xref: /minix/external/bsd/file/dist/src/funcs.c (revision 84d9c625)
1 /*	$NetBSD: funcs.c,v 1.5 2013/01/03 23:05:38 christos Exp $	*/
2 
3 /*
4  * Copyright (c) Christos Zoulas 2003.
5  * All Rights Reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice immediately at the beginning of the file, without modification,
12  *    this list of conditions, and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
21  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  */
29 #include "file.h"
30 
31 #ifndef	lint
32 #if 0
33 FILE_RCSID("@(#)$File: funcs.c,v 1.61 2012/10/30 23:11:51 christos Exp $")
34 #else
35 __RCSID("$NetBSD: funcs.c,v 1.5 2013/01/03 23:05:38 christos Exp $");
36 #endif
37 #endif	/* lint */
38 
39 #include "magic.h"
40 #include <stdarg.h>
41 #include <stdlib.h>
42 #include <string.h>
43 #include <ctype.h>
44 #if defined(HAVE_WCHAR_H)
45 #include <wchar.h>
46 #endif
47 #if defined(HAVE_WCTYPE_H)
48 #include <wctype.h>
49 #endif
50 #if defined(HAVE_LIMITS_H)
51 #include <limits.h>
52 #endif
53 
54 #ifndef SIZE_MAX
55 #define SIZE_MAX	((size_t)~0)
56 #endif
57 
58 /*
59  * Like printf, only we append to a buffer.
60  */
61 protected int
62 file_vprintf(struct magic_set *ms, const char *fmt, va_list ap)
63 {
64 	int len;
65 	char *buf, *newstr;
66 
67 	len = vasprintf(&buf, fmt, ap);
68 	if (len < 0)
69 		goto out;
70 
71 	if (ms->o.buf != NULL) {
72 		len = asprintf(&newstr, "%s%s", ms->o.buf, buf);
73 		free(buf);
74 		if (len < 0)
75 			goto out;
76 		free(ms->o.buf);
77 		buf = newstr;
78 	}
79 	ms->o.buf = buf;
80 	return 0;
81 out:
82 	file_error(ms, errno, "vasprintf failed");
83 	return -1;
84 }
85 
86 protected int
87 file_printf(struct magic_set *ms, const char *fmt, ...)
88 {
89 	int rv;
90 	va_list ap;
91 
92 	va_start(ap, fmt);
93 	rv = file_vprintf(ms, fmt, ap);
94 	va_end(ap);
95 	return rv;
96 }
97 
98 /*
99  * error - print best error message possible
100  */
101 /*VARARGS*/
102 private void
103 file_error_core(struct magic_set *ms, int error, const char *f, va_list va,
104     size_t lineno)
105 {
106 	/* Only the first error is ok */
107 	if (ms->event_flags & EVENT_HAD_ERR)
108 		return;
109 	if (lineno != 0) {
110 		free(ms->o.buf);
111 		ms->o.buf = NULL;
112 		file_printf(ms, "line %" SIZE_T_FORMAT "u: ", lineno);
113 	}
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 /*ARGSUSED*/
165 protected int
166 file_buffer(struct magic_set *ms, int fd, const char *inname __attribute__ ((__unused__)),
167     const void *buf, size_t nb)
168 {
169 	int m = 0, rv = 0, looks_text = 0;
170 	int mime = ms->flags & MAGIC_MIME;
171 	const unsigned char *ubuf = CAST(const unsigned char *, buf);
172 	unichar *u8buf = NULL;
173 	size_t ulen;
174 	const char *code = NULL;
175 	const char *code_mime = "binary";
176 	const char *type = NULL;
177 
178 
179 
180 	if (nb == 0) {
181 		if ((!mime || (mime & MAGIC_MIME_TYPE)) &&
182 		    file_printf(ms, mime ? "application/x-empty" :
183 		    "empty") == -1)
184 			return -1;
185 		return 1;
186 	} else if (nb == 1) {
187 		if ((!mime || (mime & MAGIC_MIME_TYPE)) &&
188 		    file_printf(ms, mime ? "application/octet-stream" :
189 		    "very short file (no magic)") == -1)
190 			return -1;
191 		return 1;
192 	}
193 
194 	if ((ms->flags & MAGIC_NO_CHECK_ENCODING) == 0) {
195 		looks_text = file_encoding(ms, ubuf, nb, &u8buf, &ulen,
196 		    &code, &code_mime, &type);
197 	}
198 
199 #ifdef __EMX__
200 	if ((ms->flags & MAGIC_NO_CHECK_APPTYPE) == 0 && inname) {
201 		switch (file_os2_apptype(ms, inname, buf, nb)) {
202 		case -1:
203 			return -1;
204 		case 0:
205 			break;
206 		default:
207 			return 1;
208 		}
209 	}
210 #endif
211 #if HAVE_FORK
212 	/* try compression stuff */
213 	if ((ms->flags & MAGIC_NO_CHECK_COMPRESS) == 0)
214 		if ((m = file_zmagic(ms, fd, inname, ubuf, nb)) != 0) {
215 			if ((ms->flags & MAGIC_DEBUG) != 0)
216 				(void)fprintf(stderr, "zmagic %d\n", m);
217 			goto done;
218 		}
219 #endif
220 	/* Check if we have a tar file */
221 	if ((ms->flags & MAGIC_NO_CHECK_TAR) == 0)
222 		if ((m = file_is_tar(ms, ubuf, nb)) != 0) {
223 			if ((ms->flags & MAGIC_DEBUG) != 0)
224 				(void)fprintf(stderr, "tar %d\n", m);
225 			goto done;
226 		}
227 
228 	/* Check if we have a CDF file */
229 	if ((ms->flags & MAGIC_NO_CHECK_CDF) == 0)
230 		if ((m = file_trycdf(ms, fd, ubuf, nb)) != 0) {
231 			if ((ms->flags & MAGIC_DEBUG) != 0)
232 				(void)fprintf(stderr, "cdf %d\n", m);
233 			goto done;
234 		}
235 
236 	/* try soft magic tests */
237 	if ((ms->flags & MAGIC_NO_CHECK_SOFT) == 0)
238 		if ((m = file_softmagic(ms, ubuf, nb, BINTEST,
239 		    looks_text)) != 0) {
240 			if ((ms->flags & MAGIC_DEBUG) != 0)
241 				(void)fprintf(stderr, "softmagic %d\n", m);
242 #ifdef BUILTIN_ELF
243 			if ((ms->flags & MAGIC_NO_CHECK_ELF) == 0 && m == 1 &&
244 			    nb > 5 && fd != -1) {
245 				/*
246 				 * We matched something in the file, so this
247 				 * *might* be an ELF file, and the file is at
248 				 * least 5 bytes long, so if it's an ELF file
249 				 * it has at least one byte past the ELF magic
250 				 * number - try extracting information from the
251 				 * ELF headers that cannot easily * be
252 				 * extracted with rules in the magic file.
253 				 */
254 				if ((m = file_tryelf(ms, fd, ubuf, nb)) != 0)
255 					if ((ms->flags & MAGIC_DEBUG) != 0)
256 						(void)fprintf(stderr,
257 						    "elf %d\n", m);
258 			}
259 #endif
260 			goto done;
261 		}
262 
263 	/* try text properties */
264 	if ((ms->flags & MAGIC_NO_CHECK_TEXT) == 0) {
265 
266 		if ((m = file_ascmagic(ms, ubuf, nb, looks_text)) != 0) {
267 			if ((ms->flags & MAGIC_DEBUG) != 0)
268 				(void)fprintf(stderr, "ascmagic %d\n", m);
269 			goto done;
270 		}
271 
272 		/* try to discover text encoding */
273 		if ((ms->flags & MAGIC_NO_CHECK_ENCODING) == 0) {
274 			if (looks_text == 0)
275 				if ((m = file_ascmagic_with_encoding( ms, ubuf,
276 				    nb, u8buf, ulen, code, type, looks_text))
277 				    != 0) {
278 					if ((ms->flags & MAGIC_DEBUG) != 0)
279 						(void)fprintf(stderr,
280 						    "ascmagic/enc %d\n", m);
281 					goto done;
282 				}
283 		}
284 	}
285 
286 	/* give up */
287 	m = 1;
288 	if ((!mime || (mime & MAGIC_MIME_TYPE)) &&
289 	    file_printf(ms, mime ? "application/octet-stream" : "data") == -1) {
290 	    rv = -1;
291 	}
292  done:
293 	if ((ms->flags & MAGIC_MIME_ENCODING) != 0) {
294 		if (ms->flags & MAGIC_MIME_TYPE)
295 			if (file_printf(ms, "; charset=") == -1)
296 				rv = -1;
297 		if (file_printf(ms, "%s", code_mime) == -1)
298 			rv = -1;
299 	}
300 	free(u8buf);
301 	if (rv)
302 		return rv;
303 
304 	return m;
305 }
306 #endif
307 
308 protected int
309 file_reset(struct magic_set *ms)
310 {
311 	if (ms->mlist[0] == NULL) {
312 		file_error(ms, 0, "no magic files loaded");
313 		return -1;
314 	}
315 	if (ms->o.buf) {
316 		free(ms->o.buf);
317 		ms->o.buf = NULL;
318 	}
319 	if (ms->o.pbuf) {
320 		free(ms->o.pbuf);
321 		ms->o.pbuf = NULL;
322 	}
323 	ms->event_flags &= ~EVENT_HAD_ERR;
324 	ms->error = -1;
325 	return 0;
326 }
327 
328 #define OCTALIFY(n, o)	\
329 	/*LINTED*/ \
330 	(void)(*(n)++ = '\\', \
331 	*(n)++ = (((uint32_t)*(o) >> 6) & 3) + '0', \
332 	*(n)++ = (((uint32_t)*(o) >> 3) & 7) + '0', \
333 	*(n)++ = (((uint32_t)*(o) >> 0) & 7) + '0', \
334 	(o)++)
335 
336 protected const char *
337 file_getbuffer(struct magic_set *ms)
338 {
339 	char *pbuf, *op, *np;
340 	size_t psize, len;
341 
342 	if (ms->event_flags & EVENT_HAD_ERR)
343 		return NULL;
344 
345 	if (ms->flags & MAGIC_RAW)
346 		return ms->o.buf;
347 
348 	if (ms->o.buf == NULL)
349 		return NULL;
350 
351 	/* * 4 is for octal representation, + 1 is for NUL */
352 	len = strlen(ms->o.buf);
353 	if (len > (SIZE_MAX - 1) / 4) {
354 		file_oomem(ms, len);
355 		return NULL;
356 	}
357 	psize = len * 4 + 1;
358 	if ((pbuf = CAST(char *, realloc(ms->o.pbuf, psize))) == NULL) {
359 		file_oomem(ms, psize);
360 		return NULL;
361 	}
362 	ms->o.pbuf = pbuf;
363 
364 #if defined(HAVE_WCHAR_H) && defined(HAVE_MBRTOWC) && defined(HAVE_WCWIDTH)
365 	{
366 		mbstate_t state;
367 		wchar_t nextchar;
368 		int mb_conv = 1;
369 		size_t bytesconsumed;
370 		char *eop;
371 		(void)memset(&state, 0, sizeof(mbstate_t));
372 
373 		np = ms->o.pbuf;
374 		op = ms->o.buf;
375 		eop = op + len;
376 
377 		while (op < eop) {
378 			bytesconsumed = mbrtowc(&nextchar, op,
379 			    (size_t)(eop - op), &state);
380 			if (bytesconsumed == (size_t)(-1) ||
381 			    bytesconsumed == (size_t)(-2)) {
382 				mb_conv = 0;
383 				break;
384 			}
385 
386 			if (iswprint(nextchar)) {
387 				(void)memcpy(np, op, bytesconsumed);
388 				op += bytesconsumed;
389 				np += bytesconsumed;
390 			} else {
391 				while (bytesconsumed-- > 0)
392 					OCTALIFY(np, op);
393 			}
394 		}
395 		*np = '\0';
396 
397 		/* Parsing succeeded as a multi-byte sequence */
398 		if (mb_conv != 0)
399 			return ms->o.pbuf;
400 	}
401 #endif
402 
403 	for (np = ms->o.pbuf, op = ms->o.buf; *op;) {
404 		if (isprint((unsigned char)*op)) {
405 			*np++ = *op++;
406 		} else {
407 			OCTALIFY(np, op);
408 		}
409 	}
410 	*np = '\0';
411 	return ms->o.pbuf;
412 }
413 
414 protected int
415 file_check_mem(struct magic_set *ms, unsigned int level)
416 {
417 	size_t len;
418 
419 	if (level >= ms->c.len) {
420 		len = (ms->c.len += 20) * sizeof(*ms->c.li);
421 		ms->c.li = CAST(struct level_info *, (ms->c.li == NULL) ?
422 		    malloc(len) :
423 		    realloc(ms->c.li, len));
424 		if (ms->c.li == NULL) {
425 			file_oomem(ms, len);
426 			return -1;
427 		}
428 	}
429 	ms->c.li[level].got_match = 0;
430 #ifdef ENABLE_CONDITIONALS
431 	ms->c.li[level].last_match = 0;
432 	ms->c.li[level].last_cond = COND_NONE;
433 #endif /* ENABLE_CONDITIONALS */
434 	return 0;
435 }
436 
437 protected size_t
438 file_printedlen(const struct magic_set *ms)
439 {
440 	return ms->o.buf == NULL ? 0 : strlen(ms->o.buf);
441 }
442 
443 protected int
444 file_replace(struct magic_set *ms, const char *pat, const char *rep)
445 {
446 	regex_t rx;
447 	int rc;
448 
449 	rc = regcomp(&rx, pat, REG_EXTENDED);
450 	if (rc) {
451 		char errmsg[512];
452 		(void)regerror(rc, &rx, errmsg, sizeof(errmsg));
453 		file_magerror(ms, "regex error %d, (%s)", rc, errmsg);
454 		return -1;
455 	} else {
456 		regmatch_t rm;
457 		int nm = 0;
458 		while (regexec(&rx, ms->o.buf, 1, &rm, 0) == 0) {
459 			ms->o.buf[rm.rm_so] = '\0';
460 			if (file_printf(ms, "%s%s", rep,
461 			    rm.rm_eo != 0 ? ms->o.buf + rm.rm_eo : "") == -1)
462 				return -1;
463 			nm++;
464 		}
465 		regfree(&rx);
466 		return nm;
467 	}
468 }
469