f);
1411 }
1412
1413 /*
1414 * parse a file or directory of files
1415 * const char *fn: name of magic file or directory
1416 */
1417 file_private int
cmpstrp(const void * p1,const void * p2)1418 cmpstrp(const void *p1, const void *p2)
1419 {
1420 return strcmp(*RCAST(char *const *, p1), *RCAST(char *const *, p2));
1421 }
1422
1423
1424 file_private uint32_t
set_text_binary(struct magic_set * ms,struct magic_entry * me,uint32_t nme,uint32_t starttest)1425 set_text_binary(struct magic_set *ms, struct magic_entry *me, uint32_t nme,
1426 uint32_t starttest)
1427 {
1428 static const char text[] = "text";
1429 static const char binary[] = "binary";
1430 static const size_t len = sizeof(text);
1431
1432 uint32_t i = starttest;
1433
1434 do {
1435 set_test_type(me[starttest].mp, me[i].mp);
1436 if ((ms->flags & MAGIC_DEBUG) == 0)
1437 continue;
1438 (void)fprintf(stderr, "%s%s%s: %s\n",
1439 me[i].mp->mimetype,
1440 me[i].mp->mimetype[0] == '\0' ? "" : "; ",
1441 me[i].mp->desc[0] ? me[i].mp->desc : "(no description)",
1442 me[i].mp->flag & BINTEST ? binary : text);
1443 if (me[i].mp->flag & BINTEST) {
1444 char *p = strstr(me[i].mp->desc, text);
1445 if (p && (p == me[i].mp->desc ||
1446 isspace(CAST(unsigned char, p[-1]))) &&
1447 (p + len - me[i].mp->desc == MAXstring
1448 || (p[len] == '\0' ||
1449 isspace(CAST(unsigned char, p[len])))))
1450 (void)fprintf(stderr, "*** Possible "
1451 "binary test for text type\n");
1452 }
1453 } while (++i < nme && me[i].mp->cont_level != 0);
1454 return i;
1455 }
1456
1457 file_private void
set_last_default(struct magic_set * ms,struct magic_entry * me,uint32_t nme)1458 set_last_default(struct magic_set *ms, struct magic_entry *me, uint32_t nme)
1459 {
1460 uint32_t i;
1461 for (i = 0; i < nme; i++) {
1462 if (me[i].mp->cont_level == 0 &&
1463 me[i].mp->type == FILE_DEFAULT) {
1464 while (++i < nme)
1465 if (me[i].mp->cont_level == 0)
1466 break;
1467 if (i != nme) {
1468 /* XXX - Ugh! */
1469 ms->line = me[i].mp->lineno;
1470 file_magwarn(ms,
1471 "level 0 \"default\" did not sort last");
1472 }
1473 return;
1474 }
1475 }
1476 }
1477
1478 file_private int
coalesce_entries(struct magic_set * ms,struct magic_entry * me,uint32_t nme,struct magic ** ma,uint32_t * nma)1479 coalesce_entries(struct magic_set *ms, struct magic_entry *me, uint32_t nme,
1480 struct magic **ma, uint32_t *nma)
1481 {
1482 uint32_t i, mentrycount = 0;
1483 size_t slen;
1484
1485 for (i = 0; i < nme; i++)
1486 mentrycount += me[i].cont_count;
1487
1488 if (mentrycount == 0) {
1489 *ma = NULL;
1490 *nma = 0;
1491 return 0;
1492 }
1493
1494 slen = sizeof(**ma) * mentrycount;
1495 if ((*ma = CAST(struct magic *, malloc(slen))) == NULL) {
1496 file_oomem(ms, slen);
1497 return -1;
1498 }
1499
1500 mentrycount = 0;
1501 for (i = 0; i < nme; i++) {
1502 (void)memcpy(*ma + mentrycount, me[i].mp,
1503 me[i].cont_count * sizeof(**ma));
1504 mentrycount += me[i].cont_count;
1505 }
1506 *nma = mentrycount;
1507 return 0;
1508 }
1509
1510 file_private void
magic_entry_free(struct magic_entry * me,uint32_t nme)1511 magic_entry_free(struct magic_entry *me, uint32_t nme)
1512 {
1513 uint32_t i;
1514 if (me == NULL)
1515 return;
1516 for (i = 0; i < nme; i++)
1517 free(me[i].mp);
1518 free(me);
1519 }
1520
1521 file_private struct magic_map *
apprentice_load(struct magic_set * ms,const char * fn,int action)1522 apprentice_load(struct magic_set *ms, const char *fn, int action)
1523 {
1524 int errs = 0;
1525 uint32_t i, j;
1526 size_t files = 0, maxfiles = 0;
1527 char **filearr = NULL, *mfn;
1528 struct stat st;
1529 struct magic_map *map;
1530 struct magic_entry_set mset[MAGIC_SETS];
1531 DIR *dir;
1532 struct dirent *d;
1533
1534 memset(mset, 0, sizeof(mset));
1535 ms->flags |= MAGIC_CHECK; /* Enable checks for parsed files */
1536
1537
1538 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL)
1539 {
1540 file_oomem(ms, sizeof(*map));
1541 return NULL;
1542 }
1543 map->type = MAP_TYPE_MALLOC;
1544
1545 /* print silly verbose header for USG compat. */
1546 if (action == FILE_CHECK)
1547 (void)fprintf(stderr, "%s\n", usg_hdr);
1548
1549 /* load directory or file */
1550 if (stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) {
1551 dir = opendir(fn);
1552 if (!dir) {
1553 errs++;
1554 goto out;
1555 }
1556 while ((d = readdir(dir)) != NULL) {
1557 if (d->d_name[0] == '.')
1558 continue;
1559 if (asprintf(&mfn, "%s/%s", fn, d->d_name) < 0) {
1560 file_oomem(ms,
1561 strlen(fn) + strlen(d->d_name) + 2);
1562 errs++;
1563 closedir(dir);
1564 goto out;
1565 }
1566 if (stat(mfn, &st) == -1 || !S_ISREG(st.st_mode)) {
1567 free(mfn);
1568 continue;
1569 }
1570 if (files >= maxfiles) {
1571 size_t mlen;
1572 char **nfilearr;
1573 maxfiles = (maxfiles + 1) * 2;
1574 mlen = maxfiles * sizeof(*filearr);
1575 if ((nfilearr = CAST(char **,
1576 realloc(filearr, mlen))) == NULL) {
1577 file_oomem(ms, mlen);
1578 free(mfn);
1579 closedir(dir);
1580 errs++;
1581 goto out;
1582 }
1583 filearr = nfilearr;
1584 }
1585 filearr[files++] = mfn;
1586 }
1587 closedir(dir);
1588 if (filearr) {
1589 qsort(filearr, files, sizeof(*filearr), cmpstrp);
1590 for (i = 0; i < files; i++) {
1591 load_1(ms, action, filearr[i], &errs, mset);
1592 free(filearr[i]);
1593 }
1594 free(filearr);
1595 filearr = NULL;
1596 }
1597 } else
1598 load_1(ms, action, fn, &errs, mset);
1599 if (errs)
1600 goto out;
1601
1602 for (j = 0; j < MAGIC_SETS; j++) {
1603 /* Set types of tests */
1604 for (i = 0; i < mset[j].count; ) {
1605 if (mset[j].me[i].mp->cont_level != 0) {
1606 i++;
1607 continue;
1608 }
1609 i = set_text_binary(ms, mset[j].me, mset[j].count, i);
1610 }
1611 if (mset[j].me)
1612 qsort(mset[j].me, mset[j].count, sizeof(*mset[0].me),
1613 apprentice_sort);
1614
1615 /*
1616 * Make sure that any level 0 "default" line is last
1617 * (if one exists).
1618 */
1619 set_last_default(ms, mset[j].me, mset[j].count);
1620
1621 /* coalesce per file arrays into a single one, if needed */
1622 if (mset[j].count == 0)
1623 continue;
1624
1625 if (coalesce_entries(ms, mset[j].me, mset[j].count,
1626 &map->magic[j], &map->nmagic[j]) == -1) {
1627 errs++;
1628 goto out;
1629 }
1630 }
1631
1632 out:
1633 free(filearr);
1634 for (j = 0; j < MAGIC_SETS; j++)
1635 magic_entry_free(mset[j].me, mset[j].count);
1636
1637 if (errs) {
1638 apprentice_unmap(map);
1639 return NULL;
1640 }
1641 return map;
1642 }
1643
1644 /*
1645 * extend the sign bit if the comparison is to be signed
1646 */
1647 file_protected uint64_t
file_signextend(struct magic_set * ms,struct magic * m,uint64_t v)1648 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v)
1649 {
1650 if (!(m->flag & UNSIGNED)) {
1651 switch(m->type) {
1652 /*
1653 * Do not remove the casts below. They are
1654 * vital. When later compared with the data,
1655 * the sign extension must have happened.
1656 */
1657 case FILE_BYTE:
1658 v = CAST(signed char, v);
1659 break;
1660 case FILE_SHORT:
1661 case FILE_BESHORT:
1662 case FILE_LESHORT:
1663 v = CAST(short, v);
1664 break;
1665 case FILE_DATE:
1666 case FILE_BEDATE:
1667 case FILE_LEDATE:
1668 case FILE_MEDATE:
1669 case FILE_LDATE:
1670 case FILE_BELDATE:
1671 case FILE_LELDATE:
1672 case FILE_MELDATE:
1673 case FILE_LONG:
1674 case FILE_BELONG:
1675 case FILE_LELONG:
1676 case FILE_MELONG:
1677 case FILE_FLOAT:
1678 case FILE_BEFLOAT:
1679 case FILE_LEFLOAT:
1680 case FILE_MSDOSDATE:
1681 case FILE_BEMSDOSDATE:
1682 case FILE_LEMSDOSDATE:
1683 case FILE_MSDOSTIME:
1684 case FILE_BEMSDOSTIME:
1685 case FILE_LEMSDOSTIME:
1686 v = CAST(int32_t, v);
1687 break;
1688 case FILE_QUAD:
1689 case FILE_BEQUAD:
1690 case FILE_LEQUAD:
1691 case FILE_QDATE:
1692 case FILE_QLDATE:
1693 case FILE_QWDATE:
1694 case FILE_BEQDATE:
1695 case FILE_BEQLDATE:
1696 case FILE_BEQWDATE:
1697 case FILE_LEQDATE:
1698 case FILE_LEQLDATE:
1699 case FILE_LEQWDATE:
1700 case FILE_DOUBLE:
1701 case FILE_BEDOUBLE:
1702 case FILE_LEDOUBLE:
1703 case FILE_OFFSET:
1704 case FILE_BEVARINT:
1705 case FILE_LEVARINT:
1706 v = CAST(int64_t, v);
1707 break;
1708 case FILE_STRING:
1709 case FILE_PSTRING:
1710 case FILE_BESTRING16:
1711 case FILE_LESTRING16:
1712 case FILE_REGEX:
1713 case FILE_SEARCH:
1714 case FILE_DEFAULT:
1715 case FILE_INDIRECT:
1716 case FILE_NAME:
1717 case FILE_USE:
1718 case FILE_CLEAR:
1719 case FILE_DER:
1720 case FILE_GUID:
1721 case FILE_OCTAL:
1722 break;
1723 default:
1724 if (ms->flags & MAGIC_CHECK)
1725 file_magwarn(ms, "cannot happen: m->type=%d\n",
1726 m->type);
1727 return FILE_BADSIZE;
1728 }
1729 }
1730 return v;
1731 }
1732
1733 file_private int
string_modifier_check(struct magic_set * ms,struct magic * m)1734 string_modifier_check(struct magic_set *ms, struct magic *m)
1735 {
1736 if ((ms->flags & MAGIC_CHECK) == 0)
1737 return 0;
1738
1739 if ((m->type != FILE_REGEX || (m->str_flags & REGEX_LINE_COUNT) == 0) &&
1740 (m->type != FILE_PSTRING && (m->str_flags & PSTRING_LEN) != 0)) {
1741 file_magwarn(ms,
1742 "'/BHhLl' modifiers are only allowed for pascal strings\n");
1743 return -1;
1744 }
1745 switch (m->type) {
1746 case FILE_BESTRING16:
1747 case FILE_LESTRING16:
1748 if (m->str_flags != 0) {
1749 file_magwarn(ms,
1750 "no modifiers allowed for 16-bit strings\n");
1751 return -1;
1752 }
1753 break;
1754 case FILE_STRING:
1755 case FILE_PSTRING:
1756 if ((m->str_flags & REGEX_OFFSET_START) != 0) {
1757 file_magwarn(ms,
1758 "'/%c' only allowed on regex and search\n",
1759 CHAR_REGEX_OFFSET_START);
1760 return -1;
1761 }
1762 break;
1763 case FILE_SEARCH:
1764 if (m->str_range == 0) {
1765 file_magwarn(ms,
1766 "missing range; defaulting to %d\n",
1767 STRING_DEFAULT_RANGE);
1768 m->str_range = STRING_DEFAULT_RANGE;
1769 return -1;
1770 }
1771 break;
1772 case FILE_REGEX:
1773 if ((m->str_flags & STRING_COMPACT_WHITESPACE) != 0) {
1774 file_magwarn(ms, "'/%c' not allowed on regex\n",
1775 CHAR_COMPACT_WHITESPACE);
1776 return -1;
1777 }
1778 if ((m->str_flags & STRING_COMPACT_OPTIONAL_WHITESPACE) != 0) {
1779 file_magwarn(ms, "'/%c' not allowed on regex\n",
1780 CHAR_COMPACT_OPTIONAL_WHITESPACE);
1781 return -1;
1782 }
1783 break;
1784 default:
1785 file_magwarn(ms, "coding error: m->type=%d\n",
1786 m->type);
1787 return -1;
1788 }
1789 return 0;
1790 }
1791
1792 file_private int
get_op(char c)1793 get_op(char c)
1794 {
1795 switch (c) {
1796 case '&':
1797 return FILE_OPAND;
1798 case '|':
1799 return FILE_OPOR;
1800 case '^':
1801 return FILE_OPXOR;
1802 case '+':
1803 return FILE_OPADD;
1804 case '-':
1805 return FILE_OPMINUS;
1806 case '*':
1807 return FILE_OPMULTIPLY;
1808 case '/':
1809 return FILE_OPDIVIDE;
1810 case '%':
1811 return FILE_OPMODULO;
1812 default:
1813 return -1;
1814 }
1815 }
1816
1817 #ifdef ENABLE_CONDITIONALS
1818 file_private int
get_cond(const char * l,const char ** t)1819 get_cond(const char *l, const char **t)
1820 {
1821 static const struct cond_tbl_s {
1822 char name[8];
1823 size_t len;
1824 int cond;
1825 } cond_tbl[] = {
1826 { "if", 2, COND_IF },
1827 { "elif", 4, COND_ELIF },
1828 { "else", 4, COND_ELSE },
1829 { "", 0, COND_NONE },
1830 };
1831 const struct cond_tbl_s *p;
1832
1833 for (p = cond_tbl; p->len; p++) {
1834 if (strncmp(l, p->name, p->len) == 0 &&
1835 isspace(CAST(unsigned char, l[p->len]))) {
1836 if (t)
1837 *t = l + p->len;
1838 break;
1839 }
1840 }
1841 return p->cond;
1842 }
1843
1844 file_private int
check_cond(struct magic_set * ms,int cond,uint32_t cont_level)1845 check_cond(struct magic_set *ms, int cond, uint32_t cont_level)
1846 {
1847 int last_cond;
1848 last_cond = ms->c.li[cont_level].last_cond;
1849
1850 switch (cond) {
1851 case COND_IF:
1852 if (last_cond != COND_NONE && last_cond != COND_ELIF) {
1853 if (ms->flags & MAGIC_CHECK)
1854 file_magwarn(ms, "syntax error: `if'");
1855 return -1;
1856 }
1857 last_cond = COND_IF;
1858 break;
1859
1860 case COND_ELIF:
1861 if (last_cond != COND_IF && last_cond != COND_ELIF) {
1862 if (ms->flags & MAGIC_CHECK)
1863 file_magwarn(ms, "syntax error: `elif'");
1864 return -1;
1865 }
1866 last_cond = COND_ELIF;
1867 break;
1868
1869 case COND_ELSE:
1870 if (last_cond != COND_IF && last_cond != COND_ELIF) {
1871 if (ms->flags & MAGIC_CHECK)
1872 file_magwarn(ms, "syntax error: `else'");
1873 return -1;
1874 }
1875 last_cond = COND_NONE;
1876 break;
1877
1878 case COND_NONE:
1879 last_cond = COND_NONE;
1880 break;
1881 }
1882
1883 ms->c.li[cont_level].last_cond = last_cond;
1884 return 0;
1885 }
1886 #endif /* ENABLE_CONDITIONALS */
1887
1888 file_private int
parse_indirect_modifier(struct magic_set * ms,struct magic * m,const char ** lp)1889 parse_indirect_modifier(struct magic_set *ms, struct magic *m, const char **lp)
1890 {
1891 const char *l = *lp;
1892
1893 while (!isspace(CAST(unsigned char, *++l)))
1894 switch (*l) {
1895 case CHAR_INDIRECT_RELATIVE:
1896 m->str_flags |= INDIRECT_RELATIVE;
1897 break;
1898 default:
1899 if (ms->flags & MAGIC_CHECK)
1900 file_magwarn(ms, "indirect modifier `%c' "
1901 "invalid", *l);
1902 *lp = l;
1903 return -1;
1904 }
1905 *lp = l;
1906 return 0;
1907 }
1908
1909 file_private void
parse_op_modifier(struct magic_set * ms,struct magic * m,const char ** lp,int op)1910 parse_op_modifier(struct magic_set *ms, struct magic *m, const char **lp,
1911 int op)
1912 {
1913 const char *l = *lp;
1914 char *t;
1915 uint64_t val;
1916
1917 ++l;
1918 m->mask_op |= op;
1919 val = CAST(uint64_t, strtoull(l, &t, 0));
1920 l = t;
1921 m->num_mask = file_signextend(ms, m, val);
1922 eatsize(&l);
1923 *lp = l;
1924 }
1925
1926 file_private int
parse_string_modifier(struct magic_set * ms,struct magic * m,const char ** lp)1927 parse_string_modifier(struct magic_set *ms, struct magic *m, const char **lp)
1928 {
1929 const char *l = *lp;
1930 char *t;
1931 int have_range = 0;
1932
1933 while (!isspace(CAST(unsigned char, *++l))) {
1934 switch (*l) {
1935 case '0': case '1': case '2':
1936 case '3': case '4': case '5':
1937 case '6': case '7': case '8':
1938 case '9':
1939 if (have_range && (ms->flags & MAGIC_CHECK))
1940 file_magwarn(ms, "multiple ranges");
1941 have_range = 1;
1942 m->str_range = CAST(uint32_t, strtoul(l, &t, 0));
1943 if (m->str_range == 0)
1944 file_magwarn(ms, "zero range");
1945 l = t - 1;
1946 break;
1947 case CHAR_COMPACT_WHITESPACE:
1948 m->str_flags |= STRING_COMPACT_WHITESPACE;
1949 break;
1950 case CHAR_COMPACT_OPTIONAL_WHITESPACE:
1951 m->str_flags |= STRING_COMPACT_OPTIONAL_WHITESPACE;
1952 break;
1953 case CHAR_IGNORE_LOWERCASE:
1954 m->str_flags |= STRING_IGNORE_LOWERCASE;
1955 break;
1956 case CHAR_IGNORE_UPPERCASE:
1957 m->str_flags |= STRING_IGNORE_UPPERCASE;
1958 break;
1959 case CHAR_REGEX_OFFSET_START:
1960 m->str_flags |= REGEX_OFFSET_START;
1961 break;
1962 case CHAR_BINTEST:
1963 m->str_flags |= STRING_BINTEST;
1964 break;
1965 case CHAR_TEXTTEST:
1966 m->str_flags |= STRING_TEXTTEST;
1967 break;
1968 case CHAR_TRIM:
1969 m->str_flags |= STRING_TRIM;
1970 break;
1971 case CHAR_FULL_WORD:
1972 m->str_flags |= STRING_FULL_WORD;
1973 break;
1974 case CHAR_PSTRING_1_LE:
1975 #define SET_LENGTH(a) m->str_flags = (m->str_flags & ~PSTRING_LEN) | (a)
1976 if (m->type != FILE_PSTRING)
1977 goto bad;
1978 SET_LENGTH(PSTRING_1_LE);
1979 break;
1980 case CHAR_PSTRING_2_BE:
1981 if (m->type != FILE_PSTRING)
1982 goto bad;
1983 SET_LENGTH(PSTRING_2_BE);
1984 break;
1985 case CHAR_PSTRING_2_LE:
1986 if (m->type != FILE_PSTRING)
1987 goto bad;
1988 SET_LENGTH(PSTRING_2_LE);
1989 break;
1990 case CHAR_PSTRING_4_BE:
1991 if (m->type != FILE_PSTRING)
1992 goto bad;
1993 SET_LENGTH(PSTRING_4_BE);
1994 break;
1995 case CHAR_PSTRING_4_LE:
1996 switch (m->type) {
1997 case FILE_PSTRING:
1998 case FILE_REGEX:
1999 break;
2000 default:
2001 goto bad;
2002 }
2003 SET_LENGTH(PSTRING_4_LE);
2004 break;
2005 case CHAR_PSTRING_LENGTH_INCLUDES_ITSELF:
2006 if (m->type != FILE_PSTRING)
2007 goto bad;
2008 m->str_flags |= PSTRING_LENGTH_INCLUDES_ITSELF;
2009 break;
2010 default:
2011 bad:
2012 if (ms->flags & MAGIC_CHECK)
2013 file_magwarn(ms, "string modifier `%c' "
2014 "invalid", *l);
2015 goto out;
2016 }
2017 /* allow multiple '/' for readability */
2018 if (l[1] == '/' && !isspace(CAST(unsigned char, l[2])))
2019 l++;
2020 }
2021 if (string_modifier_check(ms, m) == -1)
2022 goto out;
2023 *lp = l;
2024 return 0;
2025 out:
2026 *lp = l;
2027 return -1;
2028 }
2029
2030 /*
2031 * parse one line from magic file, put into magic[index++] if valid
2032 */
2033 file_private int
parse(struct magic_set * ms,struct magic_entry * me,const char * file,const char * line,size_t lineno,int action)2034 parse(struct magic_set *ms, struct magic_entry *me, const char *file,
2035 const char *line, size_t lineno, int action)
2036 {
2037 #ifdef ENABLE_CONDITIONALS
2038 static uint32_t last_cont_level = 0;
2039 #endif
2040 size_t i;
2041 struct magic *m;
2042 const char *l = line;
2043 char *t;
2044 int op;
2045 uint32_t cont_level;
2046 int32_t diff;
2047
2048 cont_level = 0;
2049
2050 /*
2051 * Parse the offset.
2052 */
2053 while (*l == '>') {
2054 ++l; /* step over */
2055 cont_level++;
2056 }
2057 #ifdef ENABLE_CONDITIONALS
2058 if (cont_level == 0 || cont_level > last_cont_level)
2059 if (file_check_mem(ms, cont_level) == -1)
2060 return -1;
2061 last_cont_level = cont_level;
2062 #endif
2063 if (cont_level != 0) {
2064 if (me->mp == NULL) {
2065 file_magerror(ms, "No current entry for continuation");
2066 return -1;
2067 }
2068 if (me->cont_count == 0) {
2069 file_magerror(ms, "Continuations present with 0 count");
2070 return -1;
2071 }
2072 m = &me->mp[me->cont_count - 1];
2073 diff = CAST(int32_t, cont_level) - CAST(int32_t, m->cont_level);
2074 if (diff > 1)
2075 file_magwarn(ms, "New continuation level %u is more "
2076 "than one larger than current level %u", cont_level,
2077 m->cont_level);
2078 if (me->cont_count == me->max_count) {
2079 struct magic *nm;
2080 size_t cnt = me->max_count + ALLOC_CHUNK;
2081 if ((nm = CAST(struct magic *, realloc(me->mp,
2082 sizeof(*nm) * cnt))) == NULL) {
2083 file_oomem(ms, sizeof(*nm) * cnt);
2084 return -1;
2085 }
2086 me->mp = nm;
2087 me->max_count = CAST(uint32_t, cnt);
2088 }
2089 m = &me->mp[me->cont_count++];
2090 (void)memset(m, 0, sizeof(*m));
2091 m->cont_level = cont_level;
2092 } else {
2093 static const size_t len = sizeof(*m) * ALLOC_CHUNK;
2094 if (me->mp != NULL)
2095 return 1;
2096 if ((m = CAST(struct magic *, malloc(len))) == NULL) {
2097 file_oomem(ms, len);
2098 return -1;
2099 }
2100 me->mp = m;
2101 me->max_count = ALLOC_CHUNK;
2102 (void)memset(m, 0, sizeof(*m));
2103 m->factor_op = FILE_FACTOR_OP_NONE;
2104 m->cont_level = 0;
2105 me->cont_count = 1;
2106 }
2107 m->lineno = CAST(uint32_t, lineno);
2108
2109 if (*l == '&') { /* m->cont_level == 0 checked below. */
2110 ++l; /* step over */
2111 m->flag |= OFFADD;
2112 }
2113 if (*l == '(') {
2114 ++l; /* step over */
2115 m->flag |= INDIR;
2116 if (m->flag & OFFADD)
2117 m->flag = (m->flag & ~OFFADD) | INDIROFFADD;
2118
2119 if (*l == '&') { /* m->cont_level == 0 checked below */
2120 ++l; /* step over */
2121 m->flag |= OFFADD;
2122 }
2123 }
2124 /* Indirect offsets are not valid at level 0. */
2125 if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD))) {
2126 if (ms->flags & MAGIC_CHECK)
2127 file_magwarn(ms, "relative offset at level 0");
2128 return -1;
2129 }
2130
2131 /* get offset, then skip over it */
2132 if (*l == '-' || *l == '+') {
2133 ++l; /* step over */
2134 m->flag |= l[-1] == '-' ? OFFNEGATIVE : OFFPOSITIVE;
2135 }
2136 m->offset = CAST(int32_t, strtol(l, &t, 0));
2137 if (l == t) {
2138 if (ms->flags & MAGIC_CHECK)
2139 file_magwarn(ms, "offset `%s' invalid", l);
2140 return -1;
2141 }
2142
2143 l = t;
2144
2145 if (m->flag & INDIR) {
2146 m->in_type = FILE_LONG;
2147 m->in_offset = 0;
2148 m->in_op = 0;
2149 /*
2150 * read [.,lbs][+-]nnnnn)
2151 */
2152 if (*l == '.' || *l == ',') {
2153 if (*l == ',')
2154 m->in_op |= FILE_OPSIGNED;
2155 l++;
2156 switch (*l) {
2157 case 'l':
2158 m->in_type = FILE_LELONG;
2159 break;
2160 case 'L':
2161 m->in_type = FILE_BELONG;
2162 break;
2163 case 'm':
2164 m->in_type = FILE_MELONG;
2165 break;
2166 case 'h':
2167 case 's':
2168 m->in_type = FILE_LESHORT;
2169 break;
2170 case 'H':
2171 case 'S':
2172 m->in_type = FILE_BESHORT;
2173 break;
2174 case 'c':
2175 case 'b':
2176 case 'C':
2177 case 'B':
2178 m->in_type = FILE_BYTE;
2179 break;
2180 case 'e':
2181 case 'f':
2182 case 'g':
2183 m->in_type = FILE_LEDOUBLE;
2184 break;
2185 case 'E':
2186 case 'F':
2187 case 'G':
2188 m->in_type = FILE_BEDOUBLE;
2189 break;
2190 case 'i':
2191 m->in_type = FILE_LEID3;
2192 break;
2193 case 'I':
2194 m->in_type = FILE_BEID3;
2195 break;
2196 case 'o':
2197 m->in_type = FILE_OCTAL;
2198 break;
2199 case 'q':
2200 m->in_type = FILE_LEQUAD;
2201 break;
2202 case 'Q':
2203 m->in_type = FILE_BEQUAD;
2204 break;
2205 default:
2206 if (ms->flags & MAGIC_CHECK)
2207 file_magwarn(ms,
2208 "indirect offset type `%c' invalid",
2209 *l);
2210 return -1;
2211 }
2212 l++;
2213 }
2214
2215 if (*l == '~') {
2216 m->in_op |= FILE_OPINVERSE;
2217 l++;
2218 }
2219 if ((op = get_op(*l)) != -1) {
2220 m->in_op |= op;
2221 l++;
2222 }
2223 if (*l == '(') {
2224 m->in_op |= FILE_OPINDIRECT;
2225 l++;
2226 }
2227 if (isdigit(CAST(unsigned char, *l)) || *l == '-') {
2228 m->in_offset = CAST(int32_t, strtol(l, &t, 0));
2229 if (l == t) {
2230 if (ms->flags & MAGIC_CHECK)
2231 file_magwarn(ms,
2232 "in_offset `%s' invalid", l);
2233 return -1;
2234 }
2235 l = t;
2236 }
2237 if (*l++ != ')' ||
2238 ((m->in_op & FILE_OPINDIRECT) && *l++ != ')')) {
2239 if (ms->flags & MAGIC_CHECK)
2240 file_magwarn(ms,
2241 "missing ')' in indirect offset");
2242 return -1;
2243 }
2244 }
2245 EATAB;
2246
2247 #ifdef ENABLE_CONDITIONALS
2248 m->cond = get_cond(l, &l);
2249 if (check_cond(ms, m->cond, cont_level) == -1)
2250 return -1;
2251
2252 EATAB;
2253 #endif
2254
2255 /*
2256 * Parse the type.
2257 */
2258 if (*l == 'u') {
2259 /*
2260 * Try it as a keyword type prefixed by "u"; match what
2261 * follows the "u". If that fails, try it as an SUS
2262 * integer type.
2263 */
2264 m->type = get_type(type_tbl, l + 1, &l);
2265 if (m->type == FILE_INVALID) {
2266 /*
2267 * Not a keyword type; parse it as an SUS type,
2268 * 'u' possibly followed by a number or C/S/L.
2269 */
2270 m->type = get_standard_integer_type(l, &l);
2271 }
2272 /* It's unsigned. */
2273 if (m->type != FILE_INVALID)
2274 m->flag |= UNSIGNED;
2275 } else {
2276 /*
2277 * Try it as a keyword type. If that fails, try it as
2278 * an SUS integer type if it begins with "d" or as an
2279 * SUS string type if it begins with "s". In any case,
2280 * it's not unsigned.
2281 */
2282 m->type = get_type(type_tbl, l, &l);
2283 if (m->type == FILE_INVALID) {
2284 /*
2285 * Not a keyword type; parse it as an SUS type,
2286 * either 'd' possibly followed by a number or
2287 * C/S/L, or just 's'.
2288 */
2289 if (*l == 'd')
2290 m->type = get_standard_integer_type(l, &l);
2291 else if (*l == 's'
2292 && !isalpha(CAST(unsigned char, l[1]))) {
2293 m->type = FILE_STRING;
2294 ++l;
2295 }
2296 }
2297 }
2298
2299 if (m->type == FILE_INVALID) {
2300 /* Not found - try it as a special keyword. */
2301 m->type = get_type(special_tbl, l, &l);
2302 }
2303
2304 if (m->type == FILE_INVALID) {
2305 if (ms->flags & MAGIC_CHECK)
2306 file_magwarn(ms, "type `%s' invalid", l);
2307 return -1;
2308 }
2309
2310 if (m->type == FILE_NAME && cont_level != 0) {
2311 if (ms->flags & MAGIC_CHECK)
2312 file_magwarn(ms, "`name%s' entries can only be "
2313 "declared at top level", l);
2314 return -1;
2315 }
2316
2317 /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */
2318 /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */
2319
2320 m->mask_op = 0;
2321 if (*l == '~') {
2322 if (!IS_STRING(m->type))
2323 m->mask_op |= FILE_OPINVERSE;
2324 else if (ms->flags & MAGIC_CHECK)
2325 file_magwarn(ms, "'~' invalid for string types");
2326 ++l;
2327 }
2328 m->str_range = 0;
2329 m->str_flags = m->type == FILE_PSTRING ? PSTRING_1_LE : 0;
2330 if ((op = get_op(*l)) != -1) {
2331 if (IS_STRING(m->type)) {
2332 int r;
2333
2334 if (op != FILE_OPDIVIDE) {
2335 if (ms->flags & MAGIC_CHECK)
2336 file_magwarn(ms,
2337 "invalid string/indirect op: "
2338 "`%c'", *t);
2339 return -1;
2340 }
2341
2342 if (m->type == FILE_INDIRECT)
2343 r = parse_indirect_modifier(ms, m, &l);
2344 else
2345 r = parse_string_modifier(ms, m, &l);
2346 if (r == -1)
2347 return -1;
2348 } else
2349 parse_op_modifier(ms, m, &l, op);
2350 }
2351
2352 /*
2353 * We used to set mask to all 1's here, instead let's just not do
2354 * anything if mask = 0 (unless you have a better idea)
2355 */
2356 EATAB;
2357
2358 switch (*l) {
2359 case '>':
2360 case '<':
2361 m->reln = *l;
2362 ++l;
2363 if (*l == '=') {
2364 if (ms->flags & MAGIC_CHECK) {
2365 file_magwarn(ms, "%c= not supported",
2366 m->reln);
2367 return -1;
2368 }
2369 ++l;
2370 }
2371 break;
2372 /* Old-style anding: "0 byte &0x80 dynamically linked" */
2373 case '&':
2374 case '^':
2375 case '=':
2376 m->reln = *l;
2377 ++l;
2378 if (*l == '=') {
2379 /* HP compat: ignore &= etc. */
2380 ++l;
2381 }
2382 break;
2383 case '!':
2384 m->reln = *l;
2385 ++l;
2386 break;
2387 default:
2388 m->reln = '='; /* the default relation */
2389 if (*l == 'x' && ((isascii(CAST(unsigned char, l[1])) &&
2390 isspace(CAST(unsigned char, l[1]))) || !l[1])) {
2391 m->reln = *l;
2392 ++l;
2393 }
2394 break;
2395 }
2396 /*
2397 * Grab the value part, except for an 'x' reln.
2398 */
2399 if (m->reln != 'x' && getvalue(ms, m, &l, action))
2400 return -1;
2401
2402 /*
2403 * TODO finish this macro and start using it!
2404 * #define offsetcheck {if (offset > ms->bytes_max -1)
2405 * magwarn("offset too big"); }
2406 */
2407
2408 /*
2409 * Now get last part - the description
2410 */
2411 EATAB;
2412 if (l[0] == '\b') {
2413 ++l;
2414 m->flag |= NOSPACE;
2415 } else if ((l[0] == '\\') && (l[1] == 'b')) {
2416 ++l;
2417 ++l;
2418 m->flag |= NOSPACE;
2419 }
2420 for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); )
2421 continue;
2422 if (m->desc[0] == '\0') {
2423 // Tuck in the filename for debugging.
2424 strlcpy(m->desc + 1, file, sizeof(m->desc) - 1);
2425 }
2426 if (i == sizeof(m->desc)) {
2427 m->desc[sizeof(m->desc) - 1] = '\0';
2428 if (ms->flags & MAGIC_CHECK)
2429 file_magwarn(ms, "description `%s' truncated", m->desc);
2430 }
2431
2432 /*
2433 * We only do this check while compiling, or if any of the magic
2434 * files were not compiled.
2435 */
2436 if (ms->flags & MAGIC_CHECK) {
2437 if (check_format(ms, m) == -1)
2438 return -1;
2439 }
2440 #ifndef COMPILE_ONLY
2441 if (action == FILE_CHECK) {
2442 file_mdump(m);
2443 }
2444 #endif
2445 m->mimetype[0] = '\0'; /* initialise MIME type to none */
2446 return 0;
2447 }
2448
2449 /*
2450 * parse a STRENGTH annotation line from magic file, put into magic[index - 1]
2451 * if valid
2452 */
2453 /*ARGSUSED*/
2454 file_private int
parse_strength(struct magic_set * ms,struct magic_entry * me,const char * line,size_t len)2455 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line,
2456 size_t len __attribute__((__unused__)))
2457 {
2458 const char *l = line;
2459 char *el;
2460 unsigned long factor;
2461 char sbuf[512];
2462 struct magic *m = &me->mp[0];
2463
2464 if (m->factor_op != FILE_FACTOR_OP_NONE) {
2465 file_magwarn(ms,
2466 "Current entry already has a strength type: %c %d",
2467 m->factor_op, m->factor);
2468 return -1;
2469 }
2470 if (m->type == FILE_NAME) {
2471 file_magwarn(ms, "%s: Strength setting is not supported in "
2472 "\"name\" magic entries",
2473 file_printable(ms, sbuf, sizeof(sbuf), m->value.s,
2474 sizeof(m->value.s)));
2475 return -1;
2476 }
2477 EATAB;
2478 switch (*l) {
2479 case FILE_FACTOR_OP_NONE:
2480 break;
2481 case FILE_FACTOR_OP_PLUS:
2482 case FILE_FACTOR_OP_MINUS:
2483 case FILE_FACTOR_OP_TIMES:
2484 case FILE_FACTOR_OP_DIV:
2485 m->factor_op = *l++;
2486 break;
2487 default:
2488 file_magwarn(ms, "Unknown factor op `%c'", *l);
2489 return -1;
2490 }
2491 EATAB;
2492 factor = strtoul(l, &el, 0);
2493 if (factor > 255) {
2494 file_magwarn(ms, "Too large factor `%lu'", factor);
2495 goto out;
2496 }
2497 if (*el && !isspace(CAST(unsigned char, *el))) {
2498 file_magwarn(ms, "Bad factor `%s'", l);
2499 goto out;
2500 }
2501 m->factor = CAST(uint8_t, factor);
2502 if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) {
2503 file_magwarn(ms, "Cannot have factor op `%c' and factor %u",
2504 m->factor_op, m->factor);
2505 goto out;
2506 }
2507 return 0;
2508 out:
2509 m->factor_op = FILE_FACTOR_OP_NONE;
2510 m->factor = 0;
2511 return -1;
2512 }
2513
2514 file_private int
goodchar(unsigned char x,const char * extra)2515 goodchar(unsigned char x, const char *extra)
2516 {
2517 return (isascii(x) && isalnum(x)) || strchr(extra, x);
2518 }
2519
2520 file_private int
parse_extra(struct magic_set * ms,struct magic_entry * me,const char * line,size_t llen,off_t off,size_t len,const char * name,const char * extra,int nt)2521 parse_extra(struct magic_set *ms, struct magic_entry *me, const char *line,
2522 size_t llen, off_t off, size_t len, const char *name, const char *extra,
2523 int nt)
2524 {
2525 size_t i;
2526 const char *l = line;
2527 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1];
2528 char *buf = CAST(char *, CAST(void *, m)) + off;
2529
2530 if (buf[0] != '\0') {
2531 len = nt ? strlen(buf) : len;
2532 file_magwarn(ms, "Current entry already has a %s type "
2533 "`%.*s', new type `%s'", name, CAST(int, len), buf, l);
2534 return -1;
2535 }
2536
2537 if (*m->desc == '\0') {
2538 file_magwarn(ms, "Current entry does not yet have a "
2539 "description for adding a %s type", name);
2540 return -1;
2541 }
2542
2543 EATAB;
2544 for (i = 0; *l && i < llen && i < len && goodchar(*l, extra);
2545 buf[i++] = *l++)
2546 continue;
2547
2548 if (i == len && *l) {
2549 if (nt)
2550 buf[len - 1] = '\0';
2551 if (ms->flags & MAGIC_CHECK)
2552 file_magwarn(ms, "%s type `%s' truncated %"
2553 SIZE_T_FORMAT "u", name, line, i);
2554 } else {
2555 if (!isspace(CAST(unsigned char, *l)) && !goodchar(*l, extra))
2556 file_magwarn(ms, "%s type `%s' has bad char '%c'",
2557 name, line, *l);
2558 if (nt)
2559 buf[i] = '\0';
2560 }
2561
2562 if (i > 0)
2563 return 0;
2564
2565 file_magerror(ms, "Bad magic entry '%s'", line);
2566 return -1;
2567 }
2568
2569 /*
2570 * Parse an Apple CREATOR/TYPE annotation from magic file and put it into
2571 * magic[index - 1]
2572 */
2573 file_private int
parse_apple(struct magic_set * ms,struct magic_entry * me,const char * line,size_t len)2574 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line,
2575 size_t len)
2576 {
2577 return parse_extra(ms, me, line, len,
2578 CAST(off_t, offsetof(struct magic, apple)),
2579 sizeof(me->mp[me->cont_count - 1].apple), "APPLE", "!+-./?", 0);
2580 }
2581
2582 /*
2583 * Parse a comma-separated list of extensions
2584 */
2585 file_private int
parse_ext(struct magic_set * ms,struct magic_entry * me,const char * line,size_t len)2586 parse_ext(struct magic_set *ms, struct magic_entry *me, const char *line,
2587 size_t len)
2588 {
2589 return parse_extra(ms, me, line, len,
2590 CAST(off_t, offsetof(struct magic, ext)),
2591 sizeof(me->mp[me->cont_count - 1].ext), "EXTENSION", ",!+-/@?_$&~",
2592 0);
2593 /* & for b&w */
2594 /* ~ for journal~ */
2595 }
2596
2597 /*
2598 * parse a MIME annotation line from magic file, put into magic[index - 1]
2599 * if valid
2600 */
2601 file_private int
parse_mime(struct magic_set * ms,struct magic_entry * me,const char * line,size_t len)2602 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line,
2603 size_t len)
2604 {
2605 return parse_extra(ms, me, line, len,
2606 CAST(off_t, offsetof(struct magic, mimetype)),
2607 sizeof(me->mp[me->cont_count - 1].mimetype), "MIME", "+-/.$?:{};=",
2608 1);
2609 }
2610
2611 file_private int
check_format_type(const char * ptr,int type,const char ** estr)2612 check_format_type(const char *ptr, int type, const char **estr)
2613 {
2614 int quad = 0, h;
2615 size_t len, cnt;
2616 if (*ptr == '\0') {
2617 /* Missing format string; bad */
2618 *estr = "missing format spec";
2619 return -1;
2620 }
2621
2622 switch (file_formats[type]) {
2623 case FILE_FMT_QUAD:
2624 quad = 1;
2625 /*FALLTHROUGH*/
2626 case FILE_FMT_NUM:
2627 if (quad == 0) {
2628 switch (type) {
2629 case FILE_BYTE:
2630 h = 2;
2631 break;
2632 case FILE_SHORT:
2633 case FILE_BESHORT:
2634 case FILE_LESHORT:
2635 h = 1;
2636 break;
2637 case FILE_LONG:
2638 case FILE_BELONG:
2639 case FILE_LELONG:
2640 case FILE_MELONG:
2641 case FILE_LEID3:
2642 case FILE_BEID3:
2643 case FILE_INDIRECT:
2644 h = 0;
2645 break;
2646 default:
2647 fprintf(stderr, "Bad number format %d", type);
2648 abort();
2649 }
2650 } else
2651 h = 0;
2652 while (*ptr && strchr("-.#", *ptr) != NULL)
2653 ptr++;
2654 #define CHECKLEN() do { \
2655 for (len = cnt = 0; isdigit(CAST(unsigned char, *ptr)); ptr++, cnt++) \
2656 len = len * 10 + (*ptr - '0'); \
2657 if (cnt > 5 || len > 1024) \
2658 goto toolong; \
2659 } while (/*CONSTCOND*/0)
2660
2661 CHECKLEN();
2662 if (*ptr == '.')
2663 ptr++;
2664 CHECKLEN();
2665 if (quad) {
2666 if (*ptr++ != 'l')
2667 goto invalid;
2668 if (*ptr++ != 'l')
2669 goto invalid;
2670 }
2671
2672 switch (*ptr++) {
2673 #ifdef STRICT_FORMAT /* "long" formats are int formats for us */
2674 /* so don't accept the 'l' modifier */
2675 case 'l':
2676 switch (*ptr++) {
2677 case 'i':
2678 case 'd':
2679 case 'u':
2680 case 'o':
2681 case 'x':
2682 case 'X':
2683 if (h == 0)
2684 return 0;
2685 /*FALLTHROUGH*/
2686 default:
2687 goto invalid;
2688 }
2689
2690 /*
2691 * Don't accept h and hh modifiers. They make writing
2692 * magic entries more complicated, for very little benefit
2693 */
2694 case 'h':
2695 if (h-- <= 0)
2696 goto invalid;
2697 switch (*ptr++) {
2698 case 'h':
2699 if (h-- <= 0)
2700 goto invalid;
2701 switch (*ptr++) {
2702 case 'i':
2703 case 'd':
2704 case 'u':
2705 case 'o':
2706 case 'x':
2707 case 'X':
2708 return 0;
2709 default:
2710 goto invalid;
2711 }
2712 case 'i':
2713 case 'd':
2714 case 'u':
2715 case 'o':
2716 case 'x':
2717 case 'X':
2718 if (h == 0)
2719 return 0;
2720 /*FALLTHROUGH*/
2721 default:
2722 goto invalid;
2723 }
2724 #endif
2725 case 'c':
2726 if (h == 2)
2727 return 0;
2728 goto invalid;
2729 case 'i':
2730 case 'd':
2731 case 'u':
2732 case 'o':
2733 case 'x':
2734 case 'X':
2735 #ifdef STRICT_FORMAT
2736 if (h == 0)
2737 return 0;
2738 /*FALLTHROUGH*/
2739 #else
2740 return 0;
2741 #endif
2742 default:
2743 goto invalid;
2744 }
2745
2746 case FILE_FMT_FLOAT:
2747 case FILE_FMT_DOUBLE:
2748 if (*ptr == '-')
2749 ptr++;
2750 if (*ptr == '.')
2751 ptr++;
2752 CHECKLEN();
2753 if (*ptr == '.')
2754 ptr++;
2755 CHECKLEN();
2756 switch (*ptr++) {
2757 case 'e':
2758 case 'E':
2759 case 'f':
2760 case 'F':
2761 case 'g':
2762 case 'G':
2763 return 0;
2764
2765 default:
2766 goto invalid;
2767 }
2768
2769
2770 case FILE_FMT_STR:
2771 if (*ptr == '-')
2772 ptr++;
2773 while (isdigit(CAST(unsigned char, *ptr)))
2774 ptr++;
2775 if (*ptr == '.') {
2776 ptr++;
2777 while (isdigit(CAST(unsigned char , *ptr)))
2778 ptr++;
2779 }
2780
2781 switch (*ptr++) {
2782 case 's':
2783 return 0;
2784 default:
2785 goto invalid;
2786 }
2787
2788 default:
2789 /* internal error */
2790 fprintf(stderr, "Bad file format %d", type);
2791 abort();
2792 }
2793 invalid:
2794 *estr = "not valid";
2795 return -1;
2796 toolong:
2797 *estr = "too long";
2798 return -1;
2799 }
2800
2801 /*
2802 * Check that the optional printf format in description matches
2803 * the type of the magic.
2804 */
2805 file_private int
check_format(struct magic_set * ms,struct magic * m)2806 check_format(struct magic_set *ms, struct magic *m)
2807 {
2808 char *ptr;
2809 const char *estr;
2810
2811 for (ptr = m->desc; *ptr; ptr++)
2812 if (*ptr == '%')
2813 break;
2814 if (*ptr == '\0') {
2815 /* No format string; ok */
2816 return 1;
2817 }
2818
2819 assert(file_nformats == file_nnames);
2820
2821 if (m->type >= file_nformats) {
2822 file_magwarn(ms, "Internal error inconsistency between "
2823 "m->type and format strings");
2824 return -1;
2825 }
2826 if (file_formats[m->type] == FILE_FMT_NONE) {
2827 file_magwarn(ms, "No format string for `%s' with description "
2828 "`%s'", m->desc, file_names[m->type]);
2829 return -1;
2830 }
2831
2832 ptr++;
2833 if (check_format_type(ptr, m->type, &estr) == -1) {
2834 /*
2835 * TODO: this error message is unhelpful if the format
2836 * string is not one character long
2837 */
2838 file_magwarn(ms, "Printf format is %s for type "
2839 "`%s' in description `%s'", estr,
2840 file_names[m->type], m->desc);
2841 return -1;
2842 }
2843
2844 for (; *ptr; ptr++) {
2845 if (*ptr == '%') {
2846 file_magwarn(ms,
2847 "Too many format strings (should have at most one) "
2848 "for `%s' with description `%s'",
2849 file_names[m->type], m->desc);
2850 return -1;
2851 }
2852 }
2853 return 0;
2854 }
2855
2856 /*
2857 * Read a numeric value from a pointer, into the value union of a magic
2858 * pointer, according to the magic type. Update the string pointer to point
2859 * just after the number read. Return 0 for success, non-zero for failure.
2860 */
2861 file_private int
getvalue(struct magic_set * ms,struct magic * m,const char ** p,int action)2862 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action)
2863 {
2864 char *ep;
2865 uint64_t ull;
2866 int y;
2867
2868 switch (m->type) {
2869 case FILE_BESTRING16:
2870 case FILE_LESTRING16:
2871 case FILE_STRING:
2872 case FILE_PSTRING:
2873 case FILE_REGEX:
2874 case FILE_SEARCH:
2875 case FILE_NAME:
2876 case FILE_USE:
2877 case FILE_DER:
2878 case FILE_OCTAL:
2879 *p = getstr(ms, m, *p, action == FILE_COMPILE);
2880 if (*p == NULL) {
2881 if (ms->flags & MAGIC_CHECK)
2882 file_magwarn(ms, "cannot get string from `%s'",
2883 m->value.s);
2884 return -1;
2885 }
2886 if (m->type == FILE_REGEX) {
2887 file_regex_t rx;
2888 int rc =
2889 file_regcomp(ms, &rx, m->value.s, REG_EXTENDED);
2890 if (rc == 0) {
2891 file_regfree(&rx);
2892 }
2893 return rc ? -1 : 0;
2894 }
2895 return 0;
2896 default:
2897 if (m->reln == 'x')
2898 return 0;
2899 break;
2900 }
2901
2902 switch (m->type) {
2903 case FILE_FLOAT:
2904 case FILE_BEFLOAT:
2905 case FILE_LEFLOAT:
2906 errno = 0;
2907 #ifdef HAVE_STRTOF
2908 m->value.f = strtof(*p, &ep);
2909 #else
2910 m->value.f = (float)strtod(*p, &ep);
2911 #endif
2912 if (errno == 0)
2913 *p = ep;
2914 return 0;
2915 case FILE_DOUBLE:
2916 case FILE_BEDOUBLE:
2917 case FILE_LEDOUBLE:
2918 errno = 0;
2919 m->value.d = strtod(*p, &ep);
2920 if (errno == 0)
2921 *p = ep;
2922 return 0;
2923 case FILE_GUID:
2924 if (file_parse_guid(*p, m->value.guid) == -1)
2925 return -1;
2926 *p += FILE_GUID_SIZE - 1;
2927 return 0;
2928 default:
2929 errno = 0;
2930 ull = CAST(uint64_t, strtoull(*p, &ep, 0));
2931 m->value.q = file_signextend(ms, m, ull);
2932 if (*p == ep) {
2933 file_magwarn(ms, "Unparsable number `%s'", *p);
2934 return -1;
2935 } else {
2936 size_t ts = typesize(m->type);
2937 uint64_t x;
2938 const char *q;
2939
2940 if (ts == FILE_BADSIZE) {
2941 file_magwarn(ms,
2942 "Expected numeric type got `%s'",
2943 type_tbl[m->type].name);
2944 return -1;
2945 }
2946 for (q = *p; isspace(CAST(unsigned char, *q)); q++)
2947 continue;
2948 if (*q == '-' && ull != UINT64_MAX)
2949 ull = -CAST(int64_t, ull);
2950 switch (ts) {
2951 case 1:
2952 x = CAST(uint64_t, ull & ~0xffULL);
2953 y = (x & ~0xffULL) != ~0xffULL;
2954 break;
2955 case 2:
2956 x = CAST(uint64_t, ull & ~0xffffULL);
2957 y = (x & ~0xffffULL) != ~0xffffULL;
2958 break;
2959 case 4:
2960 x = CAST(uint64_t, ull & ~0xffffffffULL);
2961 y = (x & ~0xffffffffULL) != ~0xffffffffULL;
2962 break;
2963 case 8:
2964 x = 0;
2965 y = 0;
2966 break;
2967 default:
2968 fprintf(stderr, "Bad width %zu", ts);
2969 abort();
2970 }
2971 if (x && y) {
2972 file_magwarn(ms, "Overflow for numeric"
2973 " type `%s' value %#" PRIx64,
2974 type_tbl[m->type].name, ull);
2975 return -1;
2976 }
2977 }
2978 if (errno == 0) {
2979 *p = ep;
2980 eatsize(p);
2981 }
2982 return 0;
2983 }
2984 }
2985
2986 /*
2987 * Convert a string containing C character escapes. Stop at an unescaped
2988 * space or tab.
2989 * Copy the converted version to "m->value.s", and the length in m->vallen.
2990 * Return updated scan pointer as function result. Warn if set.
2991 */
2992 file_private const char *
getstr(struct magic_set * ms,struct magic * m,const char * s,int warn)2993 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn)
2994 {
2995 const char *origs = s;
2996 char *p = m->value.s;
2997 size_t plen = sizeof(m->value.s);
2998 char *origp = p;
2999 char *pmax = p + plen - 1;
3000 int c;
3001 int val;
3002 size_t bracket_nesting = 0;
3003
3004 while ((c = *s++) != '\0') {
3005 if (isspace(CAST(unsigned char, c)))
3006 break;
3007 if (p >= pmax) {
3008 file_error(ms, 0, "string too long: `%s'", origs);
3009 return NULL;
3010 }
3011 if (c != '\\') {
3012 if (c == '[') {
3013 bracket_nesting++;
3014 }
3015 if (c == ']' && bracket_nesting > 0) {
3016 bracket_nesting--;
3017 }
3018 *p++ = CAST(char, c);
3019 continue;
3020 }
3021 switch(c = *s++) {
3022
3023 case '\0':
3024 if (warn)
3025 file_magwarn(ms, "incomplete escape");
3026 s--;
3027 goto out;
3028 case '.':
3029 if (m->type == FILE_REGEX &&
3030 bracket_nesting == 0 && warn) {
3031 file_magwarn(ms, "escaped dot ('.') found, "
3032 "use \\\\. instead");
3033 }
3034 warn = 0; /* already did */
3035 /*FALLTHROUGH*/
3036 case '\t':
3037 if (warn) {
3038 file_magwarn(ms,
3039 "escaped tab found, use \\\\t instead");
3040 warn = 0; /* already did */
3041 }
3042 /*FALLTHROUGH*/
3043 default:
3044 if (warn) {
3045 if (isprint(CAST(unsigned char, c))) {
3046 /* Allow escaping of
3047 * ``relations'' */
3048 if (strchr("<>&^=!", c) == NULL
3049 && (m->type != FILE_REGEX ||
3050 strchr("[]().*?^$|{}", c)
3051 == NULL)) {
3052 file_magwarn(ms, "no "
3053 "need to escape "
3054 "`%c'", c);
3055 }
3056 } else {
3057 file_magwarn(ms,
3058 "unknown escape sequence: "
3059 "\\%03o", c);
3060 }
3061 }
3062 /*FALLTHROUGH*/
3063 /* space, perhaps force people to use \040? */
3064 case ' ':
3065 #if 0
3066 /*
3067 * Other things people escape, but shouldn't need to,
3068 * so we disallow them
3069 */
3070 case '\'':
3071 case '"':
3072 case '?':
3073 #endif
3074 /* Relations */
3075 case '>':
3076 case '<':
3077 case '&':
3078 case '^':
3079 case '=':
3080 case '!':
3081 /* and backslash itself */
3082 case '\\':
3083 *p++ = CAST(char, c);
3084 break;
3085
3086 case 'a':
3087 *p++ = '\a';
3088 break;
3089
3090 case 'b':
3091 *p++ = '\b';
3092 break;
3093
3094 case 'f':
3095 *p++ = '\f';
3096 break;
3097
3098 case 'n':
3099 *p++ = '\n';
3100 break;
3101
3102 case 'r':
3103 *p++ = '\r';
3104 break;
3105
3106 case 't':
3107 *p++ = '\t';
3108 break;
3109
3110 case 'v':
3111 *p++ = '\v';
3112 break;
3113
3114 /* \ and up to 3 octal digits */
3115 case '0':
3116 case '1':
3117 case '2':
3118 case '3':
3119 case '4':
3120 case '5':
3121 case '6':
3122 case '7':
3123 val = c - '0';
3124 c = *s++; /* try for 2 */
3125 if (c >= '0' && c <= '7') {
3126 val = (val << 3) | (c - '0');
3127 c = *s++; /* try for 3 */
3128 if (c >= '0' && c <= '7')
3129 val = (val << 3) | (c-'0');
3130 else
3131 --s;
3132 }
3133 else
3134 --s;
3135 *p++ = CAST(char, val);
3136 break;
3137
3138 /* \x and up to 2 hex digits */
3139 case 'x':
3140 val = 'x'; /* Default if no digits */
3141 c = hextoint(*s++); /* Get next char */
3142 if (c >= 0) {
3143 val = c;
3144 c = hextoint(*s++);
3145 if (c >= 0)
3146 val = (val << 4) + c;
3147 else
3148 --s;
3149 } else
3150 --s;
3151 *p++ = CAST(char, val);
3152 break;
3153 }
3154 }
3155 --s;
3156 out:
3157 *p = '\0';
3158 m->vallen = CAST(unsigned char, (p - origp));
3159 if (m->type == FILE_PSTRING) {
3160 size_t l = file_pstring_length_size(ms, m);
3161 if (l == FILE_BADSIZE)
3162 return NULL;
3163 m->vallen += CAST(unsigned char, l);
3164 }
3165 return s;
3166 }
3167
3168
3169 /* Single hex char to int; -1 if not a hex char. */
3170 file_private int
hextoint(int c)3171 hextoint(int c)
3172 {
3173 if (!isascii(CAST(unsigned char, c)))
3174 return -1;
3175 if (isdigit(CAST(unsigned char, c)))
3176 return c - '0';
3177 if ((c >= 'a') && (c <= 'f'))
3178 return c + 10 - 'a';
3179 if (( c>= 'A') && (c <= 'F'))
3180 return c + 10 - 'A';
3181 return -1;
3182 }
3183
3184
3185 /*
3186 * Print a string containing C character escapes.
3187 */
3188 file_protected void
file_showstr(FILE * fp,const char * s,size_t len)3189 file_showstr(FILE *fp, const char *s, size_t len)
3190 {
3191 char c;
3192
3193 for (;;) {
3194 if (len == FILE_BADSIZE) {
3195 c = *s++;
3196 if (c == '\0')
3197 break;
3198 }
3199 else {
3200 if (len-- == 0)
3201 break;
3202 c = *s++;
3203 }
3204 if (c >= 040 && c <= 0176) /* TODO isprint && !iscntrl */
3205 (void) fputc(c, fp);
3206 else {
3207 (void) fputc('\\', fp);
3208 switch (c) {
3209 case '\a':
3210 (void) fputc('a', fp);
3211 break;
3212
3213 case '\b':
3214 (void) fputc('b', fp);
3215 break;
3216
3217 case '\f':
3218 (void) fputc('f', fp);
3219 break;
3220
3221 case '\n':
3222 (void) fputc('n', fp);
3223 break;
3224
3225 case '\r':
3226 (void) fputc('r', fp);
3227 break;
3228
3229 case '\t':
3230 (void) fputc('t', fp);
3231 break;
3232
3233 case '\v':
3234 (void) fputc('v', fp);
3235 break;
3236
3237 default:
3238 (void) fprintf(fp, "%.3o", c & 0377);
3239 break;
3240 }
3241 }
3242 }
3243 }
3244
3245 /*
3246 * eatsize(): Eat the size spec from a number [eg. 10UL]
3247 */
3248 file_private void
eatsize(const char ** p)3249 eatsize(const char **p)
3250 {
3251 const char *l = *p;
3252
3253 if (LOWCASE(*l) == 'u')
3254 l++;
3255
3256 switch (LOWCASE(*l)) {
3257 case 'l': /* long */
3258 case 's': /* short */
3259 case 'h': /* short */
3260 case 'b': /* char/byte */
3261 case 'c': /* char/byte */
3262 l++;
3263 /*FALLTHROUGH*/
3264 default:
3265 break;
3266 }
3267
3268 *p = l;
3269 }
3270
3271 /*
3272 * handle a buffer containing a compiled file.
3273 */
3274 file_private struct magic_map *
apprentice_buf(struct magic_set * ms,struct magic * buf,size_t len)3275 apprentice_buf(struct magic_set *ms, struct magic *buf, size_t len)
3276 {
3277 struct magic_map *map;
3278
3279 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) {
3280 file_oomem(ms, sizeof(*map));
3281 return NULL;
3282 }
3283 map->len = len;
3284 map->p = buf;
3285 map->type = MAP_TYPE_USER;
3286 if (check_buffer(ms, map, "buffer") != 0) {
3287 apprentice_unmap(map);
3288 return NULL;
3289 }
3290 return map;
3291 }
3292
3293 /*
3294 * handle a compiled file.
3295 */
3296
3297 file_private struct magic_map *
apprentice_map(struct magic_set * ms,const char * fn)3298 apprentice_map(struct magic_set *ms, const char *fn)
3299 {
3300 int fd;
3301 struct stat st;
3302 char *dbname = NULL;
3303 struct magic_map *map;
3304 struct magic_map *rv = NULL;
3305
3306 fd = -1;
3307 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) {
3308 file_oomem(ms, sizeof(*map));
3309 goto error;
3310 }
3311 map->type = MAP_TYPE_USER; /* unspecified */
3312
3313 dbname = mkdbname(ms, fn, 0);
3314 if (dbname == NULL)
3315 goto error;
3316
3317 if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1)
3318 goto error;
3319
3320 if (fstat(fd, &st) == -1) {
3321 file_error(ms, errno, "cannot stat `%s'", dbname);
3322 goto error;
3323 }
3324 if (st.st_size < 8 || st.st_size > maxoff_t()) {
3325 file_error(ms, 0, "file `%s' is too %s", dbname,
3326 st.st_size < 8 ? "small" : "large");
3327 goto error;
3328 }
3329
3330 map->len = CAST(size_t, st.st_size);
3331 #ifdef QUICK
3332 map->type = MAP_TYPE_MMAP;
3333 if ((map->p = mmap(0, CAST(size_t, st.st_size), PROT_READ|PROT_WRITE,
3334 MAP_PRIVATE|MAP_FILE, fd, CAST(off_t, 0))) == MAP_FAILED) {
3335 file_error(ms, errno, "cannot map `%s'", dbname);
3336 goto error;
3337 }
3338 #else
3339 map->type = MAP_TYPE_MALLOC;
3340 if ((map->p = CAST(void *, malloc(map->len))) == NULL) {
3341 file_oomem(ms, map->len);
3342 goto error;
3343 }
3344 if (read(fd, map->p, map->len) != (ssize_t)map->len) {
3345 file_badread(ms);
3346 goto error;
3347 }
3348 #endif
3349 (void)close(fd);
3350 fd = -1;
3351
3352 if (check_buffer(ms, map, dbname) != 0) {
3353 goto error;
3354 }
3355 #ifdef QUICK
3356 if (mprotect(map->p, CAST(size_t, st.st_size), PROT_READ) == -1) {
3357 file_error(ms, errno, "cannot mprotect `%s'", dbname);
3358 goto error;
3359 }
3360 #endif
3361
3362 free(dbname);
3363 return map;
3364
3365 error:
3366 if (fd != -1)
3367 (void)close(fd);
3368 apprentice_unmap(map);
3369 free(dbname);
3370 return rv;
3371 }
3372
3373 file_private int
check_buffer(struct magic_set * ms,struct magic_map * map,const char * dbname)3374 check_buffer(struct magic_set *ms, struct magic_map *map, const char *dbname)
3375 {
3376 uint32_t *ptr;
3377 uint32_t entries, nentries;
3378 uint32_t version;
3379 int i, needsbyteswap;
3380
3381 entries = CAST(uint32_t, map->len / sizeof(struct magic));
3382 if (entries < MAGIC_SETS + 1) {
3383 file_error(ms, 0, "Too few magic entries %u in `%s'",
3384 entries, dbname);
3385 return -1;
3386 }
3387 if ((entries * sizeof(struct magic)) != map->len) {
3388 file_error(ms, 0, "Size of `%s' %" SIZE_T_FORMAT "u is not "
3389 "a multiple of %" SIZE_T_FORMAT "u",
3390 dbname, map->len, sizeof(struct magic));
3391 return -1;
3392 }
3393
3394 ptr = CAST(uint32_t *, map->p);
3395 if (*ptr != MAGICNO) {
3396 if (swap4(*ptr) != MAGICNO) {
3397 file_error(ms, 0, "bad magic in `%s'", dbname);
3398 return -1;
3399 }
3400 needsbyteswap = 1;
3401 } else
3402 needsbyteswap = 0;
3403 if (needsbyteswap)
3404 version = swap4(ptr[1]);
3405 else
3406 version = ptr[1];
3407 if (version != VERSIONNO) {
3408 file_error(ms, 0, "File %s supports only version %d magic "
3409 "files. `%s' is version %d", VERSION,
3410 VERSIONNO, dbname, version);
3411 return -1;
3412 }
3413 map->magic[0] = CAST(struct magic *, map->p) + 1;
3414 nentries = 0;
3415 for (i = 0; i < MAGIC_SETS; i++) {
3416 if (needsbyteswap)
3417 map->nmagic[i] = swap4(ptr[i + 2]);
3418 else
3419 map->nmagic[i] = ptr[i + 2];
3420 if (i != MAGIC_SETS - 1)
3421 map->magic[i + 1] = map->magic[i] + map->nmagic[i];
3422 nentries += map->nmagic[i];
3423 }
3424 if (entries != nentries + 1) {
3425 file_error(ms, 0, "Inconsistent entries in `%s' %u != %u",
3426 dbname, entries, nentries + 1);
3427 return -1;
3428 }
3429 if (needsbyteswap)
3430 for (i = 0; i < MAGIC_SETS; i++)
3431 byteswap(map->magic[i], map->nmagic[i]);
3432 return 0;
3433 }
3434
3435 /*
3436 * handle an mmaped file.
3437 */
3438 file_private int
apprentice_compile(struct magic_set * ms,struct magic_map * map,const char * fn)3439 apprentice_compile(struct magic_set *ms, struct magic_map *map, const char *fn)
3440 {
3441 static const size_t nm = sizeof(*map->nmagic) * MAGIC_SETS;
3442 static const size_t m = sizeof(**map->magic);
3443 int fd = -1;
3444 size_t len;
3445 char *dbname;
3446 int rv = -1;
3447 uint32_t i;
3448 union {
3449 struct magic m;
3450 uint32_t h[2 + MAGIC_SETS];
3451 } hdr;
3452
3453 dbname = mkdbname(ms, fn, 1);
3454
3455 if (dbname == NULL)
3456 goto out;
3457
3458 if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1)
3459 {
3460 file_error(ms, errno, "cannot open `%s'", dbname);
3461 goto out;
3462 }
3463 memset(&hdr, 0, sizeof(hdr));
3464 hdr.h[0] = MAGICNO;
3465 hdr.h[1] = VERSIONNO;
3466 memcpy(hdr.h + 2, map->nmagic, nm);
3467
3468 if (write(fd, &hdr, sizeof(hdr)) != CAST(ssize_t, sizeof(hdr))) {
3469 file_error(ms, errno, "error writing `%s'", dbname);
3470 goto out2;
3471 }
3472
3473 for (i = 0; i < MAGIC_SETS; i++) {
3474 len = m * map->nmagic[i];
3475 if (write(fd, map->magic[i], len) != CAST(ssize_t, len)) {
3476 file_error(ms, errno, "error writing `%s'", dbname);
3477 goto out2;
3478 }
3479 }
3480
3481 rv = 0;
3482 out2:
3483 if (fd != -1)
3484 (void)close(fd);
3485 out:
3486 apprentice_unmap(map);
3487 free(dbname);
3488 return rv;
3489 }
3490
3491 file_private const char ext[] = ".mgc";
3492 /*
3493 * make a dbname
3494 */
3495 file_private char *
mkdbname(struct magic_set * ms,const char * fn,int strip)3496 mkdbname(struct magic_set *ms, const char *fn, int strip)
3497 {
3498 const char *p, *q;
3499 char *buf;
3500
3501 if (strip) {
3502 if ((p = strrchr(fn, '/')) != NULL)
3503 fn = ++p;
3504 }
3505
3506 for (q = fn; *q; q++)
3507 continue;
3508 /* Look for .mgc */
3509 for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--)
3510 if (*p != *q)
3511 break;
3512
3513 /* Did not find .mgc, restore q */
3514 if (p >= ext)
3515 for (q = fn; *q; q++)
3516 continue;
3517
3518 q++;
3519 /* Compatibility with old code that looked in .mime */
3520 if (ms->flags & MAGIC_MIME) {
3521 if (asprintf(&buf, "%.*s.mime%s", CAST(int, q - fn), fn, ext)
3522 < 0)
3523 return NULL;
3524 if (access(buf, R_OK) != -1) {
3525 ms->flags &= MAGIC_MIME_TYPE;
3526 return buf;
3527 }
3528 free(buf);
3529 }
3530 if (asprintf(&buf, "%.*s%s", CAST(int, q - fn), fn, ext) < 0)
3531 return NULL;
3532
3533 /* Compatibility with old code that looked in .mime */
3534 if (strstr(fn, ".mime") != NULL)
3535 ms->flags &= MAGIC_MIME_TYPE;
3536 return buf;
3537 }
3538
3539 /*
3540 * Byteswap an mmap'ed file if needed
3541 */
3542 file_private void
byteswap(struct magic * magic,uint32_t nmagic)3543 byteswap(struct magic *magic, uint32_t nmagic)
3544 {
3545 uint32_t i;
3546 for (i = 0; i < nmagic; i++)
3547 bs1(&magic[i]);
3548 }
3549
3550 #if !defined(HAVE_BYTESWAP_H) && !defined(HAVE_SYS_BSWAP_H)
3551 /*
3552 * swap a short
3553 */
3554 file_private uint16_t
swap2(uint16_t sv)3555 swap2(uint16_t sv)
3556 {
3557 uint16_t rv;
3558 uint8_t *s = RCAST(uint8_t *, RCAST(void *, &sv));
3559 uint8_t *d = RCAST(uint8_t *, RCAST(void *, &rv));
3560 d[0] = s[1];
3561 d[1] = s[0];
3562 return rv;
3563 }
3564
3565 /*
3566 * swap an int
3567 */
3568 file_private uint32_t
swap4(uint32_t sv)3569 swap4(uint32_t sv)
3570 {
3571 uint32_t rv;
3572 uint8_t *s = RCAST(uint8_t *, RCAST(void *, &sv));
3573 uint8_t *d = RCAST(uint8_t *, RCAST(void *, &rv));
3574 d[0] = s[3];
3575 d[1] = s[2];
3576 d[2] = s[1];
3577 d[3] = s[0];
3578 return rv;
3579 }
3580
3581 /*
3582 * swap a quad
3583 */
3584 file_private uint64_t
swap8(uint64_t sv)3585 swap8(uint64_t sv)
3586 {
3587 uint64_t rv;
3588 uint8_t *s = RCAST(uint8_t *, RCAST(void *, &sv));
3589 uint8_t *d = RCAST(uint8_t *, RCAST(void *, &rv));
3590 # if 0
3591 d[0] = s[3];
3592 d[1] = s[2];
3593 d[2] = s[1];
3594 d[3] = s[0];
3595 d[4] = s[7];
3596 d[5] = s[6];
3597 d[6] = s[5];
3598 d[7] = s[4];
3599 # else
3600 d[0] = s[7];
3601 d[1] = s[6];
3602 d[2] = s[5];
3603 d[3] = s[4];
3604 d[4] = s[3];
3605 d[5] = s[2];
3606 d[6] = s[1];
3607 d[7] = s[0];
3608 # endif
3609 return rv;
3610 }
3611 #endif
3612
3613 file_protected uintmax_t
file_varint2uintmax_t(const unsigned char * us,int t,size_t * l)3614 file_varint2uintmax_t(const unsigned char *us, int t, size_t *l)
3615 {
3616 uintmax_t x = 0;
3617 const unsigned char *c;
3618 if (t == FILE_LEVARINT) {
3619 for (c = us; *c; c++) {
3620 if ((*c & 0x80) == 0)
3621 break;
3622 }
3623 if (l)
3624 *l = c - us + 1;
3625 for (; c >= us; c--) {
3626 x |= *c & 0x7f;
3627 x <<= 7;
3628 }
3629 } else {
3630 for (c = us; *c; c++) {
3631 x |= *c & 0x7f;
3632 if ((*c & 0x80) == 0)
3633 break;
3634 x <<= 7;
3635 }
3636 if (l)
3637 *l = c - us + 1;
3638 }
3639 return x;
3640 }
3641
3642
3643 /*
3644 * byteswap a single magic entry
3645 */
3646 file_private void
bs1(struct magic * m)3647 bs1(struct magic *m)
3648 {
3649 m->cont_level = swap2(m->cont_level);
3650 m->offset = swap4(CAST(uint32_t, m->offset));
3651 m->in_offset = swap4(CAST(uint32_t, m->in_offset));
3652 m->lineno = swap4(CAST(uint32_t, m->lineno));
3653 if (IS_STRING(m->type)) {
3654 m->str_range = swap4(m->str_range);
3655 m->str_flags = swap4(m->str_flags);
3656 }
3657 else {
3658 m->value.q = swap8(m->value.q);
3659 m->num_mask = swap8(m->num_mask);
3660 }
3661 }
3662
3663 file_protected size_t
file_pstring_length_size(struct magic_set * ms,const struct magic * m)3664 file_pstring_length_size(struct magic_set *ms, const struct magic *m)
3665 {
3666 switch (m->str_flags & PSTRING_LEN) {
3667 case PSTRING_1_LE:
3668 return 1;
3669 case PSTRING_2_LE:
3670 case PSTRING_2_BE:
3671 return 2;
3672 case PSTRING_4_LE:
3673 case PSTRING_4_BE:
3674 return 4;
3675 default:
3676 file_error(ms, 0, "corrupt magic file "
3677 "(bad pascal string length %d)",
3678 m->str_flags & PSTRING_LEN);
3679 return FILE_BADSIZE;
3680 }
3681 }
3682 file_protected size_t
file_pstring_get_length(struct magic_set * ms,const struct magic * m,const char * ss)3683 file_pstring_get_length(struct magic_set *ms, const struct magic *m,
3684 const char *ss)
3685 {
3686 size_t len = 0;
3687 const unsigned char *s = RCAST(const unsigned char *, ss);
3688 unsigned int s3, s2, s1, s0;
3689
3690 switch (m->str_flags & PSTRING_LEN) {
3691 case PSTRING_1_LE:
3692 len = *s;
3693 break;
3694 case PSTRING_2_LE:
3695 s0 = s[0];
3696 s1 = s[1];
3697 len = (s1 << 8) | s0;
3698 break;
3699 case PSTRING_2_BE:
3700 s0 = s[0];
3701 s1 = s[1];
3702 len = (s0 << 8) | s1;
3703 break;
3704 case PSTRING_4_LE:
3705 s0 = s[0];
3706 s1 = s[1];
3707 s2 = s[2];
3708 s3 = s[3];
3709 len = (s3 << 24) | (s2 << 16) | (s1 << 8) | s0;
3710 break;
3711 case PSTRING_4_BE:
3712 s0 = s[0];
3713 s1 = s[1];
3714 s2 = s[2];
3715 s3 = s[3];
3716 len = (s0 << 24) | (s1 << 16) | (s2 << 8) | s3;
3717 break;
3718 default:
3719 file_error(ms, 0, "corrupt magic file "
3720 "(bad pascal string length %d)",
3721 m->str_flags & PSTRING_LEN);
3722 return FILE_BADSIZE;
3723 }
3724
3725 if (m->str_flags & PSTRING_LENGTH_INCLUDES_ITSELF) {
3726 size_t l = file_pstring_length_size(ms, m);
3727 if (l == FILE_BADSIZE)
3728 return l;
3729 len -= l;
3730 }
3731
3732 return len;
3733 }
3734
3735 file_protected int
file_magicfind(struct magic_set * ms,const char * name,struct mlist * v)3736 file_magicfind(struct magic_set *ms, const char *name, struct mlist *v)
3737 {
3738 uint32_t i, j;
3739 struct mlist *mlist, *ml;
3740
3741 mlist = ms->mlist[1];
3742
3743 for (ml = mlist->next; ml != mlist; ml = ml->next) {
3744 struct magic *ma = ml->magic;
3745 for (i = 0; i < ml->nmagic; i++) {
3746 if (ma[i].type != FILE_NAME)
3747 continue;
3748 if (strcmp(ma[i].value.s, name) == 0) {
3749 v->magic = &ma[i];
3750 v->magic_rxcomp = &(ml->magic_rxcomp[i]);
3751 for (j = i + 1; j < ml->nmagic; j++)
3752 if (ma[j].cont_level == 0)
3753 break;
3754 v->nmagic = j - i;
3755 return 0;
3756 }
3757 }
3758 }
3759 return -1;
3760 }
3761