xref: /freebsd/contrib/file/src/funcs.c (revision 78b9f0095b4af3aca6c931b2c7b009ddb8a05125) 
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 1.95 2018/05/24 18:09:17 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 #if defined(HAVE_WCHAR_H)
40 #include <wchar.h>
41 #endif
42 #if defined(HAVE_WCTYPE_H)
43 #include <wctype.h>
44 #endif
45 #if defined(HAVE_LIMITS_H)
46 #include <limits.h>
47 #endif
48 
49 #ifndef SIZE_MAX
50 #define SIZE_MAX	((size_t)~0)
51 #endif
52 
53 /*
54  * Like printf, only we append to a buffer.
55  */
56 protected int
57 file_vprintf(struct magic_set *ms, const char *fmt, va_list ap)
58 {
59 	int len;
60 	char *buf, *newstr;
61 
62 	if (ms->event_flags & EVENT_HAD_ERR)
63 		return 0;
64 	len = vasprintf(&buf, fmt, ap);
65 	if (len < 0)
66 		goto out;
67 
68 	if (ms->o.buf != NULL) {
69 		len = asprintf(&newstr, "%s%s", ms->o.buf, buf);
70 		free(buf);
71 		if (len < 0)
72 			goto out;
73 		free(ms->o.buf);
74 		buf = newstr;
75 	}
76 	ms->o.buf = buf;
77 	return 0;
78 out:
79 	fprintf(stderr, "vasprintf failed (%s)", strerror(errno));
80 	return -1;
81 }
82 
83 protected int
84 file_printf(struct magic_set *ms, const char *fmt, ...)
85 {
86 	int rv;
87 	va_list ap;
88 
89 	va_start(ap, fmt);
90 	rv = file_vprintf(ms, fmt, ap);
91 	va_end(ap);
92 	return rv;
93 }
94 
95 /*
96  * error - print best error message possible
97  */
98 /*VARARGS*/
99 __attribute__((__format__(__printf__, 3, 0)))
100 private void
101 file_error_core(struct magic_set *ms, int error, const char *f, va_list va,
102     size_t lineno)
103 {
104 	/* Only the first error is ok */
105 	if (ms->event_flags & EVENT_HAD_ERR)
106 		return;
107 	if (lineno != 0) {
108 		free(ms->o.buf);
109 		ms->o.buf = NULL;
110 		file_printf(ms, "line %" SIZE_T_FORMAT "u:", lineno);
111 	}
112 	if (ms->o.buf && *ms->o.buf)
113 		file_printf(ms, " ");
114 	file_vprintf(ms, f, va);
115 	if (error > 0)
116 		file_printf(ms, " (%s)", strerror(error));
117 	ms->event_flags |= EVENT_HAD_ERR;
118 	ms->error = error;
119 }
120 
121 /*VARARGS*/
122 protected void
123 file_error(struct magic_set *ms, int error, const char *f, ...)
124 {
125 	va_list va;
126 	va_start(va, f);
127 	file_error_core(ms, error, f, va, 0);
128 	va_end(va);
129 }
130 
131 /*
132  * Print an error with magic line number.
133  */
134 /*VARARGS*/
135 protected void
136 file_magerror(struct magic_set *ms, const char *f, ...)
137 {
138 	va_list va;
139 	va_start(va, f);
140 	file_error_core(ms, 0, f, va, ms->line);
141 	va_end(va);
142 }
143 
144 protected void
145 file_oomem(struct magic_set *ms, size_t len)
146 {
147 	file_error(ms, errno, "cannot allocate %" SIZE_T_FORMAT "u bytes",
148 	    len);
149 }
150 
151 protected void
152 file_badseek(struct magic_set *ms)
153 {
154 	file_error(ms, errno, "error seeking");
155 }
156 
157 protected void
158 file_badread(struct magic_set *ms)
159 {
160 	file_error(ms, errno, "error reading");
161 }
162 
163 #ifndef COMPILE_ONLY
164 
165 static int
166 checkdone(struct magic_set *ms, int *rv)
167 {
168 	if ((ms->flags & MAGIC_CONTINUE) == 0)
169 		return 1;
170 	if (file_printf(ms, "\n- ") == -1)
171 		*rv = -1;
172 	return 0;
173 }
174 
175 /*ARGSUSED*/
176 protected int
177 file_buffer(struct magic_set *ms, int fd, const char *inname __attribute__ ((__unused__)),
178     const void *buf, size_t nb)
179 {
180 	int m = 0, rv = 0, looks_text = 0;
181 	const char *code = NULL;
182 	const char *code_mime = "binary";
183 	const char *type = "application/octet-stream";
184 	const char *def = "data";
185 	const char *ftype = NULL;
186 	char *rbuf = NULL;
187 	struct buffer b;
188 
189 	buffer_init(&b, fd, buf, nb);
190 	ms->mode = b.st.st_mode;
191 
192 	if (nb == 0) {
193 		def = "empty";
194 		type = "application/x-empty";
195 		goto simple;
196 	} else if (nb == 1) {
197 		def = "very short file (no magic)";
198 		goto simple;
199 	}
200 
201 	if ((ms->flags & MAGIC_NO_CHECK_ENCODING) == 0) {
202 		looks_text = file_encoding(ms, &b, NULL, 0,
203 		    &code, &code_mime, &ftype);
204 	}
205 
206 #ifdef __EMX__
207 	if ((ms->flags & MAGIC_NO_CHECK_APPTYPE) == 0 && inname) {
208 		m = file_os2_apptype(ms, inname, &b);
209 		if ((ms->flags & MAGIC_DEBUG) != 0)
210 			(void)fprintf(stderr, "[try os2_apptype %d]\n", m);
211 		switch (m) {
212 		case -1:
213 			return -1;
214 		case 0:
215 			break;
216 		default:
217 			return 1;
218 		}
219 	}
220 #endif
221 #if HAVE_FORK
222 	/* try compression stuff */
223 	if ((ms->flags & MAGIC_NO_CHECK_COMPRESS) == 0) {
224 		m = file_zmagic(ms, &b, inname);
225 		if ((ms->flags & MAGIC_DEBUG) != 0)
226 			(void)fprintf(stderr, "[try zmagic %d]\n", m);
227 		if (m) {
228 			goto done_encoding;
229 		}
230 	}
231 #endif
232 	/* Check if we have a tar file */
233 	if ((ms->flags & MAGIC_NO_CHECK_TAR) == 0) {
234 		m = file_is_tar(ms, &b);
235 		if ((ms->flags & MAGIC_DEBUG) != 0)
236 			(void)fprintf(stderr, "[try tar %d]\n", m);
237 		if (m) {
238 			if (checkdone(ms, &rv))
239 				goto done;
240 		}
241 	}
242 
243 	/* Check if we have a CDF file */
244 	if ((ms->flags & MAGIC_NO_CHECK_CDF) == 0) {
245 		m = file_trycdf(ms, &b);
246 		if ((ms->flags & MAGIC_DEBUG) != 0)
247 			(void)fprintf(stderr, "[try cdf %d]\n", m);
248 		if (m) {
249 			if (checkdone(ms, &rv))
250 				goto done;
251 		}
252 	}
253 #ifdef BUILTIN_ELF
254 	if ((ms->flags & MAGIC_NO_CHECK_ELF) == 0 && nb > 5 && fd != -1) {
255 		file_pushbuf_t *pb;
256 		/*
257 		 * We matched something in the file, so this
258 		 * *might* be an ELF file, and the file is at
259 		 * least 5 bytes long, so if it's an ELF file
260 		 * it has at least one byte past the ELF magic
261 		 * number - try extracting information from the
262 		 * ELF headers that cannot easily be  extracted
263 		 * with rules in the magic file. We we don't
264 		 * print the information yet.
265 		 */
266 		if ((pb = file_push_buffer(ms)) == NULL)
267 			return -1;
268 
269 		rv = file_tryelf(ms, &b);
270 		rbuf = file_pop_buffer(ms, pb);
271 		if (rv != 1) {
272 			free(rbuf);
273 			rbuf = NULL;
274 		}
275 		if ((ms->flags & MAGIC_DEBUG) != 0)
276 			(void)fprintf(stderr, "[try elf %d]\n", m);
277 	}
278 #endif
279 
280 	/* try soft magic tests */
281 	if ((ms->flags & MAGIC_NO_CHECK_SOFT) == 0) {
282 		m = file_softmagic(ms, &b, NULL, NULL, BINTEST, looks_text);
283 		if ((ms->flags & MAGIC_DEBUG) != 0)
284 			(void)fprintf(stderr, "[try softmagic %d]\n", m);
285 		if (m == 1 && rbuf) {
286 			if (file_printf(ms, "%s", rbuf) == -1)
287 				goto done;
288 		}
289 		if (m) {
290 			if (checkdone(ms, &rv))
291 				goto done;
292 		}
293 	}
294 
295 	/* try text properties */
296 	if ((ms->flags & MAGIC_NO_CHECK_TEXT) == 0) {
297 
298 		m = file_ascmagic(ms, &b, looks_text);
299 		if ((ms->flags & MAGIC_DEBUG) != 0)
300 			(void)fprintf(stderr, "[try ascmagic %d]\n", m);
301 		if (m) {
302 			if (checkdone(ms, &rv))
303 				goto done;
304 		}
305 	}
306 
307 simple:
308 	/* give up */
309 	m = 1;
310 	if (ms->flags & MAGIC_MIME) {
311 		if ((ms->flags & MAGIC_MIME_TYPE) &&
312 		    file_printf(ms, "%s", type) == -1)
313 			rv = -1;
314 	} else if (ms->flags & MAGIC_APPLE) {
315 		if (file_printf(ms, "UNKNUNKN") == -1)
316 			rv = -1;
317 	} else if (ms->flags & MAGIC_EXTENSION) {
318 		if (file_printf(ms, "???") == -1)
319 			rv = -1;
320 	} else {
321 		if (file_printf(ms, "%s", def) == -1)
322 			rv = -1;
323 	}
324  done:
325 	if ((ms->flags & MAGIC_MIME_ENCODING) != 0) {
326 		if (ms->flags & MAGIC_MIME_TYPE)
327 			if (file_printf(ms, "; charset=") == -1)
328 				rv = -1;
329 		if (file_printf(ms, "%s", code_mime) == -1)
330 			rv = -1;
331 	}
332 #if HAVE_FORK
333  done_encoding:
334 #endif
335 	free(rbuf);
336 	buffer_fini(&b);
337 	if (rv)
338 		return rv;
339 
340 	return m;
341 }
342 #endif
343 
344 protected int
345 file_reset(struct magic_set *ms, int checkloaded)
346 {
347 	if (checkloaded && ms->mlist[0] == NULL) {
348 		file_error(ms, 0, "no magic files loaded");
349 		return -1;
350 	}
351 	if (ms->o.buf) {
352 		free(ms->o.buf);
353 		ms->o.buf = NULL;
354 	}
355 	if (ms->o.pbuf) {
356 		free(ms->o.pbuf);
357 		ms->o.pbuf = NULL;
358 	}
359 	ms->event_flags &= ~EVENT_HAD_ERR;
360 	ms->error = -1;
361 	return 0;
362 }
363 
364 #define OCTALIFY(n, o)	\
365 	/*LINTED*/ \
366 	(void)(*(n)++ = '\\', \
367 	*(n)++ = (((uint32_t)*(o) >> 6) & 3) + '0', \
368 	*(n)++ = (((uint32_t)*(o) >> 3) & 7) + '0', \
369 	*(n)++ = (((uint32_t)*(o) >> 0) & 7) + '0', \
370 	(o)++)
371 
372 protected const char *
373 file_getbuffer(struct magic_set *ms)
374 {
375 	char *pbuf, *op, *np;
376 	size_t psize, len;
377 
378 	if (ms->event_flags & EVENT_HAD_ERR)
379 		return NULL;
380 
381 	if (ms->flags & MAGIC_RAW)
382 		return ms->o.buf;
383 
384 	if (ms->o.buf == NULL)
385 		return NULL;
386 
387 	/* * 4 is for octal representation, + 1 is for NUL */
388 	len = strlen(ms->o.buf);
389 	if (len > (SIZE_MAX - 1) / 4) {
390 		file_oomem(ms, len);
391 		return NULL;
392 	}
393 	psize = len * 4 + 1;
394 	if ((pbuf = CAST(char *, realloc(ms->o.pbuf, psize))) == NULL) {
395 		file_oomem(ms, psize);
396 		return NULL;
397 	}
398 	ms->o.pbuf = pbuf;
399 
400 #if defined(HAVE_WCHAR_H) && defined(HAVE_MBRTOWC) && defined(HAVE_WCWIDTH)
401 	{
402 		mbstate_t state;
403 		wchar_t nextchar;
404 		int mb_conv = 1;
405 		size_t bytesconsumed;
406 		char *eop;
407 		(void)memset(&state, 0, sizeof(mbstate_t));
408 
409 		np = ms->o.pbuf;
410 		op = ms->o.buf;
411 		eop = op + len;
412 
413 		while (op < eop) {
414 			bytesconsumed = mbrtowc(&nextchar, op,
415 			    (size_t)(eop - op), &state);
416 			if (bytesconsumed == (size_t)(-1) ||
417 			    bytesconsumed == (size_t)(-2)) {
418 				mb_conv = 0;
419 				break;
420 			}
421 
422 			if (iswprint(nextchar)) {
423 				(void)memcpy(np, op, bytesconsumed);
424 				op += bytesconsumed;
425 				np += bytesconsumed;
426 			} else {
427 				while (bytesconsumed-- > 0)
428 					OCTALIFY(np, op);
429 			}
430 		}
431 		*np = '\0';
432 
433 		/* Parsing succeeded as a multi-byte sequence */
434 		if (mb_conv != 0)
435 			return ms->o.pbuf;
436 	}
437 #endif
438 
439 	for (np = ms->o.pbuf, op = ms->o.buf; *op;) {
440 		if (isprint((unsigned char)*op)) {
441 			*np++ = *op++;
442 		} else {
443 			OCTALIFY(np, op);
444 		}
445 	}
446 	*np = '\0';
447 	return ms->o.pbuf;
448 }
449 
450 protected int
451 file_check_mem(struct magic_set *ms, unsigned int level)
452 {
453 	size_t len;
454 
455 	if (level >= ms->c.len) {
456 		len = (ms->c.len = 20 + level) * sizeof(*ms->c.li);
457 		ms->c.li = CAST(struct level_info *, (ms->c.li == NULL) ?
458 		    malloc(len) :
459 		    realloc(ms->c.li, len));
460 		if (ms->c.li == NULL) {
461 			file_oomem(ms, len);
462 			return -1;
463 		}
464 	}
465 	ms->c.li[level].got_match = 0;
466 #ifdef ENABLE_CONDITIONALS
467 	ms->c.li[level].last_match = 0;
468 	ms->c.li[level].last_cond = COND_NONE;
469 #endif /* ENABLE_CONDITIONALS */
470 	return 0;
471 }
472 
473 protected size_t
474 file_printedlen(const struct magic_set *ms)
475 {
476 	return ms->o.buf == NULL ? 0 : strlen(ms->o.buf);
477 }
478 
479 protected int
480 file_replace(struct magic_set *ms, const char *pat, const char *rep)
481 {
482 	file_regex_t rx;
483 	int rc, rv = -1;
484 
485 	rc = file_regcomp(&rx, pat, REG_EXTENDED);
486 	if (rc) {
487 		file_regerror(&rx, rc, ms);
488 	} else {
489 		regmatch_t rm;
490 		int nm = 0;
491 		while (file_regexec(&rx, ms->o.buf, 1, &rm, 0) == 0) {
492 			ms->o.buf[rm.rm_so] = '\0';
493 			if (file_printf(ms, "%s%s", rep,
494 			    rm.rm_eo != 0 ? ms->o.buf + rm.rm_eo : "") == -1)
495 				goto out;
496 			nm++;
497 		}
498 		rv = nm;
499 	}
500 out:
501 	file_regfree(&rx);
502 	return rv;
503 }
504 
505 protected int
506 file_regcomp(file_regex_t *rx, const char *pat, int flags)
507 {
508 #ifdef USE_C_LOCALE
509 	rx->c_lc_ctype = newlocale(LC_CTYPE_MASK, "C", 0);
510 	assert(rx->c_lc_ctype != NULL);
511 	rx->old_lc_ctype = uselocale(rx->c_lc_ctype);
512 	assert(rx->old_lc_ctype != NULL);
513 #else
514 	rx->old_lc_ctype = setlocale(LC_CTYPE, "C");
515 #endif
516 	rx->pat = pat;
517 
518 	return rx->rc = regcomp(&rx->rx, pat, flags);
519 }
520 
521 protected int
522 file_regexec(file_regex_t *rx, const char *str, size_t nmatch,
523     regmatch_t* pmatch, int eflags)
524 {
525 	assert(rx->rc == 0);
526 	/* XXX: force initialization because glibc does not always do this */
527 	memset(pmatch, 0, nmatch * sizeof(*pmatch));
528 	return regexec(&rx->rx, str, nmatch, pmatch, eflags);
529 }
530 
531 protected void
532 file_regfree(file_regex_t *rx)
533 {
534 	if (rx->rc == 0)
535 		regfree(&rx->rx);
536 #ifdef USE_C_LOCALE
537 	(void)uselocale(rx->old_lc_ctype);
538 	freelocale(rx->c_lc_ctype);
539 #else
540 	(void)setlocale(LC_CTYPE, rx->old_lc_ctype);
541 #endif
542 }
543 
544 protected void
545 file_regerror(file_regex_t *rx, int rc, struct magic_set *ms)
546 {
547 	char errmsg[512];
548 
549 	(void)regerror(rc, &rx->rx, errmsg, sizeof(errmsg));
550 	file_magerror(ms, "regex error %d for `%s', (%s)", rc, rx->pat,
551 	    errmsg);
552 }
553 
554 protected file_pushbuf_t *
555 file_push_buffer(struct magic_set *ms)
556 {
557 	file_pushbuf_t *pb;
558 
559 	if (ms->event_flags & EVENT_HAD_ERR)
560 		return NULL;
561 
562 	if ((pb = (CAST(file_pushbuf_t *, malloc(sizeof(*pb))))) == NULL)
563 		return NULL;
564 
565 	pb->buf = ms->o.buf;
566 	pb->offset = ms->offset;
567 
568 	ms->o.buf = NULL;
569 	ms->offset = 0;
570 
571 	return pb;
572 }
573 
574 protected char *
575 file_pop_buffer(struct magic_set *ms, file_pushbuf_t *pb)
576 {
577 	char *rbuf;
578 
579 	if (ms->event_flags & EVENT_HAD_ERR) {
580 		free(pb->buf);
581 		free(pb);
582 		return NULL;
583 	}
584 
585 	rbuf = ms->o.buf;
586 
587 	ms->o.buf = pb->buf;
588 	ms->offset = pb->offset;
589 
590 	free(pb);
591 	return rbuf;
592 }
593 
594 /*
595  * convert string to ascii printable format.
596  */
597 protected char *
598 file_printable(char *buf, size_t bufsiz, const char *str)
599 {
600 	char *ptr, *eptr;
601 	const unsigned char *s = (const unsigned char *)str;
602 
603 	for (ptr = buf, eptr = ptr + bufsiz - 1; ptr < eptr && *s; s++) {
604 		if (isprint(*s)) {
605 			*ptr++ = *s;
606 			continue;
607 		}
608 		if (ptr >= eptr - 3)
609 			break;
610 		*ptr++ = '\\';
611 		*ptr++ = ((CAST(unsigned int, *s) >> 6) & 7) + '0';
612 		*ptr++ = ((CAST(unsigned int, *s) >> 3) & 7) + '0';
613 		*ptr++ = ((CAST(unsigned int, *s) >> 0) & 7) + '0';
614 	}
615 	*ptr = '\0';
616 	return buf;
617 }
618