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 1.158 2009/10/19 13:10:20 christos Exp $")
36 #endif /* lint */
37
38 #include "magic.h"
39 #include "patchlevel.h"
40 #include <stdlib.h>
41 #ifdef HAVE_UNISTD_H
42 #include <unistd.h>
43 #endif
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
53 #define EATAB {while (isascii((unsigned char) *l) && \
54 isspace((unsigned char) *l)) ++l;}
55 #define LOWCASE(l) (isupper((unsigned char) (l)) ? \
56 tolower((unsigned char) (l)) : (l))
57 /*
58 * Work around a bug in headers on Digital Unix.
59 * At least confirmed for: OSF1 V4.0 878
60 */
61 #if defined(__osf__) && defined(__DECC)
62 #ifdef MAP_FAILED
63 #undef MAP_FAILED
64 #endif
65 #endif
66
67 #ifndef MAP_FAILED
68 #define MAP_FAILED (void *) -1
69 #endif
70
71 #ifndef MAP_FILE
72 #define MAP_FILE 0
73 #endif
74
75 #ifndef MAXPATHLEN
76 #define MAXPATHLEN 1024
77 #endif
78
79 struct magic_entry {
80 struct magic *mp;
81 uint32_t cont_count;
82 uint32_t max_count;
83 };
84
85 int file_formats[FILE_NAMES_SIZE];
86 const size_t file_nformats = FILE_NAMES_SIZE;
87 const char *file_names[FILE_NAMES_SIZE];
88 const size_t file_nnames = FILE_NAMES_SIZE;
89
90 private int getvalue(struct magic_set *ms, struct magic *, const char **, int);
91 private int hextoint(int);
92 private const char *getstr(struct magic_set *, struct magic *, const char *,
93 int);
94 private int parse(struct magic_set *, struct magic_entry **, uint32_t *,
95 const char *, size_t, int);
96 private void eatsize(const char **);
97 private int apprentice_1(struct magic_set *, const char *, int, struct mlist *);
98 private size_t apprentice_magic_strength(const struct magic *);
99 private int apprentice_sort(const void *, const void *);
100 private int apprentice_load(struct magic_set *, struct magic **, uint32_t *,
101 const char *, int);
102 private void byteswap(struct magic *, uint32_t);
103 private void bs1(struct magic *);
104 private uint16_t swap2(uint16_t);
105 private uint32_t swap4(uint32_t);
106 private uint64_t swap8(uint64_t);
107 private char *mkdbname(struct magic_set *, const char *, int);
108 private int apprentice_map(struct magic_set *, struct magic **, uint32_t *,
109 const char *);
110 private int apprentice_compile(struct magic_set *, struct magic **, uint32_t *,
111 const char *);
112 private int check_format_type(const char *, int);
113 private int check_format(struct magic_set *, struct magic *);
114 private int get_op(char);
115 private int parse_mime(struct magic_set *, struct magic_entry *, const char *);
116 private int parse_strength(struct magic_set *, struct magic_entry *, const char *);
117 private int parse_apple(struct magic_set *, struct magic_entry *, const char *);
118
119
120 private size_t maxmagic = 0;
121 private size_t magicsize = sizeof(struct magic);
122
123 private const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc";
124
125 private struct {
126 const char *name;
127 size_t len;
128 int (*fun)(struct magic_set *, struct magic_entry *, const char *);
129 } bang[] = {
130 #define DECLARE_FIELD(name) { # name, sizeof(# name) - 1, parse_ ## name }
131 DECLARE_FIELD(mime),
132 DECLARE_FIELD(apple),
133 DECLARE_FIELD(strength),
134 #undef DECLARE_FIELD
135 { NULL, 0, NULL }
136 };
137
138 #ifdef COMPILE_ONLY
139
140 int main(int, char *[]);
141
142 int
main(int argc,char * argv[])143 main(int argc, char *argv[])
144 {
145 int ret;
146 struct magic_set *ms;
147 char *progname;
148
149 if ((progname = strrchr(argv[0], '/')) != NULL)
150 progname++;
151 else
152 progname = argv[0];
153
154 if (argc != 2) {
155 (void)fprintf(stderr, "Usage: %s file\n", progname);
156 return 1;
157 }
158
159 if ((ms = magic_open(MAGIC_CHECK)) == NULL) {
160 (void)fprintf(stderr, "%s: %s\n", progname, strerror(errno));
161 return 1;
162 }
163 ret = magic_compile(ms, argv[1]) == -1 ? 1 : 0;
164 if (ret == 1)
165 (void)fprintf(stderr, "%s: %s\n", progname, magic_error(ms));
166 magic_close(ms);
167 return ret;
168 }
169 #endif /* COMPILE_ONLY */
170
171 static const struct type_tbl_s {
172 const char name[16];
173 const size_t len;
174 const int type;
175 const int format;
176 } type_tbl[] = {
177 # define XX(s) s, (sizeof(s) - 1)
178 # define XX_NULL "", 0
179 { XX("byte"), FILE_BYTE, FILE_FMT_NUM },
180 { XX("short"), FILE_SHORT, FILE_FMT_NUM },
181 { XX("default"), FILE_DEFAULT, FILE_FMT_STR },
182 { XX("long"), FILE_LONG, FILE_FMT_NUM },
183 { XX("string"), FILE_STRING, FILE_FMT_STR },
184 { XX("date"), FILE_DATE, FILE_FMT_STR },
185 { XX("beshort"), FILE_BESHORT, FILE_FMT_NUM },
186 { XX("belong"), FILE_BELONG, FILE_FMT_NUM },
187 { XX("bedate"), FILE_BEDATE, FILE_FMT_STR },
188 { XX("leshort"), FILE_LESHORT, FILE_FMT_NUM },
189 { XX("lelong"), FILE_LELONG, FILE_FMT_NUM },
190 { XX("ledate"), FILE_LEDATE, FILE_FMT_STR },
191 { XX("pstring"), FILE_PSTRING, FILE_FMT_STR },
192 { XX("ldate"), FILE_LDATE, FILE_FMT_STR },
193 { XX("beldate"), FILE_BELDATE, FILE_FMT_STR },
194 { XX("leldate"), FILE_LELDATE, FILE_FMT_STR },
195 { XX("regex"), FILE_REGEX, FILE_FMT_STR },
196 { XX("bestring16"), FILE_BESTRING16, FILE_FMT_STR },
197 { XX("lestring16"), FILE_LESTRING16, FILE_FMT_STR },
198 { XX("search"), FILE_SEARCH, FILE_FMT_STR },
199 { XX("medate"), FILE_MEDATE, FILE_FMT_STR },
200 { XX("meldate"), FILE_MELDATE, FILE_FMT_STR },
201 { XX("melong"), FILE_MELONG, FILE_FMT_NUM },
202 { XX("quad"), FILE_QUAD, FILE_FMT_QUAD },
203 { XX("lequad"), FILE_LEQUAD, FILE_FMT_QUAD },
204 { XX("bequad"), FILE_BEQUAD, FILE_FMT_QUAD },
205 { XX("qdate"), FILE_QDATE, FILE_FMT_STR },
206 { XX("leqdate"), FILE_LEQDATE, FILE_FMT_STR },
207 { XX("beqdate"), FILE_BEQDATE, FILE_FMT_STR },
208 { XX("qldate"), FILE_QLDATE, FILE_FMT_STR },
209 { XX("leqldate"), FILE_LEQLDATE, FILE_FMT_STR },
210 { XX("beqldate"), FILE_BEQLDATE, FILE_FMT_STR },
211 { XX("float"), FILE_FLOAT, FILE_FMT_FLOAT },
212 { XX("befloat"), FILE_BEFLOAT, FILE_FMT_FLOAT },
213 { XX("lefloat"), FILE_LEFLOAT, FILE_FMT_FLOAT },
214 { XX("double"), FILE_DOUBLE, FILE_FMT_DOUBLE },
215 { XX("bedouble"), FILE_BEDOUBLE, FILE_FMT_DOUBLE },
216 { XX("ledouble"), FILE_LEDOUBLE, FILE_FMT_DOUBLE },
217 { XX("leid3"), FILE_LEID3, FILE_FMT_NUM },
218 { XX("beid3"), FILE_BEID3, FILE_FMT_NUM },
219 { XX("indirect"), FILE_INDIRECT, FILE_FMT_NONE },
220 { XX_NULL, FILE_INVALID, FILE_FMT_NONE },
221 # undef XX
222 # undef XX_NULL
223 };
224
225 private int
get_type(const char * l,const char ** t)226 get_type(const char *l, const char **t)
227 {
228 const struct type_tbl_s *p;
229
230 for (p = type_tbl; p->len; p++) {
231 if (strncmp(l, p->name, p->len) == 0) {
232 if (t)
233 *t = l + p->len;
234 break;
235 }
236 }
237 return p->type;
238 }
239
240 private void
init_file_tables(void)241 init_file_tables(void)
242 {
243 static int done = 0;
244 const struct type_tbl_s *p;
245
246 if (done)
247 return;
248 done++;
249
250 for (p = type_tbl; p->len; p++) {
251 assert(p->type < FILE_NAMES_SIZE);
252 file_names[p->type] = p->name;
253 file_formats[p->type] = p->format;
254 }
255 }
256
257 /*
258 * Handle one file or directory.
259 */
260 private int
apprentice_1(struct magic_set * ms,const char * fn,int action,struct mlist * mlist)261 apprentice_1(struct magic_set *ms, const char *fn, int action,
262 struct mlist *mlist)
263 {
264 struct magic *magic = NULL;
265 uint32_t nmagic = 0;
266 struct mlist *ml;
267 int rv = -1;
268 int mapped;
269
270 if (magicsize != FILE_MAGICSIZE) {
271 file_error(ms, 0, "magic element size %lu != %lu",
272 (unsigned long)sizeof(*magic),
273 (unsigned long)FILE_MAGICSIZE);
274 return -1;
275 }
276
277 if (action == FILE_COMPILE) {
278 rv = apprentice_load(ms, &magic, &nmagic, fn, action);
279 if (rv != 0)
280 return -1;
281 rv = apprentice_compile(ms, &magic, &nmagic, fn);
282 free(magic);
283 return rv;
284 }
285
286 #ifndef COMPILE_ONLY
287 if ((rv = apprentice_map(ms, &magic, &nmagic, fn)) == -1) {
288 if (ms->flags & MAGIC_CHECK)
289 file_magwarn(ms, "using regular magic file `%s'", fn);
290 rv = apprentice_load(ms, &magic, &nmagic, fn, action);
291 if (rv != 0)
292 return -1;
293 }
294
295 mapped = rv;
296
297 if (magic == NULL) {
298 file_delmagic(magic, mapped, nmagic);
299 return -1;
300 }
301
302 if ((ml = CAST(struct mlist *, malloc(sizeof(*ml)))) == NULL) {
303 file_delmagic(magic, mapped, nmagic);
304 file_oomem(ms, sizeof(*ml));
305 return -1;
306 }
307
308 ml->magic = magic;
309 ml->nmagic = nmagic;
310 ml->mapped = mapped;
311
312 mlist->prev->next = ml;
313 ml->prev = mlist->prev;
314 ml->next = mlist;
315 mlist->prev = ml;
316
317 return 0;
318 #endif /* COMPILE_ONLY */
319 }
320
321 protected void
file_delmagic(struct magic * p,int type,size_t entries)322 file_delmagic(struct magic *p, int type, size_t entries)
323 {
324 if (p == NULL)
325 return;
326 switch (type) {
327 case 2:
328 #ifdef QUICK
329 p--;
330 (void)munmap((void *)p, sizeof(*p) * (entries + 1));
331 break;
332 #else
333 (void)&entries;
334 abort();
335 /*NOTREACHED*/
336 #endif
337 case 1:
338 p--;
339 /*FALLTHROUGH*/
340 case 0:
341 free(p);
342 break;
343 default:
344 abort();
345 }
346 }
347
348 /* const char *fn: list of magic files and directories */
349 protected struct mlist *
file_apprentice(struct magic_set * ms,const char * fn,int action)350 file_apprentice(struct magic_set *ms, const char *fn, int action)
351 {
352 char *p, *mfn;
353 int file_err, errs = -1;
354 struct mlist *mlist;
355
356 if ((fn = magic_getpath(fn, action)) == NULL)
357 return NULL;
358
359 init_file_tables();
360
361 if ((mfn = strdup(fn)) == NULL) {
362 file_oomem(ms, strlen(fn));
363 return NULL;
364 }
365 fn = mfn;
366
367 if ((mlist = CAST(struct mlist *, malloc(sizeof(*mlist)))) == NULL) {
368 free(mfn);
369 file_oomem(ms, sizeof(*mlist));
370 return NULL;
371 }
372 mlist->next = mlist->prev = mlist;
373
374 while (fn) {
375 p = strchr(fn, PATHSEP);
376 if (p)
377 *p++ = '\0';
378 if (*fn == '\0')
379 break;
380 file_err = apprentice_1(ms, fn, action, mlist);
381 errs = MAX(errs, file_err);
382 fn = p;
383 }
384 if (errs == -1) {
385 free(mfn);
386 free(mlist);
387 mlist = NULL;
388 file_error(ms, 0, "could not find any magic files!");
389 return NULL;
390 }
391 free(mfn);
392 return mlist;
393 }
394
395 /*
396 * Get weight of this magic entry, for sorting purposes.
397 */
398 private size_t
apprentice_magic_strength(const struct magic * m)399 apprentice_magic_strength(const struct magic *m)
400 {
401 #define MULT 10
402 size_t val = 2 * MULT; /* baseline strength */
403
404 switch (m->type) {
405 case FILE_DEFAULT: /* make sure this sorts last */
406 if (m->factor_op != FILE_FACTOR_OP_NONE)
407 abort();
408 return 0;
409
410 case FILE_BYTE:
411 val += 1 * MULT;
412 break;
413
414 case FILE_SHORT:
415 case FILE_LESHORT:
416 case FILE_BESHORT:
417 val += 2 * MULT;
418 break;
419
420 case FILE_LONG:
421 case FILE_LELONG:
422 case FILE_BELONG:
423 case FILE_MELONG:
424 val += 4 * MULT;
425 break;
426
427 case FILE_PSTRING:
428 case FILE_STRING:
429 val += m->vallen * MULT;
430 break;
431
432 case FILE_BESTRING16:
433 case FILE_LESTRING16:
434 val += m->vallen * MULT / 2;
435 break;
436
437 case FILE_SEARCH:
438 case FILE_REGEX:
439 val += m->vallen * MAX(MULT / m->vallen, 1);
440 break;
441
442 case FILE_DATE:
443 case FILE_LEDATE:
444 case FILE_BEDATE:
445 case FILE_MEDATE:
446 case FILE_LDATE:
447 case FILE_LELDATE:
448 case FILE_BELDATE:
449 case FILE_MELDATE:
450 case FILE_FLOAT:
451 case FILE_BEFLOAT:
452 case FILE_LEFLOAT:
453 val += 4 * MULT;
454 break;
455
456 case FILE_QUAD:
457 case FILE_BEQUAD:
458 case FILE_LEQUAD:
459 case FILE_QDATE:
460 case FILE_LEQDATE:
461 case FILE_BEQDATE:
462 case FILE_QLDATE:
463 case FILE_LEQLDATE:
464 case FILE_BEQLDATE:
465 case FILE_DOUBLE:
466 case FILE_BEDOUBLE:
467 case FILE_LEDOUBLE:
468 val += 8 * MULT;
469 break;
470
471 default:
472 val = 0;
473 (void)fprintf(stderr, "Bad type %d\n", m->type);
474 abort();
475 }
476
477 switch (m->reln) {
478 case 'x': /* matches anything penalize */
479 case '!': /* matches almost anything penalize */
480 val = 0;
481 break;
482
483 case '=': /* Exact match, prefer */
484 val += MULT;
485 break;
486
487 case '>':
488 case '<': /* comparison match reduce strength */
489 val -= 2 * MULT;
490 break;
491
492 case '^':
493 case '&': /* masking bits, we could count them too */
494 val -= MULT;
495 break;
496
497 default:
498 (void)fprintf(stderr, "Bad relation %c\n", m->reln);
499 abort();
500 }
501
502 if (val == 0) /* ensure we only return 0 for FILE_DEFAULT */
503 val = 1;
504
505 switch (m->factor_op) {
506 case FILE_FACTOR_OP_NONE:
507 break;
508 case FILE_FACTOR_OP_PLUS:
509 val += m->factor;
510 break;
511 case FILE_FACTOR_OP_MINUS:
512 val -= m->factor;
513 break;
514 case FILE_FACTOR_OP_TIMES:
515 val *= m->factor;
516 break;
517 case FILE_FACTOR_OP_DIV:
518 val /= m->factor;
519 break;
520 default:
521 abort();
522 }
523
524 /*
525 * Magic entries with no description get a bonus because they depend
526 * on subsequent magic entries to print something.
527 */
528 if (m->desc[0] == '\0')
529 val++;
530 return val;
531 }
532
533 /*
534 * Sort callback for sorting entries by "strength" (basically length)
535 */
536 private int
apprentice_sort(const void * a,const void * b)537 apprentice_sort(const void *a, const void *b)
538 {
539 const struct magic_entry *ma = CAST(const struct magic_entry *, a);
540 const struct magic_entry *mb = CAST(const struct magic_entry *, b);
541 size_t sa = apprentice_magic_strength(ma->mp);
542 size_t sb = apprentice_magic_strength(mb->mp);
543 if (sa == sb)
544 return 0;
545 else if (sa > sb)
546 return -1;
547 else
548 return 1;
549 }
550
551 private void
set_test_type(struct magic * mstart,struct magic * m)552 set_test_type(struct magic *mstart, struct magic *m)
553 {
554 switch (m->type) {
555 case FILE_BYTE:
556 case FILE_SHORT:
557 case FILE_LONG:
558 case FILE_DATE:
559 case FILE_BESHORT:
560 case FILE_BELONG:
561 case FILE_BEDATE:
562 case FILE_LESHORT:
563 case FILE_LELONG:
564 case FILE_LEDATE:
565 case FILE_LDATE:
566 case FILE_BELDATE:
567 case FILE_LELDATE:
568 case FILE_MEDATE:
569 case FILE_MELDATE:
570 case FILE_MELONG:
571 case FILE_QUAD:
572 case FILE_LEQUAD:
573 case FILE_BEQUAD:
574 case FILE_QDATE:
575 case FILE_LEQDATE:
576 case FILE_BEQDATE:
577 case FILE_QLDATE:
578 case FILE_LEQLDATE:
579 case FILE_BEQLDATE:
580 case FILE_FLOAT:
581 case FILE_BEFLOAT:
582 case FILE_LEFLOAT:
583 case FILE_DOUBLE:
584 case FILE_BEDOUBLE:
585 case FILE_LEDOUBLE:
586 case FILE_STRING:
587 case FILE_PSTRING:
588 case FILE_BESTRING16:
589 case FILE_LESTRING16:
590 /* binary test, set flag */
591 mstart->flag |= BINTEST;
592 break;
593 case FILE_REGEX:
594 case FILE_SEARCH:
595 /* Check for override */
596 if (mstart->str_flags & STRING_BINTEST)
597 mstart->flag |= BINTEST;
598 if (mstart->str_flags & STRING_TEXTTEST)
599 mstart->flag |= TEXTTEST;
600
601 if (mstart->flag & (TEXTTEST|BINTEST))
602 break;
603
604 /* binary test if pattern is not text */
605 if (file_looks_utf8(m->value.us, (size_t)m->vallen, NULL,
606 NULL) <= 0)
607 mstart->flag |= BINTEST;
608 else
609 mstart->flag |= TEXTTEST;
610 break;
611 case FILE_DEFAULT:
612 /* can't deduce anything; we shouldn't see this at the
613 top level anyway */
614 break;
615 case FILE_INVALID:
616 default:
617 /* invalid search type, but no need to complain here */
618 break;
619 }
620 }
621
622 /*
623 * Load and parse one file.
624 */
625 private void
load_1(struct magic_set * ms,int action,const char * fn,int * errs,struct magic_entry ** marray,uint32_t * marraycount)626 load_1(struct magic_set *ms, int action, const char *fn, int *errs,
627 struct magic_entry **marray, uint32_t *marraycount)
628 {
629 char line[BUFSIZ];
630 size_t lineno = 0;
631 FILE *f = fopen(ms->file = fn, "r");
632 if (f == NULL) {
633 if (errno != ENOENT)
634 file_error(ms, errno, "cannot read magic file `%s'",
635 fn);
636 (*errs)++;
637 } else {
638 /* read and parse this file */
639 for (ms->line = 1;
640 fgets(line, CAST(int, sizeof(line)), f) != NULL;
641 ms->line++) {
642 size_t len;
643 len = strlen(line);
644 if (len == 0) /* null line, garbage, etc */
645 continue;
646 if (line[len - 1] == '\n') {
647 lineno++;
648 line[len - 1] = '\0'; /* delete newline */
649 }
650 if (line[0] == '\0') /* empty, do not parse */
651 continue;
652 if (line[0] == '#') /* comment, do not parse */
653 continue;
654 if (line[0] == '!' && line[1] == ':') {
655 size_t i;
656
657 for (i = 0; bang[i].name != NULL; i++) {
658 if (len - 2 > bang[i].len &&
659 memcmp(bang[i].name, line + 2,
660 bang[i].len) == 0)
661 break;
662 }
663 if (bang[i].name == NULL) {
664 file_error(ms, 0,
665 "Unknown !: entry `%s'", line);
666 (*errs)++;
667 continue;
668 }
669 if (*marraycount == 0) {
670 file_error(ms, 0,
671 "No current entry for :!%s type",
672 bang[i].name);
673 (*errs)++;
674 continue;
675 }
676 if ((*bang[i].fun)(ms,
677 &(*marray)[*marraycount - 1],
678 line + bang[i].len + 2) != 0) {
679 (*errs)++;
680 continue;
681 }
682 continue;
683 }
684 if (parse(ms, marray, marraycount, line, lineno,
685 action) != 0)
686 (*errs)++;
687 }
688
689 (void)fclose(f);
690 }
691 }
692
693 /*
694 * parse a file or directory of files
695 * const char *fn: name of magic file or directory
696 */
697 private int
cmpstrp(const void * p1,const void * p2)698 cmpstrp(const void *p1, const void *p2)
699 {
700 return strcmp(*(char *const *)p1, *(char *const *)p2);
701 }
702
703 private int
apprentice_load(struct magic_set * ms,struct magic ** magicp,uint32_t * nmagicp,const char * fn,int action)704 apprentice_load(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
705 const char *fn, int action)
706 {
707 int errs = 0;
708 struct magic_entry *marray;
709 uint32_t marraycount, i, mentrycount = 0, starttest;
710 size_t slen, files = 0, maxfiles = 0;
711 char subfn[MAXPATHLEN], **filearr = NULL, *mfn;
712 struct stat st;
713 DIR *dir;
714 struct dirent *d;
715
716 ms->flags |= MAGIC_CHECK; /* Enable checks for parsed files */
717
718 maxmagic = MAXMAGIS;
719 if ((marray = CAST(struct magic_entry *, calloc(maxmagic,
720 sizeof(*marray)))) == NULL) {
721 file_oomem(ms, maxmagic * sizeof(*marray));
722 return -1;
723 }
724 marraycount = 0;
725
726 /* print silly verbose header for USG compat. */
727 if (action == FILE_CHECK)
728 (void)fprintf(stderr, "%s\n", usg_hdr);
729
730 /* load directory or file */
731 if (stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) {
732 dir = opendir(fn);
733 if (!dir) {
734 errs++;
735 goto out;
736 }
737 while ((d = readdir(dir)) != NULL) {
738 (void)snprintf(subfn, sizeof(subfn), "%s/%s",
739 fn, d->d_name);
740 if (stat(subfn, &st) == -1 || !S_ISREG(st.st_mode))
741 continue;
742 if ((mfn = strdup(subfn)) == NULL) {
743 file_oomem(ms, strlen(subfn));
744 errs++;
745 goto out;
746 }
747 if (files >= maxfiles) {
748 size_t mlen;
749 maxfiles = (maxfiles + 1) * 2;
750 mlen = maxfiles * sizeof(*filearr);
751 if ((filearr = CAST(char **,
752 realloc(filearr, mlen))) == NULL) {
753 file_oomem(ms, mlen);
754 errs++;
755 goto out;
756 }
757 }
758 filearr[files++] = mfn;
759 }
760 closedir(dir);
761 qsort(filearr, files, sizeof(*filearr), cmpstrp);
762 for (i = 0; i < files; i++) {
763 load_1(ms, action, filearr[i], &errs, &marray,
764 &marraycount);
765 free(filearr[i]);
766 }
767 free(filearr);
768 } else
769 load_1(ms, action, fn, &errs, &marray, &marraycount);
770 if (errs)
771 goto out;
772
773 /* Set types of tests */
774 for (i = 0; i < marraycount; ) {
775 if (marray[i].mp->cont_level != 0) {
776 i++;
777 continue;
778 }
779
780 starttest = i;
781 do {
782 static const char text[] = "text";
783 static const char binary[] = "binary";
784 static const size_t len = sizeof(text);
785 set_test_type(marray[starttest].mp, marray[i].mp);
786 if ((ms->flags & MAGIC_DEBUG) == 0)
787 continue;
788 (void)fprintf(stderr, "%s%s%s: %s\n",
789 marray[i].mp->mimetype,
790 marray[i].mp->mimetype[0] == '\0' ? "" : "; ",
791 marray[i].mp->desc[0] ? marray[i].mp->desc :
792 "(no description)",
793 marray[i].mp->flag & BINTEST ? binary : text);
794 if (marray[i].mp->flag & BINTEST) {
795 char *p = strstr(marray[i].mp->desc, text);
796 if (p && (p == marray[i].mp->desc ||
797 isspace((unsigned char)p[-1])) &&
798 (p + len - marray[i].mp->desc ==
799 MAXstring || (p[len] == '\0' ||
800 isspace((unsigned char)p[len]))))
801 (void)fprintf(stderr, "*** Possible "
802 "binary test for text type\n");
803 }
804 } while (++i < marraycount && marray[i].mp->cont_level != 0);
805 }
806
807 qsort(marray, marraycount, sizeof(*marray), apprentice_sort);
808
809 /*
810 * Make sure that any level 0 "default" line is last (if one exists).
811 */
812 for (i = 0; i < marraycount; i++) {
813 if (marray[i].mp->cont_level == 0 &&
814 marray[i].mp->type == FILE_DEFAULT) {
815 while (++i < marraycount)
816 if (marray[i].mp->cont_level == 0)
817 break;
818 if (i != marraycount) {
819 ms->line = marray[i].mp->lineno; /* XXX - Ugh! */
820 file_magwarn(ms,
821 "level 0 \"default\" did not sort last");
822 }
823 break;
824 }
825 }
826
827 for (i = 0; i < marraycount; i++)
828 mentrycount += marray[i].cont_count;
829
830 slen = sizeof(**magicp) * mentrycount;
831 if ((*magicp = CAST(struct magic *, malloc(slen))) == NULL) {
832 file_oomem(ms, slen);
833 errs++;
834 goto out;
835 }
836
837 mentrycount = 0;
838 for (i = 0; i < marraycount; i++) {
839 (void)memmove(*magicp + mentrycount, marray[i].mp,
840 marray[i].cont_count * sizeof(**magicp));
841 mentrycount += marray[i].cont_count;
842 }
843 out:
844 for (i = 0; i < marraycount; i++)
845 free(marray[i].mp);
846 free(marray);
847 if (errs) {
848 *magicp = NULL;
849 *nmagicp = 0;
850 return errs;
851 } else {
852 *nmagicp = mentrycount;
853 return 0;
854 }
855
856 }
857
858 /*
859 * extend the sign bit if the comparison is to be signed
860 */
861 protected uint64_t
file_signextend(struct magic_set * ms,struct magic * m,uint64_t v)862 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v)
863 {
864 if (!(m->flag & UNSIGNED)) {
865 switch(m->type) {
866 /*
867 * Do not remove the casts below. They are
868 * vital. When later compared with the data,
869 * the sign extension must have happened.
870 */
871 case FILE_BYTE:
872 v = (char) v;
873 break;
874 case FILE_SHORT:
875 case FILE_BESHORT:
876 case FILE_LESHORT:
877 v = (short) v;
878 break;
879 case FILE_DATE:
880 case FILE_BEDATE:
881 case FILE_LEDATE:
882 case FILE_MEDATE:
883 case FILE_LDATE:
884 case FILE_BELDATE:
885 case FILE_LELDATE:
886 case FILE_MELDATE:
887 case FILE_LONG:
888 case FILE_BELONG:
889 case FILE_LELONG:
890 case FILE_MELONG:
891 case FILE_FLOAT:
892 case FILE_BEFLOAT:
893 case FILE_LEFLOAT:
894 v = (int32_t) v;
895 break;
896 case FILE_QUAD:
897 case FILE_BEQUAD:
898 case FILE_LEQUAD:
899 case FILE_QDATE:
900 case FILE_QLDATE:
901 case FILE_BEQDATE:
902 case FILE_BEQLDATE:
903 case FILE_LEQDATE:
904 case FILE_LEQLDATE:
905 case FILE_DOUBLE:
906 case FILE_BEDOUBLE:
907 case FILE_LEDOUBLE:
908 v = (int64_t) v;
909 break;
910 case FILE_STRING:
911 case FILE_PSTRING:
912 case FILE_BESTRING16:
913 case FILE_LESTRING16:
914 case FILE_REGEX:
915 case FILE_SEARCH:
916 case FILE_DEFAULT:
917 case FILE_INDIRECT:
918 break;
919 default:
920 if (ms->flags & MAGIC_CHECK)
921 file_magwarn(ms, "cannot happen: m->type=%d\n",
922 m->type);
923 return ~0U;
924 }
925 }
926 return v;
927 }
928
929 private int
string_modifier_check(struct magic_set * ms,struct magic * m)930 string_modifier_check(struct magic_set *ms, struct magic *m)
931 {
932 if ((ms->flags & MAGIC_CHECK) == 0)
933 return 0;
934
935 switch (m->type) {
936 case FILE_BESTRING16:
937 case FILE_LESTRING16:
938 if (m->str_flags != 0) {
939 file_magwarn(ms,
940 "no modifiers allowed for 16-bit strings\n");
941 return -1;
942 }
943 break;
944 case FILE_STRING:
945 case FILE_PSTRING:
946 if ((m->str_flags & REGEX_OFFSET_START) != 0) {
947 file_magwarn(ms,
948 "'/%c' only allowed on regex and search\n",
949 CHAR_REGEX_OFFSET_START);
950 return -1;
951 }
952 break;
953 case FILE_SEARCH:
954 if (m->str_range == 0) {
955 file_magwarn(ms,
956 "missing range; defaulting to %d\n",
957 STRING_DEFAULT_RANGE);
958 m->str_range = STRING_DEFAULT_RANGE;
959 return -1;
960 }
961 break;
962 case FILE_REGEX:
963 if ((m->str_flags & STRING_COMPACT_WHITESPACE) != 0) {
964 file_magwarn(ms, "'/%c' not allowed on regex\n",
965 CHAR_COMPACT_WHITESPACE);
966 return -1;
967 }
968 if ((m->str_flags & STRING_COMPACT_OPTIONAL_WHITESPACE) != 0) {
969 file_magwarn(ms, "'/%c' not allowed on regex\n",
970 CHAR_COMPACT_OPTIONAL_WHITESPACE);
971 return -1;
972 }
973 break;
974 default:
975 file_magwarn(ms, "coding error: m->type=%d\n",
976 m->type);
977 return -1;
978 }
979 return 0;
980 }
981
982 private int
get_op(char c)983 get_op(char c)
984 {
985 switch (c) {
986 case '&':
987 return FILE_OPAND;
988 case '|':
989 return FILE_OPOR;
990 case '^':
991 return FILE_OPXOR;
992 case '+':
993 return FILE_OPADD;
994 case '-':
995 return FILE_OPMINUS;
996 case '*':
997 return FILE_OPMULTIPLY;
998 case '/':
999 return FILE_OPDIVIDE;
1000 case '%':
1001 return FILE_OPMODULO;
1002 default:
1003 return -1;
1004 }
1005 }
1006
1007 #ifdef ENABLE_CONDITIONALS
1008 private int
get_cond(const char * l,const char ** t)1009 get_cond(const char *l, const char **t)
1010 {
1011 static const struct cond_tbl_s {
1012 char name[8];
1013 size_t len;
1014 int cond;
1015 } cond_tbl[] = {
1016 { "if", 2, COND_IF },
1017 { "elif", 4, COND_ELIF },
1018 { "else", 4, COND_ELSE },
1019 { "", 0, COND_NONE },
1020 };
1021 const struct cond_tbl_s *p;
1022
1023 for (p = cond_tbl; p->len; p++) {
1024 if (strncmp(l, p->name, p->len) == 0 &&
1025 isspace((unsigned char)l[p->len])) {
1026 if (t)
1027 *t = l + p->len;
1028 break;
1029 }
1030 }
1031 return p->cond;
1032 }
1033
1034 private int
check_cond(struct magic_set * ms,int cond,uint32_t cont_level)1035 check_cond(struct magic_set *ms, int cond, uint32_t cont_level)
1036 {
1037 int last_cond;
1038 last_cond = ms->c.li[cont_level].last_cond;
1039
1040 switch (cond) {
1041 case COND_IF:
1042 if (last_cond != COND_NONE && last_cond != COND_ELIF) {
1043 if (ms->flags & MAGIC_CHECK)
1044 file_magwarn(ms, "syntax error: `if'");
1045 return -1;
1046 }
1047 last_cond = COND_IF;
1048 break;
1049
1050 case COND_ELIF:
1051 if (last_cond != COND_IF && last_cond != COND_ELIF) {
1052 if (ms->flags & MAGIC_CHECK)
1053 file_magwarn(ms, "syntax error: `elif'");
1054 return -1;
1055 }
1056 last_cond = COND_ELIF;
1057 break;
1058
1059 case COND_ELSE:
1060 if (last_cond != COND_IF && last_cond != COND_ELIF) {
1061 if (ms->flags & MAGIC_CHECK)
1062 file_magwarn(ms, "syntax error: `else'");
1063 return -1;
1064 }
1065 last_cond = COND_NONE;
1066 break;
1067
1068 case COND_NONE:
1069 last_cond = COND_NONE;
1070 break;
1071 }
1072
1073 ms->c.li[cont_level].last_cond = last_cond;
1074 return 0;
1075 }
1076 #endif /* ENABLE_CONDITIONALS */
1077
1078 /*
1079 * parse one line from magic file, put into magic[index++] if valid
1080 */
1081 private int
parse(struct magic_set * ms,struct magic_entry ** mentryp,uint32_t * nmentryp,const char * line,size_t lineno,int action)1082 parse(struct magic_set *ms, struct magic_entry **mentryp, uint32_t *nmentryp,
1083 const char *line, size_t lineno, int action)
1084 {
1085 #ifdef ENABLE_CONDITIONALS
1086 static uint32_t last_cont_level = 0;
1087 #endif
1088 size_t i;
1089 struct magic_entry *me;
1090 struct magic *m;
1091 const char *l = line;
1092 char *t;
1093 int op;
1094 uint32_t cont_level;
1095
1096 cont_level = 0;
1097
1098 while (*l == '>') {
1099 ++l; /* step over */
1100 cont_level++;
1101 }
1102 #ifdef ENABLE_CONDITIONALS
1103 if (cont_level == 0 || cont_level > last_cont_level)
1104 if (file_check_mem(ms, cont_level) == -1)
1105 return -1;
1106 last_cont_level = cont_level;
1107 #endif
1108
1109 #define ALLOC_CHUNK (size_t)10
1110 #define ALLOC_INCR (size_t)200
1111
1112 if (cont_level != 0) {
1113 if (*nmentryp == 0) {
1114 file_error(ms, 0, "No current entry for continuation");
1115 return -1;
1116 }
1117 me = &(*mentryp)[*nmentryp - 1];
1118 if (me->cont_count == me->max_count) {
1119 struct magic *nm;
1120 size_t cnt = me->max_count + ALLOC_CHUNK;
1121 if ((nm = CAST(struct magic *, realloc(me->mp,
1122 sizeof(*nm) * cnt))) == NULL) {
1123 file_oomem(ms, sizeof(*nm) * cnt);
1124 return -1;
1125 }
1126 me->mp = m = nm;
1127 me->max_count = CAST(uint32_t, cnt);
1128 }
1129 m = &me->mp[me->cont_count++];
1130 (void)memset(m, 0, sizeof(*m));
1131 m->cont_level = cont_level;
1132 } else {
1133 if (*nmentryp == maxmagic) {
1134 struct magic_entry *mp;
1135
1136 maxmagic += ALLOC_INCR;
1137 if ((mp = CAST(struct magic_entry *,
1138 realloc(*mentryp, sizeof(*mp) * maxmagic))) ==
1139 NULL) {
1140 file_oomem(ms, sizeof(*mp) * maxmagic);
1141 return -1;
1142 }
1143 (void)memset(&mp[*nmentryp], 0, sizeof(*mp) *
1144 ALLOC_INCR);
1145 *mentryp = mp;
1146 }
1147 me = &(*mentryp)[*nmentryp];
1148 if (me->mp == NULL) {
1149 size_t len = sizeof(*m) * ALLOC_CHUNK;
1150 if ((m = CAST(struct magic *, malloc(len))) == NULL) {
1151 file_oomem(ms, len);
1152 return -1;
1153 }
1154 me->mp = m;
1155 me->max_count = ALLOC_CHUNK;
1156 } else
1157 m = me->mp;
1158 (void)memset(m, 0, sizeof(*m));
1159 m->factor_op = FILE_FACTOR_OP_NONE;
1160 m->cont_level = 0;
1161 me->cont_count = 1;
1162 }
1163 m->lineno = CAST(uint32_t, lineno);
1164
1165 if (*l == '&') { /* m->cont_level == 0 checked below. */
1166 ++l; /* step over */
1167 m->flag |= OFFADD;
1168 }
1169 if (*l == '(') {
1170 ++l; /* step over */
1171 m->flag |= INDIR;
1172 if (m->flag & OFFADD)
1173 m->flag = (m->flag & ~OFFADD) | INDIROFFADD;
1174
1175 if (*l == '&') { /* m->cont_level == 0 checked below */
1176 ++l; /* step over */
1177 m->flag |= OFFADD;
1178 }
1179 }
1180 /* Indirect offsets are not valid at level 0. */
1181 if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD)))
1182 if (ms->flags & MAGIC_CHECK)
1183 file_magwarn(ms, "relative offset at level 0");
1184
1185 /* get offset, then skip over it */
1186 m->offset = (uint32_t)strtoul(l, &t, 0);
1187 if (l == t)
1188 if (ms->flags & MAGIC_CHECK)
1189 file_magwarn(ms, "offset `%s' invalid", l);
1190 l = t;
1191
1192 if (m->flag & INDIR) {
1193 m->in_type = FILE_LONG;
1194 m->in_offset = 0;
1195 /*
1196 * read [.lbs][+-]nnnnn)
1197 */
1198 if (*l == '.') {
1199 l++;
1200 switch (*l) {
1201 case 'l':
1202 m->in_type = FILE_LELONG;
1203 break;
1204 case 'L':
1205 m->in_type = FILE_BELONG;
1206 break;
1207 case 'm':
1208 m->in_type = FILE_MELONG;
1209 break;
1210 case 'h':
1211 case 's':
1212 m->in_type = FILE_LESHORT;
1213 break;
1214 case 'H':
1215 case 'S':
1216 m->in_type = FILE_BESHORT;
1217 break;
1218 case 'c':
1219 case 'b':
1220 case 'C':
1221 case 'B':
1222 m->in_type = FILE_BYTE;
1223 break;
1224 case 'e':
1225 case 'f':
1226 case 'g':
1227 m->in_type = FILE_LEDOUBLE;
1228 break;
1229 case 'E':
1230 case 'F':
1231 case 'G':
1232 m->in_type = FILE_BEDOUBLE;
1233 break;
1234 case 'i':
1235 m->in_type = FILE_LEID3;
1236 break;
1237 case 'I':
1238 m->in_type = FILE_BEID3;
1239 break;
1240 default:
1241 if (ms->flags & MAGIC_CHECK)
1242 file_magwarn(ms,
1243 "indirect offset type `%c' invalid",
1244 *l);
1245 break;
1246 }
1247 l++;
1248 }
1249
1250 m->in_op = 0;
1251 if (*l == '~') {
1252 m->in_op |= FILE_OPINVERSE;
1253 l++;
1254 }
1255 if ((op = get_op(*l)) != -1) {
1256 m->in_op |= op;
1257 l++;
1258 }
1259 if (*l == '(') {
1260 m->in_op |= FILE_OPINDIRECT;
1261 l++;
1262 }
1263 if (isdigit((unsigned char)*l) || *l == '-') {
1264 m->in_offset = (int32_t)strtol(l, &t, 0);
1265 if (l == t)
1266 if (ms->flags & MAGIC_CHECK)
1267 file_magwarn(ms,
1268 "in_offset `%s' invalid", l);
1269 l = t;
1270 }
1271 if (*l++ != ')' ||
1272 ((m->in_op & FILE_OPINDIRECT) && *l++ != ')'))
1273 if (ms->flags & MAGIC_CHECK)
1274 file_magwarn(ms,
1275 "missing ')' in indirect offset");
1276 }
1277 EATAB;
1278
1279 #ifdef ENABLE_CONDITIONALS
1280 m->cond = get_cond(l, &l);
1281 if (check_cond(ms, m->cond, cont_level) == -1)
1282 return -1;
1283
1284 EATAB;
1285 #endif
1286
1287 if (*l == 'u') {
1288 ++l;
1289 m->flag |= UNSIGNED;
1290 }
1291
1292 m->type = get_type(l, &l);
1293 if (m->type == FILE_INVALID) {
1294 if (ms->flags & MAGIC_CHECK)
1295 file_magwarn(ms, "type `%s' invalid", l);
1296 return -1;
1297 }
1298
1299 /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */
1300 /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */
1301
1302 m->mask_op = 0;
1303 if (*l == '~') {
1304 if (!IS_STRING(m->type))
1305 m->mask_op |= FILE_OPINVERSE;
1306 else if (ms->flags & MAGIC_CHECK)
1307 file_magwarn(ms, "'~' invalid for string types");
1308 ++l;
1309 }
1310 m->str_range = 0;
1311 m->str_flags = 0;
1312 m->num_mask = 0;
1313 if ((op = get_op(*l)) != -1) {
1314 if (!IS_STRING(m->type)) {
1315 uint64_t val;
1316 ++l;
1317 m->mask_op |= op;
1318 val = (uint64_t)strtoull(l, &t, 0);
1319 l = t;
1320 m->num_mask = file_signextend(ms, m, val);
1321 eatsize(&l);
1322 }
1323 else if (op == FILE_OPDIVIDE) {
1324 int have_range = 0;
1325 while (!isspace((unsigned char)*++l)) {
1326 switch (*l) {
1327 case '0': case '1': case '2':
1328 case '3': case '4': case '5':
1329 case '6': case '7': case '8':
1330 case '9':
1331 if (have_range &&
1332 (ms->flags & MAGIC_CHECK))
1333 file_magwarn(ms,
1334 "multiple ranges");
1335 have_range = 1;
1336 m->str_range = CAST(uint32_t,
1337 strtoul(l, &t, 0));
1338 if (m->str_range == 0)
1339 file_magwarn(ms,
1340 "zero range");
1341 l = t - 1;
1342 break;
1343 case CHAR_COMPACT_WHITESPACE:
1344 m->str_flags |= STRING_COMPACT_WHITESPACE;
1345 break;
1346 case CHAR_COMPACT_OPTIONAL_WHITESPACE:
1347 m->str_flags |=
1348 STRING_COMPACT_OPTIONAL_WHITESPACE;
1349 break;
1350 case CHAR_IGNORE_LOWERCASE:
1351 m->str_flags |= STRING_IGNORE_LOWERCASE;
1352 break;
1353 case CHAR_IGNORE_UPPERCASE:
1354 m->str_flags |= STRING_IGNORE_UPPERCASE;
1355 break;
1356 case CHAR_REGEX_OFFSET_START:
1357 m->str_flags |= REGEX_OFFSET_START;
1358 break;
1359 case CHAR_BINTEST:
1360 m->str_flags |= STRING_BINTEST;
1361 break;
1362 case CHAR_TEXTTEST:
1363 m->str_flags |= STRING_TEXTTEST;
1364 break;
1365 default:
1366 if (ms->flags & MAGIC_CHECK)
1367 file_magwarn(ms,
1368 "string extension `%c' invalid",
1369 *l);
1370 return -1;
1371 }
1372 /* allow multiple '/' for readability */
1373 if (l[1] == '/' &&
1374 !isspace((unsigned char)l[2]))
1375 l++;
1376 }
1377 if (string_modifier_check(ms, m) == -1)
1378 return -1;
1379 }
1380 else {
1381 if (ms->flags & MAGIC_CHECK)
1382 file_magwarn(ms, "invalid string op: %c", *t);
1383 return -1;
1384 }
1385 }
1386 /*
1387 * We used to set mask to all 1's here, instead let's just not do
1388 * anything if mask = 0 (unless you have a better idea)
1389 */
1390 EATAB;
1391
1392 switch (*l) {
1393 case '>':
1394 case '<':
1395 m->reln = *l;
1396 ++l;
1397 if (*l == '=') {
1398 if (ms->flags & MAGIC_CHECK) {
1399 file_magwarn(ms, "%c= not supported",
1400 m->reln);
1401 return -1;
1402 }
1403 ++l;
1404 }
1405 break;
1406 /* Old-style anding: "0 byte &0x80 dynamically linked" */
1407 case '&':
1408 case '^':
1409 case '=':
1410 m->reln = *l;
1411 ++l;
1412 if (*l == '=') {
1413 /* HP compat: ignore &= etc. */
1414 ++l;
1415 }
1416 break;
1417 case '!':
1418 m->reln = *l;
1419 ++l;
1420 break;
1421 default:
1422 m->reln = '='; /* the default relation */
1423 if (*l == 'x' && ((isascii((unsigned char)l[1]) &&
1424 isspace((unsigned char)l[1])) || !l[1])) {
1425 m->reln = *l;
1426 ++l;
1427 }
1428 break;
1429 }
1430 /*
1431 * Grab the value part, except for an 'x' reln.
1432 */
1433 if (m->reln != 'x' && getvalue(ms, m, &l, action))
1434 return -1;
1435
1436 /*
1437 * TODO finish this macro and start using it!
1438 * #define offsetcheck {if (offset > HOWMANY-1)
1439 * magwarn("offset too big"); }
1440 */
1441
1442 /*
1443 * Now get last part - the description
1444 */
1445 EATAB;
1446 if (l[0] == '\b') {
1447 ++l;
1448 m->flag |= NOSPACE;
1449 } else if ((l[0] == '\\') && (l[1] == 'b')) {
1450 ++l;
1451 ++l;
1452 m->flag |= NOSPACE;
1453 }
1454 for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); )
1455 continue;
1456 if (i == sizeof(m->desc)) {
1457 m->desc[sizeof(m->desc) - 1] = '\0';
1458 if (ms->flags & MAGIC_CHECK)
1459 file_magwarn(ms, "description `%s' truncated", m->desc);
1460 }
1461
1462 /*
1463 * We only do this check while compiling, or if any of the magic
1464 * files were not compiled.
1465 */
1466 if (ms->flags & MAGIC_CHECK) {
1467 if (check_format(ms, m) == -1)
1468 return -1;
1469 }
1470 #ifndef COMPILE_ONLY
1471 if (action == FILE_CHECK) {
1472 file_mdump(m);
1473 }
1474 #endif
1475 m->mimetype[0] = '\0'; /* initialise MIME type to none */
1476 if (m->cont_level == 0)
1477 ++(*nmentryp); /* make room for next */
1478 return 0;
1479 }
1480
1481 /*
1482 * parse a STRENGTH annotation line from magic file, put into magic[index - 1]
1483 * if valid
1484 */
1485 private int
parse_strength(struct magic_set * ms,struct magic_entry * me,const char * line)1486 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line)
1487 {
1488 const char *l = line;
1489 char *el;
1490 unsigned long factor;
1491 struct magic *m = &me->mp[0];
1492
1493 if (m->factor_op != FILE_FACTOR_OP_NONE) {
1494 file_magwarn(ms,
1495 "Current entry already has a strength type: %c %d",
1496 m->factor_op, m->factor);
1497 return -1;
1498 }
1499 EATAB;
1500 switch (*l) {
1501 case FILE_FACTOR_OP_NONE:
1502 case FILE_FACTOR_OP_PLUS:
1503 case FILE_FACTOR_OP_MINUS:
1504 case FILE_FACTOR_OP_TIMES:
1505 case FILE_FACTOR_OP_DIV:
1506 m->factor_op = *l++;
1507 break;
1508 default:
1509 file_magwarn(ms, "Unknown factor op `%c'", *l);
1510 return -1;
1511 }
1512 EATAB;
1513 factor = strtoul(l, &el, 0);
1514 if (factor > 255) {
1515 file_magwarn(ms, "Too large factor `%lu'", factor);
1516 goto out;
1517 }
1518 if (*el && !isspace((unsigned char)*el)) {
1519 file_magwarn(ms, "Bad factor `%s'", l);
1520 goto out;
1521 }
1522 m->factor = (uint8_t)factor;
1523 if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) {
1524 file_magwarn(ms, "Cannot have factor op `%c' and factor %u",
1525 m->factor_op, m->factor);
1526 goto out;
1527 }
1528 return 0;
1529 out:
1530 m->factor_op = FILE_FACTOR_OP_NONE;
1531 m->factor = 0;
1532 return -1;
1533 }
1534
1535 /*
1536 * Parse an Apple CREATOR/TYPE annotation from magic file and put it into magic[index - 1]
1537 */
1538 private int
parse_apple(struct magic_set * ms,struct magic_entry * me,const char * line)1539 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line)
1540 {
1541 size_t i;
1542 const char *l = line;
1543 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1];
1544
1545 if (m->apple[0] != '\0') {
1546 file_magwarn(ms, "Current entry already has a APPLE type `%.8s',"
1547 " new type `%s'", m->mimetype, l);
1548 return -1;
1549 }
1550
1551 EATAB;
1552 for (i = 0; *l && ((isascii((unsigned char)*l) && isalnum((unsigned char)*l))
1553 || strchr("-+/.", *l)) && i < sizeof(m->apple); m->apple[i++] = *l++)
1554 continue;
1555 if (i == sizeof(m->apple) && *l) {
1556 /* We don't need to NUL terminate here, printing handles it */
1557 if (ms->flags & MAGIC_CHECK)
1558 file_magwarn(ms, "APPLE type `%s' truncated %zu",
1559 line, i);
1560 }
1561
1562 if (i > 0)
1563 return 0;
1564 else
1565 return -1;
1566 }
1567
1568 /*
1569 * parse a MIME annotation line from magic file, put into magic[index - 1]
1570 * if valid
1571 */
1572 private int
parse_mime(struct magic_set * ms,struct magic_entry * me,const char * line)1573 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line)
1574 {
1575 size_t i;
1576 const char *l = line;
1577 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1];
1578
1579 if (m->mimetype[0] != '\0') {
1580 file_magwarn(ms, "Current entry already has a MIME type `%s',"
1581 " new type `%s'", m->mimetype, l);
1582 return -1;
1583 }
1584
1585 EATAB;
1586 for (i = 0; *l && ((isascii((unsigned char)*l) && isalnum((unsigned char)*l))
1587 || strchr("-+/.", *l)) && i < sizeof(m->mimetype); m->mimetype[i++] = *l++)
1588 continue;
1589 if (i == sizeof(m->mimetype)) {
1590 m->mimetype[sizeof(m->mimetype) - 1] = '\0';
1591 if (ms->flags & MAGIC_CHECK)
1592 file_magwarn(ms, "MIME type `%s' truncated %zu",
1593 m->mimetype, i);
1594 } else
1595 m->mimetype[i] = '\0';
1596
1597 if (i > 0)
1598 return 0;
1599 else
1600 return -1;
1601 }
1602
1603 private int
check_format_type(const char * ptr,int type)1604 check_format_type(const char *ptr, int type)
1605 {
1606 int quad = 0;
1607 if (*ptr == '\0') {
1608 /* Missing format string; bad */
1609 return -1;
1610 }
1611
1612 switch (type) {
1613 case FILE_FMT_QUAD:
1614 quad = 1;
1615 /*FALLTHROUGH*/
1616 case FILE_FMT_NUM:
1617 if (*ptr == '-')
1618 ptr++;
1619 if (*ptr == '.')
1620 ptr++;
1621 while (isdigit((unsigned char)*ptr)) ptr++;
1622 if (*ptr == '.')
1623 ptr++;
1624 while (isdigit((unsigned char)*ptr)) ptr++;
1625 if (quad) {
1626 if (*ptr++ != 'l')
1627 return -1;
1628 if (*ptr++ != 'l')
1629 return -1;
1630 }
1631
1632 switch (*ptr++) {
1633 case 'l':
1634 switch (*ptr++) {
1635 case 'i':
1636 case 'd':
1637 case 'u':
1638 case 'x':
1639 case 'X':
1640 return 0;
1641 default:
1642 return -1;
1643 }
1644
1645 case 'h':
1646 switch (*ptr++) {
1647 case 'h':
1648 switch (*ptr++) {
1649 case 'i':
1650 case 'd':
1651 case 'u':
1652 case 'x':
1653 case 'X':
1654 return 0;
1655 default:
1656 return -1;
1657 }
1658 case 'd':
1659 return 0;
1660 default:
1661 return -1;
1662 }
1663
1664 case 'i':
1665 case 'c':
1666 case 'd':
1667 case 'u':
1668 case 'x':
1669 case 'X':
1670 return 0;
1671
1672 default:
1673 return -1;
1674 }
1675
1676 case FILE_FMT_FLOAT:
1677 case FILE_FMT_DOUBLE:
1678 if (*ptr == '-')
1679 ptr++;
1680 if (*ptr == '.')
1681 ptr++;
1682 while (isdigit((unsigned char)*ptr)) ptr++;
1683 if (*ptr == '.')
1684 ptr++;
1685 while (isdigit((unsigned char)*ptr)) ptr++;
1686
1687 switch (*ptr++) {
1688 case 'e':
1689 case 'E':
1690 case 'f':
1691 case 'F':
1692 case 'g':
1693 case 'G':
1694 return 0;
1695
1696 default:
1697 return -1;
1698 }
1699
1700
1701 case FILE_FMT_STR:
1702 if (*ptr == '-')
1703 ptr++;
1704 while (isdigit((unsigned char )*ptr))
1705 ptr++;
1706 if (*ptr == '.') {
1707 ptr++;
1708 while (isdigit((unsigned char )*ptr))
1709 ptr++;
1710 }
1711
1712 switch (*ptr++) {
1713 case 's':
1714 return 0;
1715 default:
1716 return -1;
1717 }
1718
1719 default:
1720 /* internal error */
1721 abort();
1722 }
1723 /*NOTREACHED*/
1724 return -1;
1725 }
1726
1727 /*
1728 * Check that the optional printf format in description matches
1729 * the type of the magic.
1730 */
1731 private int
check_format(struct magic_set * ms,struct magic * m)1732 check_format(struct magic_set *ms, struct magic *m)
1733 {
1734 char *ptr;
1735
1736 for (ptr = m->desc; *ptr; ptr++)
1737 if (*ptr == '%')
1738 break;
1739 if (*ptr == '\0') {
1740 /* No format string; ok */
1741 return 1;
1742 }
1743
1744 assert(file_nformats == file_nnames);
1745
1746 if (m->type >= file_nformats) {
1747 file_magwarn(ms, "Internal error inconsistency between "
1748 "m->type and format strings");
1749 return -1;
1750 }
1751 if (file_formats[m->type] == FILE_FMT_NONE) {
1752 file_magwarn(ms, "No format string for `%s' with description "
1753 "`%s'", m->desc, file_names[m->type]);
1754 return -1;
1755 }
1756
1757 ptr++;
1758 if (check_format_type(ptr, file_formats[m->type]) == -1) {
1759 /*
1760 * TODO: this error message is unhelpful if the format
1761 * string is not one character long
1762 */
1763 file_magwarn(ms, "Printf format `%c' is not valid for type "
1764 "`%s' in description `%s'", *ptr ? *ptr : '?',
1765 file_names[m->type], m->desc);
1766 return -1;
1767 }
1768
1769 for (; *ptr; ptr++) {
1770 if (*ptr == '%') {
1771 file_magwarn(ms,
1772 "Too many format strings (should have at most one) "
1773 "for `%s' with description `%s'",
1774 file_names[m->type], m->desc);
1775 return -1;
1776 }
1777 }
1778 return 0;
1779 }
1780
1781 /*
1782 * Read a numeric value from a pointer, into the value union of a magic
1783 * pointer, according to the magic type. Update the string pointer to point
1784 * just after the number read. Return 0 for success, non-zero for failure.
1785 */
1786 private int
getvalue(struct magic_set * ms,struct magic * m,const char ** p,int action)1787 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action)
1788 {
1789 switch (m->type) {
1790 case FILE_BESTRING16:
1791 case FILE_LESTRING16:
1792 case FILE_STRING:
1793 case FILE_PSTRING:
1794 case FILE_REGEX:
1795 case FILE_SEARCH:
1796 *p = getstr(ms, m, *p, action == FILE_COMPILE);
1797 if (*p == NULL) {
1798 if (ms->flags & MAGIC_CHECK)
1799 file_magwarn(ms, "cannot get string from `%s'",
1800 m->value.s);
1801 return -1;
1802 }
1803 return 0;
1804 case FILE_FLOAT:
1805 case FILE_BEFLOAT:
1806 case FILE_LEFLOAT:
1807 if (m->reln != 'x') {
1808 char *ep;
1809 #ifdef HAVE_STRTOF
1810 m->value.f = strtof(*p, &ep);
1811 #else
1812 m->value.f = (float)strtod(*p, &ep);
1813 #endif
1814 *p = ep;
1815 }
1816 return 0;
1817 case FILE_DOUBLE:
1818 case FILE_BEDOUBLE:
1819 case FILE_LEDOUBLE:
1820 if (m->reln != 'x') {
1821 char *ep;
1822 m->value.d = strtod(*p, &ep);
1823 *p = ep;
1824 }
1825 return 0;
1826 default:
1827 if (m->reln != 'x') {
1828 char *ep;
1829 m->value.q = file_signextend(ms, m,
1830 (uint64_t)strtoull(*p, &ep, 0));
1831 *p = ep;
1832 eatsize(p);
1833 }
1834 return 0;
1835 }
1836 }
1837
1838 /*
1839 * Convert a string containing C character escapes. Stop at an unescaped
1840 * space or tab.
1841 * Copy the converted version to "m->value.s", and the length in m->vallen.
1842 * Return updated scan pointer as function result. Warn if set.
1843 */
1844 private const char *
getstr(struct magic_set * ms,struct magic * m,const char * s,int warn)1845 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn)
1846 {
1847 const char *origs = s;
1848 char *p = m->value.s;
1849 size_t plen = sizeof(m->value.s);
1850 char *origp = p;
1851 char *pmax = p + plen - 1;
1852 int c;
1853 int val;
1854
1855 while ((c = *s++) != '\0') {
1856 if (isspace((unsigned char) c))
1857 break;
1858 if (p >= pmax) {
1859 file_error(ms, 0, "string too long: `%s'", origs);
1860 return NULL;
1861 }
1862 if (c == '\\') {
1863 switch(c = *s++) {
1864
1865 case '\0':
1866 if (warn)
1867 file_magwarn(ms, "incomplete escape");
1868 goto out;
1869
1870 case '\t':
1871 if (warn) {
1872 file_magwarn(ms,
1873 "escaped tab found, use \\t instead");
1874 warn = 0; /* already did */
1875 }
1876 /*FALLTHROUGH*/
1877 default:
1878 if (warn) {
1879 if (isprint((unsigned char)c)) {
1880 /* Allow escaping of
1881 * ``relations'' */
1882 if (strchr("<>&^=!", c)
1883 == NULL) {
1884 file_magwarn(ms, "no "
1885 "need to escape "
1886 "`%c'", c);
1887 }
1888 } else {
1889 file_magwarn(ms,
1890 "unknown escape sequence: "
1891 "\\%03o", c);
1892 }
1893 }
1894 /*FALLTHROUGH*/
1895 /* space, perhaps force people to use \040? */
1896 case ' ':
1897 #if 0
1898 /*
1899 * Other things people escape, but shouldn't need to,
1900 * so we disallow them
1901 */
1902 case '\'':
1903 case '"':
1904 case '?':
1905 #endif
1906 /* Relations */
1907 case '>':
1908 case '<':
1909 case '&':
1910 case '^':
1911 case '=':
1912 case '!':
1913 /* and baskslash itself */
1914 case '\\':
1915 *p++ = (char) c;
1916 break;
1917
1918 case 'a':
1919 *p++ = '\a';
1920 break;
1921
1922 case 'b':
1923 *p++ = '\b';
1924 break;
1925
1926 case 'f':
1927 *p++ = '\f';
1928 break;
1929
1930 case 'n':
1931 *p++ = '\n';
1932 break;
1933
1934 case 'r':
1935 *p++ = '\r';
1936 break;
1937
1938 case 't':
1939 *p++ = '\t';
1940 break;
1941
1942 case 'v':
1943 *p++ = '\v';
1944 break;
1945
1946 /* \ and up to 3 octal digits */
1947 case '0':
1948 case '1':
1949 case '2':
1950 case '3':
1951 case '4':
1952 case '5':
1953 case '6':
1954 case '7':
1955 val = c - '0';
1956 c = *s++; /* try for 2 */
1957 if (c >= '0' && c <= '7') {
1958 val = (val << 3) | (c - '0');
1959 c = *s++; /* try for 3 */
1960 if (c >= '0' && c <= '7')
1961 val = (val << 3) | (c-'0');
1962 else
1963 --s;
1964 }
1965 else
1966 --s;
1967 *p++ = (char)val;
1968 break;
1969
1970 /* \x and up to 2 hex digits */
1971 case 'x':
1972 val = 'x'; /* Default if no digits */
1973 c = hextoint(*s++); /* Get next char */
1974 if (c >= 0) {
1975 val = c;
1976 c = hextoint(*s++);
1977 if (c >= 0)
1978 val = (val << 4) + c;
1979 else
1980 --s;
1981 } else
1982 --s;
1983 *p++ = (char)val;
1984 break;
1985 }
1986 } else
1987 *p++ = (char)c;
1988 }
1989 out:
1990 *p = '\0';
1991 m->vallen = CAST(unsigned char, (p - origp));
1992 if (m->type == FILE_PSTRING)
1993 m->vallen++;
1994 return s;
1995 }
1996
1997
1998 /* Single hex char to int; -1 if not a hex char. */
1999 private int
hextoint(int c)2000 hextoint(int c)
2001 {
2002 if (!isascii((unsigned char) c))
2003 return -1;
2004 if (isdigit((unsigned char) c))
2005 return c - '0';
2006 if ((c >= 'a') && (c <= 'f'))
2007 return c + 10 - 'a';
2008 if (( c>= 'A') && (c <= 'F'))
2009 return c + 10 - 'A';
2010 return -1;
2011 }
2012
2013
2014 /*
2015 * Print a string containing C character escapes.
2016 */
2017 protected void
file_showstr(FILE * fp,const char * s,size_t len)2018 file_showstr(FILE *fp, const char *s, size_t len)
2019 {
2020 char c;
2021
2022 for (;;) {
2023 if (len == ~0U) {
2024 c = *s++;
2025 if (c == '\0')
2026 break;
2027 }
2028 else {
2029 if (len-- == 0)
2030 break;
2031 c = *s++;
2032 }
2033 if (c >= 040 && c <= 0176) /* TODO isprint && !iscntrl */
2034 (void) fputc(c, fp);
2035 else {
2036 (void) fputc('\\', fp);
2037 switch (c) {
2038 case '\a':
2039 (void) fputc('a', fp);
2040 break;
2041
2042 case '\b':
2043 (void) fputc('b', fp);
2044 break;
2045
2046 case '\f':
2047 (void) fputc('f', fp);
2048 break;
2049
2050 case '\n':
2051 (void) fputc('n', fp);
2052 break;
2053
2054 case '\r':
2055 (void) fputc('r', fp);
2056 break;
2057
2058 case '\t':
2059 (void) fputc('t', fp);
2060 break;
2061
2062 case '\v':
2063 (void) fputc('v', fp);
2064 break;
2065
2066 default:
2067 (void) fprintf(fp, "%.3o", c & 0377);
2068 break;
2069 }
2070 }
2071 }
2072 }
2073
2074 /*
2075 * eatsize(): Eat the size spec from a number [eg. 10UL]
2076 */
2077 private void
eatsize(const char ** p)2078 eatsize(const char **p)
2079 {
2080 const char *l = *p;
2081
2082 if (LOWCASE(*l) == 'u')
2083 l++;
2084
2085 switch (LOWCASE(*l)) {
2086 case 'l': /* long */
2087 case 's': /* short */
2088 case 'h': /* short */
2089 case 'b': /* char/byte */
2090 case 'c': /* char/byte */
2091 l++;
2092 /*FALLTHROUGH*/
2093 default:
2094 break;
2095 }
2096
2097 *p = l;
2098 }
2099
2100 /*
2101 * handle a compiled file.
2102 */
2103 private int
apprentice_map(struct magic_set * ms,struct magic ** magicp,uint32_t * nmagicp,const char * fn)2104 apprentice_map(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
2105 const char *fn)
2106 {
2107 int fd;
2108 struct stat st;
2109 uint32_t *ptr;
2110 uint32_t version;
2111 int needsbyteswap;
2112 char *dbname = NULL;
2113 void *mm = NULL;
2114
2115 dbname = mkdbname(ms, fn, 0);
2116 if (dbname == NULL)
2117 goto error2;
2118
2119 if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1)
2120 goto error2;
2121
2122 if (fstat(fd, &st) == -1) {
2123 file_error(ms, errno, "cannot stat `%s'", dbname);
2124 goto error1;
2125 }
2126 if (st.st_size < 8) {
2127 file_error(ms, 0, "file `%s' is too small", dbname);
2128 goto error1;
2129 }
2130
2131 #ifdef QUICK
2132 if ((mm = mmap(0, (size_t)st.st_size, PROT_READ|PROT_WRITE,
2133 MAP_PRIVATE|MAP_FILE, fd, (off_t)0)) == MAP_FAILED) {
2134 file_error(ms, errno, "cannot map `%s'", dbname);
2135 goto error1;
2136 }
2137 #define RET 2
2138 #else
2139 if ((mm = CAST(void *, malloc((size_t)st.st_size))) == NULL) {
2140 file_oomem(ms, (size_t)st.st_size);
2141 goto error1;
2142 }
2143 if (read(fd, mm, (size_t)st.st_size) != (ssize_t)st.st_size) {
2144 file_badread(ms);
2145 goto error1;
2146 }
2147 #define RET 1
2148 #endif
2149 *magicp = CAST(struct magic *, mm);
2150 (void)close(fd);
2151 fd = -1;
2152 ptr = (uint32_t *)(void *)*magicp;
2153 if (*ptr != MAGICNO) {
2154 if (swap4(*ptr) != MAGICNO) {
2155 file_error(ms, 0, "bad magic in `%s'", dbname);
2156 goto error1;
2157 }
2158 needsbyteswap = 1;
2159 } else
2160 needsbyteswap = 0;
2161 if (needsbyteswap)
2162 version = swap4(ptr[1]);
2163 else
2164 version = ptr[1];
2165 if (version != VERSIONNO) {
2166 file_error(ms, 0, "File %d.%d supports only version %d magic "
2167 "files. `%s' is version %d", FILE_VERSION_MAJOR, patchlevel,
2168 VERSIONNO, dbname, version);
2169 goto error1;
2170 }
2171 *nmagicp = (uint32_t)(st.st_size / sizeof(struct magic));
2172 if (*nmagicp > 0)
2173 (*nmagicp)--;
2174 (*magicp)++;
2175 if (needsbyteswap)
2176 byteswap(*magicp, *nmagicp);
2177 free(dbname);
2178 return RET;
2179
2180 error1:
2181 if (fd != -1)
2182 (void)close(fd);
2183 if (mm) {
2184 #ifdef QUICK
2185 (void)munmap((void *)mm, (size_t)st.st_size);
2186 #else
2187 free(mm);
2188 #endif
2189 } else {
2190 *magicp = NULL;
2191 *nmagicp = 0;
2192 }
2193 error2:
2194 free(dbname);
2195 return -1;
2196 }
2197
2198 private const uint32_t ar[] = {
2199 MAGICNO, VERSIONNO
2200 };
2201 /*
2202 * handle an mmaped file.
2203 */
2204 private int
apprentice_compile(struct magic_set * ms,struct magic ** magicp,uint32_t * nmagicp,const char * fn)2205 apprentice_compile(struct magic_set *ms, struct magic **magicp,
2206 uint32_t *nmagicp, const char *fn)
2207 {
2208 int fd;
2209 char *dbname;
2210 int rv = -1;
2211
2212 dbname = mkdbname(ms, fn, 1);
2213
2214 if (dbname == NULL)
2215 goto out;
2216
2217 if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) {
2218 file_error(ms, errno, "cannot open `%s'", dbname);
2219 goto out;
2220 }
2221
2222 if (write(fd, ar, sizeof(ar)) != (ssize_t)sizeof(ar)) {
2223 file_error(ms, errno, "error writing `%s'", dbname);
2224 goto out;
2225 }
2226
2227 if (lseek(fd, (off_t)sizeof(struct magic), SEEK_SET)
2228 != sizeof(struct magic)) {
2229 file_error(ms, errno, "error seeking `%s'", dbname);
2230 goto out;
2231 }
2232
2233 if (write(fd, *magicp, (sizeof(struct magic) * *nmagicp))
2234 != (ssize_t)(sizeof(struct magic) * *nmagicp)) {
2235 file_error(ms, errno, "error writing `%s'", dbname);
2236 goto out;
2237 }
2238
2239 (void)close(fd);
2240 rv = 0;
2241 out:
2242 free(dbname);
2243 return rv;
2244 }
2245
2246 private const char ext[] = ".mgc";
2247 /*
2248 * make a dbname
2249 */
2250 private char *
mkdbname(struct magic_set * ms,const char * fn,int strip)2251 mkdbname(struct magic_set *ms, const char *fn, int strip)
2252 {
2253 const char *p, *q;
2254 char *buf;
2255
2256 if (strip) {
2257 if ((p = strrchr(fn, '/')) != NULL)
2258 fn = ++p;
2259 }
2260
2261 for (q = fn; *q; q++)
2262 continue;
2263 /* Look for .mgc */
2264 for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--)
2265 if (*p != *q)
2266 break;
2267
2268 /* Did not find .mgc, restore q */
2269 if (p >= ext)
2270 while (*q)
2271 q++;
2272
2273 q++;
2274 /* Compatibility with old code that looked in .mime */
2275 if (ms->flags & MAGIC_MIME) {
2276 asprintf(&buf, "%.*s.mime%s", (int)(q - fn), fn, ext);
2277 if (access(buf, R_OK) != -1) {
2278 ms->flags &= MAGIC_MIME_TYPE;
2279 return buf;
2280 }
2281 free(buf);
2282 }
2283 asprintf(&buf, "%.*s%s", (int)(q - fn), fn, ext);
2284
2285 /* Compatibility with old code that looked in .mime */
2286 if (strstr(p, ".mime") != NULL)
2287 ms->flags &= MAGIC_MIME_TYPE;
2288 return buf;
2289 }
2290
2291 /*
2292 * Byteswap an mmap'ed file if needed
2293 */
2294 private void
byteswap(struct magic * magic,uint32_t nmagic)2295 byteswap(struct magic *magic, uint32_t nmagic)
2296 {
2297 uint32_t i;
2298 for (i = 0; i < nmagic; i++)
2299 bs1(&magic[i]);
2300 }
2301
2302 /*
2303 * swap a short
2304 */
2305 private uint16_t
swap2(uint16_t sv)2306 swap2(uint16_t sv)
2307 {
2308 uint16_t rv;
2309 uint8_t *s = (uint8_t *)(void *)&sv;
2310 uint8_t *d = (uint8_t *)(void *)&rv;
2311 d[0] = s[1];
2312 d[1] = s[0];
2313 return rv;
2314 }
2315
2316 /*
2317 * swap an int
2318 */
2319 private uint32_t
swap4(uint32_t sv)2320 swap4(uint32_t sv)
2321 {
2322 uint32_t rv;
2323 uint8_t *s = (uint8_t *)(void *)&sv;
2324 uint8_t *d = (uint8_t *)(void *)&rv;
2325 d[0] = s[3];
2326 d[1] = s[2];
2327 d[2] = s[1];
2328 d[3] = s[0];
2329 return rv;
2330 }
2331
2332 /*
2333 * swap a quad
2334 */
2335 private uint64_t
swap8(uint64_t sv)2336 swap8(uint64_t sv)
2337 {
2338 uint64_t rv;
2339 uint8_t *s = (uint8_t *)(void *)&sv;
2340 uint8_t *d = (uint8_t *)(void *)&rv;
2341 #if 0
2342 d[0] = s[3];
2343 d[1] = s[2];
2344 d[2] = s[1];
2345 d[3] = s[0];
2346 d[4] = s[7];
2347 d[5] = s[6];
2348 d[6] = s[5];
2349 d[7] = s[4];
2350 #else
2351 d[0] = s[7];
2352 d[1] = s[6];
2353 d[2] = s[5];
2354 d[3] = s[4];
2355 d[4] = s[3];
2356 d[5] = s[2];
2357 d[6] = s[1];
2358 d[7] = s[0];
2359 #endif
2360 return rv;
2361 }
2362
2363 /*
2364 * byteswap a single magic entry
2365 */
2366 private void
bs1(struct magic * m)2367 bs1(struct magic *m)
2368 {
2369 m->cont_level = swap2(m->cont_level);
2370 m->offset = swap4((uint32_t)m->offset);
2371 m->in_offset = swap4((uint32_t)m->in_offset);
2372 m->lineno = swap4((uint32_t)m->lineno);
2373 if (IS_STRING(m->type)) {
2374 m->str_range = swap4(m->str_range);
2375 m->str_flags = swap4(m->str_flags);
2376 }
2377 else {
2378 m->value.q = swap8(m->value.q);
2379 m->num_mask = swap8(m->num_mask);
2380 }
2381 }
2382