xref: /netbsd-src/external/bsd/file/dist/src/file.c (revision e6c7e151de239c49d2e38720a061ed9d1fa99309)
1 /*	$NetBSD: file.c,v 1.14 2019/12/17 02:31:05 christos Exp $	*/
2 
3 /*
4  * Copyright (c) Ian F. Darwin 1986-1995.
5  * Software written by Ian F. Darwin and others;
6  * maintained 1995-present by Christos Zoulas and others.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice immediately at the beginning of the file, without modification,
13  *    this list of conditions, and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  *
18  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
19  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21  * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
22  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28  * SUCH DAMAGE.
29  */
30 /*
31  * file - find type of a file or files - main program.
32  */
33 
34 #include "file.h"
35 
36 #ifndef	lint
37 #if 0
38 FILE_RCSID("@(#)$File: file.c,v 1.184 2019/08/03 11:51:59 christos Exp $")
39 #else
40 __RCSID("$NetBSD: file.c,v 1.14 2019/12/17 02:31:05 christos Exp $");
41 #endif
42 #endif	/* lint */
43 
44 #include "magic.h"
45 
46 #include <stdlib.h>
47 #include <unistd.h>
48 #include <string.h>
49 #ifdef RESTORE_TIME
50 # if (__COHERENT__ >= 0x420)
51 #  include <sys/utime.h>
52 # else
53 #  ifdef USE_UTIMES
54 #   include <sys/time.h>
55 #  else
56 #   include <utime.h>
57 #  endif
58 # endif
59 #endif
60 #ifdef HAVE_UNISTD_H
61 #include <unistd.h>	/* for read() */
62 #endif
63 #ifdef HAVE_WCHAR_H
64 #include <wchar.h>
65 #endif
66 
67 #if defined(HAVE_GETOPT_H) && defined(HAVE_STRUCT_OPTION)
68 # include <getopt.h>
69 # ifndef HAVE_GETOPT_LONG
70 int getopt_long(int, char * const *, const char *,
71     const struct option *, int *);
72 # endif
73 # else
74 #  include "mygetopt.h"
75 #endif
76 
77 #ifdef S_IFLNK
78 # define IFLNK_h "h"
79 # define IFLNK_L "L"
80 #else
81 # define IFLNK_h ""
82 # define IFLNK_L ""
83 #endif
84 
85 #define FILE_FLAGS	"bcCdE" IFLNK_h "ik" IFLNK_L "lNnprsSvzZ0"
86 #define OPTSTRING	"bcCde:Ef:F:hiklLm:nNpP:rsSvzZ0"
87 
88 # define USAGE  \
89     "Usage: %s [-" FILE_FLAGS "] [--apple] [--extension] [--mime-encoding]\n" \
90     "            [--mime-type] [-e <testname>] [-F <separator>] " \
91     " [-f <namefile>]\n" \
92     "            [-m <magicfiles>] [-P <parameter=value>] <file> ...\n" \
93     "       %s -C [-m <magicfiles>]\n" \
94     "       %s [--help]\n"
95 
96 private int 		/* Global command-line options 		*/
97 	bflag = 0,	/* brief output format	 		*/
98 	nopad = 0,	/* Don't pad output			*/
99 	nobuffer = 0,   /* Do not buffer stdout 		*/
100 	nulsep = 0;	/* Append '\0' to the separator		*/
101 
102 private const char *separator = ":";	/* Default field separator	*/
103 private const struct option long_options[] = {
104 #define OPT_HELP		1
105 #define OPT_APPLE		2
106 #define OPT_EXTENSIONS		3
107 #define OPT_MIME_TYPE		4
108 #define OPT_MIME_ENCODING	5
109 #define OPT(shortname, longname, opt, def, doc)      \
110     {longname, opt, NULL, shortname},
111 #define OPT_LONGONLY(longname, opt, def, doc, id)        \
112     {longname, opt, NULL, id},
113 #include "file_opts.h"
114 #undef OPT
115 #undef OPT_LONGONLY
116     {0, 0, NULL, 0}
117     };
118 
119 private const struct {
120 	const char *name;
121 	int value;
122 } nv[] = {
123 	{ "apptype",	MAGIC_NO_CHECK_APPTYPE },
124 	{ "ascii",	MAGIC_NO_CHECK_ASCII },
125 	{ "cdf",	MAGIC_NO_CHECK_CDF },
126 	{ "compress",	MAGIC_NO_CHECK_COMPRESS },
127 	{ "csv",	MAGIC_NO_CHECK_CSV },
128 	{ "elf",	MAGIC_NO_CHECK_ELF },
129 	{ "encoding",	MAGIC_NO_CHECK_ENCODING },
130 	{ "soft",	MAGIC_NO_CHECK_SOFT },
131 	{ "tar",	MAGIC_NO_CHECK_TAR },
132 	{ "json",	MAGIC_NO_CHECK_JSON },
133 	{ "text",	MAGIC_NO_CHECK_TEXT },	/* synonym for ascii */
134 	{ "tokens",	MAGIC_NO_CHECK_TOKENS }, /* OBSOLETE: ignored for backwards compatibility */
135 };
136 
137 private struct {
138 	const char *name;
139 	int tag;
140 	size_t value;
141 	int set;
142 } pm[] = {
143 	{ "indir",	MAGIC_PARAM_INDIR_MAX, 0, 0 },
144 	{ "name",	MAGIC_PARAM_NAME_MAX, 0, 0 },
145 	{ "elf_phnum",	MAGIC_PARAM_ELF_PHNUM_MAX, 0, 0 },
146 	{ "elf_shnum",	MAGIC_PARAM_ELF_SHNUM_MAX, 0, 0 },
147 	{ "elf_notes",	MAGIC_PARAM_ELF_NOTES_MAX, 0, 0 },
148 	{ "regex",	MAGIC_PARAM_REGEX_MAX, 0, 0 },
149 	{ "bytes",	MAGIC_PARAM_BYTES_MAX, 0, 0 },
150 };
151 
152 private int posixly;
153 
154 #ifdef __dead
155 __dead
156 #endif
157 private void usage(void);
158 private void docprint(const char *, int);
159 #ifdef __dead
160 __dead
161 #endif
162 private void help(void);
163 
164 private int unwrap(struct magic_set *, const char *);
165 private int process(struct magic_set *ms, const char *, int);
166 private struct magic_set *load(const char *, int);
167 private void setparam(const char *);
168 private void applyparam(magic_t);
169 
170 
171 /*
172  * main - parse arguments and handle options
173  */
174 int
175 main(int argc, char *argv[])
176 {
177 	int c;
178 	size_t i;
179 	int action = 0, didsomefiles = 0, errflg = 0;
180 	int flags = 0, e = 0;
181 #ifdef HAVE_LIBSECCOMP
182 	int sandbox = 1;
183 #endif
184 	struct magic_set *magic = NULL;
185 	int longindex;
186 	const char *magicfile = NULL;		/* where the magic is	*/
187 	char *progname;
188 
189 	/* makes islower etc work for other langs */
190 	(void)setlocale(LC_CTYPE, "");
191 
192 #ifdef __EMX__
193 	/* sh-like wildcard expansion! Shouldn't hurt at least ... */
194 	_wildcard(&argc, &argv);
195 #endif
196 
197 	if ((progname = strrchr(argv[0], '/')) != NULL)
198 		progname++;
199 	else
200 		progname = argv[0];
201 
202 	file_setprogname(progname);
203 
204 
205 #ifdef S_IFLNK
206 	posixly = getenv("POSIXLY_CORRECT") != NULL;
207 	flags |=  posixly ? MAGIC_SYMLINK : 0;
208 #endif
209 	while ((c = getopt_long(argc, argv, OPTSTRING, long_options,
210 	    &longindex)) != -1)
211 		switch (c) {
212 		case OPT_HELP:
213 			help();
214 			break;
215 		case OPT_APPLE:
216 			flags |= MAGIC_APPLE;
217 			break;
218 		case OPT_EXTENSIONS:
219 			flags |= MAGIC_EXTENSION;
220 			break;
221 		case OPT_MIME_TYPE:
222 			flags |= MAGIC_MIME_TYPE;
223 			break;
224 		case OPT_MIME_ENCODING:
225 			flags |= MAGIC_MIME_ENCODING;
226 			break;
227 		case '0':
228 			nulsep++;
229 			break;
230 		case 'b':
231 			bflag++;
232 			break;
233 		case 'c':
234 			action = FILE_CHECK;
235 			break;
236 		case 'C':
237 			action = FILE_COMPILE;
238 			break;
239 		case 'd':
240 			flags |= MAGIC_DEBUG|MAGIC_CHECK;
241 			break;
242 		case 'E':
243 			flags |= MAGIC_ERROR;
244 			break;
245 		case 'e':
246 			for (i = 0; i < __arraycount(nv); i++)
247 				if (strcmp(nv[i].name, optarg) == 0)
248 					break;
249 
250 			if (i == __arraycount(nv))
251 				errflg++;
252 			else
253 				flags |= nv[i].value;
254 			break;
255 
256 		case 'f':
257 			if(action)
258 				usage();
259 			if (magic == NULL)
260 				if ((magic = load(magicfile, flags)) == NULL)
261 					return 1;
262 			applyparam(magic);
263 			e |= unwrap(magic, optarg);
264 			++didsomefiles;
265 			break;
266 		case 'F':
267 			separator = optarg;
268 			break;
269 		case 'i':
270 			flags |= MAGIC_MIME;
271 			break;
272 		case 'k':
273 			flags |= MAGIC_CONTINUE;
274 			break;
275 		case 'l':
276 			action = FILE_LIST;
277 			break;
278 		case 'm':
279 			magicfile = optarg;
280 			break;
281 		case 'n':
282 			++nobuffer;
283 			break;
284 		case 'N':
285 			++nopad;
286 			break;
287 #if defined(HAVE_UTIME) || defined(HAVE_UTIMES)
288 		case 'p':
289 			flags |= MAGIC_PRESERVE_ATIME;
290 			break;
291 #endif
292 		case 'P':
293 			setparam(optarg);
294 			break;
295 		case 'r':
296 			flags |= MAGIC_RAW;
297 			break;
298 		case 's':
299 			flags |= MAGIC_DEVICES;
300 			break;
301 		case 'S':
302 #ifdef HAVE_LIBSECCOMP
303 			sandbox = 0;
304 #endif
305 			break;
306 		case 'v':
307 			if (magicfile == NULL)
308 				magicfile = magic_getpath(magicfile, action);
309 			(void)fprintf(stdout, "%s-%s\n", file_getprogname(),
310 			    VERSION);
311 			(void)fprintf(stdout, "magic file from %s\n",
312 			    magicfile);
313 #ifdef HAVE_LIBSECCOMP
314 			(void)fprintf(stdout, "seccomp support included\n");
315 #endif
316 			return 0;
317 		case 'z':
318 			flags |= MAGIC_COMPRESS;
319 			break;
320 
321 		case 'Z':
322 			flags |= MAGIC_COMPRESS|MAGIC_COMPRESS_TRANSP;
323 			break;
324 #ifdef S_IFLNK
325 		case 'L':
326 			flags |= MAGIC_SYMLINK;
327 			break;
328 		case 'h':
329 			flags &= ~MAGIC_SYMLINK;
330 			break;
331 #endif
332 		case '?':
333 		default:
334 			errflg++;
335 			break;
336 		}
337 
338 	if (errflg) {
339 		usage();
340 	}
341 	if (e)
342 		return e;
343 
344 #ifdef HAVE_LIBSECCOMP
345 #if 0
346 	if (sandbox && enable_sandbox_basic() == -1)
347 #else
348 	if (sandbox && enable_sandbox_full() == -1)
349 #endif
350 		file_err(EXIT_FAILURE, "SECCOMP initialisation failed");
351 #endif /* HAVE_LIBSECCOMP */
352 
353 	if (MAGIC_VERSION != magic_version())
354 		file_warnx("Compiled magic version [%d] "
355 		    "does not match with shared library magic version [%d]\n",
356 		    MAGIC_VERSION, magic_version());
357 
358 	switch(action) {
359 	case FILE_CHECK:
360 	case FILE_COMPILE:
361 	case FILE_LIST:
362 		/*
363 		 * Don't try to check/compile ~/.magic unless we explicitly
364 		 * ask for it.
365 		 */
366 		magic = magic_open(flags|MAGIC_CHECK);
367 		if (magic == NULL) {
368 			file_warn("Can't create magic");
369 			return 1;
370 		}
371 
372 
373 		switch(action) {
374 		case FILE_CHECK:
375 			c = magic_check(magic, magicfile);
376 			break;
377 		case FILE_COMPILE:
378 			c = magic_compile(magic, magicfile);
379 			break;
380 		case FILE_LIST:
381 			c = magic_list(magic, magicfile);
382 			break;
383 		default:
384 			abort();
385 		}
386 		if (c == -1) {
387 			file_warnx("%s", magic_error(magic));
388 			e = 1;
389 			goto out;
390 		}
391 		goto out;
392 	default:
393 		if (magic == NULL)
394 			if ((magic = load(magicfile, flags)) == NULL)
395 				return 1;
396 		applyparam(magic);
397 	}
398 
399 	if (optind == argc) {
400 		if (!didsomefiles)
401 			usage();
402 	}
403 	else {
404 		size_t j, wid, nw;
405 		for (wid = 0, j = CAST(size_t, optind); j < CAST(size_t, argc);
406 		    j++) {
407 			nw = file_mbswidth(argv[j]);
408 			if (nw > wid)
409 				wid = nw;
410 		}
411 		/*
412 		 * If bflag is only set twice, set it depending on
413 		 * number of files [this is undocumented, and subject to change]
414 		 */
415 		if (bflag == 2) {
416 			bflag = optind >= argc - 1;
417 		}
418 		for (; optind < argc; optind++)
419 			e |= process(magic, argv[optind], wid);
420 	}
421 
422 out:
423 	if (magic)
424 		magic_close(magic);
425 	return e;
426 }
427 
428 private void
429 applyparam(magic_t magic)
430 {
431 	size_t i;
432 
433 	for (i = 0; i < __arraycount(pm); i++) {
434 		if (!pm[i].set)
435 			continue;
436 		if (magic_setparam(magic, pm[i].tag, &pm[i].value) == -1)
437 			file_err(EXIT_FAILURE, "Can't set %s", pm[i].name);
438 	}
439 }
440 
441 private void
442 setparam(const char *p)
443 {
444 	size_t i;
445 	char *s;
446 
447 	if ((s = strchr(p, '=')) == NULL)
448 		goto badparm;
449 
450 	for (i = 0; i < __arraycount(pm); i++) {
451 		if (strncmp(p, pm[i].name, s - p) != 0)
452 			continue;
453 		pm[i].value = atoi(s + 1);
454 		pm[i].set = 1;
455 		return;
456 	}
457 badparm:
458 	file_errx(EXIT_FAILURE, "Unknown param %s", p);
459 }
460 
461 private struct magic_set *
462 /*ARGSUSED*/
463 load(const char *magicfile, int flags)
464 {
465 	struct magic_set *magic = magic_open(flags);
466 	const char *e;
467 
468 	if (magic == NULL) {
469 		file_warn("Can't create magic");
470 		return NULL;
471 	}
472 	if (magic_load(magic, magicfile) == -1) {
473 		file_warn("%s", magic_error(magic));
474 		magic_close(magic);
475 		return NULL;
476 	}
477 	if ((e = magic_error(magic)) != NULL)
478 		file_warn("%s", e);
479 	return magic;
480 }
481 
482 /*
483  * unwrap -- read a file of filenames, do each one.
484  */
485 private int
486 unwrap(struct magic_set *ms, const char *fn)
487 {
488 	FILE *f;
489 	ssize_t len;
490 	char *line = NULL;
491 	size_t llen = 0;
492 	int wid = 0, cwid;
493 	int e = 0;
494 
495 	if (strcmp("-", fn) == 0) {
496 		f = stdin;
497 		wid = 1;
498 	} else {
499 		if ((f = fopen(fn, "r")) == NULL) {
500 			file_warn("Cannot open `%s'", fn);
501 			return 1;
502 		}
503 
504 		while ((len = getline(&line, &llen, f)) > 0) {
505 			if (line[len - 1] == '\n')
506 				line[len - 1] = '\0';
507 			cwid = file_mbswidth(line);
508 			if (cwid > wid)
509 				wid = cwid;
510 		}
511 
512 		rewind(f);
513 	}
514 
515 	while ((len = getline(&line, &llen, f)) > 0) {
516 		if (line[len - 1] == '\n')
517 			line[len - 1] = '\0';
518 		e |= process(ms, line, wid);
519 		if(nobuffer)
520 			(void)fflush(stdout);
521 	}
522 
523 	free(line);
524 	(void)fclose(f);
525 	return e;
526 }
527 
528 /*
529  * Called for each input file on the command line (or in a list of files)
530  */
531 private int
532 process(struct magic_set *ms, const char *inname, int wid)
533 {
534 	const char *type, c = nulsep > 1 ? '\0' : '\n';
535 	int std_in = strcmp(inname, "-") == 0;
536 
537 	if (wid > 0 && !bflag) {
538 		(void)printf("%s", std_in ? "/dev/stdin" : inname);
539 		if (nulsep)
540 			(void)putc('\0', stdout);
541 		if (nulsep < 2) {
542 			(void)printf("%s", separator);
543 			(void)printf("%*s ", CAST(int, nopad ? 0
544 			    : (wid - file_mbswidth(inname))), "");
545 		}
546 	}
547 
548 	type = magic_file(ms, std_in ? NULL : inname);
549 
550 	if (type == NULL) {
551 		(void)printf("ERROR: %s%c", magic_error(ms), c);
552 		return 1;
553 	} else {
554 		(void)printf("%s%c", type, c);
555 		return 0;
556 	}
557 }
558 
559 protected size_t
560 file_mbswidth(const char *s)
561 {
562 #if defined(HAVE_WCHAR_H) && defined(HAVE_MBRTOWC) && defined(HAVE_WCWIDTH)
563 	size_t bytesconsumed, old_n, n, width = 0;
564 	mbstate_t state;
565 	wchar_t nextchar;
566 	(void)memset(&state, 0, sizeof(mbstate_t));
567 	old_n = n = strlen(s);
568 
569 	while (n > 0) {
570 		bytesconsumed = mbrtowc(&nextchar, s, n, &state);
571 		if (bytesconsumed == CAST(size_t, -1) ||
572 		    bytesconsumed == CAST(size_t, -2)) {
573 			/* Something went wrong, return something reasonable */
574 			return old_n;
575 		}
576 		if (s[0] == '\n') {
577 			/*
578 			 * do what strlen() would do, so that caller
579 			 * is always right
580 			 */
581 			width++;
582 		} else {
583 			int w = wcwidth(nextchar);
584 			if (w > 0)
585 				width += w;
586 		}
587 
588 		s += bytesconsumed, n -= bytesconsumed;
589 	}
590 	return width;
591 #else
592 	return strlen(s);
593 #endif
594 }
595 
596 private void
597 usage(void)
598 {
599 	const char *pn = file_getprogname();
600 	(void)fprintf(stderr, USAGE, pn, pn, pn);
601 	exit(EXIT_FAILURE);
602 }
603 
604 private void
605 defprint(int def)
606 {
607 	if (!def)
608 		return;
609 	if (((def & 1) && posixly) || ((def & 2) && !posixly))
610 		fprintf(stdout, " (default)");
611 	fputc('\n', stdout);
612 }
613 
614 private void
615 docprint(const char *opts, int def)
616 {
617 	size_t i;
618 	int comma;
619 	char *sp, *p;
620 
621 	p = strstr(opts, "%o");
622 	if (p == NULL) {
623 		fprintf(stdout, "%s", opts);
624 		defprint(def);
625 		return;
626 	}
627 
628 	for (sp = p - 1; sp > opts && *sp == ' '; sp--)
629 		continue;
630 
631 	fprintf(stdout, "%.*s", CAST(int, p - opts), opts);
632 
633 	comma = 0;
634 	for (i = 0; i < __arraycount(nv); i++) {
635 		fprintf(stdout, "%s%s", comma++ ? ", " : "", nv[i].name);
636 		if (i && i % 5 == 0 && i != __arraycount(nv) - 1) {
637 			fprintf(stdout, ",\n%*s", CAST(int, p - sp - 1), "");
638 			comma = 0;
639 		}
640 	}
641 
642 	fprintf(stdout, "%s", opts + (p - opts) + 2);
643 }
644 
645 private void
646 help(void)
647 {
648 	(void)fputs(
649 "Usage: file [OPTION...] [FILE...]\n"
650 "Determine type of FILEs.\n"
651 "\n", stdout);
652 #define OPT(shortname, longname, opt, def, doc)      \
653 	fprintf(stdout, "  -%c, --" longname, shortname), \
654 	docprint(doc, def);
655 #define OPT_LONGONLY(longname, opt, def, doc, id)        \
656 	fprintf(stdout, "      --" longname),	\
657 	docprint(doc, def);
658 #include "file_opts.h"
659 #undef OPT
660 #undef OPT_LONGONLY
661 	fprintf(stdout, "\nReport bugs to https://bugs.astron.com/\n");
662 	exit(EXIT_SUCCESS);
663 }
664 
665 private const char *file_progname;
666 
667 protected void
668 file_setprogname(const char *progname)
669 {
670 	file_progname = progname;
671 }
672 
673 protected const char *
674 file_getprogname(void)
675 {
676 	return file_progname;
677 }
678 
679 protected void
680 file_err(int e, const char *fmt, ...)
681 {
682 	va_list ap;
683 	int se = errno;
684 
685 	va_start(ap, fmt);
686 	fprintf(stderr, "%s: ", file_progname);
687 	vfprintf(stderr, fmt, ap);
688 	va_end(ap);
689 	fprintf(stderr, " (%s)\n", strerror(se));
690 	exit(e);
691 }
692 
693 protected void
694 file_errx(int e, const char *fmt, ...)
695 {
696 	va_list ap;
697 
698 	va_start(ap, fmt);
699 	fprintf(stderr, "%s: ", file_progname);
700 	vfprintf(stderr, fmt, ap);
701 	va_end(ap);
702 	fprintf(stderr, "\n");
703 	exit(e);
704 }
705 
706 protected void
707 file_warn(const char *fmt, ...)
708 {
709 	va_list ap;
710 	int se = errno;
711 
712 	va_start(ap, fmt);
713 	fprintf(stderr, "%s: ", file_progname);
714 	vfprintf(stderr, fmt, ap);
715 	va_end(ap);
716 	fprintf(stderr, " (%s)\n", strerror(se));
717 	errno = se;
718 }
719 
720 protected void
721 file_warnx(const char *fmt, ...)
722 {
723 	va_list ap;
724 	int se = errno;
725 
726 	va_start(ap, fmt);
727 	fprintf(stderr, "%s: ", file_progname);
728 	vfprintf(stderr, fmt, ap);
729 	va_end(ap);
730 	fprintf(stderr, "\n");
731 	errno = se;
732 }
733