xref: /freebsd/contrib/file/src/funcs.c (revision 2e3507c25e42292b45a5482e116d278f5515d04d) 
1 /*
2  * Copyright (c) Christos Zoulas 2003.
3  * All Rights Reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice immediately at the beginning of the file, without modification,
10  *    this list of conditions, and the following disclaimer.
11  * 2. Redistributions in binary form must reproduce the above copyright
12  *    notice, this list of conditions and the following disclaimer in the
13  *    documentation and/or other materials provided with the distribution.
14  *
15  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
16  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18  * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
19  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
20  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
21  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
22  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
25  * SUCH DAMAGE.
26  */
27 #include "file.h"
28 
29 #ifndef	lint
30 FILE_RCSID("@(#)$File: funcs.c,v 1.140 2023/05/21 17:08:34 christos Exp $")
31 #endif	/* lint */
32 
33 #include "magic.h"
34 #include <assert.h>
35 #include <stdarg.h>
36 #include <stdlib.h>
37 #include <string.h>
38 #include <ctype.h>
39 #ifdef HAVE_UNISTD_H
40 #include <unistd.h>	/* for pipe2() */
41 #endif
42 #if defined(HAVE_WCHAR_H)
43 #include <wchar.h>
44 #endif
45 #if defined(HAVE_WCTYPE_H)
46 #include <wctype.h>
47 #endif
48 #include <limits.h>
49 
50 #ifndef SIZE_MAX
51 #define SIZE_MAX	((size_t)~0)
52 #endif
53 
54 file_protected char *
55 file_copystr(char *buf, size_t blen, size_t width, const char *str)
56 {
57 	if (blen == 0)
58 		return buf;
59 	if (width >= blen)
60 		width = blen - 1;
61 	memcpy(buf, str, width);
62 	buf[width] = '\0';
63 	return buf;
64 }
65 
66 file_private void
67 file_clearbuf(struct magic_set *ms)
68 {
69 	free(ms->o.buf);
70 	ms->o.buf = NULL;
71 	ms->o.blen = 0;
72 }
73 
74 file_private int
75 file_checkfield(char *msg, size_t mlen, const char *what, const char **pp)
76 {
77 	const char *p = *pp;
78 	int fw = 0;
79 
80 	while (*p && isdigit((unsigned char)*p))
81 		fw = fw * 10 + (*p++ - '0');
82 
83 	*pp = p;
84 
85 	if (fw < 1024)
86 		return 1;
87 	if (msg)
88 		snprintf(msg, mlen, "field %s too large: %d", what, fw);
89 
90 	return 0;
91 }
92 
93 file_protected int
94 file_checkfmt(char *msg, size_t mlen, const char *fmt)
95 {
96 	const char *p;
97 	for (p = fmt; *p; p++) {
98 		if (*p != '%')
99 			continue;
100 		if (*++p == '%')
101 			continue;
102 		// Skip uninteresting.
103 		while (strchr("#0.'+- ", *p) != NULL)
104 			p++;
105 		if (*p == '*') {
106 			if (msg)
107 				snprintf(msg, mlen, "* not allowed in format");
108 			return -1;
109 		}
110 
111 		if (!file_checkfield(msg, mlen, "width", &p))
112 			return -1;
113 
114 		if (*p == '.') {
115 			p++;
116 			if (!file_checkfield(msg, mlen, "precision", &p))
117 				return -1;
118 		}
119 
120 		if (!isalpha((unsigned char)*p)) {
121 			if (msg)
122 				snprintf(msg, mlen, "bad format char: %c", *p);
123 			return -1;
124 		}
125 	}
126 	return 0;
127 }
128 
129 /*
130  * Like printf, only we append to a buffer.
131  */
132 file_protected int
133 file_vprintf(struct magic_set *ms, const char *fmt, va_list ap)
134 {
135 	int len;
136 	char *buf, *newstr;
137 	char tbuf[1024];
138 
139 	if (ms->event_flags & EVENT_HAD_ERR)
140 		return 0;
141 
142 	if (file_checkfmt(tbuf, sizeof(tbuf), fmt)) {
143 		file_clearbuf(ms);
144 		file_error(ms, 0, "Bad magic format `%s' (%s)", fmt, tbuf);
145 		return -1;
146 	}
147 
148 	len = vasprintf(&buf, fmt, ap);
149 	if (len < 0 || (size_t)len > 1024 || len + ms->o.blen > 1024 * 1024) {
150 		size_t blen = ms->o.blen;
151 		free(buf);
152 		file_clearbuf(ms);
153 		file_error(ms, 0, "Output buffer space exceeded %d+%"
154 		    SIZE_T_FORMAT "u", len, blen);
155 		return -1;
156 	}
157 
158 	if (ms->o.buf != NULL) {
159 		len = asprintf(&newstr, "%s%s", ms->o.buf, buf);
160 		free(buf);
161 		if (len < 0)
162 			goto out;
163 		free(ms->o.buf);
164 		buf = newstr;
165 	}
166 	ms->o.buf = buf;
167 	ms->o.blen = len;
168 	return 0;
169 out:
170 	file_clearbuf(ms);
171 	file_error(ms, errno, "vasprintf failed");
172 	return -1;
173 }
174 
175 file_protected int
176 file_printf(struct magic_set *ms, const char *fmt, ...)
177 {
178 	int rv;
179 	va_list ap;
180 
181 	va_start(ap, fmt);
182 	rv = file_vprintf(ms, fmt, ap);
183 	va_end(ap);
184 	return rv;
185 }
186 
187 /*
188  * error - print best error message possible
189  */
190 /*VARARGS*/
191 __attribute__((__format__(__printf__, 3, 0)))
192 file_private void
193 file_error_core(struct magic_set *ms, int error, const char *f, va_list va,
194     size_t lineno)
195 {
196 	/* Only the first error is ok */
197 	if (ms->event_flags & EVENT_HAD_ERR)
198 		return;
199 	if (lineno != 0) {
200 		file_clearbuf(ms);
201 		(void)file_printf(ms, "line %" SIZE_T_FORMAT "u:", lineno);
202 	}
203 	if (ms->o.buf && *ms->o.buf)
204 		(void)file_printf(ms, " ");
205 	(void)file_vprintf(ms, f, va);
206 	if (error > 0)
207 		(void)file_printf(ms, " (%s)", strerror(error));
208 	ms->event_flags |= EVENT_HAD_ERR;
209 	ms->error = error;
210 }
211 
212 /*VARARGS*/
213 file_protected void
214 file_error(struct magic_set *ms, int error, const char *f, ...)
215 {
216 	va_list va;
217 	va_start(va, f);
218 	file_error_core(ms, error, f, va, 0);
219 	va_end(va);
220 }
221 
222 /*
223  * Print an error with magic line number.
224  */
225 /*VARARGS*/
226 file_protected void
227 file_magerror(struct magic_set *ms, const char *f, ...)
228 {
229 	va_list va;
230 	va_start(va, f);
231 	file_error_core(ms, 0, f, va, ms->line);
232 	va_end(va);
233 }
234 
235 file_protected void
236 file_oomem(struct magic_set *ms, size_t len)
237 {
238 	file_error(ms, errno, "cannot allocate %" SIZE_T_FORMAT "u bytes",
239 	    len);
240 }
241 
242 file_protected void
243 file_badseek(struct magic_set *ms)
244 {
245 	file_error(ms, errno, "error seeking");
246 }
247 
248 file_protected void
249 file_badread(struct magic_set *ms)
250 {
251 	file_error(ms, errno, "error reading");
252 }
253 
254 #ifndef COMPILE_ONLY
255 #define FILE_SEPARATOR "\n- "
256 
257 file_protected int
258 file_separator(struct magic_set *ms)
259 {
260 	return file_printf(ms, FILE_SEPARATOR);
261 }
262 
263 static void
264 trim_separator(struct magic_set *ms)
265 {
266 	size_t l;
267 
268 	if (ms->o.buf == NULL)
269 		return;
270 
271 	l = strlen(ms->o.buf);
272 	if (l < sizeof(FILE_SEPARATOR))
273 		return;
274 
275 	l -= sizeof(FILE_SEPARATOR) - 1;
276 	if (strcmp(ms->o.buf + l, FILE_SEPARATOR) != 0)
277 		return;
278 
279 	ms->o.buf[l] = '\0';
280 }
281 
282 static int
283 checkdone(struct magic_set *ms, int *rv)
284 {
285 	if ((ms->flags & MAGIC_CONTINUE) == 0)
286 		return 1;
287 	if (file_separator(ms) == -1)
288 		*rv = -1;
289 	return 0;
290 }
291 
292 file_protected int
293 file_default(struct magic_set *ms, size_t nb)
294 {
295 	if (ms->flags & MAGIC_MIME) {
296 		if ((ms->flags & MAGIC_MIME_TYPE) &&
297 		    file_printf(ms, "application/%s",
298 			nb ? "octet-stream" : "x-empty") == -1)
299 			return -1;
300 		return 1;
301 	}
302 	if (ms->flags & MAGIC_APPLE) {
303 		if (file_printf(ms, "UNKNUNKN") == -1)
304 			return -1;
305 		return 1;
306 	}
307 	if (ms->flags & MAGIC_EXTENSION) {
308 		if (file_printf(ms, "???") == -1)
309 			return -1;
310 		return 1;
311 	}
312 	return 0;
313 }
314 
315 /*
316  * The magic detection functions return:
317  *	 1: found
318  *	 0: not found
319  *	-1: error
320  */
321 /*ARGSUSED*/
322 file_protected int
323 file_buffer(struct magic_set *ms, int fd, struct stat *st,
324     const char *inname __attribute__ ((__unused__)),
325     const void *buf, size_t nb)
326 {
327 	int m = 0, rv = 0, looks_text = 0;
328 	const char *code = NULL;
329 	const char *code_mime = "binary";
330 	const char *def = "data";
331 	const char *ftype = NULL;
332 	char *rbuf = NULL;
333 	struct buffer b;
334 
335 	buffer_init(&b, fd, st, buf, nb);
336 	ms->mode = b.st.st_mode;
337 
338 	if (nb == 0) {
339 		def = "empty";
340 		goto simple;
341 	} else if (nb == 1) {
342 		def = "very short file (no magic)";
343 		goto simple;
344 	}
345 
346 	if ((ms->flags & MAGIC_NO_CHECK_ENCODING) == 0) {
347 		looks_text = file_encoding(ms, &b, NULL, 0,
348 		    &code, &code_mime, &ftype);
349 	}
350 
351 #ifdef __EMX__
352 	if ((ms->flags & MAGIC_NO_CHECK_APPTYPE) == 0 && inname) {
353 		m = file_os2_apptype(ms, inname, &b);
354 		if ((ms->flags & MAGIC_DEBUG) != 0)
355 			(void)fprintf(stderr, "[try os2_apptype %d]\n", m);
356 		switch (m) {
357 		case -1:
358 			return -1;
359 		case 0:
360 			break;
361 		default:
362 			return 1;
363 		}
364 	}
365 #endif
366 #if HAVE_FORK
367 	/* try compression stuff */
368 	if ((ms->flags & MAGIC_NO_CHECK_COMPRESS) == 0) {
369 		m = file_zmagic(ms, &b, inname);
370 		if ((ms->flags & MAGIC_DEBUG) != 0)
371 			(void)fprintf(stderr, "[try zmagic %d]\n", m);
372 		if (m) {
373 			goto done_encoding;
374 		}
375 	}
376 #endif
377 	/* Check if we have a tar file */
378 	if ((ms->flags & MAGIC_NO_CHECK_TAR) == 0) {
379 		m = file_is_tar(ms, &b);
380 		if ((ms->flags & MAGIC_DEBUG) != 0)
381 			(void)fprintf(stderr, "[try tar %d]\n", m);
382 		if (m) {
383 			if (checkdone(ms, &rv))
384 				goto done;
385 		}
386 	}
387 
388 	/* Check if we have a JSON file */
389 	if ((ms->flags & MAGIC_NO_CHECK_JSON) == 0) {
390 		m = file_is_json(ms, &b);
391 		if ((ms->flags & MAGIC_DEBUG) != 0)
392 			(void)fprintf(stderr, "[try json %d]\n", m);
393 		if (m) {
394 			if (checkdone(ms, &rv))
395 				goto done;
396 		}
397 	}
398 
399 	/* Check if we have a CSV file */
400 	if ((ms->flags & MAGIC_NO_CHECK_CSV) == 0) {
401 		m = file_is_csv(ms, &b, looks_text, code);
402 		if ((ms->flags & MAGIC_DEBUG) != 0)
403 			(void)fprintf(stderr, "[try csv %d]\n", m);
404 		if (m) {
405 			if (checkdone(ms, &rv))
406 				goto done;
407 		}
408 	}
409 
410 	/* Check if we have a SIMH tape file */
411 	if ((ms->flags & MAGIC_NO_CHECK_SIMH) == 0) {
412 		m = file_is_simh(ms, &b);
413 		if ((ms->flags & MAGIC_DEBUG) != 0)
414 			(void)fprintf(stderr, "[try simh %d]\n", m);
415 		if (m) {
416 			if (checkdone(ms, &rv))
417 				goto done;
418 		}
419 	}
420 
421 	/* Check if we have a CDF file */
422 	if ((ms->flags & MAGIC_NO_CHECK_CDF) == 0) {
423 		m = file_trycdf(ms, &b);
424 		if ((ms->flags & MAGIC_DEBUG) != 0)
425 			(void)fprintf(stderr, "[try cdf %d]\n", m);
426 		if (m) {
427 			if (checkdone(ms, &rv))
428 				goto done;
429 		}
430 	}
431 #ifdef BUILTIN_ELF
432 	if ((ms->flags & MAGIC_NO_CHECK_ELF) == 0 && nb > 5 && fd != -1) {
433 		file_pushbuf_t *pb;
434 		/*
435 		 * We matched something in the file, so this
436 		 * *might* be an ELF file, and the file is at
437 		 * least 5 bytes long, so if it's an ELF file
438 		 * it has at least one byte past the ELF magic
439 		 * number - try extracting information from the
440 		 * ELF headers that cannot easily be  extracted
441 		 * with rules in the magic file. We we don't
442 		 * print the information yet.
443 		 */
444 		if ((pb = file_push_buffer(ms)) == NULL)
445 			return -1;
446 
447 		rv = file_tryelf(ms, &b);
448 		rbuf = file_pop_buffer(ms, pb);
449 		if (rv == -1) {
450 			free(rbuf);
451 			rbuf = NULL;
452 		}
453 		if ((ms->flags & MAGIC_DEBUG) != 0)
454 			(void)fprintf(stderr, "[try elf %d]\n", m);
455 	}
456 #endif
457 
458 	/* try soft magic tests */
459 	if ((ms->flags & MAGIC_NO_CHECK_SOFT) == 0) {
460 		m = file_softmagic(ms, &b, NULL, NULL, BINTEST, looks_text);
461 		if ((ms->flags & MAGIC_DEBUG) != 0)
462 			(void)fprintf(stderr, "[try softmagic %d]\n", m);
463 		if (m == 1 && rbuf) {
464 			if (file_printf(ms, "%s", rbuf) == -1)
465 				goto done;
466 		}
467 		if (m) {
468 			if (checkdone(ms, &rv))
469 				goto done;
470 		}
471 	}
472 
473 	/* try text properties */
474 	if ((ms->flags & MAGIC_NO_CHECK_TEXT) == 0) {
475 
476 		m = file_ascmagic(ms, &b, looks_text);
477 		if ((ms->flags & MAGIC_DEBUG) != 0)
478 			(void)fprintf(stderr, "[try ascmagic %d]\n", m);
479 		if (m) {
480 			goto done;
481 		}
482 	}
483 
484 simple:
485 	/* give up */
486 	if (m == 0) {
487 		m = 1;
488 		rv = file_default(ms, nb);
489 		if (rv == 0)
490 			if (file_printf(ms, "%s", def) == -1)
491 				rv = -1;
492 	}
493  done:
494 	trim_separator(ms);
495 	if ((ms->flags & MAGIC_MIME_ENCODING) != 0) {
496 		if (ms->flags & MAGIC_MIME_TYPE)
497 			if (file_printf(ms, "; charset=") == -1)
498 				rv = -1;
499 		if (file_printf(ms, "%s", code_mime) == -1)
500 			rv = -1;
501 	}
502 #if HAVE_FORK
503  done_encoding:
504 #endif
505 	free(rbuf);
506 	buffer_fini(&b);
507 	if (rv)
508 		return rv;
509 
510 	return m;
511 }
512 #endif
513 
514 file_protected int
515 file_reset(struct magic_set *ms, int checkloaded)
516 {
517 	if (checkloaded && ms->mlist[0] == NULL) {
518 		file_error(ms, 0, "no magic files loaded");
519 		return -1;
520 	}
521 	file_clearbuf(ms);
522 	if (ms->o.pbuf) {
523 		free(ms->o.pbuf);
524 		ms->o.pbuf = NULL;
525 	}
526 	ms->event_flags &= ~EVENT_HAD_ERR;
527 	ms->error = -1;
528 	return 0;
529 }
530 
531 #define OCTALIFY(n, o)	\
532 	/*LINTED*/ \
533 	(void)(*(n)++ = '\\', \
534 	*(n)++ = ((CAST(uint32_t, *(o)) >> 6) & 3) + '0', \
535 	*(n)++ = ((CAST(uint32_t, *(o)) >> 3) & 7) + '0', \
536 	*(n)++ = ((CAST(uint32_t, *(o)) >> 0) & 7) + '0', \
537 	(o)++)
538 
539 file_protected const char *
540 file_getbuffer(struct magic_set *ms)
541 {
542 	char *pbuf, *op, *np;
543 	size_t psize, len;
544 
545 	if (ms->event_flags & EVENT_HAD_ERR)
546 		return NULL;
547 
548 	if (ms->flags & MAGIC_RAW)
549 		return ms->o.buf;
550 
551 	if (ms->o.buf == NULL)
552 		return NULL;
553 
554 	/* * 4 is for octal representation, + 1 is for NUL */
555 	len = strlen(ms->o.buf);
556 	if (len > (SIZE_MAX - 1) / 4) {
557 		file_oomem(ms, len);
558 		return NULL;
559 	}
560 	psize = len * 4 + 1;
561 	if ((pbuf = CAST(char *, realloc(ms->o.pbuf, psize))) == NULL) {
562 		file_oomem(ms, psize);
563 		return NULL;
564 	}
565 	ms->o.pbuf = pbuf;
566 
567 #if defined(HAVE_WCHAR_H) && defined(HAVE_MBRTOWC) && defined(HAVE_WCWIDTH)
568 	{
569 		mbstate_t state;
570 		wchar_t nextchar;
571 		int mb_conv = 1;
572 		size_t bytesconsumed;
573 		char *eop;
574 		(void)memset(&state, 0, sizeof(mbstate_t));
575 
576 		np = ms->o.pbuf;
577 		op = ms->o.buf;
578 		eop = op + len;
579 
580 		while (op < eop) {
581 			bytesconsumed = mbrtowc(&nextchar, op,
582 			    CAST(size_t, eop - op), &state);
583 			if (bytesconsumed == CAST(size_t, -1) ||
584 			    bytesconsumed == CAST(size_t, -2)) {
585 				mb_conv = 0;
586 				break;
587 			}
588 
589 			if (iswprint(nextchar)) {
590 				(void)memcpy(np, op, bytesconsumed);
591 				op += bytesconsumed;
592 				np += bytesconsumed;
593 			} else {
594 				while (bytesconsumed-- > 0)
595 					OCTALIFY(np, op);
596 			}
597 		}
598 		*np = '\0';
599 
600 		/* Parsing succeeded as a multi-byte sequence */
601 		if (mb_conv != 0)
602 			return ms->o.pbuf;
603 	}
604 #endif
605 
606 	for (np = ms->o.pbuf, op = ms->o.buf; *op;) {
607 		if (isprint(CAST(unsigned char, *op))) {
608 			*np++ = *op++;
609 		} else {
610 			OCTALIFY(np, op);
611 		}
612 	}
613 	*np = '\0';
614 	return ms->o.pbuf;
615 }
616 
617 file_protected int
618 file_check_mem(struct magic_set *ms, unsigned int level)
619 {
620 	size_t len;
621 
622 	if (level >= ms->c.len) {
623 		len = (ms->c.len = 20 + level) * sizeof(*ms->c.li);
624 		ms->c.li = CAST(struct level_info *, (ms->c.li == NULL) ?
625 		    malloc(len) :
626 		    realloc(ms->c.li, len));
627 		if (ms->c.li == NULL) {
628 			file_oomem(ms, len);
629 			return -1;
630 		}
631 	}
632 	ms->c.li[level].got_match = 0;
633 #ifdef ENABLE_CONDITIONALS
634 	ms->c.li[level].last_match = 0;
635 	ms->c.li[level].last_cond = COND_NONE;
636 #endif /* ENABLE_CONDITIONALS */
637 	return 0;
638 }
639 
640 file_protected size_t
641 file_printedlen(const struct magic_set *ms)
642 {
643 	return ms->o.blen;
644 }
645 
646 file_protected int
647 file_replace(struct magic_set *ms, const char *pat, const char *rep)
648 {
649 	file_regex_t rx;
650 	int rc, rv = -1;
651 
652 	rc = file_regcomp(ms, &rx, pat, REG_EXTENDED);
653 	if (rc == 0) {
654 		regmatch_t rm;
655 		int nm = 0;
656 		while (file_regexec(ms, &rx, ms->o.buf, 1, &rm, 0) == 0) {
657 			ms->o.buf[rm.rm_so] = '\0';
658 			if (file_printf(ms, "%s%s", rep,
659 			    rm.rm_eo != 0 ? ms->o.buf + rm.rm_eo : "") == -1)
660 				goto out;
661 			nm++;
662 		}
663 		rv = nm;
664 	}
665 out:
666 	file_regfree(&rx);
667 	return rv;
668 }
669 
670 file_private int
671 check_regex(struct magic_set *ms, const char *pat)
672 {
673 	char sbuf[512];
674 	unsigned char oc = '\0';
675 	const char *p;
676 
677 	for (p = pat; *p; p++) {
678 		unsigned char c = *p;
679 		// Avoid repetition
680 		if (c == oc && strchr("?*+{", c) != NULL) {
681 			size_t len = strlen(pat);
682 			file_magwarn(ms,
683 			    "repetition-operator operand `%c' "
684 			    "invalid in regex `%s'", c,
685 			    file_printable(ms, sbuf, sizeof(sbuf), pat, len));
686 			return -1;
687 		}
688 		oc = c;
689 		if (isprint(c) || isspace(c) || c == '\b'
690 		    || c == 0x8a) // XXX: apple magic fixme
691 			continue;
692 		size_t len = strlen(pat);
693 		file_magwarn(ms,
694 		    "non-ascii characters in regex \\%#o `%s'",
695 		    c, file_printable(ms, sbuf, sizeof(sbuf), pat, len));
696 		return -1;
697 	}
698 	return 0;
699 }
700 
701 file_protected int
702 file_regcomp(struct magic_set *ms file_locale_used, file_regex_t *rx,
703     const char *pat, int flags)
704 {
705 	if (check_regex(ms, pat) == -1)
706 		return -1;
707 
708 #ifdef USE_C_LOCALE
709 	locale_t old = uselocale(ms->c_lc_ctype);
710 	assert(old != NULL);
711 #else
712 	char old[1024];
713 	strlcpy(old, setlocale(LC_CTYPE, NULL), sizeof(old));
714 	(void)setlocale(LC_CTYPE, "C");
715 #endif
716 	int rc;
717 	rc = regcomp(rx, pat, flags);
718 
719 #ifdef USE_C_LOCALE
720 	uselocale(old);
721 #else
722 	(void)setlocale(LC_CTYPE, old);
723 #endif
724 	if (rc > 0 && (ms->flags & MAGIC_CHECK)) {
725 		char errmsg[512], buf[512];
726 
727 		(void)regerror(rc, rx, errmsg, sizeof(errmsg));
728 		file_magerror(ms, "regex error %d for `%s', (%s)", rc,
729 		    file_printable(ms, buf, sizeof(buf), pat, strlen(pat)),
730 		    errmsg);
731 	}
732 	return rc;
733 }
734 
735 /*ARGSUSED*/
736 file_protected int
737 file_regexec(struct magic_set *ms file_locale_used, file_regex_t *rx,
738     const char *str, size_t nmatch, regmatch_t* pmatch, int eflags)
739 {
740 #ifdef USE_C_LOCALE
741 	locale_t old = uselocale(ms->c_lc_ctype);
742 	assert(old != NULL);
743 #else
744 	char old[1024];
745 	strlcpy(old, setlocale(LC_CTYPE, NULL), sizeof(old));
746 	(void)setlocale(LC_CTYPE, "C");
747 #endif
748 	int rc;
749 	/* XXX: force initialization because glibc does not always do this */
750 	if (nmatch != 0)
751 		memset(pmatch, 0, nmatch * sizeof(*pmatch));
752 	rc = regexec(rx, str, nmatch, pmatch, eflags);
753 #ifdef USE_C_LOCALE
754 	uselocale(old);
755 #else
756 	(void)setlocale(LC_CTYPE, old);
757 #endif
758 	return rc;
759 }
760 
761 file_protected void
762 file_regfree(file_regex_t *rx)
763 {
764 	regfree(rx);
765 }
766 
767 file_protected file_pushbuf_t *
768 file_push_buffer(struct magic_set *ms)
769 {
770 	file_pushbuf_t *pb;
771 
772 	if (ms->event_flags & EVENT_HAD_ERR)
773 		return NULL;
774 
775 	if ((pb = (CAST(file_pushbuf_t *, malloc(sizeof(*pb))))) == NULL)
776 		return NULL;
777 
778 	pb->buf = ms->o.buf;
779 	pb->blen = ms->o.blen;
780 	pb->offset = ms->offset;
781 
782 	ms->o.buf = NULL;
783 	ms->o.blen = 0;
784 	ms->offset = 0;
785 
786 	return pb;
787 }
788 
789 file_protected char *
790 file_pop_buffer(struct magic_set *ms, file_pushbuf_t *pb)
791 {
792 	char *rbuf;
793 
794 	if (ms->event_flags & EVENT_HAD_ERR) {
795 		free(pb->buf);
796 		free(pb);
797 		return NULL;
798 	}
799 
800 	rbuf = ms->o.buf;
801 
802 	ms->o.buf = pb->buf;
803 	ms->o.blen = pb->blen;
804 	ms->offset = pb->offset;
805 
806 	free(pb);
807 	return rbuf;
808 }
809 
810 /*
811  * convert string to ascii printable format.
812  */
813 file_protected char *
814 file_printable(struct magic_set *ms, char *buf, size_t bufsiz,
815     const char *str, size_t slen)
816 {
817 	char *ptr, *eptr = buf + bufsiz - 1;
818 	const unsigned char *s = RCAST(const unsigned char *, str);
819 	const unsigned char *es = s + slen;
820 
821 	for (ptr = buf;  ptr < eptr && s < es && *s; s++) {
822 		if ((ms->flags & MAGIC_RAW) != 0 || isprint(*s)) {
823 			*ptr++ = *s;
824 			continue;
825 		}
826 		if (ptr >= eptr - 3)
827 			break;
828 		*ptr++ = '\\';
829 		*ptr++ = ((CAST(unsigned int, *s) >> 6) & 7) + '0';
830 		*ptr++ = ((CAST(unsigned int, *s) >> 3) & 7) + '0';
831 		*ptr++ = ((CAST(unsigned int, *s) >> 0) & 7) + '0';
832 	}
833 	*ptr = '\0';
834 	return buf;
835 }
836 
837 struct guid {
838 	uint32_t data1;
839 	uint16_t data2;
840 	uint16_t data3;
841 	uint8_t data4[8];
842 };
843 
844 file_protected int
845 file_parse_guid(const char *s, uint64_t *guid)
846 {
847 	struct guid *g = CAST(struct guid *, CAST(void *, guid));
848 #ifndef WIN32
849 	return sscanf(s,
850 	    "%8x-%4hx-%4hx-%2hhx%2hhx-%2hhx%2hhx%2hhx%2hhx%2hhx%2hhx",
851 	    &g->data1, &g->data2, &g->data3, &g->data4[0], &g->data4[1],
852 	    &g->data4[2], &g->data4[3], &g->data4[4], &g->data4[5],
853 	    &g->data4[6], &g->data4[7]) == 11 ? 0 : -1;
854 #else
855 	/* MS-Windows runtime doesn't support %hhx, except under
856 	   non-default __USE_MINGW_ANSI_STDIO.  */
857 	uint16_t data16[8];
858 	int rv = sscanf(s, "%8x-%4hx-%4hx-%2hx%2hx-%2hx%2hx%2hx%2hx%2hx%2hx",
859 	    &g->data1, &g->data2, &g->data3, &data16[0], &data16[1],
860 	    &data16[2], &data16[3], &data16[4], &data16[5],
861 	    &data16[6], &data16[7]) == 11 ? 0 : -1;
862 	int i;
863 	for (i = 0; i < 8; i++)
864 	    g->data4[i] = data16[i];
865 	return rv;
866 #endif
867 }
868 
869 file_protected int
870 file_print_guid(char *str, size_t len, const uint64_t *guid)
871 {
872 	const struct guid *g = CAST(const struct guid *,
873 	    CAST(const void *, guid));
874 
875 #ifndef WIN32
876 	return snprintf(str, len, "%.8X-%.4hX-%.4hX-%.2hhX%.2hhX-"
877 	    "%.2hhX%.2hhX%.2hhX%.2hhX%.2hhX%.2hhX",
878 	    g->data1, g->data2, g->data3, g->data4[0], g->data4[1],
879 	    g->data4[2], g->data4[3], g->data4[4], g->data4[5],
880 	    g->data4[6], g->data4[7]);
881 #else
882 	return snprintf(str, len, "%.8X-%.4hX-%.4hX-%.2hX%.2hX-"
883 	    "%.2hX%.2hX%.2hX%.2hX%.2hX%.2hX",
884 	    g->data1, g->data2, g->data3, g->data4[0], g->data4[1],
885 	    g->data4[2], g->data4[3], g->data4[4], g->data4[5],
886 	    g->data4[6], g->data4[7]);
887 #endif
888 }
889 
890 file_protected int
891 file_pipe_closexec(int *fds)
892 {
893 #ifdef __MINGW32__
894 	return 0;
895 #elif defined(HAVE_PIPE2)
896 	return pipe2(fds, O_CLOEXEC);
897 #else
898 	if (pipe(fds) == -1)
899 		return -1;
900 # ifdef F_SETFD
901 	(void)fcntl(fds[0], F_SETFD, FD_CLOEXEC);
902 	(void)fcntl(fds[1], F_SETFD, FD_CLOEXEC);
903 # endif
904 	return 0;
905 #endif
906 }
907 
908 file_protected int
909 file_clear_closexec(int fd) {
910 #ifdef F_SETFD
911 	return fcntl(fd, F_SETFD, 0);
912 #else
913 	return 0;
914 #endif
915 }
916 
917 file_protected char *
918 file_strtrim(char *str)
919 {
920 	char *last;
921 
922 	while (isspace(CAST(unsigned char, *str)))
923 		str++;
924 	last = str;
925 	while (*last)
926 		last++;
927 	--last;
928 	while (isspace(CAST(unsigned char, *last)))
929 		last--;
930 	*++last = '\0';
931 	return str;
932 }
933