xref: /freebsd-12.1/usr.bin/unzip/unzip.c (revision 1de7b4b8)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3  *
4  * Copyright (c) 2009, 2010 Joerg Sonnenberger <[email protected]>
5  * Copyright (c) 2007-2008 Dag-Erling Smørgrav
6  * All rights reserved.
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, this list of conditions and the following disclaimer
13  *    in this position and unchanged.
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
22  * FOR 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  * $FreeBSD$
31  *
32  * This file would be much shorter if we didn't care about command-line
33  * compatibility with Info-ZIP's UnZip, which requires us to duplicate
34  * parts of libarchive in order to gain more detailed control of its
35  * behaviour for the purpose of implementing the -n, -o, -L and -a
36  * options.
37  */
38 
39 #include <sys/queue.h>
40 #include <sys/stat.h>
41 
42 #include <ctype.h>
43 #include <errno.h>
44 #include <fcntl.h>
45 #include <fnmatch.h>
46 #include <stdarg.h>
47 #include <stdio.h>
48 #include <stdlib.h>
49 #include <string.h>
50 #include <unistd.h>
51 
52 #include <archive.h>
53 #include <archive_entry.h>
54 
55 /* command-line options */
56 static int		 a_opt;		/* convert EOL */
57 static int		 C_opt;		/* match case-insensitively */
58 static int		 c_opt;		/* extract to stdout */
59 static const char	*d_arg;		/* directory */
60 static int		 f_opt;		/* update existing files only */
61 static int		 j_opt;		/* junk directories */
62 static int		 L_opt;		/* lowercase names */
63 static int		 n_opt;		/* never overwrite */
64 static int		 o_opt;		/* always overwrite */
65 static int		 p_opt;		/* extract to stdout, quiet */
66 static int		 q_opt;		/* quiet */
67 static int		 t_opt;		/* test */
68 static int		 u_opt;		/* update */
69 static int		 v_opt;		/* verbose/list */
70 static const char	*y_str = "";	/* 4 digit year */
71 static int		 Z1_opt;	/* zipinfo mode list files only */
72 
73 /* debug flag */
74 static int		 unzip_debug;
75 
76 /* zipinfo mode */
77 static int		 zipinfo_mode;
78 
79 /* running on tty? */
80 static int		 tty;
81 
82 /* convenience macro */
83 /* XXX should differentiate between ARCHIVE_{WARN,FAIL,RETRY} */
84 #define ac(call)						\
85 	do {							\
86 		int acret = (call);				\
87 		if (acret != ARCHIVE_OK)			\
88 			errorx("%s", archive_error_string(a));	\
89 	} while (0)
90 
91 /*
92  * Indicates that last info() did not end with EOL.  This helps error() et
93  * al. avoid printing an error message on the same line as an incomplete
94  * informational message.
95  */
96 static int noeol;
97 
98 /* fatal error message + errno */
99 static void
error(const char * fmt,...)100 error(const char *fmt, ...)
101 {
102 	va_list ap;
103 
104 	if (noeol)
105 		fprintf(stdout, "\n");
106 	fflush(stdout);
107 	fprintf(stderr, "unzip: ");
108 	va_start(ap, fmt);
109 	vfprintf(stderr, fmt, ap);
110 	va_end(ap);
111 	fprintf(stderr, ": %s\n", strerror(errno));
112 	exit(1);
113 }
114 
115 /* fatal error message, no errno */
116 static void
errorx(const char * fmt,...)117 errorx(const char *fmt, ...)
118 {
119 	va_list ap;
120 
121 	if (noeol)
122 		fprintf(stdout, "\n");
123 	fflush(stdout);
124 	fprintf(stderr, "unzip: ");
125 	va_start(ap, fmt);
126 	vfprintf(stderr, fmt, ap);
127 	va_end(ap);
128 	fprintf(stderr, "\n");
129 	exit(1);
130 }
131 
132 /* non-fatal error message + errno */
133 static void
warning(const char * fmt,...)134 warning(const char *fmt, ...)
135 {
136 	va_list ap;
137 
138 	if (noeol)
139 		fprintf(stdout, "\n");
140 	fflush(stdout);
141 	fprintf(stderr, "unzip: ");
142 	va_start(ap, fmt);
143 	vfprintf(stderr, fmt, ap);
144 	va_end(ap);
145 	fprintf(stderr, ": %s\n", strerror(errno));
146 }
147 
148 /* non-fatal error message, no errno */
149 static void
warningx(const char * fmt,...)150 warningx(const char *fmt, ...)
151 {
152 	va_list ap;
153 
154 	if (noeol)
155 		fprintf(stdout, "\n");
156 	fflush(stdout);
157 	fprintf(stderr, "unzip: ");
158 	va_start(ap, fmt);
159 	vfprintf(stderr, fmt, ap);
160 	va_end(ap);
161 	fprintf(stderr, "\n");
162 }
163 
164 /* informational message (if not -q) */
165 static void
info(const char * fmt,...)166 info(const char *fmt, ...)
167 {
168 	va_list ap;
169 
170 	if (q_opt && !unzip_debug)
171 		return;
172 	va_start(ap, fmt);
173 	vfprintf(stdout, fmt, ap);
174 	va_end(ap);
175 	fflush(stdout);
176 
177 	if (*fmt == '\0')
178 		noeol = 1;
179 	else
180 		noeol = fmt[strlen(fmt) - 1] != '\n';
181 }
182 
183 /* debug message (if unzip_debug) */
184 static void
debug(const char * fmt,...)185 debug(const char *fmt, ...)
186 {
187 	va_list ap;
188 
189 	if (!unzip_debug)
190 		return;
191 	va_start(ap, fmt);
192 	vfprintf(stderr, fmt, ap);
193 	va_end(ap);
194 	fflush(stderr);
195 
196 	if (*fmt == '\0')
197 		noeol = 1;
198 	else
199 		noeol = fmt[strlen(fmt) - 1] != '\n';
200 }
201 
202 /* duplicate a path name, possibly converting to lower case */
203 static char *
pathdup(const char * path)204 pathdup(const char *path)
205 {
206 	char *str;
207 	size_t i, len;
208 
209 	len = strlen(path);
210 	while (len && path[len - 1] == '/')
211 		len--;
212 	if ((str = malloc(len + 1)) == NULL) {
213 		errno = ENOMEM;
214 		error("malloc()");
215 	}
216 	if (L_opt) {
217 		for (i = 0; i < len; ++i)
218 			str[i] = tolower((unsigned char)path[i]);
219 	} else {
220 		memcpy(str, path, len);
221 	}
222 	str[len] = '\0';
223 
224 	return (str);
225 }
226 
227 /* concatenate two path names */
228 static char *
pathcat(const char * prefix,const char * path)229 pathcat(const char *prefix, const char *path)
230 {
231 	char *str;
232 	size_t prelen, len;
233 
234 	prelen = prefix ? strlen(prefix) + 1 : 0;
235 	len = strlen(path) + 1;
236 	if ((str = malloc(prelen + len)) == NULL) {
237 		errno = ENOMEM;
238 		error("malloc()");
239 	}
240 	if (prefix) {
241 		memcpy(str, prefix, prelen);	/* includes zero */
242 		str[prelen - 1] = '/';		/* splat zero */
243 	}
244 	memcpy(str + prelen, path, len);	/* includes zero */
245 
246 	return (str);
247 }
248 
249 /*
250  * Pattern lists for include / exclude processing
251  */
252 struct pattern {
253 	STAILQ_ENTRY(pattern) link;
254 	char pattern[];
255 };
256 
257 STAILQ_HEAD(pattern_list, pattern);
258 static struct pattern_list include = STAILQ_HEAD_INITIALIZER(include);
259 static struct pattern_list exclude = STAILQ_HEAD_INITIALIZER(exclude);
260 
261 /*
262  * Add an entry to a pattern list
263  */
264 static void
add_pattern(struct pattern_list * list,const char * pattern)265 add_pattern(struct pattern_list *list, const char *pattern)
266 {
267 	struct pattern *entry;
268 	size_t len;
269 
270 	debug("adding pattern '%s'\n", pattern);
271 	len = strlen(pattern);
272 	if ((entry = malloc(sizeof *entry + len + 1)) == NULL) {
273 		errno = ENOMEM;
274 		error("malloc()");
275 	}
276 	memcpy(entry->pattern, pattern, len + 1);
277 	STAILQ_INSERT_TAIL(list, entry, link);
278 }
279 
280 /*
281  * Match a string against a list of patterns
282  */
283 static int
match_pattern(struct pattern_list * list,const char * str)284 match_pattern(struct pattern_list *list, const char *str)
285 {
286 	struct pattern *entry;
287 
288 	STAILQ_FOREACH(entry, list, link) {
289 		if (fnmatch(entry->pattern, str, C_opt ? FNM_CASEFOLD : 0) == 0)
290 			return (1);
291 	}
292 	return (0);
293 }
294 
295 /*
296  * Verify that a given pathname is in the include list and not in the
297  * exclude list.
298  */
299 static int
accept_pathname(const char * pathname)300 accept_pathname(const char *pathname)
301 {
302 
303 	if (!STAILQ_EMPTY(&include) && !match_pattern(&include, pathname))
304 		return (0);
305 	if (!STAILQ_EMPTY(&exclude) && match_pattern(&exclude, pathname))
306 		return (0);
307 	return (1);
308 }
309 
310 /*
311  * Create the specified directory with the specified mode, taking certain
312  * precautions on they way.
313  */
314 static void
make_dir(const char * path,int mode)315 make_dir(const char *path, int mode)
316 {
317 	struct stat sb;
318 
319 	if (lstat(path, &sb) == 0) {
320 		if (S_ISDIR(sb.st_mode))
321 			return;
322 		/*
323 		 * Normally, we should either ask the user about removing
324 		 * the non-directory of the same name as a directory we
325 		 * wish to create, or respect the -n or -o command-line
326 		 * options.  However, this may lead to a later failure or
327 		 * even compromise (if this non-directory happens to be a
328 		 * symlink to somewhere unsafe), so we don't.
329 		 */
330 
331 		/*
332 		 * Don't check unlink() result; failure will cause mkdir()
333 		 * to fail later, which we will catch.
334 		 */
335 		(void)unlink(path);
336 	}
337 	if (mkdir(path, mode) != 0 && errno != EEXIST)
338 		error("mkdir('%s')", path);
339 }
340 
341 /*
342  * Ensure that all directories leading up to (but not including) the
343  * specified path exist.
344  *
345  * XXX inefficient + modifies the file in-place
346  */
347 static void
make_parent(char * path)348 make_parent(char *path)
349 {
350 	struct stat sb;
351 	char *sep;
352 
353 	sep = strrchr(path, '/');
354 	if (sep == NULL || sep == path)
355 		return;
356 	*sep = '\0';
357 	if (lstat(path, &sb) == 0) {
358 		if (S_ISDIR(sb.st_mode)) {
359 			*sep = '/';
360 			return;
361 		}
362 		unlink(path);
363 	}
364 	make_parent(path);
365 	mkdir(path, 0755);
366 	*sep = '/';
367 
368 #if 0
369 	for (sep = path; (sep = strchr(sep, '/')) != NULL; sep++) {
370 		/* root in case of absolute d_arg */
371 		if (sep == path)
372 			continue;
373 		*sep = '\0';
374 		make_dir(path, 0755);
375 		*sep = '/';
376 	}
377 #endif
378 }
379 
380 /*
381  * Extract a directory.
382  */
383 static void
extract_dir(struct archive * a,struct archive_entry * e,const char * path)384 extract_dir(struct archive *a, struct archive_entry *e, const char *path)
385 {
386 	int mode;
387 
388 	mode = archive_entry_mode(e) & 0777;
389 	if (mode == 0)
390 		mode = 0755;
391 
392 	/*
393 	 * Some zipfiles contain directories with weird permissions such
394 	 * as 0644 or 0444.  This can cause strange issues such as being
395 	 * unable to extract files into the directory we just created, or
396 	 * the user being unable to remove the directory later without
397 	 * first manually changing its permissions.  Therefore, we whack
398 	 * the permissions into shape, assuming that the user wants full
399 	 * access and that anyone who gets read access also gets execute
400 	 * access.
401 	 */
402 	mode |= 0700;
403 	if (mode & 0040)
404 		mode |= 0010;
405 	if (mode & 0004)
406 		mode |= 0001;
407 
408 	info("   creating: %s/\n", path);
409 	make_dir(path, mode);
410 	ac(archive_read_data_skip(a));
411 }
412 
413 static unsigned char buffer[8192];
414 static char spinner[] = { '|', '/', '-', '\\' };
415 
416 static int
handle_existing_file(char ** path)417 handle_existing_file(char **path)
418 {
419 	size_t alen;
420 	ssize_t len;
421 	char buf[4];
422 
423 	for (;;) {
424 		fprintf(stderr,
425 		    "replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ",
426 		    *path);
427 		if (fgets(buf, sizeof(buf), stdin) == NULL) {
428 			clearerr(stdin);
429 			printf("NULL\n(EOF or read error, "
430 			    "treating as \"[N]one\"...)\n");
431 			n_opt = 1;
432 			return -1;
433 		}
434 		switch (*buf) {
435 		case 'A':
436 			o_opt = 1;
437 			/* FALLTHROUGH */
438 		case 'y':
439 		case 'Y':
440 			(void)unlink(*path);
441 			return 1;
442 		case 'N':
443 			n_opt = 1;
444 			/* FALLTHROUGH */
445 		case 'n':
446 			return -1;
447 		case 'r':
448 		case 'R':
449 			printf("New name: ");
450 			fflush(stdout);
451 			free(*path);
452 			*path = NULL;
453 			alen = 0;
454 			len = getdelim(path, &alen, '\n', stdin);
455 			if ((*path)[len - 1] == '\n')
456 				(*path)[len - 1] = '\0';
457 			return 0;
458 		default:
459 			break;
460 		}
461 	}
462 }
463 
464 /*
465  * Detect binary files by a combination of character white list and
466  * black list. NUL bytes and other control codes without use in text files
467  * result directly in switching the file to binary mode. Otherwise, at least
468  * one white-listed byte has to be found.
469  *
470  * Black-listed: 0..6, 14..25, 28..31
471  * 0xf3ffc07f = 11110011111111111100000001111111b
472  * White-listed: 9..10, 13, >= 32
473  * 0x00002600 = 00000000000000000010011000000000b
474  *
475  * See the proginfo/txtvsbin.txt in the zip sources for a detailed discussion.
476  */
477 #define BYTE_IS_BINARY(x)	((x) < 32 && (0xf3ffc07fU & (1U << (x))))
478 #define	BYTE_IS_TEXT(x)		((x) >= 32 || (0x00002600U & (1U << (x))))
479 
480 static int
check_binary(const unsigned char * buf,size_t len)481 check_binary(const unsigned char *buf, size_t len)
482 {
483 	int rv;
484 	for (rv = 1; len--; ++buf) {
485 		if (BYTE_IS_BINARY(*buf))
486 			return 1;
487 		if (BYTE_IS_TEXT(*buf))
488 			rv = 0;
489 	}
490 
491 	return rv;
492 }
493 
494 /*
495  * Extract to a file descriptor
496  */
497 static int
extract2fd(struct archive * a,char * pathname,int fd)498 extract2fd(struct archive *a, char *pathname, int fd)
499 {
500 	int cr, text, warn;
501 	ssize_t len;
502 	unsigned char *p, *q, *end;
503 
504 	text = a_opt;
505 	warn = 0;
506 	cr = 0;
507 
508 	/* loop over file contents and write to fd */
509 	for (int n = 0; ; n++) {
510 		if (fd != STDOUT_FILENO)
511 			if (tty && (n % 4) == 0)
512 				info(" %c\b\b", spinner[(n / 4) % sizeof spinner]);
513 
514 		len = archive_read_data(a, buffer, sizeof buffer);
515 
516 		if (len < 0)
517 			ac(len);
518 
519 		/* left over CR from previous buffer */
520 		if (a_opt && cr) {
521 			if (len == 0 || buffer[0] != '\n')
522 				if (write(fd, "\r", 1) != 1)
523 					error("write('%s')", pathname);
524 			cr = 0;
525 		}
526 
527 		/* EOF */
528 		if (len == 0)
529 			break;
530 		end = buffer + len;
531 
532 		/*
533 		 * Detect whether this is a text file.  The correct way to
534 		 * do this is to check the least significant bit of the
535 		 * "internal file attributes" field of the corresponding
536 		 * file header in the central directory, but libarchive
537 		 * does not provide access to this field, so we have to
538 		 * guess by looking for non-ASCII characters in the
539 		 * buffer.  Hopefully we won't guess wrong.  If we do
540 		 * guess wrong, we print a warning message later.
541 		 */
542 		if (a_opt && n == 0) {
543 			if (check_binary(buffer, len))
544 				text = 0;
545 		}
546 
547 		/* simple case */
548 		if (!a_opt || !text) {
549 			if (write(fd, buffer, len) != len)
550 				error("write('%s')", pathname);
551 			continue;
552 		}
553 
554 		/* hard case: convert \r\n to \n (sigh...) */
555 		for (p = buffer; p < end; p = q + 1) {
556 			for (q = p; q < end; q++) {
557 				if (!warn && BYTE_IS_BINARY(*q)) {
558 					warningx("%s may be corrupted due"
559 					    " to weak text file detection"
560 					    " heuristic", pathname);
561 					warn = 1;
562 				}
563 				if (q[0] != '\r')
564 					continue;
565 				if (&q[1] == end) {
566 					cr = 1;
567 					break;
568 				}
569 				if (q[1] == '\n')
570 					break;
571 			}
572 			if (write(fd, p, q - p) != q - p)
573 				error("write('%s')", pathname);
574 		}
575 	}
576 
577 	return text;
578 }
579 
580 /*
581  * Extract a regular file.
582  */
583 static void
extract_file(struct archive * a,struct archive_entry * e,char ** path)584 extract_file(struct archive *a, struct archive_entry *e, char **path)
585 {
586 	int mode;
587 	struct timespec mtime;
588 	struct stat sb;
589 	struct timespec ts[2];
590 	int fd, check, text;
591 	const char *linkname;
592 
593 	mode = archive_entry_mode(e) & 0777;
594 	if (mode == 0)
595 		mode = 0644;
596 	mtime.tv_sec = archive_entry_mtime(e);
597 	mtime.tv_nsec = archive_entry_mtime_nsec(e);
598 
599 	/* look for existing file of same name */
600 recheck:
601 	if (lstat(*path, &sb) == 0) {
602 		if (u_opt || f_opt) {
603 			/* check if up-to-date */
604 			if ((S_ISREG(sb.st_mode) || S_ISLNK(sb.st_mode)) &&
605 			    (sb.st_mtim.tv_sec > mtime.tv_sec ||
606 			    (sb.st_mtim.tv_sec == mtime.tv_sec &&
607 			    sb.st_mtim.tv_nsec >= mtime.tv_nsec)))
608 				return;
609 			(void)unlink(*path);
610 		} else if (o_opt) {
611 			/* overwrite */
612 			(void)unlink(*path);
613 		} else if (n_opt) {
614 			/* do not overwrite */
615 			return;
616 		} else {
617 			check = handle_existing_file(path);
618 			if (check == 0)
619 				goto recheck;
620 			if (check == -1)
621 				return; /* do not overwrite */
622 		}
623 	} else {
624 		if (f_opt)
625 			return;
626 	}
627 
628 	ts[0].tv_sec = 0;
629 	ts[0].tv_nsec = UTIME_NOW;
630 	ts[1] = mtime;
631 
632 	/* process symlinks */
633 	linkname = archive_entry_symlink(e);
634 	if (linkname != NULL) {
635 		if (symlink(linkname, *path) != 0)
636 			error("symlink('%s')", *path);
637 		info(" extracting: %s -> %s\n", *path, linkname);
638 		if (lchmod(*path, mode) != 0)
639 			warning("Cannot set mode for '%s'", *path);
640 		/* set access and modification time */
641 		if (utimensat(AT_FDCWD, *path, ts, AT_SYMLINK_NOFOLLOW) != 0)
642 			warning("utimensat('%s')", *path);
643 		return;
644 	}
645 
646 	if ((fd = open(*path, O_RDWR|O_CREAT|O_TRUNC, mode)) < 0)
647 		error("open('%s')", *path);
648 
649 	info(" extracting: %s", *path);
650 
651 	text = extract2fd(a, *path, fd);
652 
653 	if (tty)
654 		info("  \b\b");
655 	if (text)
656 		info(" (text)");
657 	info("\n");
658 
659 	/* set access and modification time */
660 	if (futimens(fd, ts) != 0)
661 		error("futimens('%s')", *path);
662 	if (close(fd) != 0)
663 		error("close('%s')", *path);
664 }
665 
666 /*
667  * Extract a zipfile entry: first perform some sanity checks to ensure
668  * that it is either a directory or a regular file and that the path is
669  * not absolute and does not try to break out of the current directory;
670  * then call either extract_dir() or extract_file() as appropriate.
671  *
672  * This is complicated a bit by the various ways in which we need to
673  * manipulate the path name.  Case conversion (if requested by the -L
674  * option) happens first, but the include / exclude patterns are applied
675  * to the full converted path name, before the directory part of the path
676  * is removed in accordance with the -j option.  Sanity checks are
677  * intentionally done earlier than they need to be, so the user will get a
678  * warning about insecure paths even for files or directories which
679  * wouldn't be extracted anyway.
680  */
681 static void
extract(struct archive * a,struct archive_entry * e)682 extract(struct archive *a, struct archive_entry *e)
683 {
684 	char *pathname, *realpathname;
685 	mode_t filetype;
686 	char *p, *q;
687 
688 	pathname = pathdup(archive_entry_pathname(e));
689 	filetype = archive_entry_filetype(e);
690 
691 	/* sanity checks */
692 	if (pathname[0] == '/' ||
693 	    strncmp(pathname, "../", 3) == 0 ||
694 	    strstr(pathname, "/../") != NULL) {
695 		warningx("skipping insecure entry '%s'", pathname);
696 		ac(archive_read_data_skip(a));
697 		free(pathname);
698 		return;
699 	}
700 
701 	/* I don't think this can happen in a zipfile.. */
702 	if (!S_ISDIR(filetype) && !S_ISREG(filetype) && !S_ISLNK(filetype)) {
703 		warningx("skipping non-regular entry '%s'", pathname);
704 		ac(archive_read_data_skip(a));
705 		free(pathname);
706 		return;
707 	}
708 
709 	/* skip directories in -j case */
710 	if (S_ISDIR(filetype) && j_opt) {
711 		ac(archive_read_data_skip(a));
712 		free(pathname);
713 		return;
714 	}
715 
716 	/* apply include / exclude patterns */
717 	if (!accept_pathname(pathname)) {
718 		ac(archive_read_data_skip(a));
719 		free(pathname);
720 		return;
721 	}
722 
723 	/* apply -j and -d */
724 	if (j_opt) {
725 		for (p = q = pathname; *p; ++p)
726 			if (*p == '/')
727 				q = p + 1;
728 		realpathname = pathcat(d_arg, q);
729 	} else {
730 		realpathname = pathcat(d_arg, pathname);
731 	}
732 
733 	/* ensure that parent directory exists */
734 	make_parent(realpathname);
735 
736 	if (S_ISDIR(filetype))
737 		extract_dir(a, e, realpathname);
738 	else
739 		extract_file(a, e, &realpathname);
740 
741 	free(realpathname);
742 	free(pathname);
743 }
744 
745 static void
extract_stdout(struct archive * a,struct archive_entry * e)746 extract_stdout(struct archive *a, struct archive_entry *e)
747 {
748 	char *pathname;
749 	mode_t filetype;
750 
751 	pathname = pathdup(archive_entry_pathname(e));
752 	filetype = archive_entry_filetype(e);
753 
754 	/* I don't think this can happen in a zipfile.. */
755 	if (!S_ISDIR(filetype) && !S_ISREG(filetype) && !S_ISLNK(filetype)) {
756 		warningx("skipping non-regular entry '%s'", pathname);
757 		ac(archive_read_data_skip(a));
758 		free(pathname);
759 		return;
760 	}
761 
762 	/* skip directories in -j case */
763 	if (S_ISDIR(filetype)) {
764 		ac(archive_read_data_skip(a));
765 		free(pathname);
766 		return;
767 	}
768 
769 	/* apply include / exclude patterns */
770 	if (!accept_pathname(pathname)) {
771 		ac(archive_read_data_skip(a));
772 		free(pathname);
773 		return;
774 	}
775 
776 	if (c_opt)
777 		info("x %s\n", pathname);
778 
779 	(void)extract2fd(a, pathname, STDOUT_FILENO);
780 
781 	free(pathname);
782 }
783 
784 /*
785  * Print the name of an entry to stdout.
786  */
787 static void
list(struct archive * a,struct archive_entry * e)788 list(struct archive *a, struct archive_entry *e)
789 {
790 	char buf[20];
791 	time_t mtime;
792 	struct tm *tm;
793 
794 	mtime = archive_entry_mtime(e);
795 	tm = localtime(&mtime);
796 	if (*y_str)
797 		strftime(buf, sizeof(buf), "%m-%d-%G %R", tm);
798 	else
799 		strftime(buf, sizeof(buf), "%m-%d-%g %R", tm);
800 
801 	if (!zipinfo_mode) {
802 		if (v_opt == 1) {
803 			printf(" %8ju  %s   %s\n",
804 			    (uintmax_t)archive_entry_size(e),
805 			    buf, archive_entry_pathname(e));
806 		} else if (v_opt == 2) {
807 			printf("%8ju  Stored  %7ju   0%%  %s  %08x  %s\n",
808 			    (uintmax_t)archive_entry_size(e),
809 			    (uintmax_t)archive_entry_size(e),
810 			    buf,
811 			    0U,
812 			    archive_entry_pathname(e));
813 		}
814 	} else {
815 		if (Z1_opt)
816 			printf("%s\n",archive_entry_pathname(e));
817 	}
818 	ac(archive_read_data_skip(a));
819 }
820 
821 /*
822  * Extract to memory to check CRC
823  */
824 static int
test(struct archive * a,struct archive_entry * e)825 test(struct archive *a, struct archive_entry *e)
826 {
827 	ssize_t len;
828 	int error_count;
829 
830 	error_count = 0;
831 	if (S_ISDIR(archive_entry_filetype(e)))
832 		return 0;
833 
834 	info("    testing: %s\t", archive_entry_pathname(e));
835 	while ((len = archive_read_data(a, buffer, sizeof buffer)) > 0)
836 		/* nothing */;
837 	if (len < 0) {
838 		info(" %s\n", archive_error_string(a));
839 		++error_count;
840 	} else {
841 		info(" OK\n");
842 	}
843 
844 	/* shouldn't be necessary, but it doesn't hurt */
845 	ac(archive_read_data_skip(a));
846 
847 	return error_count;
848 }
849 
850 /*
851  * Main loop: open the zipfile, iterate over its contents and decide what
852  * to do with each entry.
853  */
854 static void
unzip(const char * fn)855 unzip(const char *fn)
856 {
857 	struct archive *a;
858 	struct archive_entry *e;
859 	int ret;
860 	uintmax_t total_size, file_count, error_count;
861 
862 	if ((a = archive_read_new()) == NULL)
863 		error("archive_read_new failed");
864 
865 	ac(archive_read_support_format_zip(a));
866 	ac(archive_read_open_filename(a, fn, 8192));
867 
868 	if (!zipinfo_mode) {
869 		if (!p_opt && !q_opt)
870 			printf("Archive:  %s\n", fn);
871 		if (v_opt == 1) {
872 			printf("  Length     %sDate   Time    Name\n", y_str);
873 			printf(" --------    %s----   ----    ----\n", y_str);
874 		} else if (v_opt == 2) {
875 			printf(" Length   Method    Size  Ratio   %sDate   Time   CRC-32    Name\n", y_str);
876 			printf("--------  ------  ------- -----   %s----   ----   ------    ----\n", y_str);
877 		}
878 	}
879 
880 	total_size = 0;
881 	file_count = 0;
882 	error_count = 0;
883 	for (;;) {
884 		ret = archive_read_next_header(a, &e);
885 		if (ret == ARCHIVE_EOF)
886 			break;
887 		ac(ret);
888 		if (!zipinfo_mode) {
889 			if (t_opt)
890 				error_count += test(a, e);
891 			else if (v_opt)
892 				list(a, e);
893 			else if (p_opt || c_opt)
894 				extract_stdout(a, e);
895 			else
896 				extract(a, e);
897 		} else {
898 			if (Z1_opt)
899 				list(a, e);
900 		}
901 
902 		total_size += archive_entry_size(e);
903 		++file_count;
904 	}
905 
906 	if (zipinfo_mode) {
907 		if (v_opt == 1) {
908 			printf(" --------                   %s-------\n", y_str);
909 			printf(" %8ju                   %s%ju file%s\n",
910 			    total_size, y_str, file_count, file_count != 1 ? "s" : "");
911 		} else if (v_opt == 2) {
912 			printf("--------          -------  ---                            %s-------\n", y_str);
913 			printf("%8ju          %7ju   0%%                            %s%ju file%s\n",
914 			    total_size, total_size, y_str, file_count,
915 			    file_count != 1 ? "s" : "");
916 		}
917 	}
918 
919 	ac(archive_read_close(a));
920 	(void)archive_read_free(a);
921 
922 	if (t_opt) {
923 		if (error_count > 0) {
924 			errorx("%ju checksum error(s) found.", error_count);
925 		}
926 		else {
927 			printf("No errors detected in compressed data of %s.\n",
928 			       fn);
929 		}
930 	}
931 }
932 
933 static void
usage(void)934 usage(void)
935 {
936 
937 	fprintf(stderr, "Usage: unzip [-aCcfjLlnopqtuvyZ1] [-d dir] [-x pattern] "
938 		"zipfile\n");
939 	exit(1);
940 }
941 
942 static int
getopts(int argc,char * argv[])943 getopts(int argc, char *argv[])
944 {
945 	int opt;
946 
947 	optreset = optind = 1;
948 	while ((opt = getopt(argc, argv, "aCcd:fjLlnopqtuvx:yZ1")) != -1)
949 		switch (opt) {
950 		case '1':
951 			Z1_opt = 1;
952 			break;
953 		case 'a':
954 			a_opt = 1;
955 			break;
956 		case 'C':
957 			C_opt = 1;
958 			break;
959 		case 'c':
960 			c_opt = 1;
961 			break;
962 		case 'd':
963 			d_arg = optarg;
964 			break;
965 		case 'f':
966 			f_opt = 1;
967 			break;
968 		case 'j':
969 			j_opt = 1;
970 			break;
971 		case 'L':
972 			L_opt = 1;
973 			break;
974 		case 'l':
975 			if (v_opt == 0)
976 				v_opt = 1;
977 			break;
978 		case 'n':
979 			n_opt = 1;
980 			break;
981 		case 'o':
982 			o_opt = 1;
983 			q_opt = 1;
984 			break;
985 		case 'p':
986 			p_opt = 1;
987 			break;
988 		case 'q':
989 			q_opt = 1;
990 			break;
991 		case 't':
992 			t_opt = 1;
993 			break;
994 		case 'u':
995 			u_opt = 1;
996 			break;
997 		case 'v':
998 			v_opt = 2;
999 			break;
1000 		case 'x':
1001 			add_pattern(&exclude, optarg);
1002 			break;
1003 		case 'y':
1004 			y_str = "  ";
1005 			break;
1006 		case 'Z':
1007 			zipinfo_mode = 1;
1008 			break;
1009 		default:
1010 			usage();
1011 		}
1012 
1013 	return (optind);
1014 }
1015 
1016 int
main(int argc,char * argv[])1017 main(int argc, char *argv[])
1018 {
1019 	const char *zipfile;
1020 	int nopts;
1021 
1022 	if (isatty(STDOUT_FILENO))
1023 		tty = 1;
1024 
1025 	if (getenv("UNZIP_DEBUG") != NULL)
1026 		unzip_debug = 1;
1027 	for (int i = 0; i < argc; ++i)
1028 		debug("%s%c", argv[i], (i < argc - 1) ? ' ' : '\n');
1029 
1030 	/*
1031 	 * Info-ZIP's unzip(1) expects certain options to come before the
1032 	 * zipfile name, and others to come after - though it does not
1033 	 * enforce this.  For simplicity, we accept *all* options both
1034 	 * before and after the zipfile name.
1035 	 */
1036 	nopts = getopts(argc, argv);
1037 
1038 	/*
1039 	 * When more of the zipinfo mode options are implemented, this
1040 	 * will need to change.
1041 	 */
1042 	if (zipinfo_mode && !Z1_opt) {
1043 		printf("Zipinfo mode needs additional options\n");
1044 		exit(1);
1045 	}
1046 
1047 	if (argc <= nopts)
1048 		usage();
1049 	zipfile = argv[nopts++];
1050 
1051 	if (strcmp(zipfile, "-") == 0)
1052 		zipfile = NULL; /* STDIN */
1053 
1054 	while (nopts < argc && *argv[nopts] != '-')
1055 		add_pattern(&include, argv[nopts++]);
1056 
1057 	nopts--; /* fake argv[0] */
1058 	nopts += getopts(argc - nopts, argv + nopts);
1059 
1060 	if (n_opt + o_opt + u_opt > 1)
1061 		errorx("-n, -o and -u are contradictory");
1062 
1063 	unzip(zipfile);
1064 
1065 	exit(0);
1066 }
1067