xref: /minix3/usr.bin/unzip/unzip.c (revision 40b23ce476b0e85e5ffcf4697b70823258d11d2c)
1 /* $NetBSD: unzip.c,v 1.19 2011/09/06 18:43:41 joerg Exp $ */
2 
3 /*-
4  * Copyright (c) 2009, 2010 Joerg Sonnenberger <joerg@NetBSD.org>
5  * Copyright (c) 2007-2008 Dag-Erling Co�dan 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  * $FreeBSD: revision 180124$
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/cdefs.h>
40 __RCSID("$NetBSD: unzip.c,v 1.19 2011/09/06 18:43:41 joerg Exp $");
41 
42 #include <sys/queue.h>
43 #include <sys/stat.h>
44 
45 #include <ctype.h>
46 #include <errno.h>
47 #include <fcntl.h>
48 #include <fnmatch.h>
49 #include <stdarg.h>
50 #include <stdio.h>
51 #include <stdlib.h>
52 #include <string.h>
53 #include <unistd.h>
54 
55 #include <archive.h>
56 #include <archive_entry.h>
57 
58 /* command-line options */
59 static int		 a_opt;		/* convert EOL */
60 static int		 C_opt;		/* match case-insensitively */
61 static int		 c_opt;		/* extract to stdout */
62 static const char	*d_arg;		/* directory */
63 static int		 f_opt;		/* update existing files only */
64 static int		 j_opt;		/* junk directories */
65 static int		 L_opt;		/* lowercase names */
66 static int		 n_opt;		/* never overwrite */
67 static int		 o_opt;		/* always overwrite */
68 static int		 p_opt;		/* extract to stdout, quiet */
69 static int		 q_opt;		/* quiet */
70 static int		 t_opt;		/* test */
71 static int		 u_opt;		/* update */
72 static int		 v_opt;		/* verbose/list */
73 static const char *	 y_str = "";	/* 4 digit year */
74 
75 /* time when unzip started */
76 static time_t		 now;
77 
78 /* debug flag */
79 static int		 unzip_debug;
80 
81 /* running on tty? */
82 static int		 tty;
83 
84 /* convenience macro */
85 /* XXX should differentiate between ARCHIVE_{WARN,FAIL,RETRY} */
86 #define ac(call)						\
87 	do {							\
88 		int acret = (call);				\
89 		if (acret != ARCHIVE_OK)			\
90 			errorx("%s", archive_error_string(a));	\
91 	} while (0)
92 
93 /*
94  * Indicates that last info() did not end with EOL.  This helps error() et
95  * al. avoid printing an error message on the same line as an incomplete
96  * informational message.
97  */
98 static int noeol;
99 
100 /* fatal error message + errno */
101 __dead __printflike(1, 2) static void
error(const char * fmt,...)102 error(const char *fmt, ...)
103 {
104 	va_list ap;
105 
106 	if (noeol)
107 		fprintf(stdout, "\n");
108 	fflush(stdout);
109 	fprintf(stderr, "unzip: ");
110 	va_start(ap, fmt);
111 	vfprintf(stderr, fmt, ap);
112 	va_end(ap);
113 	fprintf(stderr, ": %s\n", strerror(errno));
114 	exit(1);
115 }
116 
117 /* fatal error message, no errno */
118 __dead __printflike(1, 2) static void
errorx(const char * fmt,...)119 errorx(const char *fmt, ...)
120 {
121 	va_list ap;
122 
123 	if (noeol)
124 		fprintf(stdout, "\n");
125 	fflush(stdout);
126 	fprintf(stderr, "unzip: ");
127 	va_start(ap, fmt);
128 	vfprintf(stderr, fmt, ap);
129 	va_end(ap);
130 	fprintf(stderr, "\n");
131 	exit(1);
132 }
133 
134 #if 0
135 /* non-fatal error message + errno */
136 __printflike(1, 2) static void
137 warning(const char *fmt, ...)
138 {
139 	va_list ap;
140 
141 	if (noeol)
142 		fprintf(stdout, "\n");
143 	fflush(stdout);
144 	fprintf(stderr, "unzip: ");
145 	va_start(ap, fmt);
146 	vfprintf(stderr, fmt, ap);
147 	va_end(ap);
148 	fprintf(stderr, ": %s\n", strerror(errno));
149 }
150 #endif
151 /* non-fatal error message, no errno */
152 __printflike(1, 2) static void
warningx(const char * fmt,...)153 warningx(const char *fmt, ...)
154 {
155 	va_list ap;
156 
157 	if (noeol)
158 		fprintf(stdout, "\n");
159 	fflush(stdout);
160 	fprintf(stderr, "unzip: ");
161 	va_start(ap, fmt);
162 	vfprintf(stderr, fmt, ap);
163 	va_end(ap);
164 	fprintf(stderr, "\n");
165 }
166 
167 /* informational message (if not -q) */
168 __printflike(1, 2) static void
info(const char * fmt,...)169 info(const char *fmt, ...)
170 {
171 	va_list ap;
172 
173 	if (q_opt && !unzip_debug)
174 		return;
175 	va_start(ap, fmt);
176 	vfprintf(stdout, fmt, ap);
177 	va_end(ap);
178 	fflush(stdout);
179 
180 	if (*fmt == '\0')
181 		noeol = 1;
182 	else
183 		noeol = fmt[strlen(fmt) - 1] != '\n';
184 }
185 
186 /* debug message (if unzip_debug) */
187 __printflike(1, 2) static void
debug(const char * fmt,...)188 debug(const char *fmt, ...)
189 {
190 	va_list ap;
191 
192 	if (!unzip_debug)
193 		return;
194 	va_start(ap, fmt);
195 	vfprintf(stderr, fmt, ap);
196 	va_end(ap);
197 	fflush(stderr);
198 
199 	if (*fmt == '\0')
200 		noeol = 1;
201 	else
202 		noeol = fmt[strlen(fmt) - 1] != '\n';
203 }
204 
205 /* duplicate a path name, possibly converting to lower case */
206 static char *
pathdup(const char * path)207 pathdup(const char *path)
208 {
209 	char *str;
210 	size_t i, len;
211 
212 	len = strlen(path);
213 	while (len && path[len - 1] == '/')
214 		len--;
215 	if ((str = malloc(len + 1)) == NULL) {
216 		errno = ENOMEM;
217 		error("malloc()");
218 	}
219 	if (L_opt) {
220 		for (i = 0; i < len; ++i)
221 			str[i] = tolower((unsigned char)path[i]);
222 	} else {
223 		memcpy(str, path, len);
224 	}
225 	str[len] = '\0';
226 
227 	return (str);
228 }
229 
230 /* concatenate two path names */
231 static char *
pathcat(const char * prefix,const char * path)232 pathcat(const char *prefix, const char *path)
233 {
234 	char *str;
235 	size_t prelen, len;
236 
237 	prelen = prefix ? strlen(prefix) + 1 : 0;
238 	len = strlen(path) + 1;
239 	if ((str = malloc(prelen + len)) == NULL) {
240 		errno = ENOMEM;
241 		error("malloc()");
242 	}
243 	if (prefix) {
244 		memcpy(str, prefix, prelen);	/* includes zero */
245 		str[prelen - 1] = '/';		/* splat zero */
246 	}
247 	memcpy(str + prelen, path, len);	/* includes zero */
248 
249 	return (str);
250 }
251 
252 /*
253  * Pattern lists for include / exclude processing
254  */
255 struct pattern {
256 	STAILQ_ENTRY(pattern) link;
257 	char pattern[];
258 };
259 
260 STAILQ_HEAD(pattern_list, pattern);
261 static struct pattern_list include = STAILQ_HEAD_INITIALIZER(include);
262 static struct pattern_list exclude = STAILQ_HEAD_INITIALIZER(exclude);
263 
264 /*
265  * Add an entry to a pattern list
266  */
267 static void
add_pattern(struct pattern_list * list,const char * pattern)268 add_pattern(struct pattern_list *list, const char *pattern)
269 {
270 	struct pattern *entry;
271 	size_t len;
272 
273 	debug("adding pattern '%s'\n", pattern);
274 	len = strlen(pattern);
275 	if ((entry = malloc(sizeof *entry + len + 1)) == NULL) {
276 		errno = ENOMEM;
277 		error("malloc()");
278 	}
279 	memcpy(entry->pattern, pattern, len + 1);
280 	STAILQ_INSERT_TAIL(list, entry, link);
281 }
282 
283 /*
284  * Match a string against a list of patterns
285  */
286 static int
match_pattern(struct pattern_list * list,const char * str)287 match_pattern(struct pattern_list *list, const char *str)
288 {
289 	struct pattern *entry;
290 
291 	STAILQ_FOREACH(entry, list, link) {
292 		if (fnmatch(entry->pattern, str, C_opt ? FNM_CASEFOLD : 0) == 0)
293 			return (1);
294 	}
295 	return (0);
296 }
297 
298 /*
299  * Verify that a given pathname is in the include list and not in the
300  * exclude list.
301  */
302 static int
accept_pathname(const char * pathname)303 accept_pathname(const char *pathname)
304 {
305 
306 	if (!STAILQ_EMPTY(&include) && !match_pattern(&include, pathname))
307 		return (0);
308 	if (!STAILQ_EMPTY(&exclude) && match_pattern(&exclude, pathname))
309 		return (0);
310 	return (1);
311 }
312 
313 /*
314  * Create the specified directory with the specified mode, taking certain
315  * precautions on they way.
316  */
317 static void
make_dir(const char * path,int mode)318 make_dir(const char *path, int mode)
319 {
320 	struct stat sb;
321 
322 	if (lstat(path, &sb) == 0) {
323 		if (S_ISDIR(sb.st_mode))
324 			return;
325 		/*
326 		 * Normally, we should either ask the user about removing
327 		 * the non-directory of the same name as a directory we
328 		 * wish to create, or respect the -n or -o command-line
329 		 * options.  However, this may lead to a later failure or
330 		 * even compromise (if this non-directory happens to be a
331 		 * symlink to somewhere unsafe), so we don't.
332 		 */
333 
334 		/*
335 		 * Don't check unlink() result; failure will cause mkdir()
336 		 * to fail later, which we will catch.
337 		 */
338 		(void)unlink(path);
339 	}
340 	if (mkdir(path, mode) != 0 && errno != EEXIST)
341 		error("mkdir('%s')", path);
342 }
343 
344 /*
345  * Ensure that all directories leading up to (but not including) the
346  * specified path exist.
347  *
348  * XXX inefficient + modifies the file in-place
349  */
350 static void
make_parent(char * path)351 make_parent(char *path)
352 {
353 	struct stat sb;
354 	char *sep;
355 
356 	sep = strrchr(path, '/');
357 	if (sep == NULL || sep == path)
358 		return;
359 	*sep = '\0';
360 	if (lstat(path, &sb) == 0) {
361 		if (S_ISDIR(sb.st_mode)) {
362 			*sep = '/';
363 			return;
364 		}
365 		unlink(path);
366 	}
367 	make_parent(path);
368 	mkdir(path, 0755);
369 	*sep = '/';
370 
371 #if 0
372 	for (sep = path; (sep = strchr(sep, '/')) != NULL; sep++) {
373 		/* root in case of absolute d_arg */
374 		if (sep == path)
375 			continue;
376 		*sep = '\0';
377 		make_dir(path, 0755);
378 		*sep = '/';
379 	}
380 #endif
381 }
382 
383 /*
384  * Extract a directory.
385  */
386 static void
extract_dir(struct archive * a,struct archive_entry * e,const char * path)387 extract_dir(struct archive *a, struct archive_entry *e, const char *path)
388 {
389 	int mode;
390 
391 	mode = archive_entry_mode(e) & 0777;
392 	if (mode == 0)
393 		mode = 0755;
394 
395 	/*
396 	 * Some zipfiles contain directories with weird permissions such
397 	 * as 0644 or 0444.  This can cause strange issues such as being
398 	 * unable to extract files into the directory we just created, or
399 	 * the user being unable to remove the directory later without
400 	 * first manually changing its permissions.  Therefore, we whack
401 	 * the permissions into shape, assuming that the user wants full
402 	 * access and that anyone who gets read access also gets execute
403 	 * access.
404 	 */
405 	mode |= 0700;
406 	if (mode & 0040)
407 		mode |= 0010;
408 	if (mode & 0004)
409 		mode |= 0001;
410 
411 	info("   creating: %s/\n", path);
412 	make_dir(path, mode);
413 	ac(archive_read_data_skip(a));
414 }
415 
416 static unsigned char buffer[8192];
417 static char spinner[] = { '|', '/', '-', '\\' };
418 
419 static int
handle_existing_file(char ** path)420 handle_existing_file(char **path)
421 {
422 	size_t alen;
423 	ssize_t len;
424 	char buf[4];
425 
426 	for (;;) {
427 		fprintf(stderr,
428 		    "replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ",
429 		    *path);
430 		fgets(buf, 4, stdin);
431 		switch (*buf) {
432 		case 'A':
433 			o_opt = 1;
434 			/* FALL THROUGH */
435 		case 'y':
436 		case 'Y':
437 			(void)unlink(*path);
438 			return 1;
439 		case 'N':
440 			n_opt = 1;
441 			/* FALL THROUGH */
442 		case 'n':
443 			return -1;
444 		case 'r':
445 		case 'R':
446 			printf("New name: ");
447 			fflush(stdout);
448 			free(*path);
449 			*path = NULL;
450 			alen = 0;
451 			len = getline(path, &alen, stdin);
452 			if ((*path)[len - 1] == '\n')
453 				(*path)[len - 1] = '\0';
454 			return 0;
455 		default:
456 			break;
457 		}
458 	}
459 }
460 
461 /*
462  * Detect binary files by a combination of character white list and
463  * black list. NUL bytes and other control codes without use in text files
464  * result directly in switching the file to binary mode. Otherwise, at least
465  * one white-listed byte has to be found.
466  *
467  * Black-listed: 0..6, 14..25, 28..31
468  * White-listed: 9..10, 13, >= 32
469  *
470  * See the proginfo/txtvsbin.txt in the zip sources for a detailed discussion.
471  */
472 #define BYTE_IS_BINARY(x)	((x) < 32 && (0xf3ffc07fU & (1U << (x))))
473 #define	BYTE_IS_TEXT(x)		((x) >= 32 || (0x00002600U & (1U << (x))))
474 
475 static int
check_binary(const unsigned char * buf,size_t len)476 check_binary(const unsigned char *buf, size_t len)
477 {
478 	int rv;
479 	for (rv = 1; len--; ++buf) {
480 		if (BYTE_IS_BINARY(*buf))
481 			return 1;
482 		if (BYTE_IS_TEXT(*buf))
483 			rv = 0;
484 	}
485 
486 	return rv;
487 }
488 
489 /*
490  * Extract a regular file.
491  */
492 static void
extract_file(struct archive * a,struct archive_entry * e,char ** path)493 extract_file(struct archive *a, struct archive_entry *e, char **path)
494 {
495 	int mode;
496 	time_t mtime;
497 	struct stat sb;
498 	struct timeval tv[2];
499 	int cr, fd, text, warn, check;
500 	ssize_t len;
501 	unsigned char *p, *q, *end;
502 
503 	mode = archive_entry_mode(e) & 0777;
504 	if (mode == 0)
505 		mode = 0644;
506 	mtime = archive_entry_mtime(e);
507 
508 	/* look for existing file of same name */
509 recheck:
510 	if (lstat(*path, &sb) == 0) {
511 		if (u_opt || f_opt) {
512 			/* check if up-to-date */
513 			if (S_ISREG(sb.st_mode) && sb.st_mtime >= mtime)
514 				return;
515 			(void)unlink(*path);
516 		} else if (o_opt) {
517 			/* overwrite */
518 			(void)unlink(*path);
519 		} else if (n_opt) {
520 			/* do not overwrite */
521 			return;
522 		} else {
523 			check = handle_existing_file(path);
524 			if (check == 0)
525 				goto recheck;
526 			if (check == -1)
527 				return; /* do not overwrite */
528 		}
529 	} else {
530 		if (f_opt)
531 			return;
532 	}
533 
534 	if ((fd = open(*path, O_RDWR|O_CREAT|O_TRUNC, mode)) < 0)
535 		error("open('%s')", *path);
536 
537 	/* loop over file contents and write to disk */
538 	info(" extracting: %s", *path);
539 	text = a_opt;
540 	warn = 0;
541 	cr = 0;
542 	for (int n = 0; ; n++) {
543 		if (tty && (n % 4) == 0)
544 			info(" %c\b\b", spinner[(n / 4) % sizeof spinner]);
545 
546 		len = archive_read_data(a, buffer, sizeof buffer);
547 
548 		if (len < 0)
549 			ac(len);
550 
551 		/* left over CR from previous buffer */
552 		if (a_opt && cr) {
553 			if (len == 0 || buffer[0] != '\n')
554 				if (write(fd, "\r", 1) != 1)
555 					error("write('%s')", *path);
556 			cr = 0;
557 		}
558 
559 		/* EOF */
560 		if (len == 0)
561 			break;
562 		end = buffer + len;
563 
564 		/*
565 		 * Detect whether this is a text file.  The correct way to
566 		 * do this is to check the least significant bit of the
567 		 * "internal file attributes" field of the corresponding
568 		 * file header in the central directory, but libarchive
569 		 * does not read the central directory, so we have to
570 		 * guess by looking for non-ASCII characters in the
571 		 * buffer.  Hopefully we won't guess wrong.  If we do
572 		 * guess wrong, we print a warning message later.
573 		 */
574 		if (a_opt && n == 0) {
575 			if (check_binary(buffer, len))
576 				text = 0;
577 		}
578 
579 		/* simple case */
580 		if (!a_opt || !text) {
581 			if (write(fd, buffer, len) != len)
582 				error("write('%s')", *path);
583 			continue;
584 		}
585 
586 		/* hard case: convert \r\n to \n (sigh...) */
587 		for (p = buffer; p < end; p = q + 1) {
588 			for (q = p; q < end; q++) {
589 				if (!warn && BYTE_IS_BINARY(*q)) {
590 					warningx("%s may be corrupted due"
591 					    " to weak text file detection"
592 					    " heuristic", *path);
593 					warn = 1;
594 				}
595 				if (q[0] != '\r')
596 					continue;
597 				if (&q[1] == end) {
598 					cr = 1;
599 					break;
600 				}
601 				if (q[1] == '\n')
602 					break;
603 			}
604 			if (write(fd, p, q - p) != q - p)
605 				error("write('%s')", *path);
606 		}
607 	}
608 	if (tty)
609 		info("  \b\b");
610 	if (text)
611 		info(" (text)");
612 	info("\n");
613 
614 	/* set access and modification time */
615 	tv[0].tv_sec = now;
616 	tv[0].tv_usec = 0;
617 	tv[1].tv_sec = mtime;
618 	tv[1].tv_usec = 0;
619 	if (futimes(fd, tv) != 0)
620 		error("utimes('%s')", *path);
621 	if (close(fd) != 0)
622 		error("close('%s')", *path);
623 }
624 
625 /*
626  * Extract a zipfile entry: first perform some sanity checks to ensure
627  * that it is either a directory or a regular file and that the path is
628  * not absolute and does not try to break out of the current directory;
629  * then call either extract_dir() or extract_file() as appropriate.
630  *
631  * This is complicated a bit by the various ways in which we need to
632  * manipulate the path name.  Case conversion (if requested by the -L
633  * option) happens first, but the include / exclude patterns are applied
634  * to the full converted path name, before the directory part of the path
635  * is removed in accordance with the -j option.  Sanity checks are
636  * intentionally done earlier than they need to be, so the user will get a
637  * warning about insecure paths even for files or directories which
638  * wouldn't be extracted anyway.
639  */
640 static void
extract(struct archive * a,struct archive_entry * e)641 extract(struct archive *a, struct archive_entry *e)
642 {
643 	char *pathname, *realpathname;
644 	mode_t filetype;
645 	char *p, *q;
646 
647 	pathname = pathdup(archive_entry_pathname(e));
648 	filetype = archive_entry_filetype(e);
649 
650 	/* sanity checks */
651 	if (pathname[0] == '/' ||
652 	    strncmp(pathname, "../", 3) == 0 ||
653 	    strstr(pathname, "/../") != NULL) {
654 		warningx("skipping insecure entry '%s'", pathname);
655 		ac(archive_read_data_skip(a));
656 		free(pathname);
657 		return;
658 	}
659 
660 	/* I don't think this can happen in a zipfile.. */
661 	if (!S_ISDIR(filetype) && !S_ISREG(filetype)) {
662 		warningx("skipping non-regular entry '%s'", pathname);
663 		ac(archive_read_data_skip(a));
664 		free(pathname);
665 		return;
666 	}
667 
668 	/* skip directories in -j case */
669 	if (S_ISDIR(filetype) && j_opt) {
670 		ac(archive_read_data_skip(a));
671 		free(pathname);
672 		return;
673 	}
674 
675 	/* apply include / exclude patterns */
676 	if (!accept_pathname(pathname)) {
677 		ac(archive_read_data_skip(a));
678 		free(pathname);
679 		return;
680 	}
681 
682 	/* apply -j and -d */
683 	if (j_opt) {
684 		for (p = q = pathname; *p; ++p)
685 			if (*p == '/')
686 				q = p + 1;
687 		realpathname = pathcat(d_arg, q);
688 	} else {
689 		realpathname = pathcat(d_arg, pathname);
690 	}
691 
692 	/* ensure that parent directory exists */
693 	make_parent(realpathname);
694 
695 	if (S_ISDIR(filetype))
696 		extract_dir(a, e, realpathname);
697 	else
698 		extract_file(a, e, &realpathname);
699 
700 	free(realpathname);
701 	free(pathname);
702 }
703 
704 static void
extract_stdout(struct archive * a,struct archive_entry * e)705 extract_stdout(struct archive *a, struct archive_entry *e)
706 {
707 	char *pathname;
708 	mode_t filetype;
709 	int cr, text, warn;
710 	ssize_t len;
711 	unsigned char *p, *q, *end;
712 
713 	pathname = pathdup(archive_entry_pathname(e));
714 	filetype = archive_entry_filetype(e);
715 
716 	/* I don't think this can happen in a zipfile.. */
717 	if (!S_ISDIR(filetype) && !S_ISREG(filetype)) {
718 		warningx("skipping non-regular entry '%s'", pathname);
719 		ac(archive_read_data_skip(a));
720 		free(pathname);
721 		return;
722 	}
723 
724 	/* skip directories in -j case */
725 	if (S_ISDIR(filetype)) {
726 		ac(archive_read_data_skip(a));
727 		free(pathname);
728 		return;
729 	}
730 
731 	/* apply include / exclude patterns */
732 	if (!accept_pathname(pathname)) {
733 		ac(archive_read_data_skip(a));
734 		free(pathname);
735 		return;
736 	}
737 
738 	if (c_opt)
739 		info("x %s\n", pathname);
740 
741 	text = a_opt;
742 	warn = 0;
743 	cr = 0;
744 	for (int n = 0; ; n++) {
745 		len = archive_read_data(a, buffer, sizeof buffer);
746 
747 		if (len < 0)
748 			ac(len);
749 
750 		/* left over CR from previous buffer */
751 		if (a_opt && cr) {
752 			if (len == 0 || buffer[0] != '\n') {
753 				if (fwrite("\r", 1, 1, stderr) != 1)
754 					error("write('%s')", pathname);
755 			}
756 			cr = 0;
757 		}
758 
759 		/* EOF */
760 		if (len == 0)
761 			break;
762 		end = buffer + len;
763 
764 		/*
765 		 * Detect whether this is a text file.  The correct way to
766 		 * do this is to check the least significant bit of the
767 		 * "internal file attributes" field of the corresponding
768 		 * file header in the central directory, but libarchive
769 		 * does not read the central directory, so we have to
770 		 * guess by looking for non-ASCII characters in the
771 		 * buffer.  Hopefully we won't guess wrong.  If we do
772 		 * guess wrong, we print a warning message later.
773 		 */
774 		if (a_opt && n == 0) {
775 			for (p = buffer; p < end; ++p) {
776 				if (!isascii((unsigned char)*p)) {
777 					text = 0;
778 					break;
779 				}
780 			}
781 		}
782 
783 		/* simple case */
784 		if (!a_opt || !text) {
785 			if (fwrite(buffer, 1, len, stdout) != (size_t)len)
786 				error("write('%s')", pathname);
787 			continue;
788 		}
789 
790 		/* hard case: convert \r\n to \n (sigh...) */
791 		for (p = buffer; p < end; p = q + 1) {
792 			for (q = p; q < end; q++) {
793 				if (!warn && !isascii(*q)) {
794 					warningx("%s may be corrupted due"
795 					    " to weak text file detection"
796 					    " heuristic", pathname);
797 					warn = 1;
798 				}
799 				if (q[0] != '\r')
800 					continue;
801 				if (&q[1] == end) {
802 					cr = 1;
803 					break;
804 				}
805 				if (q[1] == '\n')
806 					break;
807 			}
808 			if (fwrite(p, 1, q - p, stdout) != (size_t)(q - p))
809 				error("write('%s')", pathname);
810 		}
811 	}
812 
813 	free(pathname);
814 }
815 
816 /*
817  * Print the name of an entry to stdout.
818  */
819 static void
list(struct archive * a,struct archive_entry * e)820 list(struct archive *a, struct archive_entry *e)
821 {
822 	char buf[20];
823 	time_t mtime;
824 	struct tm *tm;
825 
826 	mtime = archive_entry_mtime(e);
827 	tm = localtime(&mtime);
828 	if (*y_str)
829 		strftime(buf, sizeof(buf), "%m-%d-%G %R", tm);
830 	else
831 		strftime(buf, sizeof(buf), "%m-%d-%g %R", tm);
832 
833 	if (v_opt == 1) {
834 		printf(" %8ju  %s   %s\n",
835 		    (uintmax_t)archive_entry_size(e),
836 		    buf, archive_entry_pathname(e));
837 	} else if (v_opt == 2) {
838 		printf("%8ju  Stored  %7ju   0%%  %s  %08x  %s\n",
839 		    (uintmax_t)archive_entry_size(e),
840 		    (uintmax_t)archive_entry_size(e),
841 		    buf,
842 		    0U,
843 		    archive_entry_pathname(e));
844 	}
845 	ac(archive_read_data_skip(a));
846 }
847 
848 /*
849  * Extract to memory to check CRC
850  */
851 static int
test(struct archive * a,struct archive_entry * e)852 test(struct archive *a, struct archive_entry *e)
853 {
854 	ssize_t len;
855 	int error_count;
856 
857 	error_count = 0;
858 	if (S_ISDIR(archive_entry_filetype(e)))
859 		return 0;
860 
861 	info("    testing: %s\t", archive_entry_pathname(e));
862 	while ((len = archive_read_data(a, buffer, sizeof buffer)) > 0)
863 		/* nothing */;
864 	if (len < 0) {
865 		info(" %s\n", archive_error_string(a));
866 		++error_count;
867 	} else {
868 		info(" OK\n");
869 	}
870 
871 	/* shouldn't be necessary, but it doesn't hurt */
872 	ac(archive_read_data_skip(a));
873 
874 	return error_count;
875 }
876 
877 
878 /*
879  * Main loop: open the zipfile, iterate over its contents and decide what
880  * to do with each entry.
881  */
882 static void
unzip(const char * fn)883 unzip(const char *fn)
884 {
885 	struct archive *a;
886 	struct archive_entry *e;
887 	int fd, ret;
888 	uintmax_t total_size, file_count, error_count;
889 
890 	if ((fd = open(fn, O_RDONLY)) < 0)
891 		error("%s", fn);
892 
893 	a = archive_read_new();
894 	ac(archive_read_support_format_zip(a));
895 	ac(archive_read_open_fd(a, fd, 8192));
896 
897 	if (!q_opt && !p_opt)
898 	    printf("Archive:  %s\n", fn);
899 
900 	if (v_opt == 1) {
901 		printf("  Length     %sDate   Time    Name\n", y_str);
902 		printf(" --------    %s----   ----    ----\n", y_str);
903 	} else if (v_opt == 2) {
904 		printf(" Length   Method    Size  Ratio   %sDate   Time   CRC-32    Name\n", y_str);
905 		printf("--------  ------  ------- -----   %s----   ----   ------    ----\n", y_str);
906 	}
907 
908 	total_size = 0;
909 	file_count = 0;
910 	error_count = 0;
911 	for (;;) {
912 		ret = archive_read_next_header(a, &e);
913 		if (ret == ARCHIVE_EOF)
914 			break;
915 		ac(ret);
916 		if (t_opt)
917 			error_count += test(a, e);
918 		else if (v_opt)
919 			list(a, e);
920 		else if (p_opt || c_opt)
921 			extract_stdout(a, e);
922 		else
923 			extract(a, e);
924 
925 		total_size += archive_entry_size(e);
926 		++file_count;
927 	}
928 
929 	if (v_opt == 1) {
930 		printf(" --------                   %s-------\n", y_str);
931 		printf(" %8ju                   %s%ju file%s\n",
932 		    total_size, y_str, file_count, file_count != 1 ? "s" : "");
933 	} else if (v_opt == 2) {
934 		printf("--------          -------  ---                            %s-------\n", y_str);
935 		printf("%8ju          %7ju   0%%                            %s%ju file%s\n",
936 		    total_size, total_size, y_str, file_count,
937 		    file_count != 1 ? "s" : "");
938 	}
939 
940 	ac(archive_read_close(a));
941 	(void)archive_read_finish(a);
942 
943 	if (close(fd) != 0)
944 		error("%s", fn);
945 
946 	if (t_opt) {
947 		if (error_count > 0) {
948 			errorx("%ju checksum error(s) found.", error_count);
949 		}
950 		else {
951 			printf("No errors detected in compressed data of %s.\n",
952 			       fn);
953 		}
954 	}
955 }
956 
957 static void __dead
usage(void)958 usage(void)
959 {
960 
961 	fprintf(stderr, "Usage: %s [-aCcfjLlnopqtuvy] [-d dir] [-x pattern] "
962 	    "zipfile\n", getprogname());
963 	exit(1);
964 }
965 
966 static int
getopts(int argc,char * argv[])967 getopts(int argc, char *argv[])
968 {
969 	int opt;
970 
971 	optreset = optind = 1;
972 	while ((opt = getopt(argc, argv, "aCcd:fjLlnopqtuvyx:")) != -1)
973 		switch (opt) {
974 		case 'a':
975 			a_opt = 1;
976 			break;
977 		case 'C':
978 			C_opt = 1;
979 			break;
980 		case 'c':
981 			c_opt = 1;
982 			break;
983 		case 'd':
984 			d_arg = optarg;
985 			break;
986 		case 'f':
987 			f_opt = 1;
988 			break;
989 		case 'j':
990 			j_opt = 1;
991 			break;
992 		case 'L':
993 			L_opt = 1;
994 			break;
995 		case 'l':
996 			if (v_opt == 0)
997 				v_opt = 1;
998 			break;
999 		case 'n':
1000 			n_opt = 1;
1001 			break;
1002 		case 'o':
1003 			o_opt = 1;
1004 			q_opt = 1;
1005 			break;
1006 		case 'p':
1007 			p_opt = 1;
1008 			break;
1009 		case 'q':
1010 			q_opt = 1;
1011 			break;
1012 		case 't':
1013 			t_opt = 1;
1014 			break;
1015 		case 'u':
1016 			u_opt = 1;
1017 			break;
1018 		case 'v':
1019 			v_opt = 2;
1020 			break;
1021 		case 'x':
1022 			add_pattern(&exclude, optarg);
1023 			break;
1024 		case 'y':
1025 			y_str = "  ";
1026 			break;
1027 		default:
1028 			usage();
1029 		}
1030 
1031 	return (optind);
1032 }
1033 
1034 int
main(int argc,char * argv[])1035 main(int argc, char *argv[])
1036 {
1037 	const char *zipfile;
1038 	int nopts;
1039 
1040 	if (isatty(STDOUT_FILENO))
1041 		tty = 1;
1042 
1043 	if (getenv("UNZIP_DEBUG") != NULL)
1044 		unzip_debug = 1;
1045 	for (int i = 0; i < argc; ++i)
1046 		debug("%s%c", argv[i], (i < argc - 1) ? ' ' : '\n');
1047 
1048 	/*
1049 	 * Info-ZIP's unzip(1) expects certain options to come before the
1050 	 * zipfile name, and others to come after - though it does not
1051 	 * enforce this.  For simplicity, we accept *all* options both
1052 	 * before and after the zipfile name.
1053 	 */
1054 	nopts = getopts(argc, argv);
1055 
1056 	if (argc <= nopts)
1057 		usage();
1058 	zipfile = argv[nopts++];
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 	time(&now);
1070 
1071 	unzip(zipfile);
1072 
1073 	exit(0);
1074 }
1075