xref: /freebsd/contrib/libarchive/unzip/bsdunzip.c (revision 401026e4825a05abba6f945cf1b74b3328876fa2)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause
3  *
4  * Copyright (c) 2009, 2010 Joerg Sonnenberger <joerg@NetBSD.org>
5  * Copyright (c) 2007-2008 Dag-Erling Smørgrav
6  * All rights reserved.
7  */
8 
9 #include "bsdunzip_platform.h"
10 
11 #include "la_queue.h"
12 #include "la_getline.h"
13 #ifdef HAVE_SYS_STAT_H
14 #include <sys/stat.h>
15 #endif
16 
17 #ifdef HAVE_CTYPE_H
18 #include <ctype.h>
19 #endif
20 #ifdef HAVE_ERRNO_H
21 #include <errno.h>
22 #endif
23 #ifdef HAVE_FCNTL_H
24 #include <fcntl.h>
25 #endif
26 #ifdef HAVE_FNMATCH_H
27 #include <fnmatch.h>
28 #endif
29 #ifdef HAVE_LOCALE_H
30 #include <locale.h>
31 #endif
32 #ifdef HAVE_SIGNAL_H
33 #include <signal.h>
34 #endif
35 #ifdef HAVE_STDARG_H
36 #include <stdarg.h>
37 #endif
38 #include <stdio.h>
39 #ifdef HAVE_STDLIB_H
40 #include <stdlib.h>
41 #endif
42 #ifdef HAVE_STRING_H
43 #include <string.h>
44 #endif
45 #ifdef HAVE_UNISTD_H
46 #include <unistd.h>
47 #endif
48 #if ((!defined(HAVE_UTIMENSAT) && defined(HAVE_LUTIMES)) || \
49     (!defined(HAVE_FUTIMENS) && defined(HAVE_FUTIMES)))
50 #ifdef HAVE_SYS_TIME_H
51 #include <sys/time.h>
52 #endif
53 #endif
54 #ifdef HAVE_GETOPT_OPTRESET
55 #include <getopt.h>
56 #endif
57 
58 #include "bsdunzip.h"
59 #include "passphrase.h"
60 #include "lafe_err.h"
61 
62 /* command-line options */
63 static int		 a_opt;		/* convert EOL */
64 static int		 C_opt;		/* match case-insensitively */
65 static int		 c_opt;		/* extract to stdout */
66 static const char	*d_arg;		/* directory */
67 static int		 f_opt;		/* update existing files only */
68 static const char	*O_arg;		/* encoding */
69 static int		 j_opt;		/* junk directories */
70 static int		 L_opt;		/* lowercase names */
71 static int		 n_opt;		/* never overwrite */
72 static int		 o_opt;		/* always overwrite */
73 static int		 p_opt;		/* extract to stdout, quiet */
74 static const char	*P_arg;		/* passphrase */
75 static int		 q_opt;		/* quiet */
76 static int		 t_opt;		/* test */
77 static int		 u_opt;		/* update */
78 static int		 v_opt;		/* verbose/list */
79 static const char	*y_str = "";	/* 4 digit year */
80 static int		 Z1_opt;	/* zipinfo mode list files only */
81 static int		 version_opt;	/* version string */
82 
83 /* debug flag */
84 static int		 unzip_debug;
85 
86 /* zipinfo mode */
87 static int		 zipinfo_mode;
88 
89 /* running on tty? */
90 static int		 tty;
91 
92 /* processing exclude list */
93 static int		 unzip_exclude_mode = 0;
94 
95 int bsdunzip_optind;
96 
97 /* convenience macro */
98 /* XXX should differentiate between ARCHIVE_{WARN,FAIL,RETRY} */
99 #define ac(call)						\
100 	do {							\
101 		int acret = (call);				\
102 		if (acret != ARCHIVE_OK)			\
103 			errorx("%s", archive_error_string(a));	\
104 	} while (0)
105 
106 /*
107  * Indicates that last info() did not end with EOL.  This helps error() et
108  * al. avoid printing an error message on the same line as an incomplete
109  * informational message.
110  */
111 static int noeol;
112 
113 /* for an interactive passphrase input */
114 static char *passphrase_buf;
115 
116 /* fatal error message + errno */
117 static void __LA_NORETURN
error(const char * fmt,...)118 error(const char *fmt, ...)
119 {
120 	va_list ap;
121 
122 	if (noeol)
123 		fprintf(stdout, "\n");
124 	fflush(stdout);
125 	fprintf(stderr, "unzip: ");
126 	va_start(ap, fmt);
127 	vfprintf(stderr, fmt, ap);
128 	va_end(ap);
129 	fprintf(stderr, ": %s\n", strerror(errno));
130 	exit(EXIT_FAILURE);
131 }
132 
133 /* fatal error message, no errno */
134 static void __LA_NORETURN
errorx(const char * fmt,...)135 errorx(const char *fmt, ...)
136 {
137 	va_list ap;
138 
139 	if (noeol)
140 		fprintf(stdout, "\n");
141 	fflush(stdout);
142 	fprintf(stderr, "unzip: ");
143 	va_start(ap, fmt);
144 	vfprintf(stderr, fmt, ap);
145 	va_end(ap);
146 	fprintf(stderr, "\n");
147 	exit(EXIT_FAILURE);
148 }
149 
150 /* non-fatal error message + errno */
151 static void
warning(const char * fmt,...)152 warning(const char *fmt, ...)
153 {
154 	va_list ap;
155 
156 	if (noeol)
157 		fprintf(stdout, "\n");
158 	fflush(stdout);
159 	fprintf(stderr, "unzip: ");
160 	va_start(ap, fmt);
161 	vfprintf(stderr, fmt, ap);
162 	va_end(ap);
163 	fprintf(stderr, ": %s\n", strerror(errno));
164 }
165 
166 /* non-fatal error message, no errno */
167 static void
warningx(const char * fmt,...)168 warningx(const char *fmt, ...)
169 {
170 	va_list ap;
171 
172 	if (noeol)
173 		fprintf(stdout, "\n");
174 	fflush(stdout);
175 	fprintf(stderr, "unzip: ");
176 	va_start(ap, fmt);
177 	vfprintf(stderr, fmt, ap);
178 	va_end(ap);
179 	fprintf(stderr, "\n");
180 }
181 
182 /* informational message (if not -q) */
183 static void
info(const char * fmt,...)184 info(const char *fmt, ...)
185 {
186 	va_list ap;
187 
188 	if (q_opt && !unzip_debug)
189 		return;
190 	va_start(ap, fmt);
191 	vfprintf(stdout, fmt, ap);
192 	va_end(ap);
193 	fflush(stdout);
194 
195 	if (*fmt == '\0')
196 		noeol = 1;
197 	else
198 		noeol = fmt[strlen(fmt) - 1] != '\n';
199 }
200 
201 /* debug message (if unzip_debug) */
202 static void
debug(const char * fmt,...)203 debug(const char *fmt, ...)
204 {
205 	va_list ap;
206 
207 	if (!unzip_debug)
208 		return;
209 	va_start(ap, fmt);
210 	vfprintf(stderr, fmt, ap);
211 	va_end(ap);
212 	fflush(stderr);
213 
214 	if (*fmt == '\0')
215 		noeol = 1;
216 	else
217 		noeol = fmt[strlen(fmt) - 1] != '\n';
218 }
219 
220 /* duplicate a path name, possibly converting to lower case */
221 static char *
pathdup(const char * path)222 pathdup(const char *path)
223 {
224 	char *str;
225 	size_t i, len;
226 
227 	if (path == NULL || path[0] == '\0')
228 		return (NULL);
229 
230 	len = strlen(path);
231 	while (len && path[len - 1] == '/')
232 		len--;
233 	if ((str = malloc(len + 1)) == NULL) {
234 		errno = ENOMEM;
235 		error("malloc()");
236 	}
237 	if (L_opt) {
238 		for (i = 0; i < len; ++i)
239 			str[i] = (char)tolower((unsigned char)path[i]);
240 	} else {
241 		memcpy(str, path, len);
242 	}
243 	str[len] = '\0';
244 
245 	return (str);
246 }
247 
248 /* concatenate two path names */
249 static char *
pathcat(const char * prefix,const char * path)250 pathcat(const char *prefix, const char *path)
251 {
252 	char *str;
253 	size_t prelen, len;
254 
255 	prelen = prefix ? strlen(prefix) + 1 : 0;
256 	len = strlen(path) + 1;
257 	if ((str = malloc(prelen + len)) == NULL) {
258 		errno = ENOMEM;
259 		error("malloc()");
260 	}
261 	if (prefix) {
262 		memcpy(str, prefix, prelen);	/* includes zero */
263 		str[prelen - 1] = '/';		/* splat zero */
264 	}
265 	memcpy(str + prelen, path, len);	/* includes zero */
266 
267 	return (str);
268 }
269 
270 /*
271  * Pattern lists for include / exclude processing
272  */
273 struct pattern {
274 	STAILQ_ENTRY(pattern) link;
275 	char pattern[];
276 };
277 
278 STAILQ_HEAD(pattern_list, pattern);
279 static struct pattern_list include = STAILQ_HEAD_INITIALIZER(include);
280 static struct pattern_list exclude = STAILQ_HEAD_INITIALIZER(exclude);
281 
282 /*
283  * Add an entry to a pattern list
284  */
285 static void
add_pattern(struct pattern_list * list,const char * pattern)286 add_pattern(struct pattern_list *list, const char *pattern)
287 {
288 	struct pattern *entry;
289 	size_t len;
290 
291 	debug("adding pattern '%s'\n", pattern);
292 	len = strlen(pattern);
293 	if ((entry = malloc(sizeof *entry + len + 1)) == NULL) {
294 		errno = ENOMEM;
295 		error("malloc()");
296 	}
297 	memcpy(entry->pattern, pattern, len + 1);
298 	STAILQ_INSERT_TAIL(list, entry, link);
299 }
300 
301 /*
302  * Match a string against a list of patterns
303  */
304 static int
match_pattern(struct pattern_list * list,const char * str)305 match_pattern(struct pattern_list *list, const char *str)
306 {
307 	struct pattern *entry;
308 
309 	STAILQ_FOREACH(entry, list, link) {
310 #ifdef HAVE_FNMATCH
311 		if (fnmatch(entry->pattern, str, C_opt ? FNM_CASEFOLD : 0) == 0)
312 			return (1);
313 #else
314 #error "Unsupported platform: fnmatch() is required"
315 #endif
316 	}
317 	return (0);
318 }
319 
320 /*
321  * Verify that a given pathname is in the include list and not in the
322  * exclude list.
323  */
324 static int
accept_pathname(const char * pathname)325 accept_pathname(const char *pathname)
326 {
327 
328 	if (!STAILQ_EMPTY(&include) && !match_pattern(&include, pathname))
329 		return (0);
330 	if (!STAILQ_EMPTY(&exclude) && match_pattern(&exclude, pathname))
331 		return (0);
332 	return (1);
333 }
334 
335 /*
336  * Create the specified directory with the specified mode, taking certain
337  * precautions on they way.
338  */
339 static void
make_dir(const char * path,int mode)340 make_dir(const char *path, int mode)
341 {
342 	struct stat sb;
343 
344 	if (lstat(path, &sb) == 0) {
345 		if (S_ISDIR(sb.st_mode))
346 			return;
347 		/*
348 		 * Normally, we should either ask the user about removing
349 		 * the non-directory of the same name as a directory we
350 		 * wish to create, or respect the -n or -o command-line
351 		 * options.  However, this may lead to a later failure or
352 		 * even compromise (if this non-directory happens to be a
353 		 * symlink to somewhere unsafe), so we don't.
354 		 */
355 
356 		/*
357 		 * Don't check unlink() result; failure will cause mkdir()
358 		 * to fail later, which we will catch.
359 		 */
360 		(void)unlink(path);
361 	}
362 	if (mkdir(path, (mode_t)mode) != 0 && errno != EEXIST)
363 		error("mkdir('%s')", path);
364 }
365 
366 /*
367  * Ensure that all directories leading up to (but not including) the
368  * specified path exist.
369  *
370  * XXX inefficient + modifies the file in-place
371  */
372 static void
make_parent(char * path)373 make_parent(char *path)
374 {
375 	struct stat sb;
376 	char *sep;
377 
378 	sep = strrchr(path, '/');
379 	if (sep == NULL || sep == path)
380 		return;
381 	*sep = '\0';
382 	if (lstat(path, &sb) == 0) {
383 		if (S_ISDIR(sb.st_mode)) {
384 			*sep = '/';
385 			return;
386 		}
387 		unlink(path);
388 	}
389 	make_parent(path);
390 	mkdir(path, 0755);
391 	*sep = '/';
392 
393 #if 0
394 	for (sep = path; (sep = strchr(sep, '/')) != NULL; sep++) {
395 		/* root in case of absolute d_arg */
396 		if (sep == path)
397 			continue;
398 		*sep = '\0';
399 		make_dir(path, 0755);
400 		*sep = '/';
401 	}
402 #endif
403 }
404 
405 /*
406  * Extract a directory.
407  */
408 static void
extract_dir(struct archive * a,struct archive_entry * e,const char * path)409 extract_dir(struct archive *a, struct archive_entry *e, const char *path)
410 {
411 	int mode;
412 
413 	/*
414 	 * Dropbox likes to create '/' directory entries, just ignore
415 	 * such junk.
416 	 */
417 	if (*path == '\0')
418 		return;
419 
420 	mode = archive_entry_mode(e) & 0777;
421 	if (mode == 0)
422 		mode = 0755;
423 
424 	/*
425 	 * Some zipfiles contain directories with weird permissions such
426 	 * as 0644 or 0444.  This can cause strange issues such as being
427 	 * unable to extract files into the directory we just created, or
428 	 * the user being unable to remove the directory later without
429 	 * first manually changing its permissions.  Therefore, we whack
430 	 * the permissions into shape, assuming that the user wants full
431 	 * access and that anyone who gets read access also gets execute
432 	 * access.
433 	 */
434 	mode |= 0700;
435 	if (mode & 0040)
436 		mode |= 0010;
437 	if (mode & 0004)
438 		mode |= 0001;
439 
440 	info("   creating: %s/\n", path);
441 	make_dir(path, mode);
442 	ac(archive_read_data_skip(a));
443 }
444 
445 static unsigned char buffer[8192];
446 static char spinner[] = { '|', '/', '-', '\\' };
447 
448 static int
handle_existing_file(char ** path)449 handle_existing_file(char **path)
450 {
451 	size_t alen;
452 	ssize_t len;
453 	char buf[4];
454 
455 	for (;;) {
456 		fprintf(stderr,
457 		    "replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ",
458 		    *path);
459 		if (fgets(buf, sizeof(buf), stdin) == NULL)
460 			goto stdin_err;
461 		switch (*buf) {
462 		case 'A':
463 			o_opt = 1;
464 			/* FALLTHROUGH */
465 		case 'y':
466 		case 'Y':
467 			(void)unlink(*path);
468 			return 1;
469 		case 'N':
470 			n_opt = 1;
471 			/* FALLTHROUGH */
472 		case 'n':
473 			return -1;
474 		case 'r':
475 		case 'R':
476 			printf("New name: ");
477 			fflush(stdout);
478 			free(*path);
479 			*path = NULL;
480 			alen = 0;
481 			len = getline(path, &alen, stdin);
482 			if (len < 1)
483 				goto stdin_err;
484 			if ((*path)[len - 1] == '\n')
485 				(*path)[len - 1] = '\0';
486 			return 0;
487 		default:
488 			break;
489 		}
490 	}
491 stdin_err:
492 	clearerr(stdin);
493 	printf("NULL\n(EOF or read error, "
494 		"treating as \"[N]one\"...)\n");
495 	n_opt = 1;
496 	return -1;
497 }
498 
499 /*
500  * Detect binary files by a combination of character white list and
501  * black list. NUL bytes and other control codes without use in text files
502  * result directly in switching the file to binary mode. Otherwise, at least
503  * one white-listed byte has to be found.
504  *
505  * Black-listed: 0..6, 14..25, 28..31
506  * 0xf3ffc07f = 11110011111111111100000001111111b
507  * White-listed: 9..10, 13, >= 32
508  * 0x00002600 = 00000000000000000010011000000000b
509  *
510  * See the proginfo/txtvsbin.txt in the zip sources for a detailed discussion.
511  */
512 #define BYTE_IS_BINARY(x)	((x) < 32 && (0xf3ffc07fU & (1U << (x))))
513 #define	BYTE_IS_TEXT(x)		((x) >= 32 || (0x00002600U & (1U << (x))))
514 
515 static int
check_binary(const unsigned char * buf,size_t len)516 check_binary(const unsigned char *buf, size_t len)
517 {
518 	int rv;
519 	for (rv = 1; len--; ++buf) {
520 		if (BYTE_IS_BINARY(*buf))
521 			return 1;
522 		if (BYTE_IS_TEXT(*buf))
523 			rv = 0;
524 	}
525 
526 	return rv;
527 }
528 
529 /*
530  * Extract to a file descriptor
531  */
532 static int
extract2fd(struct archive * a,char * pathname,int fd)533 extract2fd(struct archive *a, char *pathname, int fd)
534 {
535 	int cr, text, warn;
536 	ssize_t len;
537 	unsigned char *p, *q, *end;
538 
539 	text = a_opt;
540 	warn = 0;
541 	cr = 0;
542 
543 	/* loop over file contents and write to fd */
544 	for (int n = 0; ; n++) {
545 		if (fd != STDOUT_FILENO)
546 			if (tty && (n % 4) == 0)
547 				info(" %c\b\b", spinner[(n / 4) % sizeof spinner]);
548 
549 		len = archive_read_data(a, buffer, sizeof buffer);
550 
551 		if (len < 0)
552 			ac(len);
553 
554 		/* left over CR from previous buffer */
555 		if (a_opt && cr) {
556 			if (len == 0 || buffer[0] != '\n')
557 				if (write(fd, "\r", 1) != 1)
558 					error("write('%s')", pathname);
559 			cr = 0;
560 		}
561 
562 		/* EOF */
563 		if (len == 0)
564 			break;
565 		end = buffer + len;
566 
567 		/*
568 		 * Detect whether this is a text file.  The correct way to
569 		 * do this is to check the least significant bit of the
570 		 * "internal file attributes" field of the corresponding
571 		 * file header in the central directory, but libarchive
572 		 * does not provide access to this field, so we have to
573 		 * guess by looking for non-ASCII characters in the
574 		 * buffer.  Hopefully we won't guess wrong.  If we do
575 		 * guess wrong, we print a warning message later.
576 		 */
577 		if (a_opt && n == 0) {
578 			if (check_binary(buffer, len))
579 				text = 0;
580 		}
581 
582 		/* simple case */
583 		if (!a_opt || !text) {
584 			if (write(fd, buffer, len) != len)
585 				error("write('%s')", pathname);
586 			continue;
587 		}
588 
589 		/* hard case: convert \r\n to \n (sigh...) */
590 		for (p = buffer; p < end; p = q + 1) {
591 			for (q = p; q < end; q++) {
592 				if (!warn && BYTE_IS_BINARY(*q)) {
593 					warningx("%s may be corrupted due"
594 					    " to weak text file detection"
595 					    " heuristic", pathname);
596 					warn = 1;
597 				}
598 				if (q[0] != '\r')
599 					continue;
600 				if (&q[1] == end) {
601 					cr = 1;
602 					break;
603 				}
604 				if (q[1] == '\n')
605 					break;
606 			}
607 			if (write(fd, p, q - p) != q - p)
608 				error("write('%s')", pathname);
609 		}
610 	}
611 
612 	return text;
613 }
614 
615 /*
616  * Extract a regular file.
617  */
618 static void
extract_file(struct archive * a,struct archive_entry * e,char ** path)619 extract_file(struct archive *a, struct archive_entry *e, char **path)
620 {
621 	int mode;
622 	struct timespec mtime;
623 	struct stat sb;
624 	int fd, check, text;
625 	const char *linkname;
626 #if defined(HAVE_UTIMENSAT) || defined(HAVE_FUTIMENS)
627 	struct timespec ts[2];
628 #endif
629 #if ((!defined(HAVE_UTIMENSAT) && defined(HAVE_LUTIMES)) || \
630     (!defined(HAVE_FUTIMENS) && defined(HAVE_FUTIMES)))
631 	struct timeval times[2];
632 #endif
633 
634 	mode = archive_entry_mode(e) & 0777;
635 	if (mode == 0)
636 		mode = 0644;
637 	mtime.tv_sec = archive_entry_mtime(e);
638 	mtime.tv_nsec = archive_entry_mtime_nsec(e);
639 
640 	/* look for existing file of same name */
641 recheck:
642 	if (lstat(*path, &sb) == 0) {
643 		if (u_opt || f_opt) {
644 			/* check if up-to-date */
645 			if (S_ISREG(sb.st_mode) && (
646 #if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC
647 			    sb.st_mtimespec.tv_sec > mtime.tv_sec ||
648 			    (sb.st_mtimespec.tv_sec == mtime.tv_sec &&
649 			    sb.st_mtimespec.tv_nsec >= mtime.tv_nsec)
650 #elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC
651 			    sb.st_mtim.tv_sec > mtime.tv_sec ||
652 			    (sb.st_mtim.tv_sec == mtime.tv_sec &&
653 			    sb.st_mtim.tv_nsec >= mtime.tv_nsec)
654 #elif HAVE_STRUCT_STAT_ST_MTIME_N
655 			    sb.st_mtime > mtime.tv_sec ||
656 			    (sb.st_mtime == mtime.tv_sec &&
657 			    sb.st_mtime_n => mtime.tv_nsec)
658 #elif HAVE_STRUCT_STAT_ST_MTIME_USEC
659 			    sb.st_mtime > mtime.tv_sec ||
660 			    (sb.st_mtime == mtime.tv_sec &&
661 			    sb.st_mtime_usec => mtime.tv_nsec / 1000)
662 #else
663 			    sb.st_mtime > mtime.tv_sec
664 #endif
665 			    ))
666 				return;
667 			(void)unlink(*path);
668 		} else if (o_opt) {
669 			/* overwrite */
670 			(void)unlink(*path);
671 		} else if (n_opt) {
672 			/* do not overwrite */
673 			return;
674 		} else {
675 			check = handle_existing_file(path);
676 			if (check == 0)
677 				goto recheck;
678 			if (check == -1)
679 				return; /* do not overwrite */
680 		}
681 	} else {
682 		if (f_opt)
683 			return;
684 	}
685 
686 #if defined(HAVE_UTIMENSAT) || defined(HAVE_FUTIMENS)
687 	ts[0].tv_sec = 0;
688 	ts[0].tv_nsec = UTIME_NOW;
689 	ts[1] = mtime;
690 #endif
691 #if ((!defined(HAVE_UTIMENSAT) && defined(HAVE_LUTIMES)) || \
692     (!defined(HAVE_FUTIMENS) && defined(HAVE_FUTIMES)))
693 	times[0].tv_sec = 0;
694 	times[0].tv_usec = -1;
695 	times[1].tv_sec = mtime.tv_sec;
696 	times[1].tv_usec = mtime.tv_nsec / 1000;
697 #endif
698 
699 	/* process symlinks */
700 	linkname = archive_entry_symlink(e);
701 	if (linkname != NULL) {
702 		if (symlink(linkname, *path) != 0)
703 			error("symlink('%s')", *path);
704 		info(" extracting: %s -> %s\n", *path, linkname);
705 #ifdef HAVE_LCHMOD
706 		if (lchmod(*path, (mode_t)mode) != 0)
707 			warning("Cannot set mode for '%s'", *path);
708 #endif
709 		/* set access and modification time */
710 #if defined(HAVE_UTIMENSAT)
711 		if (utimensat(AT_FDCWD, *path, ts, AT_SYMLINK_NOFOLLOW) != 0)
712 			warning("utimensat('%s')", *path);
713 #elif defined(HAVE_LUTIMES)
714 		gettimeofday(&times[0], NULL);
715 		if (lutimes(*path, times) != 0)
716 			warning("lutimes('%s')", *path);
717 #endif
718 		return;
719 	}
720 
721 	if ((fd = open(*path, O_RDWR|O_CREAT|O_TRUNC, mode)) < 0)
722 		error("open('%s')", *path);
723 
724 	info(" extracting: %s", *path);
725 
726 	text = extract2fd(a, *path, fd);
727 
728 	if (tty)
729 		info("  \b\b");
730 	if (text)
731 		info(" (text)");
732 	info("\n");
733 
734 	/* set access and modification time */
735 #if defined(HAVE_FUTIMENS)
736 	if (futimens(fd, ts) != 0)
737 		error("futimens('%s')", *path);
738 #elif defined(HAVE_FUTIMES)
739 	gettimeofday(&times[0], NULL);
740 	if (futimes(fd, times) != 0)
741 		error("futimes('%s')", *path);
742 #endif
743 	if (close(fd) != 0)
744 		error("close('%s')", *path);
745 }
746 
747 /*
748  * Extract a zipfile entry: first perform some sanity checks to ensure
749  * that it is either a directory or a regular file and that the path is
750  * not absolute and does not try to break out of the current directory;
751  * then call either extract_dir() or extract_file() as appropriate.
752  *
753  * This is complicated a bit by the various ways in which we need to
754  * manipulate the path name.  Case conversion (if requested by the -L
755  * option) happens first, but the include / exclude patterns are applied
756  * to the full converted path name, before the directory part of the path
757  * is removed in accordance with the -j option.  Sanity checks are
758  * intentionally done earlier than they need to be, so the user will get a
759  * warning about insecure paths even for files or directories which
760  * wouldn't be extracted anyway.
761  */
762 static void
extract(struct archive * a,struct archive_entry * e)763 extract(struct archive *a, struct archive_entry *e)
764 {
765 	char *pathname, *realpathname;
766 	mode_t filetype;
767 	char *p, *q;
768 
769 	if ((pathname = pathdup(archive_entry_pathname(e))) == NULL) {
770 		warningx("skipping empty or unreadable filename entry");
771 		ac(archive_read_data_skip(a));
772 		return;
773 	}
774 	filetype = archive_entry_filetype(e);
775 
776 	/* sanity checks */
777 	if (pathname[0] == '/' ||
778 	    strncmp(pathname, "../", 3) == 0 ||
779 	    strstr(pathname, "/../") != NULL) {
780 		warningx("skipping insecure entry '%s'", pathname);
781 		ac(archive_read_data_skip(a));
782 		free(pathname);
783 		return;
784 	}
785 
786 	/* I don't think this can happen in a zipfile.. */
787 	if (!S_ISDIR(filetype) && !S_ISREG(filetype) && !S_ISLNK(filetype)) {
788 		warningx("skipping non-regular entry '%s'", pathname);
789 		ac(archive_read_data_skip(a));
790 		free(pathname);
791 		return;
792 	}
793 
794 	/* skip directories in -j case */
795 	if (S_ISDIR(filetype) && j_opt) {
796 		ac(archive_read_data_skip(a));
797 		free(pathname);
798 		return;
799 	}
800 
801 	/* apply include / exclude patterns */
802 	if (!accept_pathname(pathname)) {
803 		ac(archive_read_data_skip(a));
804 		free(pathname);
805 		return;
806 	}
807 
808 	/* apply -j and -d */
809 	if (j_opt) {
810 		for (p = q = pathname; *p; ++p)
811 			if (*p == '/')
812 				q = p + 1;
813 		realpathname = pathcat(d_arg, q);
814 	} else {
815 		realpathname = pathcat(d_arg, pathname);
816 	}
817 
818 	/* ensure that parent directory exists */
819 	make_parent(realpathname);
820 
821 	if (S_ISDIR(filetype))
822 		extract_dir(a, e, realpathname);
823 	else
824 		extract_file(a, e, &realpathname);
825 
826 	free(realpathname);
827 	free(pathname);
828 }
829 
830 static void
extract_stdout(struct archive * a,struct archive_entry * e)831 extract_stdout(struct archive *a, struct archive_entry *e)
832 {
833 	char *pathname;
834 	mode_t filetype;
835 
836 	if ((pathname = pathdup(archive_entry_pathname(e))) == NULL) {
837 		warningx("skipping empty or unreadable filename entry");
838 		ac(archive_read_data_skip(a));
839 		return;
840 	}
841 	filetype = archive_entry_filetype(e);
842 
843 	/* I don't think this can happen in a zipfile.. */
844 	if (!S_ISDIR(filetype) && !S_ISREG(filetype) && !S_ISLNK(filetype)) {
845 		warningx("skipping non-regular entry '%s'", pathname);
846 		ac(archive_read_data_skip(a));
847 		free(pathname);
848 		return;
849 	}
850 
851 	/* skip directories in -j case */
852 	if (S_ISDIR(filetype)) {
853 		ac(archive_read_data_skip(a));
854 		free(pathname);
855 		return;
856 	}
857 
858 	/* apply include / exclude patterns */
859 	if (!accept_pathname(pathname)) {
860 		ac(archive_read_data_skip(a));
861 		free(pathname);
862 		return;
863 	}
864 
865 	if (c_opt)
866 		info("x %s\n", pathname);
867 
868 	(void)extract2fd(a, pathname, STDOUT_FILENO);
869 
870 	free(pathname);
871 }
872 
873 /*
874  * Print the name of an entry to stdout.
875  */
876 static void
list(struct archive * a,struct archive_entry * e)877 list(struct archive *a, struct archive_entry *e)
878 {
879 	char buf[20];
880 	time_t mtime;
881 	struct tm *tm;
882 	const char *pathname;
883 
884 	mtime = archive_entry_mtime(e);
885 	tm = localtime(&mtime);
886 	if (*y_str)
887 		strftime(buf, sizeof(buf), "%m-%d-%G %R", tm);
888 	else
889 		strftime(buf, sizeof(buf), "%m-%d-%g %R", tm);
890 
891 	pathname = archive_entry_pathname(e);
892 	if (!pathname)
893 		pathname = "";
894 	if (!zipinfo_mode) {
895 		if (v_opt == 1) {
896 			printf(" %8ju  %s   %s\n",
897 			    (uintmax_t)archive_entry_size(e),
898 			    buf, pathname);
899 		} else if (v_opt == 2) {
900 			printf("%8ju  Stored  %7ju   0%%  %s  %08x  %s\n",
901 			    (uintmax_t)archive_entry_size(e),
902 			    (uintmax_t)archive_entry_size(e),
903 			    buf,
904 			    0U,
905 			    pathname);
906 		}
907 	} else {
908 		if (Z1_opt)
909 			printf("%s\n", pathname);
910 	}
911 	ac(archive_read_data_skip(a));
912 }
913 
914 /*
915  * Extract to memory to check CRC
916  */
917 static int
test(struct archive * a,struct archive_entry * e)918 test(struct archive *a, struct archive_entry *e)
919 {
920 	ssize_t len;
921 	int error_count;
922 
923 	error_count = 0;
924 	if (S_ISDIR(archive_entry_filetype(e)))
925 		return 0;
926 
927 	info("    testing: %s\t", archive_entry_pathname(e));
928 	while ((len = archive_read_data(a, buffer, sizeof buffer)) > 0)
929 		/* nothing */;
930 	if (len < 0) {
931 		info(" %s\n", archive_error_string(a));
932 		++error_count;
933 	} else {
934 		info(" OK\n");
935 	}
936 
937 	/* shouldn't be necessary, but it doesn't hurt */
938 	ac(archive_read_data_skip(a));
939 
940 	return error_count;
941 }
942 
943 /*
944  * Callback function for reading passphrase.
945  * Originally from cpio.c and passphrase.c, libarchive.
946  */
947 #define PPBUFF_SIZE 1024
948 static const char *
passphrase_callback(struct archive * a,void * _client_data)949 passphrase_callback(struct archive *a, void *_client_data)
950 {
951 	char *p;
952 
953 	(void)a; /* UNUSED */
954 	(void)_client_data; /* UNUSED */
955 
956 	if (passphrase_buf == NULL) {
957 		passphrase_buf = malloc(PPBUFF_SIZE);
958 		if (passphrase_buf == NULL) {
959 			errno = ENOMEM;
960 			error("malloc()");
961 		}
962 	}
963 
964 	p = lafe_readpassphrase("\nEnter password: ", passphrase_buf,
965 		PPBUFF_SIZE);
966 
967 	if (p == NULL && errno != EINTR)
968 		error("Error reading password");
969 
970 	return p;
971 }
972 
973 /*
974  * Main loop: open the zipfile, iterate over its contents and decide what
975  * to do with each entry.
976  */
977 static void
unzip(const char * fn)978 unzip(const char *fn)
979 {
980 	struct archive *a;
981 	struct archive_entry *e;
982 	int ret;
983 	uintmax_t total_size, file_count, error_count;
984 
985 	if ((a = archive_read_new()) == NULL)
986 		error("archive_read_new failed");
987 
988 	ac(archive_read_support_format_zip(a));
989 
990 	if (O_arg)
991 		ac(archive_read_set_format_option(a, "zip", "hdrcharset", O_arg));
992 
993 	if (P_arg)
994 		archive_read_add_passphrase(a, P_arg);
995 	else
996 		archive_read_set_passphrase_callback(a, NULL,
997 			&passphrase_callback);
998 
999 	ac(archive_read_open_filename(a, fn, 8192));
1000 
1001 	if (!zipinfo_mode) {
1002 		if (!p_opt && !q_opt)
1003 			printf("Archive:  %s\n", fn);
1004 		if (v_opt == 1) {
1005 			printf("  Length     %sDate   Time    Name\n", y_str);
1006 			printf(" --------    %s----   ----    ----\n", y_str);
1007 		} else if (v_opt == 2) {
1008 			printf(" Length   Method    Size  Ratio   %sDate   Time   CRC-32    Name\n", y_str);
1009 			printf("--------  ------  ------- -----   %s----   ----   ------    ----\n", y_str);
1010 		}
1011 	}
1012 
1013 	total_size = 0;
1014 	file_count = 0;
1015 	error_count = 0;
1016 	for (;;) {
1017 		ret = archive_read_next_header(a, &e);
1018 		if (ret == ARCHIVE_EOF)
1019 			break;
1020 		ac(ret);
1021 		if (!zipinfo_mode) {
1022 			if (t_opt)
1023 				error_count += test(a, e);
1024 			else if (v_opt)
1025 				list(a, e);
1026 			else if (p_opt || c_opt)
1027 				extract_stdout(a, e);
1028 			else
1029 				extract(a, e);
1030 		} else {
1031 			if (Z1_opt)
1032 				list(a, e);
1033 		}
1034 
1035 		total_size += archive_entry_size(e);
1036 		++file_count;
1037 	}
1038 
1039 	if (zipinfo_mode) {
1040 		if (v_opt == 1) {
1041 			printf(" --------                   %s-------\n", y_str);
1042 			printf(" %8ju                   %s%ju file%s\n",
1043 			    total_size, y_str, file_count, file_count != 1 ? "s" : "");
1044 		} else if (v_opt == 2) {
1045 			printf("--------          -------  ---                            %s-------\n", y_str);
1046 			printf("%8ju          %7ju   0%%                            %s%ju file%s\n",
1047 			    total_size, total_size, y_str, file_count,
1048 			    file_count != 1 ? "s" : "");
1049 		}
1050 	}
1051 
1052 	ac(archive_read_free(a));
1053 
1054 	if (passphrase_buf != NULL) {
1055 		memset(passphrase_buf, 0, PPBUFF_SIZE);
1056 		free(passphrase_buf);
1057 	}
1058 
1059 	if (t_opt) {
1060 		if (error_count > 0) {
1061 			errorx("%ju checksum error(s) found.", error_count);
1062 		}
1063 		else {
1064 			printf("No errors detected in compressed data of %s.\n",
1065 			       fn);
1066 		}
1067 	}
1068 }
1069 
1070 static void
usage(void)1071 usage(void)
1072 {
1073 
1074 	fprintf(stderr,
1075 "Usage: unzip [-aCcfjLlnopqtuvyZ1] [{-O|-I} encoding] [-d dir] [-x pattern] [-P password] zipfile\n"
1076 "             [member ...]\n");
1077 	exit(EXIT_FAILURE);
1078 }
1079 
1080 static void
version(void)1081 version(void)
1082 {
1083         printf("bsdunzip %s - %s \n",
1084             BSDUNZIP_VERSION_STRING,
1085             archive_version_details());
1086         exit(0);
1087 }
1088 
1089 static int
getopts(int argc,char * argv[])1090 getopts(int argc, char *argv[])
1091 {
1092 	struct bsdunzip *bsdunzip, bsdunzip_storage;
1093 	int opt;
1094 	bsdunzip_optind = 1;
1095 
1096 	bsdunzip = &bsdunzip_storage;
1097 	memset(bsdunzip, 0, sizeof(*bsdunzip));
1098 
1099         bsdunzip->argv = argv;
1100         bsdunzip->argc = argc;
1101 
1102 	while ((opt = bsdunzip_getopt(bsdunzip)) != -1) {
1103 		unzip_exclude_mode = 0;
1104 		switch (opt) {
1105 		case 'a':
1106 			a_opt = 1;
1107 			break;
1108 		case 'C':
1109 			C_opt = 1;
1110 			break;
1111 		case 'c':
1112 			c_opt = 1;
1113 			break;
1114 		case 'd':
1115 			d_arg = bsdunzip->argument;
1116 			break;
1117 		case 'f':
1118 			f_opt = 1;
1119 			break;
1120 		case 'I':
1121 		case 'O':
1122 			O_arg = bsdunzip->argument;
1123 			break;
1124 		case 'j':
1125 			j_opt = 1;
1126 			break;
1127 		case 'L':
1128 			L_opt = 1;
1129 			break;
1130 		case 'l':
1131 			if (v_opt == 0)
1132 				v_opt = 1;
1133 			break;
1134 		case 'n':
1135 			n_opt = 1;
1136 			break;
1137 		case 'o':
1138 			o_opt = 1;
1139 			q_opt = 1;
1140 			break;
1141 		case 'p':
1142 			p_opt = 1;
1143 			break;
1144 		case 'P':
1145 			P_arg = bsdunzip->argument;
1146 			break;
1147 		case 'q':
1148 			q_opt = 1;
1149 			break;
1150 		case 't':
1151 			t_opt = 1;
1152 			break;
1153 		case 'u':
1154 			u_opt = 1;
1155 			break;
1156 		case 'v':
1157 			v_opt = 2;
1158 			break;
1159 		case 'x':
1160 			add_pattern(&exclude, bsdunzip->argument);
1161 			unzip_exclude_mode = 1;
1162 			break;
1163 		case 'y':
1164 			y_str = "  ";
1165 			break;
1166 		case 'Z':
1167 			zipinfo_mode = 1;
1168 			if (bsdunzip->argument != NULL &&
1169 			    strcmp(bsdunzip->argument, "1") == 0) {
1170 				Z1_opt = 1;
1171 			}
1172 			break;
1173 		case OPTION_VERSION:
1174 			version_opt = 1;
1175 			break;
1176 		case OPTION_NONE:
1177 			break;
1178 		default:
1179 			usage();
1180 		}
1181 		if (opt == OPTION_NONE)
1182 			break;
1183 	}
1184 	return (bsdunzip_optind);
1185 }
1186 
1187 int
main(int argc,char * argv[])1188 main(int argc, char *argv[])
1189 {
1190 	const char *zipfile;
1191 	int nopts;
1192 
1193 #if defined(HAVE_SIGACTION) && defined(SIGCHLD)
1194 	{ /* Do not ignore SIGCHLD. */
1195 		struct sigaction sa;
1196 		sa.sa_handler = SIG_DFL;
1197 		sigemptyset(&sa.sa_mask);
1198 		sa.sa_flags = 0;
1199 		sigaction(SIGCHLD, &sa, NULL);
1200 	}
1201 #endif
1202 
1203 	lafe_setprogname(*argv, "bsdunzip");
1204 
1205 #if HAVE_SETLOCALE
1206 	if (setlocale(LC_ALL, "") == NULL)
1207 		lafe_warnc(0, "Failed to set default locale");
1208 #endif
1209 
1210 	if (isatty(STDOUT_FILENO))
1211 		tty = 1;
1212 
1213 	if (getenv("UNZIP_DEBUG") != NULL)
1214 		unzip_debug = 1;
1215 	for (int i = 0; i < argc; ++i)
1216 		debug("%s%c", argv[i], (i < argc - 1) ? ' ' : '\n');
1217 
1218 #ifdef __GLIBC__
1219 	/* Prevent GNU getopt(3) from rearranging options. */
1220 	setenv("POSIXLY_CORRECT", "", 1);
1221 #endif
1222 	/*
1223 	 * Info-ZIP's unzip(1) expects certain options to come before the
1224 	 * zipfile name, and others to come after - though it does not
1225 	 * enforce this.  For simplicity, we accept *all* options both
1226 	 * before and after the zipfile name.
1227 	 */
1228 	nopts = getopts(argc, argv);
1229 
1230 	if (version_opt == 1)
1231 		version();
1232 
1233 	/*
1234 	 * When more of the zipinfo mode options are implemented, this
1235 	 * will need to change.
1236 	 */
1237 	if (zipinfo_mode && !Z1_opt) {
1238 		printf("Zipinfo mode needs additional options\n");
1239 		exit(EXIT_FAILURE);
1240 	}
1241 
1242 	if (argc <= nopts)
1243 		usage();
1244 	zipfile = argv[nopts++];
1245 
1246 	if (strcmp(zipfile, "-") == 0)
1247 		zipfile = NULL; /* STDIN */
1248 
1249 	unzip_exclude_mode = 0;
1250 
1251 	while (nopts < argc && *argv[nopts] != '-')
1252 		add_pattern(&include, argv[nopts++]);
1253 
1254 	nopts--; /* fake argv[0] */
1255 	nopts += getopts(argc - nopts, argv + nopts);
1256 
1257 	/*
1258 	 * For compatibility with Info-ZIP's unzip(1) we need to treat
1259 	 * non-option arguments following an -x after the zipfile as
1260 	 * exclude list members.
1261 	 */
1262 	if (unzip_exclude_mode) {
1263 		while (nopts < argc && *argv[nopts] != '-')
1264 			add_pattern(&exclude, argv[nopts++]);
1265 		nopts--; /* fake argv[0] */
1266 		nopts += getopts(argc - nopts, argv + nopts);
1267 	}
1268 
1269 	/* There may be residual arguments if we encountered -- */
1270 	while (nopts < argc)
1271 		add_pattern(&include, argv[nopts++]);
1272 
1273 	if (n_opt + o_opt + u_opt > 1)
1274 		errorx("-n, -o and -u are contradictory");
1275 
1276 	unzip(zipfile);
1277 
1278 	exit(EXIT_SUCCESS);
1279 }
1280