1/*-
2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3 *
4 * Copyright (c) 2009, 2010 Joerg Sonnenberger <joerg@NetBSD.org>
5 * Copyright (c) 2007-2008 Dag-Erling Sm��rgrav
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 */
56static int		 a_opt;		/* convert EOL */
57static int		 C_opt;		/* match case-insensitively */
58static int		 c_opt;		/* extract to stdout */
59static const char	*d_arg;		/* directory */
60static int		 f_opt;		/* update existing files only */
61static int		 j_opt;		/* junk directories */
62static int		 L_opt;		/* lowercase names */
63static int		 n_opt;		/* never overwrite */
64static int		 o_opt;		/* always overwrite */
65static int		 p_opt;		/* extract to stdout, quiet */
66static int		 q_opt;		/* quiet */
67static int		 t_opt;		/* test */
68static int		 u_opt;		/* update */
69static int		 v_opt;		/* verbose/list */
70static const char	*y_str = "";	/* 4 digit year */
71static int		 Z1_opt;	/* zipinfo mode list files only */
72
73/* debug flag */
74static int		 unzip_debug;
75
76/* zipinfo mode */
77static int		 zipinfo_mode;
78
79/* running on tty? */
80static 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 */
96static int noeol;
97
98/* fatal error message + errno */
99static void
100error(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 */
116static void
117errorx(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 */
133static void
134warning(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 */
149static void
150warningx(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) */
165static void
166info(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) */
184static void
185debug(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 */
203static char *
204pathdup(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 */
228static char *
229pathcat(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 */
252struct pattern {
253	STAILQ_ENTRY(pattern) link;
254	char pattern[];
255};
256
257STAILQ_HEAD(pattern_list, pattern);
258static struct pattern_list include = STAILQ_HEAD_INITIALIZER(include);
259static struct pattern_list exclude = STAILQ_HEAD_INITIALIZER(exclude);
260
261/*
262 * Add an entry to a pattern list
263 */
264static void
265add_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 */
283static int
284match_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 */
299static int
300accept_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 */
314static void
315make_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 */
347static void
348make_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 */
383static void
384extract_dir(struct archive *a, struct archive_entry *e, const char *path)
385{
386	int mode;
387
388	/*
389	 * Dropbox likes to create '/' directory entries, just ignore
390	 * such junk.
391	 */
392	if (*path == '\0')
393		return;
394
395	mode = archive_entry_mode(e) & 0777;
396	if (mode == 0)
397		mode = 0755;
398
399	/*
400	 * Some zipfiles contain directories with weird permissions such
401	 * as 0644 or 0444.  This can cause strange issues such as being
402	 * unable to extract files into the directory we just created, or
403	 * the user being unable to remove the directory later without
404	 * first manually changing its permissions.  Therefore, we whack
405	 * the permissions into shape, assuming that the user wants full
406	 * access and that anyone who gets read access also gets execute
407	 * access.
408	 */
409	mode |= 0700;
410	if (mode & 0040)
411		mode |= 0010;
412	if (mode & 0004)
413		mode |= 0001;
414
415	info("   creating: %s/\n", path);
416	make_dir(path, mode);
417	ac(archive_read_data_skip(a));
418}
419
420static unsigned char buffer[8192];
421static char spinner[] = { '|', '/', '-', '\\' };
422
423static int
424handle_existing_file(char **path)
425{
426	size_t alen;
427	ssize_t len;
428	char buf[4];
429
430	for (;;) {
431		fprintf(stderr,
432		    "replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ",
433		    *path);
434		if (fgets(buf, sizeof(buf), stdin) == NULL) {
435			clearerr(stdin);
436			printf("NULL\n(EOF or read error, "
437			    "treating as \"[N]one\"...)\n");
438			n_opt = 1;
439			return -1;
440		}
441		switch (*buf) {
442		case 'A':
443			o_opt = 1;
444			/* FALLTHROUGH */
445		case 'y':
446		case 'Y':
447			(void)unlink(*path);
448			return 1;
449		case 'N':
450			n_opt = 1;
451			/* FALLTHROUGH */
452		case 'n':
453			return -1;
454		case 'r':
455		case 'R':
456			printf("New name: ");
457			fflush(stdout);
458			free(*path);
459			*path = NULL;
460			alen = 0;
461			len = getline(path, &alen, stdin);
462			if ((*path)[len - 1] == '\n')
463				(*path)[len - 1] = '\0';
464			return 0;
465		default:
466			break;
467		}
468	}
469}
470
471/*
472 * Detect binary files by a combination of character white list and
473 * black list. NUL bytes and other control codes without use in text files
474 * result directly in switching the file to binary mode. Otherwise, at least
475 * one white-listed byte has to be found.
476 *
477 * Black-listed: 0..6, 14..25, 28..31
478 * 0xf3ffc07f = 11110011111111111100000001111111b
479 * White-listed: 9..10, 13, >= 32
480 * 0x00002600 = 00000000000000000010011000000000b
481 *
482 * See the proginfo/txtvsbin.txt in the zip sources for a detailed discussion.
483 */
484#define BYTE_IS_BINARY(x)	((x) < 32 && (0xf3ffc07fU & (1U << (x))))
485#define	BYTE_IS_TEXT(x)		((x) >= 32 || (0x00002600U & (1U << (x))))
486
487static int
488check_binary(const unsigned char *buf, size_t len)
489{
490	int rv;
491	for (rv = 1; len--; ++buf) {
492		if (BYTE_IS_BINARY(*buf))
493			return 1;
494		if (BYTE_IS_TEXT(*buf))
495			rv = 0;
496	}
497
498	return rv;
499}
500
501/*
502 * Extract to a file descriptor
503 */
504static int
505extract2fd(struct archive *a, char *pathname, int fd)
506{
507	int cr, text, warn;
508	ssize_t len;
509	unsigned char *p, *q, *end;
510
511	text = a_opt;
512	warn = 0;
513	cr = 0;
514
515	/* loop over file contents and write to fd */
516	for (int n = 0; ; n++) {
517		if (fd != STDOUT_FILENO)
518			if (tty && (n % 4) == 0)
519				info(" %c\b\b", spinner[(n / 4) % sizeof spinner]);
520
521		len = archive_read_data(a, buffer, sizeof buffer);
522
523		if (len < 0)
524			ac(len);
525
526		/* left over CR from previous buffer */
527		if (a_opt && cr) {
528			if (len == 0 || buffer[0] != '\n')
529				if (write(fd, "\r", 1) != 1)
530					error("write('%s')", pathname);
531			cr = 0;
532		}
533
534		/* EOF */
535		if (len == 0)
536			break;
537		end = buffer + len;
538
539		/*
540		 * Detect whether this is a text file.  The correct way to
541		 * do this is to check the least significant bit of the
542		 * "internal file attributes" field of the corresponding
543		 * file header in the central directory, but libarchive
544		 * does not provide access to this field, so we have to
545		 * guess by looking for non-ASCII characters in the
546		 * buffer.  Hopefully we won't guess wrong.  If we do
547		 * guess wrong, we print a warning message later.
548		 */
549		if (a_opt && n == 0) {
550			if (check_binary(buffer, len))
551				text = 0;
552		}
553
554		/* simple case */
555		if (!a_opt || !text) {
556			if (write(fd, buffer, len) != len)
557				error("write('%s')", pathname);
558			continue;
559		}
560
561		/* hard case: convert \r\n to \n (sigh...) */
562		for (p = buffer; p < end; p = q + 1) {
563			for (q = p; q < end; q++) {
564				if (!warn && BYTE_IS_BINARY(*q)) {
565					warningx("%s may be corrupted due"
566					    " to weak text file detection"
567					    " heuristic", pathname);
568					warn = 1;
569				}
570				if (q[0] != '\r')
571					continue;
572				if (&q[1] == end) {
573					cr = 1;
574					break;
575				}
576				if (q[1] == '\n')
577					break;
578			}
579			if (write(fd, p, q - p) != q - p)
580				error("write('%s')", pathname);
581		}
582	}
583
584	return text;
585}
586
587/*
588 * Extract a regular file.
589 */
590static void
591extract_file(struct archive *a, struct archive_entry *e, char **path)
592{
593	int mode;
594	struct timespec mtime;
595	struct stat sb;
596	struct timespec ts[2];
597	int fd, check, text;
598	const char *linkname;
599
600	mode = archive_entry_mode(e) & 0777;
601	if (mode == 0)
602		mode = 0644;
603	mtime.tv_sec = archive_entry_mtime(e);
604	mtime.tv_nsec = archive_entry_mtime_nsec(e);
605
606	/* look for existing file of same name */
607recheck:
608	if (lstat(*path, &sb) == 0) {
609		if (u_opt || f_opt) {
610			/* check if up-to-date */
611			if (S_ISREG(sb.st_mode) &&
612			    (sb.st_mtim.tv_sec > mtime.tv_sec ||
613			    (sb.st_mtim.tv_sec == mtime.tv_sec &&
614			    sb.st_mtim.tv_nsec >= mtime.tv_nsec)))
615				return;
616			(void)unlink(*path);
617		} else if (o_opt) {
618			/* overwrite */
619			(void)unlink(*path);
620		} else if (n_opt) {
621			/* do not overwrite */
622			return;
623		} else {
624			check = handle_existing_file(path);
625			if (check == 0)
626				goto recheck;
627			if (check == -1)
628				return; /* do not overwrite */
629		}
630	} else {
631		if (f_opt)
632			return;
633	}
634
635	ts[0].tv_sec = 0;
636	ts[0].tv_nsec = UTIME_NOW;
637	ts[1] = mtime;
638
639	/* process symlinks */
640	linkname = archive_entry_symlink(e);
641	if (linkname != NULL) {
642		if (symlink(linkname, *path) != 0)
643			error("symlink('%s')", *path);
644		info(" extracting: %s -> %s\n", *path, linkname);
645		if (lchmod(*path, mode) != 0)
646			warning("Cannot set mode for '%s'", *path);
647		/* set access and modification time */
648		if (utimensat(AT_FDCWD, *path, ts, AT_SYMLINK_NOFOLLOW) != 0)
649			warning("utimensat('%s')", *path);
650		return;
651	}
652
653	if ((fd = open(*path, O_RDWR|O_CREAT|O_TRUNC, mode)) < 0)
654		error("open('%s')", *path);
655
656	info(" extracting: %s", *path);
657
658	text = extract2fd(a, *path, fd);
659
660	if (tty)
661		info("  \b\b");
662	if (text)
663		info(" (text)");
664	info("\n");
665
666	/* set access and modification time */
667	if (futimens(fd, ts) != 0)
668		error("futimens('%s')", *path);
669	if (close(fd) != 0)
670		error("close('%s')", *path);
671}
672
673/*
674 * Extract a zipfile entry: first perform some sanity checks to ensure
675 * that it is either a directory or a regular file and that the path is
676 * not absolute and does not try to break out of the current directory;
677 * then call either extract_dir() or extract_file() as appropriate.
678 *
679 * This is complicated a bit by the various ways in which we need to
680 * manipulate the path name.  Case conversion (if requested by the -L
681 * option) happens first, but the include / exclude patterns are applied
682 * to the full converted path name, before the directory part of the path
683 * is removed in accordance with the -j option.  Sanity checks are
684 * intentionally done earlier than they need to be, so the user will get a
685 * warning about insecure paths even for files or directories which
686 * wouldn't be extracted anyway.
687 */
688static void
689extract(struct archive *a, struct archive_entry *e)
690{
691	char *pathname, *realpathname;
692	mode_t filetype;
693	char *p, *q;
694
695	pathname = pathdup(archive_entry_pathname(e));
696	filetype = archive_entry_filetype(e);
697
698	/* sanity checks */
699	if (pathname[0] == '/' ||
700	    strncmp(pathname, "../", 3) == 0 ||
701	    strstr(pathname, "/../") != NULL) {
702		warningx("skipping insecure entry '%s'", pathname);
703		ac(archive_read_data_skip(a));
704		free(pathname);
705		return;
706	}
707
708	/* I don't think this can happen in a zipfile.. */
709	if (!S_ISDIR(filetype) && !S_ISREG(filetype) && !S_ISLNK(filetype)) {
710		warningx("skipping non-regular entry '%s'", pathname);
711		ac(archive_read_data_skip(a));
712		free(pathname);
713		return;
714	}
715
716	/* skip directories in -j case */
717	if (S_ISDIR(filetype) && j_opt) {
718		ac(archive_read_data_skip(a));
719		free(pathname);
720		return;
721	}
722
723	/* apply include / exclude patterns */
724	if (!accept_pathname(pathname)) {
725		ac(archive_read_data_skip(a));
726		free(pathname);
727		return;
728	}
729
730	/* apply -j and -d */
731	if (j_opt) {
732		for (p = q = pathname; *p; ++p)
733			if (*p == '/')
734				q = p + 1;
735		realpathname = pathcat(d_arg, q);
736	} else {
737		realpathname = pathcat(d_arg, pathname);
738	}
739
740	/* ensure that parent directory exists */
741	make_parent(realpathname);
742
743	if (S_ISDIR(filetype))
744		extract_dir(a, e, realpathname);
745	else
746		extract_file(a, e, &realpathname);
747
748	free(realpathname);
749	free(pathname);
750}
751
752static void
753extract_stdout(struct archive *a, struct archive_entry *e)
754{
755	char *pathname;
756	mode_t filetype;
757
758	pathname = pathdup(archive_entry_pathname(e));
759	filetype = archive_entry_filetype(e);
760
761	/* I don't think this can happen in a zipfile.. */
762	if (!S_ISDIR(filetype) && !S_ISREG(filetype) && !S_ISLNK(filetype)) {
763		warningx("skipping non-regular entry '%s'", pathname);
764		ac(archive_read_data_skip(a));
765		free(pathname);
766		return;
767	}
768
769	/* skip directories in -j case */
770	if (S_ISDIR(filetype)) {
771		ac(archive_read_data_skip(a));
772		free(pathname);
773		return;
774	}
775
776	/* apply include / exclude patterns */
777	if (!accept_pathname(pathname)) {
778		ac(archive_read_data_skip(a));
779		free(pathname);
780		return;
781	}
782
783	if (c_opt)
784		info("x %s\n", pathname);
785
786	(void)extract2fd(a, pathname, STDOUT_FILENO);
787
788	free(pathname);
789}
790
791/*
792 * Print the name of an entry to stdout.
793 */
794static void
795list(struct archive *a, struct archive_entry *e)
796{
797	char buf[20];
798	time_t mtime;
799	struct tm *tm;
800
801	mtime = archive_entry_mtime(e);
802	tm = localtime(&mtime);
803	if (*y_str)
804		strftime(buf, sizeof(buf), "%m-%d-%G %R", tm);
805	else
806		strftime(buf, sizeof(buf), "%m-%d-%g %R", tm);
807
808	if (!zipinfo_mode) {
809		if (v_opt == 1) {
810			printf(" %8ju  %s   %s\n",
811			    (uintmax_t)archive_entry_size(e),
812			    buf, archive_entry_pathname(e));
813		} else if (v_opt == 2) {
814			printf("%8ju  Stored  %7ju   0%%  %s  %08x  %s\n",
815			    (uintmax_t)archive_entry_size(e),
816			    (uintmax_t)archive_entry_size(e),
817			    buf,
818			    0U,
819			    archive_entry_pathname(e));
820		}
821	} else {
822		if (Z1_opt)
823			printf("%s\n",archive_entry_pathname(e));
824	}
825	ac(archive_read_data_skip(a));
826}
827
828/*
829 * Extract to memory to check CRC
830 */
831static int
832test(struct archive *a, struct archive_entry *e)
833{
834	ssize_t len;
835	int error_count;
836
837	error_count = 0;
838	if (S_ISDIR(archive_entry_filetype(e)))
839		return 0;
840
841	info("    testing: %s\t", archive_entry_pathname(e));
842	while ((len = archive_read_data(a, buffer, sizeof buffer)) > 0)
843		/* nothing */;
844	if (len < 0) {
845		info(" %s\n", archive_error_string(a));
846		++error_count;
847	} else {
848		info(" OK\n");
849	}
850
851	/* shouldn't be necessary, but it doesn't hurt */
852	ac(archive_read_data_skip(a));
853
854	return error_count;
855}
856
857/*
858 * Main loop: open the zipfile, iterate over its contents and decide what
859 * to do with each entry.
860 */
861static void
862unzip(const char *fn)
863{
864	struct archive *a;
865	struct archive_entry *e;
866	int ret;
867	uintmax_t total_size, file_count, error_count;
868
869	if ((a = archive_read_new()) == NULL)
870		error("archive_read_new failed");
871
872	ac(archive_read_support_format_zip(a));
873	ac(archive_read_open_filename(a, fn, 8192));
874
875	if (!zipinfo_mode) {
876		if (!p_opt && !q_opt)
877			printf("Archive:  %s\n", fn);
878		if (v_opt == 1) {
879			printf("  Length     %sDate   Time    Name\n", y_str);
880			printf(" --------    %s----   ----    ----\n", y_str);
881		} else if (v_opt == 2) {
882			printf(" Length   Method    Size  Ratio   %sDate   Time   CRC-32    Name\n", y_str);
883			printf("--------  ------  ------- -----   %s----   ----   ------    ----\n", y_str);
884		}
885	}
886
887	total_size = 0;
888	file_count = 0;
889	error_count = 0;
890	for (;;) {
891		ret = archive_read_next_header(a, &e);
892		if (ret == ARCHIVE_EOF)
893			break;
894		ac(ret);
895		if (!zipinfo_mode) {
896			if (t_opt)
897				error_count += test(a, e);
898			else if (v_opt)
899				list(a, e);
900			else if (p_opt || c_opt)
901				extract_stdout(a, e);
902			else
903				extract(a, e);
904		} else {
905			if (Z1_opt)
906				list(a, e);
907		}
908
909		total_size += archive_entry_size(e);
910		++file_count;
911	}
912
913	if (zipinfo_mode) {
914		if (v_opt == 1) {
915			printf(" --------                   %s-------\n", y_str);
916			printf(" %8ju                   %s%ju file%s\n",
917			    total_size, y_str, file_count, file_count != 1 ? "s" : "");
918		} else if (v_opt == 2) {
919			printf("--------          -------  ---                            %s-------\n", y_str);
920			printf("%8ju          %7ju   0%%                            %s%ju file%s\n",
921			    total_size, total_size, y_str, file_count,
922			    file_count != 1 ? "s" : "");
923		}
924	}
925
926	ac(archive_read_free(a));
927
928	if (t_opt) {
929		if (error_count > 0) {
930			errorx("%ju checksum error(s) found.", error_count);
931		}
932		else {
933			printf("No errors detected in compressed data of %s.\n",
934			       fn);
935		}
936	}
937}
938
939static void
940usage(void)
941{
942
943	fprintf(stderr, "Usage: unzip [-aCcfjLlnopqtuvyZ1] [-d dir] [-x pattern] "
944		"zipfile\n");
945	exit(1);
946}
947
948static int
949getopts(int argc, char *argv[])
950{
951	int opt;
952
953	optreset = optind = 1;
954	while ((opt = getopt(argc, argv, "aCcd:fjLlnopqtuvx:yZ1")) != -1)
955		switch (opt) {
956		case '1':
957			Z1_opt = 1;
958			break;
959		case 'a':
960			a_opt = 1;
961			break;
962		case 'C':
963			C_opt = 1;
964			break;
965		case 'c':
966			c_opt = 1;
967			break;
968		case 'd':
969			d_arg = optarg;
970			break;
971		case 'f':
972			f_opt = 1;
973			break;
974		case 'j':
975			j_opt = 1;
976			break;
977		case 'L':
978			L_opt = 1;
979			break;
980		case 'l':
981			if (v_opt == 0)
982				v_opt = 1;
983			break;
984		case 'n':
985			n_opt = 1;
986			break;
987		case 'o':
988			o_opt = 1;
989			q_opt = 1;
990			break;
991		case 'p':
992			p_opt = 1;
993			break;
994		case 'q':
995			q_opt = 1;
996			break;
997		case 't':
998			t_opt = 1;
999			break;
1000		case 'u':
1001			u_opt = 1;
1002			break;
1003		case 'v':
1004			v_opt = 2;
1005			break;
1006		case 'x':
1007			add_pattern(&exclude, optarg);
1008			break;
1009		case 'y':
1010			y_str = "  ";
1011			break;
1012		case 'Z':
1013			zipinfo_mode = 1;
1014			break;
1015		default:
1016			usage();
1017		}
1018
1019	return (optind);
1020}
1021
1022int
1023main(int argc, char *argv[])
1024{
1025	const char *zipfile;
1026	int nopts;
1027
1028	if (isatty(STDOUT_FILENO))
1029		tty = 1;
1030
1031	if (getenv("UNZIP_DEBUG") != NULL)
1032		unzip_debug = 1;
1033	for (int i = 0; i < argc; ++i)
1034		debug("%s%c", argv[i], (i < argc - 1) ? ' ' : '\n');
1035
1036	/*
1037	 * Info-ZIP's unzip(1) expects certain options to come before the
1038	 * zipfile name, and others to come after - though it does not
1039	 * enforce this.  For simplicity, we accept *all* options both
1040	 * before and after the zipfile name.
1041	 */
1042	nopts = getopts(argc, argv);
1043
1044	/*
1045	 * When more of the zipinfo mode options are implemented, this
1046	 * will need to change.
1047	 */
1048	if (zipinfo_mode && !Z1_opt) {
1049		printf("Zipinfo mode needs additional options\n");
1050		exit(1);
1051	}
1052
1053	if (argc <= nopts)
1054		usage();
1055	zipfile = argv[nopts++];
1056
1057	if (strcmp(zipfile, "-") == 0)
1058		zipfile = NULL; /* STDIN */
1059
1060	while (nopts < argc && *argv[nopts] != '-')
1061		add_pattern(&include, argv[nopts++]);
1062
1063	nopts--; /* fake argv[0] */
1064	nopts += getopts(argc - nopts, argv + nopts);
1065
1066	if (n_opt + o_opt + u_opt > 1)
1067		errorx("-n, -o and -u are contradictory");
1068
1069	unzip(zipfile);
1070
1071	exit(0);
1072}
1073