xref: /dflybsd-src/contrib/file/src/apprentice.c (revision 79343712f725e22dd7010730453b5f3a7b17bb4f)
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 1.151 2009/03/18 15:19:23 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
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
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
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
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
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 *
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 	init_file_tables();
357 
358 	if (fn == NULL)
359 		fn = getenv("MAGIC");
360 	if (fn == NULL)
361 		fn = MAGIC;
362 
363 	if ((mfn = strdup(fn)) == NULL) {
364 		file_oomem(ms, strlen(fn));
365 		return NULL;
366 	}
367 	fn = mfn;
368 
369 	if ((mlist = CAST(struct mlist *, malloc(sizeof(*mlist)))) == NULL) {
370 		free(mfn);
371 		file_oomem(ms, sizeof(*mlist));
372 		return NULL;
373 	}
374 	mlist->next = mlist->prev = mlist;
375 
376 	while (fn) {
377 		p = strchr(fn, PATHSEP);
378 		if (p)
379 			*p++ = '\0';
380 		if (*fn == '\0')
381 			break;
382 		file_err = apprentice_1(ms, fn, action, mlist);
383 		errs = MAX(errs, file_err);
384 		fn = p;
385 	}
386 	if (errs == -1) {
387 		free(mfn);
388 		free(mlist);
389 		mlist = NULL;
390 		file_error(ms, 0, "could not find any magic files!");
391 		return NULL;
392 	}
393 	free(mfn);
394 	return mlist;
395 }
396 
397 /*
398  * Get weight of this magic entry, for sorting purposes.
399  */
400 private size_t
401 apprentice_magic_strength(const struct magic *m)
402 {
403 #define MULT 10
404 	size_t val = 2 * MULT;	/* baseline strength */
405 
406 	switch (m->type) {
407 	case FILE_DEFAULT:	/* make sure this sorts last */
408 		if (m->factor_op != FILE_FACTOR_OP_NONE)
409 			abort();
410 		return 0;
411 
412 	case FILE_BYTE:
413 		val += 1 * MULT;
414 		break;
415 
416 	case FILE_SHORT:
417 	case FILE_LESHORT:
418 	case FILE_BESHORT:
419 		val += 2 * MULT;
420 		break;
421 
422 	case FILE_LONG:
423 	case FILE_LELONG:
424 	case FILE_BELONG:
425 	case FILE_MELONG:
426 		val += 4 * MULT;
427 		break;
428 
429 	case FILE_PSTRING:
430 	case FILE_STRING:
431 		val += m->vallen * MULT;
432 		break;
433 
434 	case FILE_BESTRING16:
435 	case FILE_LESTRING16:
436 		val += m->vallen * MULT / 2;
437 		break;
438 
439 	case FILE_SEARCH:
440 	case FILE_REGEX:
441 		val += m->vallen * MAX(MULT / m->vallen, 1);
442 		break;
443 
444 	case FILE_DATE:
445 	case FILE_LEDATE:
446 	case FILE_BEDATE:
447 	case FILE_MEDATE:
448 	case FILE_LDATE:
449 	case FILE_LELDATE:
450 	case FILE_BELDATE:
451 	case FILE_MELDATE:
452 	case FILE_FLOAT:
453 	case FILE_BEFLOAT:
454 	case FILE_LEFLOAT:
455 		val += 4 * MULT;
456 		break;
457 
458 	case FILE_QUAD:
459 	case FILE_BEQUAD:
460 	case FILE_LEQUAD:
461 	case FILE_QDATE:
462 	case FILE_LEQDATE:
463 	case FILE_BEQDATE:
464 	case FILE_QLDATE:
465 	case FILE_LEQLDATE:
466 	case FILE_BEQLDATE:
467 	case FILE_DOUBLE:
468 	case FILE_BEDOUBLE:
469 	case FILE_LEDOUBLE:
470 		val += 8 * MULT;
471 		break;
472 
473 	default:
474 		val = 0;
475 		(void)fprintf(stderr, "Bad type %d\n", m->type);
476 		abort();
477 	}
478 
479 	switch (m->reln) {
480 	case 'x':	/* matches anything penalize */
481 	case '!':       /* matches almost anything penalize */
482 		val = 0;
483 		break;
484 
485 	case '=':	/* Exact match, prefer */
486 		val += MULT;
487 		break;
488 
489 	case '>':
490 	case '<':	/* comparison match reduce strength */
491 		val -= 2 * MULT;
492 		break;
493 
494 	case '^':
495 	case '&':	/* masking bits, we could count them too */
496 		val -= MULT;
497 		break;
498 
499 	default:
500 		(void)fprintf(stderr, "Bad relation %c\n", m->reln);
501 		abort();
502 	}
503 
504 	if (val == 0)	/* ensure we only return 0 for FILE_DEFAULT */
505 		val = 1;
506 
507 	switch (m->factor_op) {
508 	case FILE_FACTOR_OP_NONE:
509 		break;
510 	case FILE_FACTOR_OP_PLUS:
511 		val += m->factor;
512 		break;
513 	case FILE_FACTOR_OP_MINUS:
514 		val -= m->factor;
515 		break;
516 	case FILE_FACTOR_OP_TIMES:
517 		val *= m->factor;
518 		break;
519 	case FILE_FACTOR_OP_DIV:
520 		val /= m->factor;
521 		break;
522 	default:
523 		abort();
524 	}
525 
526 	/*
527 	 * Magic entries with no description get a bonus because they depend
528 	 * on subsequent magic entries to print something.
529 	 */
530 	if (m->desc[0] == '\0')
531 		val++;
532 	return val;
533 }
534 
535 /*
536  * Sort callback for sorting entries by "strength" (basically length)
537  */
538 private int
539 apprentice_sort(const void *a, const void *b)
540 {
541 	const struct magic_entry *ma = CAST(const struct magic_entry *, a);
542 	const struct magic_entry *mb = CAST(const struct magic_entry *, b);
543 	size_t sa = apprentice_magic_strength(ma->mp);
544 	size_t sb = apprentice_magic_strength(mb->mp);
545 	if (sa == sb)
546 		return 0;
547 	else if (sa > sb)
548 		return -1;
549 	else
550 		return 1;
551 }
552 
553 private void
554 set_test_type(struct magic *mstart, struct magic *m)
555 {
556 	switch (m->type) {
557 	case FILE_BYTE:
558 	case FILE_SHORT:
559 	case FILE_LONG:
560 	case FILE_DATE:
561 	case FILE_BESHORT:
562 	case FILE_BELONG:
563 	case FILE_BEDATE:
564 	case FILE_LESHORT:
565 	case FILE_LELONG:
566 	case FILE_LEDATE:
567 	case FILE_LDATE:
568 	case FILE_BELDATE:
569 	case FILE_LELDATE:
570 	case FILE_MEDATE:
571 	case FILE_MELDATE:
572 	case FILE_MELONG:
573 	case FILE_QUAD:
574 	case FILE_LEQUAD:
575 	case FILE_BEQUAD:
576 	case FILE_QDATE:
577 	case FILE_LEQDATE:
578 	case FILE_BEQDATE:
579 	case FILE_QLDATE:
580 	case FILE_LEQLDATE:
581 	case FILE_BEQLDATE:
582 	case FILE_FLOAT:
583 	case FILE_BEFLOAT:
584 	case FILE_LEFLOAT:
585 	case FILE_DOUBLE:
586 	case FILE_BEDOUBLE:
587 	case FILE_LEDOUBLE:
588 	case FILE_STRING:
589 	case FILE_PSTRING:
590 	case FILE_BESTRING16:
591 	case FILE_LESTRING16:
592 		/* binary test, set flag */
593 		mstart->flag |= BINTEST;
594 		break;
595 	case FILE_REGEX:
596 	case FILE_SEARCH:
597 		/* binary test if pattern is not text */
598 		if (file_looks_utf8(m->value.us, (size_t)m->vallen, NULL,
599 		    NULL) <= 0)
600 			mstart->flag |= BINTEST;
601 		break;
602 	case FILE_DEFAULT:
603 		/* can't deduce anything; we shouldn't see this at the
604 		   top level anyway */
605 		break;
606 	case FILE_INVALID:
607 	default:
608 		/* invalid search type, but no need to complain here */
609 		break;
610 	}
611 }
612 
613 /*
614  * Load and parse one file.
615  */
616 private void
617 load_1(struct magic_set *ms, int action, const char *fn, int *errs,
618    struct magic_entry **marray, uint32_t *marraycount)
619 {
620 	char line[BUFSIZ];
621 	size_t lineno = 0;
622 	FILE *f = fopen(ms->file = fn, "r");
623 	if (f == NULL) {
624 		if (errno != ENOENT)
625 			file_error(ms, errno, "cannot read magic file `%s'",
626 				   fn);
627 		(*errs)++;
628 	} else {
629 		/* read and parse this file */
630 		for (ms->line = 1; fgets(line, sizeof(line), f) != NULL; ms->line++) {
631 			size_t len;
632 			len = strlen(line);
633 			if (len == 0) /* null line, garbage, etc */
634 				continue;
635 			if (line[len - 1] == '\n') {
636 				lineno++;
637 				line[len - 1] = '\0'; /* delete newline */
638 			}
639 			if (line[0] == '\0')	/* empty, do not parse */
640 				continue;
641 			if (line[0] == '#')	/* comment, do not parse */
642 				continue;
643 			if (line[0] == '!' && line[1] == ':') {
644 				size_t i;
645 
646 				for (i = 0; bang[i].name != NULL; i++) {
647 					if (len - 2 > bang[i].len &&
648 					    memcmp(bang[i].name, line + 2,
649 					    bang[i].len) == 0)
650 						break;
651 				}
652 				if (bang[i].name == NULL) {
653 					file_error(ms, 0,
654 					    "Unknown !: entry `%s'", line);
655 					(*errs)++;
656 					continue;
657 				}
658 				if (*marraycount == 0) {
659 					file_error(ms, 0,
660 					    "No current entry for :!%s type",
661 						bang[i].name);
662 					(*errs)++;
663 					continue;
664 				}
665 				if ((*bang[i].fun)(ms,
666 				    &(*marray)[*marraycount - 1],
667 				    line + bang[i].len + 2) != 0) {
668 					(*errs)++;
669 					continue;
670 				}
671 				continue;
672 			}
673 			if (parse(ms, marray, marraycount, line, lineno,
674 			    action) != 0)
675 				(*errs)++;
676 		}
677 
678 		(void)fclose(f);
679 	}
680 }
681 
682 /*
683  * parse a file or directory of files
684  * const char *fn: name of magic file or directory
685  */
686 private int
687 apprentice_load(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
688     const char *fn, int action)
689 {
690 	int errs = 0;
691 	struct magic_entry *marray;
692 	uint32_t marraycount, i, mentrycount = 0, starttest;
693 	size_t slen;
694 	char subfn[MAXPATHLEN];
695 	struct stat st;
696 	DIR *dir;
697 	struct dirent *d;
698 
699 	ms->flags |= MAGIC_CHECK;	/* Enable checks for parsed files */
700 
701         maxmagic = MAXMAGIS;
702 	if ((marray = CAST(struct magic_entry *, calloc(maxmagic,
703 	    sizeof(*marray)))) == NULL) {
704 		file_oomem(ms, maxmagic * sizeof(*marray));
705 		return -1;
706 	}
707 	marraycount = 0;
708 
709 	/* print silly verbose header for USG compat. */
710 	if (action == FILE_CHECK)
711 		(void)fprintf(stderr, "%s\n", usg_hdr);
712 
713 	/* load directory or file */
714         /* FIXME: Read file names and sort them to prevent
715            non-determinism. See Debian bug #488562. */
716 	if (stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) {
717 		dir = opendir(fn);
718 		if (dir) {
719 			while ((d = readdir(dir)) != NULL) {
720 				snprintf(subfn, sizeof(subfn), "%s/%s",
721 				    fn, d->d_name);
722 				if (stat(subfn, &st) == 0 &&
723 				    S_ISREG(st.st_mode)) {
724 					load_1(ms, action, subfn, &errs,
725 					    &marray, &marraycount);
726 				}
727 			}
728 			closedir(dir);
729 		} else
730 			errs++;
731 	} else
732 		load_1(ms, action, fn, &errs, &marray, &marraycount);
733 	if (errs)
734 		goto out;
735 
736 	/* Set types of tests */
737 	for (i = 0; i < marraycount; ) {
738 		if (marray[i].mp->cont_level != 0) {
739 			i++;
740 			continue;
741 		}
742 
743 		starttest = i;
744 		do {
745 			static const char text[] = "text";
746 			static const char binary[] = "binary";
747 			static const size_t len = sizeof(text);
748 			set_test_type(marray[starttest].mp, marray[i].mp);
749 			if ((ms->flags & MAGIC_DEBUG) == 0)
750 				continue;
751 			(void)fprintf(stderr, "%s%s%s: %s\n",
752 			    marray[i].mp->mimetype,
753 			    marray[i].mp->mimetype[0] == '\0' ? "" : "; ",
754 			    marray[i].mp->desc[0] ? marray[i].mp->desc :
755 			    "(no description)",
756 			    marray[i].mp->flag & BINTEST ? binary : text);
757 			if (marray[i].mp->flag & BINTEST) {
758 				char *p = strstr(marray[i].mp->desc, text);
759 				if (p && (p == marray[i].mp->desc ||
760 				    isspace((unsigned char)p[-1])) &&
761 				    (p + len - marray[i].mp->desc ==
762 				    MAXstring || (p[len] == '\0' ||
763 				    isspace((unsigned char)p[len]))))
764 					(void)fprintf(stderr, "*** Possible "
765 					    "binary test for text type\n");
766 			}
767 		} while (++i < marraycount && marray[i].mp->cont_level != 0);
768 	}
769 
770 	qsort(marray, marraycount, sizeof(*marray), apprentice_sort);
771 
772 	/*
773 	 * Make sure that any level 0 "default" line is last (if one exists).
774 	 */
775 	for (i = 0; i < marraycount; i++) {
776 		if (marray[i].mp->cont_level == 0 &&
777 		    marray[i].mp->type == FILE_DEFAULT) {
778 			while (++i < marraycount)
779 				if (marray[i].mp->cont_level == 0)
780 					break;
781 			if (i != marraycount) {
782 				ms->line = marray[i].mp->lineno; /* XXX - Ugh! */
783 				file_magwarn(ms,
784 				    "level 0 \"default\" did not sort last");
785 			}
786 			break;
787 		}
788 	}
789 
790 	for (i = 0; i < marraycount; i++)
791 		mentrycount += marray[i].cont_count;
792 
793 	slen = sizeof(**magicp) * mentrycount;
794 	if ((*magicp = CAST(struct magic *, malloc(slen))) == NULL) {
795 		file_oomem(ms, slen);
796 		errs++;
797 		goto out;
798 	}
799 
800 	mentrycount = 0;
801 	for (i = 0; i < marraycount; i++) {
802 		(void)memcpy(*magicp + mentrycount, marray[i].mp,
803 		    marray[i].cont_count * sizeof(**magicp));
804 		mentrycount += marray[i].cont_count;
805 	}
806 out:
807 	for (i = 0; i < marraycount; i++)
808 		free(marray[i].mp);
809 	free(marray);
810 	if (errs) {
811 		*magicp = NULL;
812 		*nmagicp = 0;
813 		return errs;
814 	} else {
815 		*nmagicp = mentrycount;
816 		return 0;
817 	}
818 
819 }
820 
821 /*
822  * extend the sign bit if the comparison is to be signed
823  */
824 protected uint64_t
825 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v)
826 {
827 	if (!(m->flag & UNSIGNED)) {
828 		switch(m->type) {
829 		/*
830 		 * Do not remove the casts below.  They are
831 		 * vital.  When later compared with the data,
832 		 * the sign extension must have happened.
833 		 */
834 		case FILE_BYTE:
835 			v = (char) v;
836 			break;
837 		case FILE_SHORT:
838 		case FILE_BESHORT:
839 		case FILE_LESHORT:
840 			v = (short) v;
841 			break;
842 		case FILE_DATE:
843 		case FILE_BEDATE:
844 		case FILE_LEDATE:
845 		case FILE_MEDATE:
846 		case FILE_LDATE:
847 		case FILE_BELDATE:
848 		case FILE_LELDATE:
849 		case FILE_MELDATE:
850 		case FILE_LONG:
851 		case FILE_BELONG:
852 		case FILE_LELONG:
853 		case FILE_MELONG:
854 		case FILE_FLOAT:
855 		case FILE_BEFLOAT:
856 		case FILE_LEFLOAT:
857 			v = (int32_t) v;
858 			break;
859 		case FILE_QUAD:
860 		case FILE_BEQUAD:
861 		case FILE_LEQUAD:
862 		case FILE_QDATE:
863 		case FILE_QLDATE:
864 		case FILE_BEQDATE:
865 		case FILE_BEQLDATE:
866 		case FILE_LEQDATE:
867 		case FILE_LEQLDATE:
868 		case FILE_DOUBLE:
869 		case FILE_BEDOUBLE:
870 		case FILE_LEDOUBLE:
871 			v = (int64_t) v;
872 			break;
873 		case FILE_STRING:
874 		case FILE_PSTRING:
875 		case FILE_BESTRING16:
876 		case FILE_LESTRING16:
877 		case FILE_REGEX:
878 		case FILE_SEARCH:
879 		case FILE_DEFAULT:
880 		case FILE_INDIRECT:
881 			break;
882 		default:
883 			if (ms->flags & MAGIC_CHECK)
884 			    file_magwarn(ms, "cannot happen: m->type=%d\n",
885 				    m->type);
886 			return ~0U;
887 		}
888 	}
889 	return v;
890 }
891 
892 private int
893 string_modifier_check(struct magic_set *ms, struct magic *m)
894 {
895 	if ((ms->flags & MAGIC_CHECK) == 0)
896 		return 0;
897 
898 	switch (m->type) {
899 	case FILE_BESTRING16:
900 	case FILE_LESTRING16:
901 		if (m->str_flags != 0) {
902 			file_magwarn(ms,
903 			    "no modifiers allowed for 16-bit strings\n");
904 			return -1;
905 		}
906 		break;
907 	case FILE_STRING:
908 	case FILE_PSTRING:
909 		if ((m->str_flags & REGEX_OFFSET_START) != 0) {
910 			file_magwarn(ms,
911 			    "'/%c' only allowed on regex and search\n",
912 			    CHAR_REGEX_OFFSET_START);
913 			return -1;
914 		}
915 		break;
916 	case FILE_SEARCH:
917 		if (m->str_range == 0) {
918 			file_magwarn(ms,
919 			    "missing range; defaulting to %d\n",
920                             STRING_DEFAULT_RANGE);
921 			m->str_range = STRING_DEFAULT_RANGE;
922 			return -1;
923 		}
924 		break;
925 	case FILE_REGEX:
926 		if ((m->str_flags & STRING_COMPACT_BLANK) != 0) {
927 			file_magwarn(ms, "'/%c' not allowed on regex\n",
928 			    CHAR_COMPACT_BLANK);
929 			return -1;
930 		}
931 		if ((m->str_flags & STRING_COMPACT_OPTIONAL_BLANK) != 0) {
932 			file_magwarn(ms, "'/%c' not allowed on regex\n",
933 			    CHAR_COMPACT_OPTIONAL_BLANK);
934 			return -1;
935 		}
936 		break;
937 	default:
938 		file_magwarn(ms, "coding error: m->type=%d\n",
939 		    m->type);
940 		return -1;
941 	}
942 	return 0;
943 }
944 
945 private int
946 get_op(char c)
947 {
948 	switch (c) {
949 	case '&':
950 		return FILE_OPAND;
951 	case '|':
952 		return FILE_OPOR;
953 	case '^':
954 		return FILE_OPXOR;
955 	case '+':
956 		return FILE_OPADD;
957 	case '-':
958 		return FILE_OPMINUS;
959 	case '*':
960 		return FILE_OPMULTIPLY;
961 	case '/':
962 		return FILE_OPDIVIDE;
963 	case '%':
964 		return FILE_OPMODULO;
965 	default:
966 		return -1;
967 	}
968 }
969 
970 #ifdef ENABLE_CONDITIONALS
971 private int
972 get_cond(const char *l, const char **t)
973 {
974 	static const struct cond_tbl_s {
975 		char name[8];
976 		size_t len;
977 		int cond;
978 	} cond_tbl[] = {
979 		{ "if",		2,	COND_IF },
980 		{ "elif",	4,	COND_ELIF },
981 		{ "else",	4,	COND_ELSE },
982 		{ "",		0,	COND_NONE },
983 	};
984 	const struct cond_tbl_s *p;
985 
986 	for (p = cond_tbl; p->len; p++) {
987 		if (strncmp(l, p->name, p->len) == 0 &&
988 		    isspace((unsigned char)l[p->len])) {
989 			if (t)
990 				*t = l + p->len;
991 			break;
992 		}
993 	}
994 	return p->cond;
995 }
996 
997 private int
998 check_cond(struct magic_set *ms, int cond, uint32_t cont_level)
999 {
1000 	int last_cond;
1001 	last_cond = ms->c.li[cont_level].last_cond;
1002 
1003 	switch (cond) {
1004 	case COND_IF:
1005 		if (last_cond != COND_NONE && last_cond != COND_ELIF) {
1006 			if (ms->flags & MAGIC_CHECK)
1007 				file_magwarn(ms, "syntax error: `if'");
1008 			return -1;
1009 		}
1010 		last_cond = COND_IF;
1011 		break;
1012 
1013 	case COND_ELIF:
1014 		if (last_cond != COND_IF && last_cond != COND_ELIF) {
1015 			if (ms->flags & MAGIC_CHECK)
1016 				file_magwarn(ms, "syntax error: `elif'");
1017 			return -1;
1018 		}
1019 		last_cond = COND_ELIF;
1020 		break;
1021 
1022 	case COND_ELSE:
1023 		if (last_cond != COND_IF && last_cond != COND_ELIF) {
1024 			if (ms->flags & MAGIC_CHECK)
1025 				file_magwarn(ms, "syntax error: `else'");
1026 			return -1;
1027 		}
1028 		last_cond = COND_NONE;
1029 		break;
1030 
1031 	case COND_NONE:
1032 		last_cond = COND_NONE;
1033 		break;
1034 	}
1035 
1036 	ms->c.li[cont_level].last_cond = last_cond;
1037 	return 0;
1038 }
1039 #endif /* ENABLE_CONDITIONALS */
1040 
1041 /*
1042  * parse one line from magic file, put into magic[index++] if valid
1043  */
1044 private int
1045 parse(struct magic_set *ms, struct magic_entry **mentryp, uint32_t *nmentryp,
1046     const char *line, size_t lineno, int action)
1047 {
1048 #ifdef ENABLE_CONDITIONALS
1049 	static uint32_t last_cont_level = 0;
1050 #endif
1051 	size_t i;
1052 	struct magic_entry *me;
1053 	struct magic *m;
1054 	const char *l = line;
1055 	char *t;
1056 	int op;
1057 	uint32_t cont_level;
1058 
1059 	cont_level = 0;
1060 
1061 	while (*l == '>') {
1062 		++l;		/* step over */
1063 		cont_level++;
1064 	}
1065 #ifdef ENABLE_CONDITIONALS
1066 	if (cont_level == 0 || cont_level > last_cont_level)
1067 		if (file_check_mem(ms, cont_level) == -1)
1068 			return -1;
1069 	last_cont_level = cont_level;
1070 #endif
1071 
1072 #define ALLOC_CHUNK	(size_t)10
1073 #define ALLOC_INCR	(size_t)200
1074 
1075 	if (cont_level != 0) {
1076 		if (*nmentryp == 0) {
1077 			file_error(ms, 0, "No current entry for continuation");
1078 			return -1;
1079 		}
1080 		me = &(*mentryp)[*nmentryp - 1];
1081 		if (me->cont_count == me->max_count) {
1082 			struct magic *nm;
1083 			size_t cnt = me->max_count + ALLOC_CHUNK;
1084 			if ((nm = CAST(struct magic *, realloc(me->mp,
1085 			    sizeof(*nm) * cnt))) == NULL) {
1086 				file_oomem(ms, sizeof(*nm) * cnt);
1087 				return -1;
1088 			}
1089 			me->mp = m = nm;
1090 			me->max_count = cnt;
1091 		}
1092 		m = &me->mp[me->cont_count++];
1093 		(void)memset(m, 0, sizeof(*m));
1094 		m->cont_level = cont_level;
1095 	} else {
1096 		if (*nmentryp == maxmagic) {
1097 			struct magic_entry *mp;
1098 
1099 			maxmagic += ALLOC_INCR;
1100 			if ((mp = CAST(struct magic_entry *,
1101 			    realloc(*mentryp, sizeof(*mp) * maxmagic))) ==
1102 			    NULL) {
1103 				file_oomem(ms, sizeof(*mp) * maxmagic);
1104 				return -1;
1105 			}
1106 			(void)memset(&mp[*nmentryp], 0, sizeof(*mp) *
1107 			    ALLOC_INCR);
1108 			*mentryp = mp;
1109 		}
1110 		me = &(*mentryp)[*nmentryp];
1111 		if (me->mp == NULL) {
1112 			size_t len = sizeof(*m) * ALLOC_CHUNK;
1113 			if ((m = CAST(struct magic *, malloc(len))) == NULL) {
1114 				file_oomem(ms, len);
1115 				return -1;
1116 			}
1117 			me->mp = m;
1118 			me->max_count = ALLOC_CHUNK;
1119 		} else
1120 			m = me->mp;
1121 		(void)memset(m, 0, sizeof(*m));
1122 		m->factor_op = FILE_FACTOR_OP_NONE;
1123 		m->cont_level = 0;
1124 		me->cont_count = 1;
1125 	}
1126 	m->lineno = lineno;
1127 
1128 	if (*l == '&') {  /* m->cont_level == 0 checked below. */
1129                 ++l;            /* step over */
1130                 m->flag |= OFFADD;
1131         }
1132 	if (*l == '(') {
1133 		++l;		/* step over */
1134 		m->flag |= INDIR;
1135 		if (m->flag & OFFADD)
1136 			m->flag = (m->flag & ~OFFADD) | INDIROFFADD;
1137 
1138 		if (*l == '&') {  /* m->cont_level == 0 checked below */
1139 			++l;            /* step over */
1140 			m->flag |= OFFADD;
1141 		}
1142 	}
1143 	/* Indirect offsets are not valid at level 0. */
1144 	if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD)))
1145 		if (ms->flags & MAGIC_CHECK)
1146 			file_magwarn(ms, "relative offset at level 0");
1147 
1148 	/* get offset, then skip over it */
1149 	m->offset = (uint32_t)strtoul(l, &t, 0);
1150         if (l == t)
1151 		if (ms->flags & MAGIC_CHECK)
1152 			file_magwarn(ms, "offset `%s' invalid", l);
1153         l = t;
1154 
1155 	if (m->flag & INDIR) {
1156 		m->in_type = FILE_LONG;
1157 		m->in_offset = 0;
1158 		/*
1159 		 * read [.lbs][+-]nnnnn)
1160 		 */
1161 		if (*l == '.') {
1162 			l++;
1163 			switch (*l) {
1164 			case 'l':
1165 				m->in_type = FILE_LELONG;
1166 				break;
1167 			case 'L':
1168 				m->in_type = FILE_BELONG;
1169 				break;
1170 			case 'm':
1171 				m->in_type = FILE_MELONG;
1172 				break;
1173 			case 'h':
1174 			case 's':
1175 				m->in_type = FILE_LESHORT;
1176 				break;
1177 			case 'H':
1178 			case 'S':
1179 				m->in_type = FILE_BESHORT;
1180 				break;
1181 			case 'c':
1182 			case 'b':
1183 			case 'C':
1184 			case 'B':
1185 				m->in_type = FILE_BYTE;
1186 				break;
1187 			case 'e':
1188 			case 'f':
1189 			case 'g':
1190 				m->in_type = FILE_LEDOUBLE;
1191 				break;
1192 			case 'E':
1193 			case 'F':
1194 			case 'G':
1195 				m->in_type = FILE_BEDOUBLE;
1196 				break;
1197 			case 'i':
1198 				m->in_type = FILE_LEID3;
1199 				break;
1200 			case 'I':
1201 				m->in_type = FILE_BEID3;
1202 				break;
1203 			default:
1204 				if (ms->flags & MAGIC_CHECK)
1205 					file_magwarn(ms,
1206 					    "indirect offset type `%c' invalid",
1207 					    *l);
1208 				break;
1209 			}
1210 			l++;
1211 		}
1212 
1213 		m->in_op = 0;
1214 		if (*l == '~') {
1215 			m->in_op |= FILE_OPINVERSE;
1216 			l++;
1217 		}
1218 		if ((op = get_op(*l)) != -1) {
1219 			m->in_op |= op;
1220 			l++;
1221 		}
1222 		if (*l == '(') {
1223 			m->in_op |= FILE_OPINDIRECT;
1224 			l++;
1225 		}
1226 		if (isdigit((unsigned char)*l) || *l == '-') {
1227 			m->in_offset = (int32_t)strtol(l, &t, 0);
1228 			if (l == t)
1229 				if (ms->flags & MAGIC_CHECK)
1230 					file_magwarn(ms,
1231 					    "in_offset `%s' invalid", l);
1232 			l = t;
1233 		}
1234 		if (*l++ != ')' ||
1235 		    ((m->in_op & FILE_OPINDIRECT) && *l++ != ')'))
1236 			if (ms->flags & MAGIC_CHECK)
1237 				file_magwarn(ms,
1238 				    "missing ')' in indirect offset");
1239 	}
1240 	EATAB;
1241 
1242 #ifdef ENABLE_CONDITIONALS
1243 	m->cond = get_cond(l, &l);
1244 	if (check_cond(ms, m->cond, cont_level) == -1)
1245 		return -1;
1246 
1247 	EATAB;
1248 #endif
1249 
1250 	if (*l == 'u') {
1251 		++l;
1252 		m->flag |= UNSIGNED;
1253 	}
1254 
1255 	m->type = get_type(l, &l);
1256 	if (m->type == FILE_INVALID) {
1257 		if (ms->flags & MAGIC_CHECK)
1258 			file_magwarn(ms, "type `%s' invalid", l);
1259 		return -1;
1260 	}
1261 
1262 	/* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */
1263 	/* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */
1264 
1265 	m->mask_op = 0;
1266 	if (*l == '~') {
1267 		if (!IS_STRING(m->type))
1268 			m->mask_op |= FILE_OPINVERSE;
1269 		else if (ms->flags & MAGIC_CHECK)
1270 			file_magwarn(ms, "'~' invalid for string types");
1271 		++l;
1272 	}
1273 	m->str_range = 0;
1274 	m->str_flags = 0;
1275 	m->num_mask = 0;
1276 	if ((op = get_op(*l)) != -1) {
1277 		if (!IS_STRING(m->type)) {
1278 			uint64_t val;
1279 			++l;
1280 			m->mask_op |= op;
1281 			val = (uint64_t)strtoull(l, &t, 0);
1282 			l = t;
1283 			m->num_mask = file_signextend(ms, m, val);
1284 			eatsize(&l);
1285 		}
1286 		else if (op == FILE_OPDIVIDE) {
1287 			int have_range = 0;
1288 			while (!isspace((unsigned char)*++l)) {
1289 				switch (*l) {
1290 				case '0':  case '1':  case '2':
1291 				case '3':  case '4':  case '5':
1292 				case '6':  case '7':  case '8':
1293 				case '9':
1294 					if (have_range &&
1295 					    (ms->flags & MAGIC_CHECK))
1296 						file_magwarn(ms,
1297 						    "multiple ranges");
1298 					have_range = 1;
1299 					m->str_range = strtoul(l, &t, 0);
1300 					if (m->str_range == 0)
1301 						file_magwarn(ms,
1302 						    "zero range");
1303 					l = t - 1;
1304 					break;
1305 				case CHAR_COMPACT_BLANK:
1306 					m->str_flags |= STRING_COMPACT_BLANK;
1307 					break;
1308 				case CHAR_COMPACT_OPTIONAL_BLANK:
1309 					m->str_flags |=
1310 					    STRING_COMPACT_OPTIONAL_BLANK;
1311 					break;
1312 				case CHAR_IGNORE_LOWERCASE:
1313 					m->str_flags |= STRING_IGNORE_LOWERCASE;
1314 					break;
1315 				case CHAR_IGNORE_UPPERCASE:
1316 					m->str_flags |= STRING_IGNORE_UPPERCASE;
1317 					break;
1318 				case CHAR_REGEX_OFFSET_START:
1319 					m->str_flags |= REGEX_OFFSET_START;
1320 					break;
1321 				default:
1322 					if (ms->flags & MAGIC_CHECK)
1323 						file_magwarn(ms,
1324 						"string extension `%c' invalid",
1325 						*l);
1326 					return -1;
1327 				}
1328 				/* allow multiple '/' for readability */
1329 				if (l[1] == '/' &&
1330 				    !isspace((unsigned char)l[2]))
1331 					l++;
1332 			}
1333 			if (string_modifier_check(ms, m) == -1)
1334 				return -1;
1335 		}
1336 		else {
1337 			if (ms->flags & MAGIC_CHECK)
1338 				file_magwarn(ms, "invalid string op: %c", *t);
1339 			return -1;
1340 		}
1341 	}
1342 	/*
1343 	 * We used to set mask to all 1's here, instead let's just not do
1344 	 * anything if mask = 0 (unless you have a better idea)
1345 	 */
1346 	EATAB;
1347 
1348 	switch (*l) {
1349 	case '>':
1350 	case '<':
1351   		m->reln = *l;
1352   		++l;
1353 		if (*l == '=') {
1354 			if (ms->flags & MAGIC_CHECK) {
1355 				file_magwarn(ms, "%c= not supported",
1356 				    m->reln);
1357 				return -1;
1358 			}
1359 		   ++l;
1360 		}
1361 		break;
1362 	/* Old-style anding: "0 byte &0x80 dynamically linked" */
1363 	case '&':
1364 	case '^':
1365 	case '=':
1366   		m->reln = *l;
1367   		++l;
1368 		if (*l == '=') {
1369 		   /* HP compat: ignore &= etc. */
1370 		   ++l;
1371 		}
1372 		break;
1373 	case '!':
1374 		m->reln = *l;
1375 		++l;
1376 		break;
1377 	default:
1378   		m->reln = '=';	/* the default relation */
1379 		if (*l == 'x' && ((isascii((unsigned char)l[1]) &&
1380 		    isspace((unsigned char)l[1])) || !l[1])) {
1381 			m->reln = *l;
1382 			++l;
1383 		}
1384 		break;
1385 	}
1386 	/*
1387 	 * Grab the value part, except for an 'x' reln.
1388 	 */
1389 	if (m->reln != 'x' && getvalue(ms, m, &l, action))
1390 		return -1;
1391 
1392 	/*
1393 	 * TODO finish this macro and start using it!
1394 	 * #define offsetcheck {if (offset > HOWMANY-1)
1395 	 *	magwarn("offset too big"); }
1396 	 */
1397 
1398 	/*
1399 	 * Now get last part - the description
1400 	 */
1401 	EATAB;
1402 	if (l[0] == '\b') {
1403 		++l;
1404 		m->flag |= NOSPACE;
1405 	} else if ((l[0] == '\\') && (l[1] == 'b')) {
1406 		++l;
1407 		++l;
1408 		m->flag |= NOSPACE;
1409 	}
1410 	for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); )
1411 		continue;
1412 	if (i == sizeof(m->desc)) {
1413 		m->desc[sizeof(m->desc) - 1] = '\0';
1414 		if (ms->flags & MAGIC_CHECK)
1415 			file_magwarn(ms, "description `%s' truncated", m->desc);
1416 	}
1417 
1418         /*
1419 	 * We only do this check while compiling, or if any of the magic
1420 	 * files were not compiled.
1421          */
1422         if (ms->flags & MAGIC_CHECK) {
1423 		if (check_format(ms, m) == -1)
1424 			return -1;
1425 	}
1426 #ifndef COMPILE_ONLY
1427 	if (action == FILE_CHECK) {
1428 		file_mdump(m);
1429 	}
1430 #endif
1431 	m->mimetype[0] = '\0';		/* initialise MIME type to none */
1432 	if (m->cont_level == 0)
1433 		++(*nmentryp);		/* make room for next */
1434 	return 0;
1435 }
1436 
1437 /*
1438  * parse a STRENGTH annotation line from magic file, put into magic[index - 1]
1439  * if valid
1440  */
1441 private int
1442 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line)
1443 {
1444 	const char *l = line;
1445 	char *el;
1446 	unsigned long factor;
1447 	struct magic *m = &me->mp[0];
1448 
1449 	if (m->factor_op != FILE_FACTOR_OP_NONE) {
1450 		file_magwarn(ms,
1451 		    "Current entry already has a strength type: %c %d",
1452 		    m->factor_op, m->factor);
1453 		return -1;
1454 	}
1455 	EATAB;
1456 	switch (*l) {
1457 	case FILE_FACTOR_OP_NONE:
1458 	case FILE_FACTOR_OP_PLUS:
1459 	case FILE_FACTOR_OP_MINUS:
1460 	case FILE_FACTOR_OP_TIMES:
1461 	case FILE_FACTOR_OP_DIV:
1462 		m->factor_op = *l++;
1463 		break;
1464 	default:
1465 		file_magwarn(ms, "Unknown factor op `%c'", *l);
1466 		return -1;
1467 	}
1468 	EATAB;
1469 	factor = strtoul(l, &el, 0);
1470 	if (factor > 255) {
1471 		file_magwarn(ms, "Too large factor `%lu'", factor);
1472 		goto out;
1473 	}
1474 	if (*el && !isspace((unsigned char)*el)) {
1475 		file_magwarn(ms, "Bad factor `%s'", l);
1476 		goto out;
1477 	}
1478 	m->factor = (uint8_t)factor;
1479 	if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) {
1480 		file_magwarn(ms, "Cannot have factor op `%c' and factor %u",
1481 		    m->factor_op, m->factor);
1482 		goto out;
1483 	}
1484 	return 0;
1485 out:
1486 	m->factor_op = FILE_FACTOR_OP_NONE;
1487 	m->factor = 0;
1488 	return -1;
1489 }
1490 
1491 /*
1492  * Parse an Apple CREATOR/TYPE annotation from magic file and put it into magic[index - 1]
1493  */
1494 private int
1495 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line)
1496 {
1497 	size_t i;
1498 	const char *l = line;
1499 	struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1];
1500 
1501 	if (m->apple[0] != '\0') {
1502 		file_magwarn(ms, "Current entry already has a APPLE type `%.8s',"
1503 		    " new type `%s'", m->mimetype, l);
1504 		return -1;
1505 	}
1506 
1507 	EATAB;
1508 	for (i = 0; *l && ((isascii((unsigned char)*l) && isalnum((unsigned char)*l))
1509 	     || strchr("-+/.", *l)) && i < sizeof(m->apple); m->apple[i++] = *l++)
1510 		continue;
1511 	if (i == sizeof(m->apple) && *l) {
1512 		if (ms->flags & MAGIC_CHECK)
1513 			file_magwarn(ms, "APPLE type `%s' truncated %zu",
1514 			    line, i);
1515 	}
1516 
1517 	if (i > 0)
1518 		return 0;
1519 	else
1520 		return -1;
1521 }
1522 
1523 /*
1524  * parse a MIME annotation line from magic file, put into magic[index - 1]
1525  * if valid
1526  */
1527 private int
1528 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line)
1529 {
1530 	size_t i;
1531 	const char *l = line;
1532 	struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1];
1533 
1534 	if (m->mimetype[0] != '\0') {
1535 		file_magwarn(ms, "Current entry already has a MIME type `%s',"
1536 		    " new type `%s'", m->mimetype, l);
1537 		return -1;
1538 	}
1539 
1540 	EATAB;
1541 	for (i = 0; *l && ((isascii((unsigned char)*l) && isalnum((unsigned char)*l))
1542 	     || strchr("-+/.", *l)) && i < sizeof(m->mimetype); m->mimetype[i++] = *l++)
1543 		continue;
1544 	if (i == sizeof(m->mimetype)) {
1545 		m->desc[sizeof(m->mimetype) - 1] = '\0';
1546 		if (ms->flags & MAGIC_CHECK)
1547 			file_magwarn(ms, "MIME type `%s' truncated %zu",
1548 			    m->mimetype, i);
1549 	} else
1550 		m->mimetype[i] = '\0';
1551 
1552 	if (i > 0)
1553 		return 0;
1554 	else
1555 		return -1;
1556 }
1557 
1558 private int
1559 check_format_type(const char *ptr, int type)
1560 {
1561 	int quad = 0;
1562 	if (*ptr == '\0') {
1563 		/* Missing format string; bad */
1564 		return -1;
1565 	}
1566 
1567 	switch (type) {
1568 	case FILE_FMT_QUAD:
1569 		quad = 1;
1570 		/*FALLTHROUGH*/
1571 	case FILE_FMT_NUM:
1572 		if (*ptr == '-')
1573 			ptr++;
1574 		if (*ptr == '.')
1575 			ptr++;
1576 		while (isdigit((unsigned char)*ptr)) ptr++;
1577 		if (*ptr == '.')
1578 			ptr++;
1579 		while (isdigit((unsigned char)*ptr)) ptr++;
1580 		if (quad) {
1581 			if (*ptr++ != 'l')
1582 				return -1;
1583 			if (*ptr++ != 'l')
1584 				return -1;
1585 		}
1586 
1587 		switch (*ptr++) {
1588 		case 'l':
1589 			switch (*ptr++) {
1590 			case 'i':
1591 			case 'd':
1592 			case 'u':
1593 			case 'x':
1594 			case 'X':
1595 				return 0;
1596 			default:
1597 				return -1;
1598 			}
1599 
1600 		case 'h':
1601 			switch (*ptr++) {
1602 			case 'h':
1603 				switch (*ptr++) {
1604 				case 'i':
1605 				case 'd':
1606 				case 'u':
1607 				case 'x':
1608 				case 'X':
1609 					return 0;
1610 				default:
1611 					return -1;
1612 				}
1613 			case 'd':
1614 				return 0;
1615 			default:
1616 				return -1;
1617 			}
1618 
1619 		case 'i':
1620 		case 'c':
1621 		case 'd':
1622 		case 'u':
1623 		case 'x':
1624 		case 'X':
1625 			return 0;
1626 
1627 		default:
1628 			return -1;
1629 		}
1630 
1631 	case FILE_FMT_FLOAT:
1632 	case FILE_FMT_DOUBLE:
1633 		if (*ptr == '-')
1634 			ptr++;
1635 		if (*ptr == '.')
1636 			ptr++;
1637 		while (isdigit((unsigned char)*ptr)) ptr++;
1638 		if (*ptr == '.')
1639 			ptr++;
1640 		while (isdigit((unsigned char)*ptr)) ptr++;
1641 
1642 		switch (*ptr++) {
1643 		case 'e':
1644 		case 'E':
1645 		case 'f':
1646 		case 'F':
1647 		case 'g':
1648 		case 'G':
1649 			return 0;
1650 
1651 		default:
1652 			return -1;
1653 		}
1654 
1655 
1656 	case FILE_FMT_STR:
1657 		if (*ptr == '-')
1658 			ptr++;
1659 		while (isdigit((unsigned char )*ptr))
1660 			ptr++;
1661 		if (*ptr == '.') {
1662 			ptr++;
1663 			while (isdigit((unsigned char )*ptr))
1664 				ptr++;
1665 		}
1666 
1667 		switch (*ptr++) {
1668 		case 's':
1669 			return 0;
1670 		default:
1671 			return -1;
1672 		}
1673 
1674 	default:
1675 		/* internal error */
1676 		abort();
1677 	}
1678 	/*NOTREACHED*/
1679 	return -1;
1680 }
1681 
1682 /*
1683  * Check that the optional printf format in description matches
1684  * the type of the magic.
1685  */
1686 private int
1687 check_format(struct magic_set *ms, struct magic *m)
1688 {
1689 	char *ptr;
1690 
1691 	for (ptr = m->desc; *ptr; ptr++)
1692 		if (*ptr == '%')
1693 			break;
1694 	if (*ptr == '\0') {
1695 		/* No format string; ok */
1696 		return 1;
1697 	}
1698 
1699 	assert(file_nformats == file_nnames);
1700 
1701 	if (m->type >= file_nformats) {
1702 		file_magwarn(ms, "Internal error inconsistency between "
1703 		    "m->type and format strings");
1704 		return -1;
1705 	}
1706 	if (file_formats[m->type] == FILE_FMT_NONE) {
1707 		file_magwarn(ms, "No format string for `%s' with description "
1708 		    "`%s'", m->desc, file_names[m->type]);
1709 		return -1;
1710 	}
1711 
1712 	ptr++;
1713 	if (check_format_type(ptr, file_formats[m->type]) == -1) {
1714 		/*
1715 		 * TODO: this error message is unhelpful if the format
1716 		 * string is not one character long
1717 		 */
1718 		file_magwarn(ms, "Printf format `%c' is not valid for type "
1719 		    "`%s' in description `%s'", *ptr ? *ptr : '?',
1720 		    file_names[m->type], m->desc);
1721 		return -1;
1722 	}
1723 
1724 	for (; *ptr; ptr++) {
1725 		if (*ptr == '%') {
1726 			file_magwarn(ms,
1727 			    "Too many format strings (should have at most one) "
1728 			    "for `%s' with description `%s'",
1729 			    file_names[m->type], m->desc);
1730 			return -1;
1731 		}
1732 	}
1733 	return 0;
1734 }
1735 
1736 /*
1737  * Read a numeric value from a pointer, into the value union of a magic
1738  * pointer, according to the magic type.  Update the string pointer to point
1739  * just after the number read.  Return 0 for success, non-zero for failure.
1740  */
1741 private int
1742 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action)
1743 {
1744 	switch (m->type) {
1745 	case FILE_BESTRING16:
1746 	case FILE_LESTRING16:
1747 	case FILE_STRING:
1748 	case FILE_PSTRING:
1749 	case FILE_REGEX:
1750 	case FILE_SEARCH:
1751 		*p = getstr(ms, m, *p, action == FILE_COMPILE);
1752 		if (*p == NULL) {
1753 			if (ms->flags & MAGIC_CHECK)
1754 				file_magwarn(ms, "cannot get string from `%s'",
1755 				    m->value.s);
1756 			return -1;
1757 		}
1758 		return 0;
1759 	case FILE_FLOAT:
1760 	case FILE_BEFLOAT:
1761 	case FILE_LEFLOAT:
1762 		if (m->reln != 'x') {
1763 			char *ep;
1764 #ifdef HAVE_STRTOF
1765 			m->value.f = strtof(*p, &ep);
1766 #else
1767 			m->value.f = (float)strtod(*p, &ep);
1768 #endif
1769 			*p = ep;
1770 		}
1771 		return 0;
1772 	case FILE_DOUBLE:
1773 	case FILE_BEDOUBLE:
1774 	case FILE_LEDOUBLE:
1775 		if (m->reln != 'x') {
1776 			char *ep;
1777 			m->value.d = strtod(*p, &ep);
1778 			*p = ep;
1779 		}
1780 		return 0;
1781 	default:
1782 		if (m->reln != 'x') {
1783 			char *ep;
1784 			m->value.q = file_signextend(ms, m,
1785 			    (uint64_t)strtoull(*p, &ep, 0));
1786 			*p = ep;
1787 			eatsize(p);
1788 		}
1789 		return 0;
1790 	}
1791 }
1792 
1793 /*
1794  * Convert a string containing C character escapes.  Stop at an unescaped
1795  * space or tab.
1796  * Copy the converted version to "m->value.s", and the length in m->vallen.
1797  * Return updated scan pointer as function result. Warn if set.
1798  */
1799 private const char *
1800 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn)
1801 {
1802 	const char *origs = s;
1803 	char	*p = m->value.s;
1804 	size_t  plen = sizeof(m->value.s);
1805 	char 	*origp = p;
1806 	char	*pmax = p + plen - 1;
1807 	int	c;
1808 	int	val;
1809 
1810 	while ((c = *s++) != '\0') {
1811 		if (isspace((unsigned char) c))
1812 			break;
1813 		if (p >= pmax) {
1814 			file_error(ms, 0, "string too long: `%s'", origs);
1815 			return NULL;
1816 		}
1817 		if (c == '\\') {
1818 			switch(c = *s++) {
1819 
1820 			case '\0':
1821 				if (warn)
1822 					file_magwarn(ms, "incomplete escape");
1823 				goto out;
1824 
1825 			case '\t':
1826 				if (warn) {
1827 					file_magwarn(ms,
1828 					    "escaped tab found, use \\t instead");
1829 					warn = 0;	/* already did */
1830 				}
1831 				/*FALLTHROUGH*/
1832 			default:
1833 				if (warn) {
1834 					if (isprint((unsigned char)c)) {
1835 						/* Allow escaping of
1836 						 * ``relations'' */
1837 						if (strchr("<>&^=!", c)
1838 						    == NULL) {
1839 							file_magwarn(ms, "no "
1840 							    "need to escape "
1841 							    "`%c'", c);
1842 						}
1843 					} else {
1844 						file_magwarn(ms,
1845 						    "unknown escape sequence: "
1846 						    "\\%03o", c);
1847 					}
1848 				}
1849 				/*FALLTHROUGH*/
1850 			/* space, perhaps force people to use \040? */
1851 			case ' ':
1852 #if 0
1853 			/*
1854 			 * Other things people escape, but shouldn't need to,
1855 			 * so we disallow them
1856 			 */
1857 			case '\'':
1858 			case '"':
1859 			case '?':
1860 #endif
1861 			/* Relations */
1862 			case '>':
1863 			case '<':
1864 			case '&':
1865 			case '^':
1866 			case '=':
1867 			case '!':
1868 			/* and baskslash itself */
1869 			case '\\':
1870 				*p++ = (char) c;
1871 				break;
1872 
1873 			case 'a':
1874 				*p++ = '\a';
1875 				break;
1876 
1877 			case 'b':
1878 				*p++ = '\b';
1879 				break;
1880 
1881 			case 'f':
1882 				*p++ = '\f';
1883 				break;
1884 
1885 			case 'n':
1886 				*p++ = '\n';
1887 				break;
1888 
1889 			case 'r':
1890 				*p++ = '\r';
1891 				break;
1892 
1893 			case 't':
1894 				*p++ = '\t';
1895 				break;
1896 
1897 			case 'v':
1898 				*p++ = '\v';
1899 				break;
1900 
1901 			/* \ and up to 3 octal digits */
1902 			case '0':
1903 			case '1':
1904 			case '2':
1905 			case '3':
1906 			case '4':
1907 			case '5':
1908 			case '6':
1909 			case '7':
1910 				val = c - '0';
1911 				c = *s++;  /* try for 2 */
1912 				if (c >= '0' && c <= '7') {
1913 					val = (val << 3) | (c - '0');
1914 					c = *s++;  /* try for 3 */
1915 					if (c >= '0' && c <= '7')
1916 						val = (val << 3) | (c-'0');
1917 					else
1918 						--s;
1919 				}
1920 				else
1921 					--s;
1922 				*p++ = (char)val;
1923 				break;
1924 
1925 			/* \x and up to 2 hex digits */
1926 			case 'x':
1927 				val = 'x';	/* Default if no digits */
1928 				c = hextoint(*s++);	/* Get next char */
1929 				if (c >= 0) {
1930 					val = c;
1931 					c = hextoint(*s++);
1932 					if (c >= 0)
1933 						val = (val << 4) + c;
1934 					else
1935 						--s;
1936 				} else
1937 					--s;
1938 				*p++ = (char)val;
1939 				break;
1940 			}
1941 		} else
1942 			*p++ = (char)c;
1943 	}
1944 out:
1945 	*p = '\0';
1946 	m->vallen = p - origp;
1947 	if (m->type == FILE_PSTRING)
1948 		m->vallen++;
1949 	return s;
1950 }
1951 
1952 
1953 /* Single hex char to int; -1 if not a hex char. */
1954 private int
1955 hextoint(int c)
1956 {
1957 	if (!isascii((unsigned char) c))
1958 		return -1;
1959 	if (isdigit((unsigned char) c))
1960 		return c - '0';
1961 	if ((c >= 'a') && (c <= 'f'))
1962 		return c + 10 - 'a';
1963 	if (( c>= 'A') && (c <= 'F'))
1964 		return c + 10 - 'A';
1965 	return -1;
1966 }
1967 
1968 
1969 /*
1970  * Print a string containing C character escapes.
1971  */
1972 protected void
1973 file_showstr(FILE *fp, const char *s, size_t len)
1974 {
1975 	char	c;
1976 
1977 	for (;;) {
1978 		c = *s++;
1979 		if (len == ~0U) {
1980 			if (c == '\0')
1981 				break;
1982 		}
1983 		else  {
1984 			if (len-- == 0)
1985 				break;
1986 		}
1987 		if (c >= 040 && c <= 0176)	/* TODO isprint && !iscntrl */
1988 			(void) fputc(c, fp);
1989 		else {
1990 			(void) fputc('\\', fp);
1991 			switch (c) {
1992 			case '\a':
1993 				(void) fputc('a', fp);
1994 				break;
1995 
1996 			case '\b':
1997 				(void) fputc('b', fp);
1998 				break;
1999 
2000 			case '\f':
2001 				(void) fputc('f', fp);
2002 				break;
2003 
2004 			case '\n':
2005 				(void) fputc('n', fp);
2006 				break;
2007 
2008 			case '\r':
2009 				(void) fputc('r', fp);
2010 				break;
2011 
2012 			case '\t':
2013 				(void) fputc('t', fp);
2014 				break;
2015 
2016 			case '\v':
2017 				(void) fputc('v', fp);
2018 				break;
2019 
2020 			default:
2021 				(void) fprintf(fp, "%.3o", c & 0377);
2022 				break;
2023 			}
2024 		}
2025 	}
2026 }
2027 
2028 /*
2029  * eatsize(): Eat the size spec from a number [eg. 10UL]
2030  */
2031 private void
2032 eatsize(const char **p)
2033 {
2034 	const char *l = *p;
2035 
2036 	if (LOWCASE(*l) == 'u')
2037 		l++;
2038 
2039 	switch (LOWCASE(*l)) {
2040 	case 'l':    /* long */
2041 	case 's':    /* short */
2042 	case 'h':    /* short */
2043 	case 'b':    /* char/byte */
2044 	case 'c':    /* char/byte */
2045 		l++;
2046 		/*FALLTHROUGH*/
2047 	default:
2048 		break;
2049 	}
2050 
2051 	*p = l;
2052 }
2053 
2054 /*
2055  * handle a compiled file.
2056  */
2057 private int
2058 apprentice_map(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
2059     const char *fn)
2060 {
2061 	int fd;
2062 	struct stat st;
2063 	uint32_t *ptr;
2064 	uint32_t version;
2065 	int needsbyteswap;
2066 	char *dbname = NULL;
2067 	void *mm = NULL;
2068 
2069 	dbname = mkdbname(ms, fn, 0);
2070 	if (dbname == NULL)
2071 		goto error2;
2072 
2073 	if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1)
2074 		goto error2;
2075 
2076 	if (fstat(fd, &st) == -1) {
2077 		file_error(ms, errno, "cannot stat `%s'", dbname);
2078 		goto error1;
2079 	}
2080 	if (st.st_size < 8) {
2081 		file_error(ms, 0, "file `%s' is too small", dbname);
2082 		goto error1;
2083 	}
2084 
2085 #ifdef QUICK
2086 	if ((mm = mmap(0, (size_t)st.st_size, PROT_READ|PROT_WRITE,
2087 	    MAP_PRIVATE|MAP_FILE, fd, (off_t)0)) == MAP_FAILED) {
2088 		file_error(ms, errno, "cannot map `%s'", dbname);
2089 		goto error1;
2090 	}
2091 #define RET	2
2092 #else
2093 	if ((mm = CAST(void *, malloc((size_t)st.st_size))) == NULL) {
2094 		file_oomem(ms, (size_t)st.st_size);
2095 		goto error1;
2096 	}
2097 	if (read(fd, mm, (size_t)st.st_size) != (ssize_t)st.st_size) {
2098 		file_badread(ms);
2099 		goto error1;
2100 	}
2101 #define RET	1
2102 #endif
2103 	*magicp = CAST(struct magic *, mm);
2104 	(void)close(fd);
2105 	fd = -1;
2106 	ptr = (uint32_t *)(void *)*magicp;
2107 	if (*ptr != MAGICNO) {
2108 		if (swap4(*ptr) != MAGICNO) {
2109 			file_error(ms, 0, "bad magic in `%s'", dbname);
2110 			goto error1;
2111 		}
2112 		needsbyteswap = 1;
2113 	} else
2114 		needsbyteswap = 0;
2115 	if (needsbyteswap)
2116 		version = swap4(ptr[1]);
2117 	else
2118 		version = ptr[1];
2119 	if (version != VERSIONNO) {
2120 		file_error(ms, 0, "File %d.%d supports only version %d magic "
2121 		    "files. `%s' is version %d", FILE_VERSION_MAJOR, patchlevel,
2122 		    VERSIONNO, dbname, version);
2123 		goto error1;
2124 	}
2125 	*nmagicp = (uint32_t)(st.st_size / sizeof(struct magic));
2126 	if (*nmagicp > 0)
2127 		(*nmagicp)--;
2128 	(*magicp)++;
2129 	if (needsbyteswap)
2130 		byteswap(*magicp, *nmagicp);
2131 	free(dbname);
2132 	return RET;
2133 
2134 error1:
2135 	if (fd != -1)
2136 		(void)close(fd);
2137 	if (mm) {
2138 #ifdef QUICK
2139 		(void)munmap((void *)mm, (size_t)st.st_size);
2140 #else
2141 		free(mm);
2142 #endif
2143 	} else {
2144 		*magicp = NULL;
2145 		*nmagicp = 0;
2146 	}
2147 error2:
2148 	free(dbname);
2149 	return -1;
2150 }
2151 
2152 private const uint32_t ar[] = {
2153     MAGICNO, VERSIONNO
2154 };
2155 /*
2156  * handle an mmaped file.
2157  */
2158 private int
2159 apprentice_compile(struct magic_set *ms, struct magic **magicp,
2160     uint32_t *nmagicp, const char *fn)
2161 {
2162 	int fd;
2163 	char *dbname;
2164 	int rv = -1;
2165 
2166 	dbname = mkdbname(ms, fn, 1);
2167 
2168 	if (dbname == NULL)
2169 		goto out;
2170 
2171 	if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) {
2172 		file_error(ms, errno, "cannot open `%s'", dbname);
2173 		goto out;
2174 	}
2175 
2176 	if (write(fd, ar, sizeof(ar)) != (ssize_t)sizeof(ar)) {
2177 		file_error(ms, errno, "error writing `%s'", dbname);
2178 		goto out;
2179 	}
2180 
2181 	if (lseek(fd, (off_t)sizeof(struct magic), SEEK_SET)
2182 	    != sizeof(struct magic)) {
2183 		file_error(ms, errno, "error seeking `%s'", dbname);
2184 		goto out;
2185 	}
2186 
2187 	if (write(fd, *magicp, (sizeof(struct magic) * *nmagicp))
2188 	    != (ssize_t)(sizeof(struct magic) * *nmagicp)) {
2189 		file_error(ms, errno, "error writing `%s'", dbname);
2190 		goto out;
2191 	}
2192 
2193 	(void)close(fd);
2194 	rv = 0;
2195 out:
2196 	free(dbname);
2197 	return rv;
2198 }
2199 
2200 private const char ext[] = ".mgc";
2201 /*
2202  * make a dbname
2203  */
2204 private char *
2205 mkdbname(struct magic_set *ms, const char *fn, int strip)
2206 {
2207 	const char *p, *q;
2208 	char *buf;
2209 
2210 	if (strip) {
2211 		if ((p = strrchr(fn, '/')) != NULL)
2212 			fn = ++p;
2213 	}
2214 
2215 	for (q = fn; *q; q++)
2216 		continue;
2217 	/* Look for .mgc */
2218 	for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--)
2219 		if (*p != *q)
2220 			break;
2221 
2222 	/* Did not find .mgc, restore q */
2223 	if (p >= ext)
2224 		while (*q)
2225 			q++;
2226 
2227 	q++;
2228 	/* Compatibility with old code that looked in .mime */
2229 	if (ms->flags & MAGIC_MIME) {
2230 		asprintf(&buf, "%.*s.mime%s", (int)(q - fn), fn, ext);
2231 		if (access(buf, R_OK) != -1) {
2232 			ms->flags &= MAGIC_MIME_TYPE;
2233 			return buf;
2234 		}
2235 		free(buf);
2236 	}
2237 	asprintf(&buf, "%.*s%s", (int)(q - fn), fn, ext);
2238 
2239 	/* Compatibility with old code that looked in .mime */
2240 	if (strstr(p, ".mime") != NULL)
2241 		ms->flags &= MAGIC_MIME_TYPE;
2242 	return buf;
2243 }
2244 
2245 /*
2246  * Byteswap an mmap'ed file if needed
2247  */
2248 private void
2249 byteswap(struct magic *magic, uint32_t nmagic)
2250 {
2251 	uint32_t i;
2252 	for (i = 0; i < nmagic; i++)
2253 		bs1(&magic[i]);
2254 }
2255 
2256 /*
2257  * swap a short
2258  */
2259 private uint16_t
2260 swap2(uint16_t sv)
2261 {
2262 	uint16_t rv;
2263 	uint8_t *s = (uint8_t *)(void *)&sv;
2264 	uint8_t *d = (uint8_t *)(void *)&rv;
2265 	d[0] = s[1];
2266 	d[1] = s[0];
2267 	return rv;
2268 }
2269 
2270 /*
2271  * swap an int
2272  */
2273 private uint32_t
2274 swap4(uint32_t sv)
2275 {
2276 	uint32_t rv;
2277 	uint8_t *s = (uint8_t *)(void *)&sv;
2278 	uint8_t *d = (uint8_t *)(void *)&rv;
2279 	d[0] = s[3];
2280 	d[1] = s[2];
2281 	d[2] = s[1];
2282 	d[3] = s[0];
2283 	return rv;
2284 }
2285 
2286 /*
2287  * swap a quad
2288  */
2289 private uint64_t
2290 swap8(uint64_t sv)
2291 {
2292 	uint64_t rv;
2293 	uint8_t *s = (uint8_t *)(void *)&sv;
2294 	uint8_t *d = (uint8_t *)(void *)&rv;
2295 #if 0
2296 	d[0] = s[3];
2297 	d[1] = s[2];
2298 	d[2] = s[1];
2299 	d[3] = s[0];
2300 	d[4] = s[7];
2301 	d[5] = s[6];
2302 	d[6] = s[5];
2303 	d[7] = s[4];
2304 #else
2305 	d[0] = s[7];
2306 	d[1] = s[6];
2307 	d[2] = s[5];
2308 	d[3] = s[4];
2309 	d[4] = s[3];
2310 	d[5] = s[2];
2311 	d[6] = s[1];
2312 	d[7] = s[0];
2313 #endif
2314 	return rv;
2315 }
2316 
2317 /*
2318  * byteswap a single magic entry
2319  */
2320 private void
2321 bs1(struct magic *m)
2322 {
2323 	m->cont_level = swap2(m->cont_level);
2324 	m->offset = swap4((uint32_t)m->offset);
2325 	m->in_offset = swap4((uint32_t)m->in_offset);
2326 	m->lineno = swap4((uint32_t)m->lineno);
2327 	if (IS_STRING(m->type)) {
2328 		m->str_range = swap4(m->str_range);
2329 		m->str_flags = swap4(m->str_flags);
2330 	}
2331 	else {
2332 		m->value.q = swap8(m->value.q);
2333 		m->num_mask = swap8(m->num_mask);
2334 	}
2335 }
2336