xref: /PHP-5.5/ext/fileinfo/libmagic/magic.c (revision cdc1a637)
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 #include "file.h"
29 
30 #ifndef	lint
31 FILE_RCSID("@(#)$File: magic.c,v 1.78 2013/01/07 18:20:19 christos Exp $")
32 #endif	/* lint */
33 
34 #include "magic.h"
35 
36 #include <stdlib.h>
37 #ifdef PHP_WIN32
38 #include "win32/unistd.h"
39 #else
40 #include <unistd.h>
41 #endif
42 #include <string.h>
43 #ifdef PHP_WIN32
44 # include "config.w32.h"
45 #else
46 # include "php_config.h"
47 #endif
48 
49 #ifdef PHP_WIN32
50 #include <shlwapi.h>
51 #endif
52 
53 #include <limits.h>	/* for PIPE_BUF */
54 
55 #if defined(HAVE_UTIMES)
56 # include <sys/time.h>
57 #elif defined(HAVE_UTIME)
58 # if defined(HAVE_SYS_UTIME_H)
59 #  include <sys/utime.h>
60 # elif defined(HAVE_UTIME_H)
61 #  include <utime.h>
62 # endif
63 #endif
64 
65 #ifdef HAVE_UNISTD_H
66 #include <unistd.h>	/* for read() */
67 #endif
68 
69 #ifndef PIPE_BUF
70 /* Get the PIPE_BUF from pathconf */
71 #ifdef _PC_PIPE_BUF
72 #define PIPE_BUF pathconf(".", _PC_PIPE_BUF)
73 #else
74 #define PIPE_BUF 512
75 #endif
76 #endif
77 
78 #ifdef PHP_WIN32
79 # undef S_IFLNK
80 # undef S_IFIFO
81 #endif
82 
83 private void close_and_restore(const struct magic_set *, const char *, int,
84     const struct stat *);
85 private int unreadable_info(struct magic_set *, mode_t, const char *);
86 #if 0
87 private const char* get_default_magic(void);
88 #endif
89 private const char *file_or_stream(struct magic_set *, const char *, php_stream *);
90 
91 #ifndef	STDIN_FILENO
92 #define	STDIN_FILENO	0
93 #endif
94 
95 /* XXX this functionality is excluded in php, enable it in apprentice.c:340 */
96 #if 0
97 private const char *
98 get_default_magic(void)
99 {
100 	static const char hmagic[] = "/.magic/magic.mgc";
101 	static char *default_magic;
102 	char *home, *hmagicpath;
103 
104 #ifndef PHP_WIN32
105 	struct stat st;
106 
107 	if (default_magic) {
108 		free(default_magic);
109 		default_magic = NULL;
110 	}
111 	if ((home = getenv("HOME")) == NULL)
112 		return MAGIC;
113 
114 	if (asprintf(&hmagicpath, "%s/.magic.mgc", home) < 0)
115 		return MAGIC;
116 	if (stat(hmagicpath, &st) == -1) {
117 		free(hmagicpath);
118 	if (asprintf(&hmagicpath, "%s/.magic", home) < 0)
119 		return MAGIC;
120 	if (stat(hmagicpath, &st) == -1)
121 		goto out;
122 	if (S_ISDIR(st.st_mode)) {
123 		free(hmagicpath);
124 		if (asprintf(&hmagicpath, "%s/%s", home, hmagic) < 0)
125 			return MAGIC;
126 		if (access(hmagicpath, R_OK) == -1)
127 			goto out;
128 	}
129 	}
130 
131 	if (asprintf(&default_magic, "%s:%s", hmagicpath, MAGIC) < 0)
132 		goto out;
133 	free(hmagicpath);
134 	return default_magic;
135 out:
136 	default_magic = NULL;
137 	free(hmagicpath);
138 	return MAGIC;
139 #else
140 	char *hmagicp = hmagicpath;
141 	char *tmppath = NULL;
142 	LPTSTR dllpath;
143 
144 #define APPENDPATH() \
145 	do { \
146 		if (tmppath && access(tmppath, R_OK) != -1) { \
147 			if (hmagicpath == NULL) \
148 				hmagicpath = tmppath; \
149 			else { \
150 				if (asprintf(&hmagicp, "%s%c%s", hmagicpath, \
151 				    PATHSEP, tmppath) >= 0) { \
152 					free(hmagicpath); \
153 					hmagicpath = hmagicp; \
154 				} \
155 				free(tmppath); \
156 			} \
157 			tmppath = NULL; \
158 		} \
159 	} while (/*CONSTCOND*/0)
160 
161 	if (default_magic) {
162 		free(default_magic);
163 		default_magic = NULL;
164 	}
165 
166 	/* First, try to get user-specific magic file */
167 	if ((home = getenv("LOCALAPPDATA")) == NULL) {
168 		if ((home = getenv("USERPROFILE")) != NULL)
169 			if (asprintf(&tmppath,
170 			    "%s/Local Settings/Application Data%s", home,
171 			    hmagic) < 0)
172 				tmppath = NULL;
173 	} else {
174 		if (asprintf(&tmppath, "%s%s", home, hmagic) < 0)
175 			tmppath = NULL;
176 	}
177 
178 	APPENDPATH();
179 
180 	/* Second, try to get a magic file from Common Files */
181 	if ((home = getenv("COMMONPROGRAMFILES")) != NULL) {
182 		if (asprintf(&tmppath, "%s%s", home, hmagic) >= 0)
183 			APPENDPATH();
184 	}
185 
186 	/* Third, try to get magic file relative to dll location */
187 	dllpath = malloc(sizeof(*dllpath) * (MAX_PATH + 1));
188 	dllpath[MAX_PATH] = 0;	/* just in case long path gets truncated and not null terminated */
189 	if (GetModuleFileNameA(NULL, dllpath, MAX_PATH)){
190 		PathRemoveFileSpecA(dllpath);
191 		if (strlen(dllpath) > 3 &&
192 		    stricmp(&dllpath[strlen(dllpath) - 3], "bin") == 0) {
193 			if (asprintf(&tmppath,
194 			    "%s/../share/misc/magic.mgc", dllpath) >= 0)
195 				APPENDPATH();
196 		} else {
197 			if (asprintf(&tmppath,
198 			    "%s/share/misc/magic.mgc", dllpath) >= 0)
199 				APPENDPATH();
200 			else if (asprintf(&tmppath,
201 			    "%s/magic.mgc", dllpath) >= 0)
202 				APPENDPATH();
203 		}
204 	}
205 
206 	/* Don't put MAGIC constant - it likely points to a file within MSys
207 	tree */
208 	default_magic = hmagicpath;
209 	return default_magic;
210 #endif
211 }
212 
213 public const char *
214 magic_getpath(const char *magicfile, int action)
215 {
216 	if (magicfile != NULL)
217 		return magicfile;
218 
219 	magicfile = getenv("MAGIC");
220 	if (magicfile != NULL)
221 		return magicfile;
222 
223 	return action == FILE_LOAD ? get_default_magic() : MAGIC;
224 }
225 #endif
226 
227 public struct magic_set *
magic_open(int flags)228 magic_open(int flags)
229 {
230 	return file_ms_alloc(flags);
231 }
232 
233 private int
unreadable_info(struct magic_set * ms,mode_t md,const char * file)234 unreadable_info(struct magic_set *ms, mode_t md, const char *file)
235 {
236 	/* We cannot open it, but we were able to stat it. */
237 	if (access(file, W_OK) == 0)
238 		if (file_printf(ms, "writable, ") == -1)
239 			return -1;
240 	if (access(file, X_OK) == 0)
241 		if (file_printf(ms, "executable, ") == -1)
242 			return -1;
243 	if (S_ISREG(md))
244 		if (file_printf(ms, "regular file, ") == -1)
245 			return -1;
246 	if (file_printf(ms, "no read permission") == -1)
247 		return -1;
248 	return 0;
249 }
250 
251 public void
magic_close(struct magic_set * ms)252 magic_close(struct magic_set *ms)
253 {
254 	if (ms == NULL)
255 		return;
256 	file_ms_free(ms);
257 }
258 
259 /*
260  * load a magic file
261  */
262 public int
magic_load(struct magic_set * ms,const char * magicfile)263 magic_load(struct magic_set *ms, const char *magicfile)
264 {
265 	if (ms == NULL)
266 	return -1;
267 	return file_apprentice(ms, magicfile, FILE_LOAD);
268 }
269 
270 public int
magic_compile(struct magic_set * ms,const char * magicfile)271 magic_compile(struct magic_set *ms, const char *magicfile)
272 {
273 	if (ms == NULL)
274 		return -1;
275 	return file_apprentice(ms, magicfile, FILE_COMPILE);
276 }
277 
278 
279 public int
magic_list(struct magic_set * ms,const char * magicfile)280 magic_list(struct magic_set *ms, const char *magicfile)
281 {
282 	if (ms == NULL)
283 		return -1;
284 	return file_apprentice(ms, magicfile, FILE_LIST);
285 }
286 
287 private void
close_and_restore(const struct magic_set * ms,const char * name,int fd,const struct stat * sb)288 close_and_restore(const struct magic_set *ms, const char *name, int fd,
289     const struct stat *sb)
290 {
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 
318 /*
319  * find type of descriptor
320  */
321 public const char *
magic_descriptor(struct magic_set * ms,int fd)322 magic_descriptor(struct magic_set *ms, int fd)
323 {
324 	if (ms == NULL)
325 		return NULL;
326 	return file_or_stream(ms, NULL, NULL);
327 }
328 
329 /*
330  * find type of named file
331  */
332 public const char *
magic_file(struct magic_set * ms,const char * inname)333 magic_file(struct magic_set *ms, const char *inname)
334 {
335 	if (ms == NULL)
336 		return NULL;
337 	return file_or_stream(ms, inname, NULL);
338 }
339 
340 public const char *
magic_stream(struct magic_set * ms,php_stream * stream)341 magic_stream(struct magic_set *ms, php_stream *stream)
342 {
343 	if (ms == NULL)
344 		return NULL;
345 	return file_or_stream(ms, NULL, stream);
346 }
347 
348 private const char *
file_or_stream(struct magic_set * ms,const char * inname,php_stream * stream)349 file_or_stream(struct magic_set *ms, const char *inname, php_stream *stream)
350 {
351 	int	rv = -1;
352 	unsigned char *buf;
353 	struct stat	sb;
354 	ssize_t nbytes = 0;	/* number of bytes read from a datafile */
355 	int no_in_stream = 0;
356 	TSRMLS_FETCH();
357 
358 	if (!inname && !stream) {
359 		return NULL;
360 	}
361 
362 	/*
363 	 * one extra for terminating '\0', and
364 	 * some overlapping space for matches near EOF
365 	 */
366 #define SLOP (1 + sizeof(union VALUETYPE))
367 	buf = emalloc(HOWMANY + SLOP);
368 
369 	if (file_reset(ms) == -1)
370 		goto done;
371 
372 	switch (file_fsmagic(ms, inname, &sb, stream)) {
373 	case -1:		/* error */
374 		goto done;
375 	case 0:			/* nothing found */
376 		break;
377 	default:		/* matched it and printed type */
378 		rv = 0;
379 		goto done;
380 	}
381 
382 	errno = 0;
383 
384 	if (!stream && inname) {
385 		no_in_stream = 1;
386 #if PHP_API_VERSION < 20100412
387 		stream = php_stream_open_wrapper((char *)inname, "rb", REPORT_ERRORS|ENFORCE_SAFE_MODE, NULL);
388 #else
389 		stream = php_stream_open_wrapper((char *)inname, "rb", REPORT_ERRORS, NULL);
390 #endif
391 	}
392 
393 	if (!stream) {
394 		if (unreadable_info(ms, sb.st_mode, inname) == -1)
395 			goto done;
396 		rv = 0;
397 		goto done;
398 	}
399 
400 #ifdef O_NONBLOCK
401 /* we should be already be in non blocking mode for network socket */
402 #endif
403 
404 	/*
405 	 * try looking at the first HOWMANY bytes
406 	 */
407 	if ((nbytes = php_stream_read(stream, (char *)buf, HOWMANY)) < 0) {
408 		file_error(ms, errno, "cannot read `%s'", inname);
409 		goto done;
410 	}
411 
412 	(void)memset(buf + nbytes, 0, SLOP); /* NUL terminate */
413 	if (file_buffer(ms, stream, inname, buf, (size_t)nbytes) == -1)
414 		goto done;
415 	rv = 0;
416 done:
417 	efree(buf);
418 
419 	if (no_in_stream && stream) {
420 		php_stream_close(stream);
421 	}
422 
423 	close_and_restore(ms, inname, 0, &sb);
424 	return rv == 0 ? file_getbuffer(ms) : NULL;
425 }
426 
427 
428 public const char *
magic_buffer(struct magic_set * ms,const void * buf,size_t nb)429 magic_buffer(struct magic_set *ms, const void *buf, size_t nb)
430 {
431 	if (ms == NULL)
432 		return NULL;
433 	if (file_reset(ms) == -1)
434 		return NULL;
435 	/*
436 	 * The main work is done here!
437 	 * We have the file name and/or the data buffer to be identified.
438 	 */
439 	if (file_buffer(ms, NULL, NULL, buf, nb) == -1) {
440 		return NULL;
441 	}
442 	return file_getbuffer(ms);
443 }
444 
445 public const char *
magic_error(struct magic_set * ms)446 magic_error(struct magic_set *ms)
447 {
448 	if (ms == NULL)
449 		return "Magic database is not open";
450 	return (ms->event_flags & EVENT_HAD_ERR) ? ms->o.buf : NULL;
451 }
452 
453 public int
magic_errno(struct magic_set * ms)454 magic_errno(struct magic_set *ms)
455 {
456 	if (ms == NULL)
457 		return EINVAL;
458 	return (ms->event_flags & EVENT_HAD_ERR) ? ms->error : 0;
459 }
460 
461 public int
magic_setflags(struct magic_set * ms,int flags)462 magic_setflags(struct magic_set *ms, int flags)
463 {
464 	if (ms == NULL)
465 		return -1;
466 #if !defined(HAVE_UTIME) && !defined(HAVE_UTIMES)
467 	if (flags & MAGIC_PRESERVE_ATIME)
468 		return -1;
469 #endif
470 	ms->flags = flags;
471 	return 0;
472 }
473 
474 public int
magic_version(void)475 magic_version(void)
476 {
477 	return MAGIC_VERSION;
478 }
479