xref: /freebsd/contrib/file/src/magic.c (revision 57718be8fa0bd5edc11ab9a72e68cc71982939a6)
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 
28 #ifdef WIN32
29 #include <windows.h>
30 #include <shlwapi.h>
31 #endif
32 
33 #include "file.h"
34 
35 #ifndef	lint
36 FILE_RCSID("@(#)$File: magic.c,v 1.84 2014/05/14 23:15:42 christos Exp $")
37 #endif	/* lint */
38 
39 #include "magic.h"
40 
41 #include <stdlib.h>
42 #include <unistd.h>
43 #include <string.h>
44 #ifdef QUICK
45 #include <sys/mman.h>
46 #endif
47 #ifdef HAVE_LIMITS_H
48 #include <limits.h>	/* for PIPE_BUF */
49 #endif
50 
51 #if defined(HAVE_UTIMES)
52 # include <sys/time.h>
53 #elif defined(HAVE_UTIME)
54 # if defined(HAVE_SYS_UTIME_H)
55 #  include <sys/utime.h>
56 # elif defined(HAVE_UTIME_H)
57 #  include <utime.h>
58 # endif
59 #endif
60 
61 #ifdef HAVE_UNISTD_H
62 #include <unistd.h>	/* for read() */
63 #endif
64 
65 #ifndef PIPE_BUF
66 /* Get the PIPE_BUF from pathconf */
67 #ifdef _PC_PIPE_BUF
68 #define PIPE_BUF pathconf(".", _PC_PIPE_BUF)
69 #else
70 #define PIPE_BUF 512
71 #endif
72 #endif
73 
74 private void close_and_restore(const struct magic_set *, const char *, int,
75     const struct stat *);
76 private int unreadable_info(struct magic_set *, mode_t, const char *);
77 private const char* get_default_magic(void);
78 #ifndef COMPILE_ONLY
79 private const char *file_or_fd(struct magic_set *, const char *, int);
80 #endif
81 
82 #ifndef	STDIN_FILENO
83 #define	STDIN_FILENO	0
84 #endif
85 
86 private const char *
87 get_default_magic(void)
88 {
89 	static const char hmagic[] = "/.magic/magic.mgc";
90 	static char *default_magic;
91 	char *home, *hmagicpath;
92 
93 #ifndef WIN32
94 	struct stat st;
95 
96 	if (default_magic) {
97 		free(default_magic);
98 		default_magic = NULL;
99 	}
100 	if ((home = getenv("HOME")) == NULL)
101 		return MAGIC;
102 
103 	if (asprintf(&hmagicpath, "%s/.magic.mgc", home) < 0)
104 		return MAGIC;
105 	if (stat(hmagicpath, &st) == -1) {
106 		free(hmagicpath);
107 		if (asprintf(&hmagicpath, "%s/.magic", home) < 0)
108 			return MAGIC;
109 		if (stat(hmagicpath, &st) == -1)
110 			goto out;
111 		if (S_ISDIR(st.st_mode)) {
112 			free(hmagicpath);
113 			if (asprintf(&hmagicpath, "%s/%s", home, hmagic) < 0)
114 				return MAGIC;
115 			if (access(hmagicpath, R_OK) == -1)
116 				goto out;
117 		}
118 	}
119 
120 	if (asprintf(&default_magic, "%s:%s", hmagicpath, MAGIC) < 0)
121 		goto out;
122 	free(hmagicpath);
123 	return default_magic;
124 out:
125 	default_magic = NULL;
126 	free(hmagicpath);
127 	return MAGIC;
128 #else
129 	char *hmagicp;
130 	char *tmppath = NULL;
131 	hmagicpath = NULL;
132 
133 #define APPENDPATH() \
134 	do { \
135 		if (tmppath && access(tmppath, R_OK) != -1) { \
136 			if (hmagicpath == NULL) \
137 				hmagicpath = tmppath; \
138 			else { \
139 				if (asprintf(&hmagicp, "%s%c%s", hmagicpath, \
140 				    PATHSEP, tmppath) >= 0) { \
141 					free(hmagicpath); \
142 					hmagicpath = hmagicp; \
143 				} \
144 				free(tmppath); \
145 			} \
146 			tmppath = NULL; \
147 		} \
148 	} while (/*CONSTCOND*/0)
149 
150 	if (default_magic) {
151 		free(default_magic);
152 		default_magic = NULL;
153 	}
154 
155 	/* First, try to get user-specific magic file */
156 	if ((home = getenv("LOCALAPPDATA")) == NULL) {
157 		if ((home = getenv("USERPROFILE")) != NULL)
158 			if (asprintf(&tmppath,
159 			    "%s/Local Settings/Application Data%s", home,
160 			    hmagic) < 0)
161 				tmppath = NULL;
162 	} else {
163 		if (asprintf(&tmppath, "%s%s", home, hmagic) < 0)
164 			tmppath = NULL;
165 	}
166 
167 	APPENDPATH();
168 
169 	/* Second, try to get a magic file from Common Files */
170 	if ((home = getenv("COMMONPROGRAMFILES")) != NULL) {
171 		if (asprintf(&tmppath, "%s%s", home, hmagic) >= 0)
172 			APPENDPATH();
173 	}
174 
175 	/* Third, try to get magic file relative to dll location */
176 	LPTSTR dllpath = malloc(sizeof(*dllpath) * (MAX_PATH + 1));
177 	dllpath[MAX_PATH] = 0;	/* just in case long path gets truncated and not null terminated */
178 	if (GetModuleFileNameA(NULL, dllpath, MAX_PATH)){
179 		PathRemoveFileSpecA(dllpath);
180 		if (strlen(dllpath) > 3 &&
181 		    stricmp(&dllpath[strlen(dllpath) - 3], "bin") == 0) {
182 			if (asprintf(&tmppath,
183 			    "%s/../share/misc/magic.mgc", dllpath) >= 0)
184 				APPENDPATH();
185 		} else {
186 			if (asprintf(&tmppath,
187 			    "%s/share/misc/magic.mgc", dllpath) >= 0)
188 				APPENDPATH();
189 			else if (asprintf(&tmppath,
190 			    "%s/magic.mgc", dllpath) >= 0)
191 				APPENDPATH();
192 		}
193 	}
194 
195 	/* Don't put MAGIC constant - it likely points to a file within MSys
196 	tree */
197 	default_magic = hmagicpath;
198 	return default_magic;
199 #endif
200 }
201 
202 public const char *
203 magic_getpath(const char *magicfile, int action)
204 {
205 	if (magicfile != NULL)
206 		return magicfile;
207 
208 	magicfile = getenv("MAGIC");
209 	if (magicfile != NULL)
210 		return magicfile;
211 
212 	return action == FILE_LOAD ? get_default_magic() : MAGIC;
213 }
214 
215 public struct magic_set *
216 magic_open(int flags)
217 {
218 	return file_ms_alloc(flags);
219 }
220 
221 private int
222 unreadable_info(struct magic_set *ms, mode_t md, const char *file)
223 {
224 	if (file) {
225 		/* We cannot open it, but we were able to stat it. */
226 		if (access(file, W_OK) == 0)
227 			if (file_printf(ms, "writable, ") == -1)
228 				return -1;
229 		if (access(file, X_OK) == 0)
230 			if (file_printf(ms, "executable, ") == -1)
231 				return -1;
232 	}
233 	if (S_ISREG(md))
234 		if (file_printf(ms, "regular file, ") == -1)
235 			return -1;
236 	if (file_printf(ms, "no read permission") == -1)
237 		return -1;
238 	return 0;
239 }
240 
241 public void
242 magic_close(struct magic_set *ms)
243 {
244 	if (ms == NULL)
245 		return;
246 	file_ms_free(ms);
247 }
248 
249 /*
250  * load a magic file
251  */
252 public int
253 magic_load(struct magic_set *ms, const char *magicfile)
254 {
255 	if (ms == NULL)
256 		return -1;
257 	return file_apprentice(ms, magicfile, FILE_LOAD);
258 }
259 
260 public int
261 magic_compile(struct magic_set *ms, const char *magicfile)
262 {
263 	if (ms == NULL)
264 		return -1;
265 	return file_apprentice(ms, magicfile, FILE_COMPILE);
266 }
267 
268 public int
269 magic_check(struct magic_set *ms, const char *magicfile)
270 {
271 	if (ms == NULL)
272 		return -1;
273 	return file_apprentice(ms, magicfile, FILE_CHECK);
274 }
275 
276 public int
277 magic_list(struct magic_set *ms, const char *magicfile)
278 {
279 	if (ms == NULL)
280 		return -1;
281 	return file_apprentice(ms, magicfile, FILE_LIST);
282 }
283 
284 private void
285 close_and_restore(const struct magic_set *ms, const char *name, int fd,
286     const struct stat *sb)
287 {
288 	if (fd == STDIN_FILENO || name == NULL)
289 		return;
290 	(void) close(fd);
291 
292 	if ((ms->flags & MAGIC_PRESERVE_ATIME) != 0) {
293 		/*
294 		 * Try to restore access, modification times if read it.
295 		 * This is really *bad* because it will modify the status
296 		 * time of the file... And of course this will affect
297 		 * backup programs
298 		 */
299 #ifdef HAVE_UTIMES
300 		struct timeval  utsbuf[2];
301 		(void)memset(utsbuf, 0, sizeof(utsbuf));
302 		utsbuf[0].tv_sec = sb->st_atime;
303 		utsbuf[1].tv_sec = sb->st_mtime;
304 
305 		(void) utimes(name, utsbuf); /* don't care if loses */
306 #elif defined(HAVE_UTIME_H) || defined(HAVE_SYS_UTIME_H)
307 		struct utimbuf  utbuf;
308 
309 		(void)memset(&utbuf, 0, sizeof(utbuf));
310 		utbuf.actime = sb->st_atime;
311 		utbuf.modtime = sb->st_mtime;
312 		(void) utime(name, &utbuf); /* don't care if loses */
313 #endif
314 	}
315 }
316 
317 #ifndef COMPILE_ONLY
318 
319 /*
320  * find type of descriptor
321  */
322 public const char *
323 magic_descriptor(struct magic_set *ms, int fd)
324 {
325 	if (ms == NULL)
326 		return NULL;
327 	return file_or_fd(ms, NULL, fd);
328 }
329 
330 /*
331  * find type of named file
332  */
333 public const char *
334 magic_file(struct magic_set *ms, const char *inname)
335 {
336 	if (ms == NULL)
337 		return NULL;
338 	return file_or_fd(ms, inname, STDIN_FILENO);
339 }
340 
341 private const char *
342 file_or_fd(struct magic_set *ms, const char *inname, int fd)
343 {
344 	int	rv = -1;
345 	unsigned char *buf;
346 	struct stat	sb;
347 	ssize_t nbytes = 0;	/* number of bytes read from a datafile */
348 	int	ispipe = 0;
349 	off_t	pos = (off_t)-1;
350 
351 	if (file_reset(ms) == -1)
352 		goto out;
353 
354 	/*
355 	 * one extra for terminating '\0', and
356 	 * some overlapping space for matches near EOF
357 	 */
358 #define SLOP (1 + sizeof(union VALUETYPE))
359 	if ((buf = CAST(unsigned char *, malloc(HOWMANY + SLOP))) == NULL)
360 		return NULL;
361 
362 	switch (file_fsmagic(ms, inname, &sb)) {
363 	case -1:		/* error */
364 		goto done;
365 	case 0:			/* nothing found */
366 		break;
367 	default:		/* matched it and printed type */
368 		rv = 0;
369 		goto done;
370 	}
371 
372 #ifdef WIN32
373 	/* Place stdin in binary mode, so EOF (Ctrl+Z) doesn't stop early. */
374 	if (fd == STDIN_FILENO)
375 		_setmode(STDIN_FILENO, O_BINARY);
376 #endif
377 
378 	if (inname == NULL) {
379 		if (fstat(fd, &sb) == 0 && S_ISFIFO(sb.st_mode))
380 			ispipe = 1;
381 		else
382 			pos = lseek(fd, (off_t)0, SEEK_CUR);
383 	} else {
384 		int flags = O_RDONLY|O_BINARY;
385 		int okstat = stat(inname, &sb) == 0;
386 
387 		if (okstat && S_ISFIFO(sb.st_mode)) {
388 #ifdef O_NONBLOCK
389 			flags |= O_NONBLOCK;
390 #endif
391 			ispipe = 1;
392 		}
393 
394 		errno = 0;
395 		if ((fd = open(inname, flags)) < 0) {
396 #ifdef WIN32
397 			/*
398 			 * Can't stat, can't open.  It may have been opened in
399 			 * fsmagic, so if the user doesn't have read permission,
400 			 * allow it to say so; otherwise an error was probably
401 			 * displayed in fsmagic.
402 			 */
403 			if (!okstat && errno == EACCES) {
404 				sb.st_mode = S_IFBLK;
405 				okstat = 1;
406 			}
407 #endif
408 			if (okstat &&
409 			    unreadable_info(ms, sb.st_mode, inname) == -1)
410 				goto done;
411 			rv = 0;
412 			goto done;
413 		}
414 #ifdef O_NONBLOCK
415 		if ((flags = fcntl(fd, F_GETFL)) != -1) {
416 			flags &= ~O_NONBLOCK;
417 			(void)fcntl(fd, F_SETFL, flags);
418 		}
419 #endif
420 	}
421 
422 	/*
423 	 * try looking at the first HOWMANY bytes
424 	 */
425 	if (ispipe) {
426 		ssize_t r = 0;
427 
428 		while ((r = sread(fd, (void *)&buf[nbytes],
429 		    (size_t)(HOWMANY - nbytes), 1)) > 0) {
430 			nbytes += r;
431 			if (r < PIPE_BUF) break;
432 		}
433 
434 		if (nbytes == 0) {
435 			/* We can not read it, but we were able to stat it. */
436 			if (unreadable_info(ms, sb.st_mode, inname) == -1)
437 				goto done;
438 			rv = 0;
439 			goto done;
440 		}
441 
442 	} else {
443 		/* Windows refuses to read from a big console buffer. */
444 		size_t howmany =
445 #if defined(WIN32) && HOWMANY > 8 * 1024
446 				_isatty(fd) ? 8 * 1024 :
447 #endif
448 				HOWMANY;
449 		if ((nbytes = read(fd, (char *)buf, howmany)) == -1) {
450 			if (inname == NULL && fd != STDIN_FILENO)
451 				file_error(ms, errno, "cannot read fd %d", fd);
452 			else
453 				file_error(ms, errno, "cannot read `%s'",
454 				    inname == NULL ? "/dev/stdin" : inname);
455 			goto done;
456 		}
457 	}
458 
459 	(void)memset(buf + nbytes, 0, SLOP); /* NUL terminate */
460 	if (file_buffer(ms, fd, inname, buf, (size_t)nbytes) == -1)
461 		goto done;
462 	rv = 0;
463 done:
464 	free(buf);
465 	if (pos != (off_t)-1)
466 		(void)lseek(fd, pos, SEEK_SET);
467 	close_and_restore(ms, inname, fd, &sb);
468 out:
469 	return rv == 0 ? file_getbuffer(ms) : NULL;
470 }
471 
472 
473 public const char *
474 magic_buffer(struct magic_set *ms, const void *buf, size_t nb)
475 {
476 	if (ms == NULL)
477 		return NULL;
478 	if (file_reset(ms) == -1)
479 		return NULL;
480 	/*
481 	 * The main work is done here!
482 	 * We have the file name and/or the data buffer to be identified.
483 	 */
484 	if (file_buffer(ms, -1, NULL, buf, nb) == -1) {
485 		return NULL;
486 	}
487 	return file_getbuffer(ms);
488 }
489 #endif
490 
491 public const char *
492 magic_error(struct magic_set *ms)
493 {
494 	if (ms == NULL)
495 		return "Magic database is not open";
496 	return (ms->event_flags & EVENT_HAD_ERR) ? ms->o.buf : NULL;
497 }
498 
499 public int
500 magic_errno(struct magic_set *ms)
501 {
502 	if (ms == NULL)
503 		return EINVAL;
504 	return (ms->event_flags & EVENT_HAD_ERR) ? ms->error : 0;
505 }
506 
507 public int
508 magic_setflags(struct magic_set *ms, int flags)
509 {
510 	if (ms == NULL)
511 		return -1;
512 #if !defined(HAVE_UTIME) && !defined(HAVE_UTIMES)
513 	if (flags & MAGIC_PRESERVE_ATIME)
514 		return -1;
515 #endif
516 	ms->flags = flags;
517 	return 0;
518 }
519 
520 public int
521 magic_version(void)
522 {
523 	return MAGIC_VERSION;
524 }
525