xref: /netbsd-src/external/bsd/file/dist/src/file.c (revision 6cf6fe02a981b55727c49c3d37b0d8191a98c0ee)
1 /*	$NetBSD: file.c,v 1.4 2014/06/13 02:08:06 christos Exp $	*/
2 /*
3  * Copyright (c) Ian F. Darwin 1986-1995.
4  * Software written by Ian F. Darwin and others;
5  * maintained 1995-present by Christos Zoulas and others.
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 /*
30  * file - find type of a file or files - main program.
31  */
32 
33 #include "file.h"
34 
35 #ifndef	lint
36 #if 0
37 FILE_RCSID("@(#)$File: file.c,v 1.153 2014/02/11 15:41:04 christos Exp $")
38 #else
39 __RCSID("$NetBSD: file.c,v 1.4 2014/06/13 02:08:06 christos Exp $");
40 #endif
41 #endif	/* lint */
42 
43 #include "magic.h"
44 
45 #include <stdlib.h>
46 #include <unistd.h>
47 #include <string.h>
48 #ifdef RESTORE_TIME
49 # if (__COHERENT__ >= 0x420)
50 #  include <sys/utime.h>
51 # else
52 #  ifdef USE_UTIMES
53 #   include <sys/time.h>
54 #  else
55 #   include <utime.h>
56 #  endif
57 # endif
58 #endif
59 #ifdef HAVE_UNISTD_H
60 #include <unistd.h>	/* for read() */
61 #endif
62 #ifdef HAVE_LOCALE_H
63 #include <locale.h>
64 #endif
65 #ifdef HAVE_WCHAR_H
66 #include <wchar.h>
67 #endif
68 
69 #if defined(HAVE_GETOPT_H) && defined(HAVE_STRUCT_OPTION)
70 #include <getopt.h>
71 #ifndef HAVE_GETOPT_LONG
72 int getopt_long(int argc, char * const *argv, const char *optstring, const struct option *longopts, int *longindex);
73 #endif
74 #else
75 #include "mygetopt.h"
76 #endif
77 
78 #ifdef S_IFLNK
79 #define FILE_FLAGS "-bcEhikLlNnprsvz0"
80 #else
81 #define FILE_FLAGS "-bcEiklNnprsvz0"
82 #endif
83 
84 # define USAGE  \
85     "Usage: %s [" FILE_FLAGS \
86 	"] [--apple] [--mime-encoding] [--mime-type]\n" \
87     "            [-e testname] [-F separator] [-f namefile] [-m magicfiles] " \
88     "file ...\n" \
89     "       %s -C [-m magicfiles]\n" \
90     "       %s [--help]\n"
91 
92 private int 		/* Global command-line options 		*/
93 	bflag = 0,	/* brief output format	 		*/
94 	nopad = 0,	/* Don't pad output			*/
95 	nobuffer = 0,   /* Do not buffer stdout 		*/
96 	nulsep = 0;	/* Append '\0' to the separator		*/
97 
98 private const char *separator = ":";	/* Default field separator	*/
99 private const struct option long_options[] = {
100 #define OPT(shortname, longname, opt, doc)      \
101     {longname, opt, NULL, shortname},
102 #define OPT_LONGONLY(longname, opt, doc)        \
103     {longname, opt, NULL, 0},
104 #include "file_opts.h"
105 #undef OPT
106 #undef OPT_LONGONLY
107     {0, 0, NULL, 0}
108 };
109 #define OPTSTRING	"bcCde:Ef:F:hiklLm:nNprsvz0"
110 
111 private const struct {
112 	const char *name;
113 	int value;
114 } nv[] = {
115 	{ "apptype",	MAGIC_NO_CHECK_APPTYPE },
116 	{ "ascii",	MAGIC_NO_CHECK_ASCII },
117 	{ "cdf",	MAGIC_NO_CHECK_CDF },
118 	{ "compress",	MAGIC_NO_CHECK_COMPRESS },
119 	{ "elf",	MAGIC_NO_CHECK_ELF },
120 	{ "encoding",	MAGIC_NO_CHECK_ENCODING },
121 	{ "soft",	MAGIC_NO_CHECK_SOFT },
122 	{ "tar",	MAGIC_NO_CHECK_TAR },
123 	{ "text",	MAGIC_NO_CHECK_TEXT },	/* synonym for ascii */
124 	{ "tokens",	MAGIC_NO_CHECK_TOKENS }, /* OBSOLETE: ignored for backwards compatibility */
125 };
126 
127 private char *progname;		/* used throughout 		*/
128 
129 #ifdef __dead
130 __dead
131 #endif
132 private void usage(void);
133 private void docprint(const char *);
134 #ifdef __dead
135 __dead
136 #endif
137 private void help(void);
138 
139 private int unwrap(struct magic_set *, const char *);
140 private int process(struct magic_set *ms, const char *, int);
141 private struct magic_set *load(const char *, int);
142 
143 
144 /*
145  * main - parse arguments and handle options
146  */
147 int
148 main(int argc, char *argv[])
149 {
150 	int c;
151 	size_t i;
152 	int action = 0, didsomefiles = 0, errflg = 0;
153 	int flags = 0, e = 0;
154 	struct magic_set *magic = NULL;
155 	int longindex;
156 	const char *magicfile = NULL;		/* where the magic is	*/
157 
158 	/* makes islower etc work for other langs */
159 	(void)setlocale(LC_CTYPE, "");
160 
161 #ifdef __EMX__
162 	/* sh-like wildcard expansion! Shouldn't hurt at least ... */
163 	_wildcard(&argc, &argv);
164 #endif
165 
166 	if ((progname = strrchr(argv[0], '/')) != NULL)
167 		progname++;
168 	else
169 		progname = argv[0];
170 
171 #ifdef S_IFLNK
172 	flags |= getenv("POSIXLY_CORRECT") ? MAGIC_SYMLINK : 0;
173 #endif
174 	while ((c = getopt_long(argc, argv, OPTSTRING, long_options,
175 	    &longindex)) != -1)
176 		switch (c) {
177 		case 0 :
178 			switch (longindex) {
179 			case 0:
180 				help();
181 				break;
182 			case 10:
183 				flags |= MAGIC_APPLE;
184 				break;
185 			case 11:
186 				flags |= MAGIC_MIME_TYPE;
187 				break;
188 			case 12:
189 				flags |= MAGIC_MIME_ENCODING;
190 				break;
191 			}
192 			break;
193 		case '0':
194 			nulsep = 1;
195 			break;
196 		case 'b':
197 			bflag++;
198 			break;
199 		case 'c':
200 			action = FILE_CHECK;
201 			break;
202 		case 'C':
203 			action = FILE_COMPILE;
204 			break;
205 		case 'd':
206 			flags |= MAGIC_DEBUG|MAGIC_CHECK;
207 			break;
208 		case 'E':
209 			flags |= MAGIC_ERROR;
210 			break;
211 		case 'e':
212 			for (i = 0; i < sizeof(nv) / sizeof(nv[0]); i++)
213 				if (strcmp(nv[i].name, optarg) == 0)
214 					break;
215 
216 			if (i == sizeof(nv) / sizeof(nv[0]))
217 				errflg++;
218 			else
219 				flags |= nv[i].value;
220 			break;
221 
222 		case 'f':
223 			if(action)
224 				usage();
225 			if (magic == NULL)
226 				if ((magic = load(magicfile, flags)) == NULL)
227 					return 1;
228 			e |= unwrap(magic, optarg);
229 			++didsomefiles;
230 			break;
231 		case 'F':
232 			separator = optarg;
233 			break;
234 		case 'i':
235 			flags |= MAGIC_MIME;
236 			break;
237 		case 'k':
238 			flags |= MAGIC_CONTINUE;
239 			break;
240 		case 'l':
241 			action = FILE_LIST;
242 			break;
243 		case 'm':
244 			magicfile = optarg;
245 			break;
246 		case 'n':
247 			++nobuffer;
248 			break;
249 		case 'N':
250 			++nopad;
251 			break;
252 #if defined(HAVE_UTIME) || defined(HAVE_UTIMES)
253 		case 'p':
254 			flags |= MAGIC_PRESERVE_ATIME;
255 			break;
256 #endif
257 		case 'r':
258 			flags |= MAGIC_RAW;
259 			break;
260 		case 's':
261 			flags |= MAGIC_DEVICES;
262 			break;
263 		case 'v':
264 			if (magicfile == NULL)
265 				magicfile = magic_getpath(magicfile, action);
266 			(void)fprintf(stdout, "%s-%s\n", progname, VERSION);
267 			(void)fprintf(stdout, "magic file from %s\n",
268 				       magicfile);
269 			return 0;
270 		case 'z':
271 			flags |= MAGIC_COMPRESS;
272 			break;
273 #ifdef S_IFLNK
274 		case 'L':
275 			flags |= MAGIC_SYMLINK;
276 			break;
277 		case 'h':
278 			flags &= ~MAGIC_SYMLINK;
279 			break;
280 #endif
281 		case '?':
282 		default:
283 			errflg++;
284 			break;
285 		}
286 
287 	if (errflg) {
288 		usage();
289 	}
290 	if (e)
291 		return e;
292 
293 	if (MAGIC_VERSION != magic_version())
294 		(void)fprintf(stderr, "%s: compiled magic version [%d] "
295 		    "does not match with shared library magic version [%d]\n",
296 		    progname, MAGIC_VERSION, magic_version());
297 
298 	switch(action) {
299 	case FILE_CHECK:
300 	case FILE_COMPILE:
301 	case FILE_LIST:
302 		/*
303 		 * Don't try to check/compile ~/.magic unless we explicitly
304 		 * ask for it.
305 		 */
306 		magic = magic_open(flags|MAGIC_CHECK);
307 		if (magic == NULL) {
308 			(void)fprintf(stderr, "%s: %s\n", progname,
309 			    strerror(errno));
310 			return 1;
311 		}
312 		switch(action) {
313 		case FILE_CHECK:
314 			c = magic_check(magic, magicfile);
315 			break;
316 		case FILE_COMPILE:
317 			c = magic_compile(magic, magicfile);
318 			break;
319 		case FILE_LIST:
320 			c = magic_list(magic, magicfile);
321 			break;
322 		default:
323 			abort();
324 		}
325 		if (c == -1) {
326 			(void)fprintf(stderr, "%s: %s\n", progname,
327 			    magic_error(magic));
328 			return 1;
329 		}
330 		return 0;
331 	default:
332 		if (magic == NULL)
333 			if ((magic = load(magicfile, flags)) == NULL)
334 				return 1;
335 		break;
336 	}
337 
338 	if (optind == argc) {
339 		if (!didsomefiles)
340 			usage();
341 	}
342 	else {
343 		size_t j, wid, nw;
344 		for (wid = 0, j = (size_t)optind; j < (size_t)argc; j++) {
345 			nw = file_mbswidth(argv[j]);
346 			if (nw > wid)
347 				wid = nw;
348 		}
349 		/*
350 		 * If bflag is only set twice, set it depending on
351 		 * number of files [this is undocumented, and subject to change]
352 		 */
353 		if (bflag == 2) {
354 			bflag = optind >= argc - 1;
355 		}
356 		for (; optind < argc; optind++)
357 			e |= process(magic, argv[optind], wid);
358 	}
359 
360 	if (magic)
361 		magic_close(magic);
362 	return e;
363 }
364 
365 
366 private struct magic_set *
367 /*ARGSUSED*/
368 load(const char *magicfile, int flags)
369 {
370 	struct magic_set *magic = magic_open(flags);
371 	if (magic == NULL) {
372 		(void)fprintf(stderr, "%s: %s\n", progname, strerror(errno));
373 		return NULL;
374 	}
375 	if (magic_load(magic, magicfile) == -1) {
376 		(void)fprintf(stderr, "%s: %s\n",
377 		    progname, magic_error(magic));
378 		magic_close(magic);
379 		return NULL;
380 	}
381 	return magic;
382 }
383 
384 /*
385  * unwrap -- read a file of filenames, do each one.
386  */
387 private int
388 unwrap(struct magic_set *ms, const char *fn)
389 {
390 	FILE *f;
391 	ssize_t len;
392 	char *line = NULL;
393 	size_t llen = 0;
394 	int wid = 0, cwid;
395 	int e = 0;
396 
397 	if (strcmp("-", fn) == 0) {
398 		f = stdin;
399 		wid = 1;
400 	} else {
401 		if ((f = fopen(fn, "r")) == NULL) {
402 			(void)fprintf(stderr, "%s: Cannot open `%s' (%s).\n",
403 			    progname, fn, strerror(errno));
404 			return 1;
405 		}
406 
407 		while ((len = getline(&line, &llen, f)) > 0) {
408 			if (line[len - 1] == '\n')
409 				line[len - 1] = '\0';
410 			cwid = file_mbswidth(line);
411 			if (cwid > wid)
412 				wid = cwid;
413 		}
414 
415 		rewind(f);
416 	}
417 
418 	while ((len = getline(&line, &llen, f)) > 0) {
419 		if (line[len - 1] == '\n')
420 			line[len - 1] = '\0';
421 		e |= process(ms, line, wid);
422 		if(nobuffer)
423 			(void)fflush(stdout);
424 	}
425 
426 	free(line);
427 	(void)fclose(f);
428 	return e;
429 }
430 
431 /*
432  * Called for each input file on the command line (or in a list of files)
433  */
434 private int
435 process(struct magic_set *ms, const char *inname, int wid)
436 {
437 	const char *type;
438 	int std_in = strcmp(inname, "-") == 0;
439 
440 	if (wid > 0 && !bflag) {
441 		(void)printf("%s", std_in ? "/dev/stdin" : inname);
442 		if (nulsep)
443 			(void)putc('\0', stdout);
444 		(void)printf("%s", separator);
445 		(void)printf("%*s ",
446 		    (int) (nopad ? 0 : (wid - file_mbswidth(inname))), "");
447 	}
448 
449 	type = magic_file(ms, std_in ? NULL : inname);
450 	if (type == NULL) {
451 		(void)printf("ERROR: %s\n", magic_error(ms));
452 		return 1;
453 	} else {
454 		(void)printf("%s\n", type);
455 		return 0;
456 	}
457 }
458 
459 protected size_t
460 file_mbswidth(const char *s)
461 {
462 #if defined(HAVE_WCHAR_H) && defined(HAVE_MBRTOWC) && defined(HAVE_WCWIDTH)
463 	size_t bytesconsumed, old_n, n, width = 0;
464 	mbstate_t state;
465 	wchar_t nextchar;
466 	(void)memset(&state, 0, sizeof(mbstate_t));
467 	old_n = n = strlen(s);
468 
469 	while (n > 0) {
470 		bytesconsumed = mbrtowc(&nextchar, s, n, &state);
471 		if (bytesconsumed == (size_t)(-1) ||
472 		    bytesconsumed == (size_t)(-2)) {
473 			/* Something went wrong, return something reasonable */
474 			return old_n;
475 		}
476 		if (s[0] == '\n') {
477 			/*
478 			 * do what strlen() would do, so that caller
479 			 * is always right
480 			 */
481 			width++;
482 		} else {
483 			int w = wcwidth(nextchar);
484 			if (w > 0)
485 				width += w;
486 		}
487 
488 		s += bytesconsumed, n -= bytesconsumed;
489 	}
490 	return width;
491 #else
492 	return strlen(s);
493 #endif
494 }
495 
496 private void
497 usage(void)
498 {
499 	(void)fprintf(stderr, USAGE, progname, progname, progname);
500 	exit(1);
501 }
502 
503 private void
504 docprint(const char *opts)
505 {
506 	size_t i;
507 	int comma;
508 	char *sp, *p;
509 
510 	p = strstr(opts, "%o");
511 	if (p == NULL) {
512 		fprintf(stdout, "%s", opts);
513 		return;
514 	}
515 
516 	for (sp = p - 1; sp > opts && *sp == ' '; sp--)
517 		continue;
518 
519 	fprintf(stdout, "%.*s", (int)(p - opts), opts);
520 
521 	comma = 0;
522 	for (i = 0; i < __arraycount(nv); i++) {
523 		fprintf(stdout, "%s%s", comma++ ? ", " : "", nv[i].name);
524 		if (i && i % 5 == 0) {
525 			fprintf(stdout, ",\n%*s", (int)(p - sp - 1), "");
526 			comma = 0;
527 		}
528 	}
529 
530 	fprintf(stdout, "%s", opts + (p - opts) + 2);
531 }
532 
533 private void
534 help(void)
535 {
536 	(void)fputs(
537 "Usage: file [OPTION...] [FILE...]\n"
538 "Determine type of FILEs.\n"
539 "\n", stdout);
540 #define OPT(shortname, longname, opt, doc)      \
541 	fprintf(stdout, "  -%c, --" longname, shortname), \
542 	docprint(doc);
543 #define OPT_LONGONLY(longname, opt, doc)        \
544 	fprintf(stdout, "      --" longname),	\
545 	docprint(doc);
546 #include "file_opts.h"
547 #undef OPT
548 #undef OPT_LONGONLY
549 	fprintf(stdout, "\nReport bugs to http://bugs.gw.com/\n");
550 	exit(0);
551 }
552