unzip.c revision 1.2 1 1.2 joerg /* $NetBSD: unzip.c,v 1.2 2009/08/22 02:19:42 joerg Exp $ */
2 1.1 joerg
3 1.1 joerg /*-
4 1.1 joerg * Copyright (c) 2009 Joerg Sonnenberger <joerg (at) NetBSD.org>
5 1.1 joerg * Copyright (c) 2007-2008 Dag-Erling Codan Smrgrav
6 1.1 joerg * All rights reserved.
7 1.1 joerg *
8 1.1 joerg * Redistribution and use in source and binary forms, with or without
9 1.1 joerg * modification, are permitted provided that the following conditions
10 1.1 joerg * are met:
11 1.1 joerg * 1. Redistributions of source code must retain the above copyright
12 1.1 joerg * notice, this list of conditions and the following disclaimer
13 1.1 joerg * in this position and unchanged.
14 1.1 joerg * 2. Redistributions in binary form must reproduce the above copyright
15 1.1 joerg * notice, this list of conditions and the following disclaimer in the
16 1.1 joerg * documentation and/or other materials provided with the distribution.
17 1.1 joerg *
18 1.1 joerg * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
19 1.1 joerg * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20 1.1 joerg * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 1.1 joerg * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
22 1.1 joerg * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23 1.1 joerg * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24 1.1 joerg * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25 1.1 joerg * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26 1.1 joerg * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27 1.1 joerg * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28 1.1 joerg * SUCH DAMAGE.
29 1.1 joerg *
30 1.1 joerg * $FreeBSD: revision 180124$
31 1.1 joerg *
32 1.1 joerg * This file would be much shorter if we didn't care about command-line
33 1.1 joerg * compatibility with Info-ZIP's UnZip, which requires us to duplicate
34 1.1 joerg * parts of libarchive in order to gain more detailed control of its
35 1.1 joerg * behaviour for the purpose of implementing the -n, -o, -L and -a
36 1.1 joerg * options.
37 1.1 joerg */
38 1.1 joerg
39 1.1 joerg #include <sys/cdefs.h>
40 1.2 joerg __RCSID("$NetBSD: unzip.c,v 1.2 2009/08/22 02:19:42 joerg Exp $");
41 1.1 joerg
42 1.1 joerg #include <sys/queue.h>
43 1.1 joerg #include <sys/stat.h>
44 1.1 joerg
45 1.1 joerg #include <ctype.h>
46 1.1 joerg #include <errno.h>
47 1.1 joerg #include <fcntl.h>
48 1.1 joerg #include <fnmatch.h>
49 1.1 joerg #include <stdarg.h>
50 1.1 joerg #include <stdio.h>
51 1.1 joerg #include <stdlib.h>
52 1.1 joerg #include <string.h>
53 1.1 joerg #include <unistd.h>
54 1.1 joerg
55 1.1 joerg #include <archive.h>
56 1.1 joerg #include <archive_entry.h>
57 1.1 joerg
58 1.1 joerg /* command-line options */
59 1.1 joerg static int a_opt; /* convert EOL */
60 1.1 joerg static const char *d_arg; /* directory */
61 1.1 joerg static int f_opt; /* update existing files only */
62 1.1 joerg static int j_opt; /* junk directories */
63 1.1 joerg static int L_opt; /* lowercase names */
64 1.1 joerg static int n_opt; /* never overwrite */
65 1.1 joerg static int o_opt; /* always overwrite */
66 1.2 joerg static int p_opt; /* extract to stdout */
67 1.1 joerg static int q_opt; /* quiet */
68 1.1 joerg static int t_opt; /* test */
69 1.1 joerg static int u_opt; /* update */
70 1.2 joerg static int v_opt; /* verbose */
71 1.1 joerg
72 1.1 joerg /* time when unzip started */
73 1.1 joerg static time_t now;
74 1.1 joerg
75 1.1 joerg /* debug flag */
76 1.1 joerg static int unzip_debug;
77 1.1 joerg
78 1.1 joerg /* running on tty? */
79 1.1 joerg static int tty;
80 1.1 joerg
81 1.1 joerg /* error flag for -t */
82 1.1 joerg static int test_failed;
83 1.1 joerg
84 1.1 joerg /* convenience macro */
85 1.1 joerg /* XXX should differentiate between ARCHIVE_{WARN,FAIL,RETRY} */
86 1.1 joerg #define ac(call) \
87 1.1 joerg do { \
88 1.1 joerg int acret = (call); \
89 1.1 joerg if (acret != ARCHIVE_OK) \
90 1.1 joerg errorx("%s", archive_error_string(a)); \
91 1.1 joerg } while (0)
92 1.1 joerg
93 1.1 joerg /*
94 1.1 joerg * Indicates that last info() did not end with EOL. This helps error() et
95 1.1 joerg * al. avoid printing an error message on the same line as an incomplete
96 1.1 joerg * informational message.
97 1.1 joerg */
98 1.1 joerg static int noeol;
99 1.1 joerg
100 1.1 joerg /* fatal error message + errno */
101 1.1 joerg static void
102 1.1 joerg error(const char *fmt, ...)
103 1.1 joerg {
104 1.1 joerg va_list ap;
105 1.1 joerg
106 1.1 joerg if (noeol)
107 1.1 joerg fprintf(stdout, "\n");
108 1.1 joerg fflush(stdout);
109 1.1 joerg fprintf(stderr, "unzip: ");
110 1.1 joerg va_start(ap, fmt);
111 1.1 joerg vfprintf(stderr, fmt, ap);
112 1.1 joerg va_end(ap);
113 1.1 joerg fprintf(stderr, ": %s\n", strerror(errno));
114 1.1 joerg exit(1);
115 1.1 joerg }
116 1.1 joerg
117 1.1 joerg /* fatal error message, no errno */
118 1.1 joerg static void
119 1.1 joerg errorx(const char *fmt, ...)
120 1.1 joerg {
121 1.1 joerg va_list ap;
122 1.1 joerg
123 1.1 joerg if (noeol)
124 1.1 joerg fprintf(stdout, "\n");
125 1.1 joerg fflush(stdout);
126 1.1 joerg fprintf(stderr, "unzip: ");
127 1.1 joerg va_start(ap, fmt);
128 1.1 joerg vfprintf(stderr, fmt, ap);
129 1.1 joerg va_end(ap);
130 1.1 joerg fprintf(stderr, "\n");
131 1.1 joerg exit(1);
132 1.1 joerg }
133 1.1 joerg
134 1.1 joerg #if 0
135 1.1 joerg /* non-fatal error message + errno */
136 1.1 joerg static void
137 1.1 joerg warning(const char *fmt, ...)
138 1.1 joerg {
139 1.1 joerg va_list ap;
140 1.1 joerg
141 1.1 joerg if (noeol)
142 1.1 joerg fprintf(stdout, "\n");
143 1.1 joerg fflush(stdout);
144 1.1 joerg fprintf(stderr, "unzip: ");
145 1.1 joerg va_start(ap, fmt);
146 1.1 joerg vfprintf(stderr, fmt, ap);
147 1.1 joerg va_end(ap);
148 1.1 joerg fprintf(stderr, ": %s\n", strerror(errno));
149 1.1 joerg }
150 1.1 joerg #endif
151 1.1 joerg
152 1.1 joerg /* non-fatal error message, no errno */
153 1.1 joerg static void
154 1.1 joerg warningx(const char *fmt, ...)
155 1.1 joerg {
156 1.1 joerg va_list ap;
157 1.1 joerg
158 1.1 joerg if (noeol)
159 1.1 joerg fprintf(stdout, "\n");
160 1.1 joerg fflush(stdout);
161 1.1 joerg fprintf(stderr, "unzip: ");
162 1.1 joerg va_start(ap, fmt);
163 1.1 joerg vfprintf(stderr, fmt, ap);
164 1.1 joerg va_end(ap);
165 1.1 joerg fprintf(stderr, "\n");
166 1.1 joerg }
167 1.1 joerg
168 1.1 joerg /* informational message (if not -q) */
169 1.1 joerg static void
170 1.1 joerg info(const char *fmt, ...)
171 1.1 joerg {
172 1.1 joerg va_list ap;
173 1.1 joerg
174 1.1 joerg if (q_opt && !unzip_debug)
175 1.1 joerg return;
176 1.1 joerg va_start(ap, fmt);
177 1.1 joerg vfprintf(stdout, fmt, ap);
178 1.1 joerg va_end(ap);
179 1.1 joerg fflush(stdout);
180 1.1 joerg
181 1.1 joerg if (*fmt == '\0')
182 1.1 joerg noeol = 1;
183 1.1 joerg else
184 1.1 joerg noeol = fmt[strlen(fmt) - 1] != '\n';
185 1.1 joerg }
186 1.1 joerg
187 1.1 joerg /* debug message (if unzip_debug) */
188 1.1 joerg static void
189 1.1 joerg debug(const char *fmt, ...)
190 1.1 joerg {
191 1.1 joerg va_list ap;
192 1.1 joerg
193 1.1 joerg if (!unzip_debug)
194 1.1 joerg return;
195 1.1 joerg va_start(ap, fmt);
196 1.1 joerg vfprintf(stderr, fmt, ap);
197 1.1 joerg va_end(ap);
198 1.1 joerg fflush(stderr);
199 1.1 joerg
200 1.1 joerg if (*fmt == '\0')
201 1.1 joerg noeol = 1;
202 1.1 joerg else
203 1.1 joerg noeol = fmt[strlen(fmt) - 1] != '\n';
204 1.1 joerg }
205 1.1 joerg
206 1.1 joerg /* duplicate a path name, possibly converting to lower case */
207 1.1 joerg static char *
208 1.1 joerg pathdup(const char *path)
209 1.1 joerg {
210 1.1 joerg char *str;
211 1.1 joerg size_t i, len;
212 1.1 joerg
213 1.1 joerg len = strlen(path);
214 1.1 joerg while (len && path[len - 1] == '/')
215 1.1 joerg len--;
216 1.1 joerg if ((str = malloc(len + 1)) == NULL) {
217 1.1 joerg errno = ENOMEM;
218 1.1 joerg error("malloc()");
219 1.1 joerg }
220 1.1 joerg if (L_opt) {
221 1.1 joerg for (i = 0; i < len; ++i)
222 1.1 joerg str[i] = tolower((unsigned char)path[i]);
223 1.1 joerg } else {
224 1.1 joerg memcpy(str, path, len);
225 1.1 joerg }
226 1.1 joerg str[len] = '\0';
227 1.1 joerg
228 1.1 joerg return (str);
229 1.1 joerg }
230 1.1 joerg
231 1.1 joerg /* concatenate two path names */
232 1.1 joerg static char *
233 1.1 joerg pathcat(const char *prefix, const char *path)
234 1.1 joerg {
235 1.1 joerg char *str;
236 1.1 joerg size_t prelen, len;
237 1.1 joerg
238 1.1 joerg prelen = prefix ? strlen(prefix) + 1 : 0;
239 1.1 joerg len = strlen(path) + 1;
240 1.1 joerg if ((str = malloc(prelen + len)) == NULL) {
241 1.1 joerg errno = ENOMEM;
242 1.1 joerg error("malloc()");
243 1.1 joerg }
244 1.1 joerg if (prefix) {
245 1.1 joerg memcpy(str, prefix, prelen); /* includes zero */
246 1.1 joerg str[prelen - 1] = '/'; /* splat zero */
247 1.1 joerg }
248 1.1 joerg memcpy(str + prelen, path, len); /* includes zero */
249 1.1 joerg
250 1.1 joerg return (str);
251 1.1 joerg }
252 1.1 joerg
253 1.1 joerg /*
254 1.1 joerg * Pattern lists for include / exclude processing
255 1.1 joerg */
256 1.1 joerg struct pattern {
257 1.1 joerg STAILQ_ENTRY(pattern) link;
258 1.1 joerg char pattern[];
259 1.1 joerg };
260 1.1 joerg
261 1.1 joerg STAILQ_HEAD(pattern_list, pattern);
262 1.1 joerg static struct pattern_list include = STAILQ_HEAD_INITIALIZER(include);
263 1.1 joerg static struct pattern_list exclude = STAILQ_HEAD_INITIALIZER(exclude);
264 1.1 joerg
265 1.1 joerg /*
266 1.1 joerg * Add an entry to a pattern list
267 1.1 joerg */
268 1.1 joerg static void
269 1.1 joerg add_pattern(struct pattern_list *list, const char *pattern)
270 1.1 joerg {
271 1.1 joerg struct pattern *entry;
272 1.1 joerg size_t len;
273 1.1 joerg
274 1.1 joerg debug("adding pattern '%s'\n", pattern);
275 1.1 joerg len = strlen(pattern);
276 1.1 joerg if ((entry = malloc(sizeof *entry + len + 1)) == NULL) {
277 1.1 joerg errno = ENOMEM;
278 1.1 joerg error("malloc()");
279 1.1 joerg }
280 1.1 joerg memcpy(entry->pattern, pattern, len + 1);
281 1.1 joerg STAILQ_INSERT_TAIL(list, entry, link);
282 1.1 joerg }
283 1.1 joerg
284 1.1 joerg /*
285 1.1 joerg * Match a string against a list of patterns
286 1.1 joerg */
287 1.1 joerg static int
288 1.1 joerg match_pattern(struct pattern_list *list, const char *str)
289 1.1 joerg {
290 1.1 joerg struct pattern *entry;
291 1.1 joerg
292 1.1 joerg STAILQ_FOREACH(entry, list, link) {
293 1.1 joerg if (fnmatch(entry->pattern, str, 0) == 0)
294 1.1 joerg return (1);
295 1.1 joerg }
296 1.1 joerg return (0);
297 1.1 joerg }
298 1.1 joerg
299 1.1 joerg /*
300 1.1 joerg * Verify that a given pathname is in the include list and not in the
301 1.1 joerg * exclude list.
302 1.1 joerg */
303 1.1 joerg static int
304 1.1 joerg accept_pathname(const char *pathname)
305 1.1 joerg {
306 1.1 joerg
307 1.1 joerg if (!STAILQ_EMPTY(&include) && !match_pattern(&include, pathname))
308 1.1 joerg return (0);
309 1.1 joerg if (!STAILQ_EMPTY(&exclude) && match_pattern(&exclude, pathname))
310 1.1 joerg return (0);
311 1.1 joerg return (1);
312 1.1 joerg }
313 1.1 joerg
314 1.1 joerg /*
315 1.1 joerg * Create the specified directory with the specified mode, taking certain
316 1.1 joerg * precautions on they way.
317 1.1 joerg */
318 1.1 joerg static void
319 1.1 joerg make_dir(const char *path, int mode)
320 1.1 joerg {
321 1.1 joerg struct stat sb;
322 1.1 joerg
323 1.1 joerg if (lstat(path, &sb) == 0) {
324 1.1 joerg if (S_ISDIR(sb.st_mode))
325 1.1 joerg return;
326 1.1 joerg /*
327 1.1 joerg * Normally, we should either ask the user about removing
328 1.1 joerg * the non-directory of the same name as a directory we
329 1.1 joerg * wish to create, or respect the -n or -o command-line
330 1.1 joerg * options. However, this may lead to a later failure or
331 1.1 joerg * even compromise (if this non-directory happens to be a
332 1.1 joerg * symlink to somewhere unsafe), so we don't.
333 1.1 joerg */
334 1.1 joerg
335 1.1 joerg /*
336 1.1 joerg * Don't check unlink() result; failure will cause mkdir()
337 1.1 joerg * to fail later, which we will catch.
338 1.1 joerg */
339 1.1 joerg (void)unlink(path);
340 1.1 joerg }
341 1.1 joerg if (mkdir(path, mode) != 0 && errno != EEXIST)
342 1.1 joerg error("mkdir('%s')", path);
343 1.1 joerg }
344 1.1 joerg
345 1.1 joerg /*
346 1.1 joerg * Ensure that all directories leading up to (but not including) the
347 1.1 joerg * specified path exist.
348 1.1 joerg *
349 1.1 joerg * XXX inefficient + modifies the file in-place
350 1.1 joerg */
351 1.1 joerg static void
352 1.1 joerg make_parent(char *path)
353 1.1 joerg {
354 1.1 joerg struct stat sb;
355 1.1 joerg char *sep;
356 1.1 joerg
357 1.1 joerg sep = strrchr(path, '/');
358 1.1 joerg if (sep == NULL || sep == path)
359 1.1 joerg return;
360 1.1 joerg *sep = '\0';
361 1.1 joerg if (lstat(path, &sb) == 0) {
362 1.1 joerg if (S_ISDIR(sb.st_mode)) {
363 1.1 joerg *sep = '/';
364 1.1 joerg return;
365 1.1 joerg }
366 1.1 joerg unlink(path);
367 1.1 joerg }
368 1.1 joerg make_parent(path);
369 1.1 joerg mkdir(path, 0755);
370 1.1 joerg *sep = '/';
371 1.1 joerg
372 1.1 joerg #if 0
373 1.1 joerg for (sep = path; (sep = strchr(sep, '/')) != NULL; sep++) {
374 1.1 joerg /* root in case of absolute d_arg */
375 1.1 joerg if (sep == path)
376 1.1 joerg continue;
377 1.1 joerg *sep = '\0';
378 1.1 joerg make_dir(path, 0755);
379 1.1 joerg *sep = '/';
380 1.1 joerg }
381 1.1 joerg #endif
382 1.1 joerg }
383 1.1 joerg
384 1.1 joerg /*
385 1.1 joerg * Extract a directory.
386 1.1 joerg */
387 1.1 joerg static void
388 1.1 joerg extract_dir(struct archive *a, struct archive_entry *e, const char *path)
389 1.1 joerg {
390 1.1 joerg int mode;
391 1.1 joerg
392 1.1 joerg mode = archive_entry_filetype(e) & 0777;
393 1.1 joerg if (mode == 0)
394 1.1 joerg mode = 0755;
395 1.1 joerg
396 1.1 joerg /*
397 1.1 joerg * Some zipfiles contain directories with weird permissions such
398 1.1 joerg * as 0644 or 0444. This can cause strange issues such as being
399 1.1 joerg * unable to extract files into the directory we just created, or
400 1.1 joerg * the user being unable to remove the directory later without
401 1.1 joerg * first manually changing its permissions. Therefore, we whack
402 1.1 joerg * the permissions into shape, assuming that the user wants full
403 1.1 joerg * access and that anyone who gets read access also gets execute
404 1.1 joerg * access.
405 1.1 joerg */
406 1.1 joerg mode |= 0700;
407 1.1 joerg if (mode & 0040)
408 1.1 joerg mode |= 0010;
409 1.1 joerg if (mode & 0004)
410 1.1 joerg mode |= 0001;
411 1.1 joerg
412 1.1 joerg info("d %s\n", path);
413 1.1 joerg make_dir(path, mode);
414 1.1 joerg ac(archive_read_data_skip(a));
415 1.1 joerg }
416 1.1 joerg
417 1.1 joerg static unsigned char buffer[8192];
418 1.1 joerg static char spinner[] = { '|', '/', '-', '\\' };
419 1.1 joerg
420 1.1 joerg /*
421 1.1 joerg * Extract a regular file.
422 1.1 joerg */
423 1.1 joerg static void
424 1.1 joerg extract_file(struct archive *a, struct archive_entry *e, const char *path)
425 1.1 joerg {
426 1.1 joerg int mode;
427 1.1 joerg time_t mtime;
428 1.1 joerg struct stat sb;
429 1.1 joerg struct timeval tv[2];
430 1.1 joerg int cr, fd, text, warn;
431 1.1 joerg ssize_t len;
432 1.1 joerg unsigned char *p, *q, *end;
433 1.1 joerg
434 1.1 joerg mode = archive_entry_filetype(e) & 0777;
435 1.1 joerg if (mode == 0)
436 1.1 joerg mode = 0644;
437 1.1 joerg mtime = archive_entry_mtime(e);
438 1.1 joerg
439 1.1 joerg /* look for existing file of same name */
440 1.1 joerg if (lstat(path, &sb) == 0) {
441 1.1 joerg if (u_opt || f_opt) {
442 1.1 joerg /* check if up-to-date */
443 1.1 joerg if (S_ISREG(sb.st_mode) && sb.st_mtime >= mtime)
444 1.1 joerg return;
445 1.1 joerg (void)unlink(path);
446 1.1 joerg } else if (o_opt) {
447 1.1 joerg /* overwrite */
448 1.1 joerg (void)unlink(path);
449 1.1 joerg } else if (n_opt) {
450 1.1 joerg /* do not overwrite */
451 1.1 joerg return;
452 1.1 joerg } else {
453 1.1 joerg /* XXX ask user */
454 1.1 joerg errorx("not implemented");
455 1.1 joerg }
456 1.1 joerg } else {
457 1.1 joerg if (f_opt)
458 1.1 joerg return;
459 1.1 joerg }
460 1.1 joerg
461 1.1 joerg if ((fd = open(path, O_RDWR|O_CREAT|O_TRUNC, mode)) < 0)
462 1.1 joerg error("open('%s')", path);
463 1.1 joerg
464 1.1 joerg /* loop over file contents and write to disk */
465 1.1 joerg info("x %s", path);
466 1.1 joerg text = a_opt;
467 1.1 joerg warn = 0;
468 1.1 joerg cr = 0;
469 1.1 joerg for (int n = 0; ; n++) {
470 1.1 joerg if (tty && (n % 4) == 0)
471 1.1 joerg info(" %c\b\b", spinner[(n / 4) % sizeof spinner]);
472 1.1 joerg
473 1.1 joerg len = archive_read_data(a, buffer, sizeof buffer);
474 1.1 joerg
475 1.1 joerg if (len < 0)
476 1.1 joerg ac(len);
477 1.1 joerg
478 1.1 joerg /* left over CR from previous buffer */
479 1.1 joerg if (a_opt && cr) {
480 1.1 joerg if (len == 0 || buffer[0] != '\n')
481 1.1 joerg if (write(fd, "\r", 1) != 1)
482 1.1 joerg error("write('%s')", path);
483 1.1 joerg cr = 0;
484 1.1 joerg }
485 1.1 joerg
486 1.1 joerg /* EOF */
487 1.1 joerg if (len == 0)
488 1.1 joerg break;
489 1.1 joerg end = buffer + len;
490 1.1 joerg
491 1.1 joerg /*
492 1.1 joerg * Detect whether this is a text file. The correct way to
493 1.1 joerg * do this is to check the least significant bit of the
494 1.1 joerg * "internal file attributes" field of the corresponding
495 1.1 joerg * file header in the central directory, but libarchive
496 1.1 joerg * does not read the central directory, so we have to
497 1.1 joerg * guess by looking for non-ASCII characters in the
498 1.1 joerg * buffer. Hopefully we won't guess wrong. If we do
499 1.1 joerg * guess wrong, we print a warning message later.
500 1.1 joerg */
501 1.1 joerg if (a_opt && n == 0) {
502 1.1 joerg for (p = buffer; p < end; ++p) {
503 1.1 joerg if (!isascii((unsigned char)*p)) {
504 1.1 joerg text = 0;
505 1.1 joerg break;
506 1.1 joerg }
507 1.1 joerg }
508 1.1 joerg }
509 1.1 joerg
510 1.1 joerg /* simple case */
511 1.1 joerg if (!a_opt || !text) {
512 1.1 joerg if (write(fd, buffer, len) != len)
513 1.1 joerg error("write('%s')", path);
514 1.1 joerg continue;
515 1.1 joerg }
516 1.1 joerg
517 1.1 joerg /* hard case: convert \r\n to \n (sigh...) */
518 1.1 joerg for (p = buffer; p < end; p = q + 1) {
519 1.1 joerg for (q = p; q < end; q++) {
520 1.1 joerg if (!warn && !isascii(*q)) {
521 1.1 joerg warningx("%s may be corrupted due"
522 1.1 joerg " to weak text file detection"
523 1.1 joerg " heuristic", path);
524 1.1 joerg warn = 1;
525 1.1 joerg }
526 1.1 joerg if (q[0] != '\r')
527 1.1 joerg continue;
528 1.1 joerg if (&q[1] == end) {
529 1.1 joerg cr = 1;
530 1.1 joerg break;
531 1.1 joerg }
532 1.1 joerg if (q[1] == '\n')
533 1.1 joerg break;
534 1.1 joerg }
535 1.1 joerg if (write(fd, p, q - p) != q - p)
536 1.1 joerg error("write('%s')", path);
537 1.1 joerg }
538 1.1 joerg }
539 1.1 joerg if (tty)
540 1.1 joerg info(" \b\b");
541 1.1 joerg if (text)
542 1.1 joerg info(" (text)");
543 1.1 joerg info("\n");
544 1.1 joerg
545 1.1 joerg /* set access and modification time */
546 1.1 joerg tv[0].tv_sec = now;
547 1.1 joerg tv[0].tv_usec = 0;
548 1.1 joerg tv[1].tv_sec = mtime;
549 1.1 joerg tv[1].tv_usec = 0;
550 1.1 joerg if (futimes(fd, tv) != 0)
551 1.1 joerg error("utimes('%s')", path);
552 1.1 joerg if (close(fd) != 0)
553 1.1 joerg error("close('%s')", path);
554 1.1 joerg }
555 1.1 joerg
556 1.1 joerg /*
557 1.1 joerg * Extract a zipfile entry: first perform some sanity checks to ensure
558 1.1 joerg * that it is either a directory or a regular file and that the path is
559 1.1 joerg * not absolute and does not try to break out of the current directory;
560 1.1 joerg * then call either extract_dir() or extract_file() as appropriate.
561 1.1 joerg *
562 1.1 joerg * This is complicated a bit by the various ways in which we need to
563 1.1 joerg * manipulate the path name. Case conversion (if requested by the -L
564 1.1 joerg * option) happens first, but the include / exclude patterns are applied
565 1.1 joerg * to the full converted path name, before the directory part of the path
566 1.1 joerg * is removed in accordance with the -j option. Sanity checks are
567 1.1 joerg * intentionally done earlier than they need to be, so the user will get a
568 1.1 joerg * warning about insecure paths even for files or directories which
569 1.1 joerg * wouldn't be extracted anyway.
570 1.1 joerg */
571 1.1 joerg static void
572 1.1 joerg extract(struct archive *a, struct archive_entry *e)
573 1.1 joerg {
574 1.1 joerg char *pathname, *realpathname;
575 1.1 joerg mode_t filetype;
576 1.1 joerg char *p, *q;
577 1.1 joerg
578 1.1 joerg pathname = pathdup(archive_entry_pathname(e));
579 1.1 joerg filetype = archive_entry_filetype(e);
580 1.1 joerg
581 1.1 joerg /* sanity checks */
582 1.1 joerg if (pathname[0] == '/' ||
583 1.1 joerg strncmp(pathname, "../", 3) == 0 ||
584 1.1 joerg strstr(pathname, "/../") != NULL) {
585 1.1 joerg warningx("skipping insecure entry '%s'", pathname);
586 1.1 joerg ac(archive_read_data_skip(a));
587 1.1 joerg free(pathname);
588 1.1 joerg return;
589 1.1 joerg }
590 1.1 joerg
591 1.1 joerg /* I don't think this can happen in a zipfile.. */
592 1.1 joerg if (!S_ISDIR(filetype) && !S_ISREG(filetype)) {
593 1.1 joerg warningx("skipping non-regular entry '%s'", pathname);
594 1.1 joerg ac(archive_read_data_skip(a));
595 1.1 joerg free(pathname);
596 1.1 joerg return;
597 1.1 joerg }
598 1.1 joerg
599 1.1 joerg /* skip directories in -j case */
600 1.1 joerg if (S_ISDIR(filetype) && j_opt) {
601 1.1 joerg ac(archive_read_data_skip(a));
602 1.1 joerg free(pathname);
603 1.1 joerg return;
604 1.1 joerg }
605 1.1 joerg
606 1.1 joerg /* apply include / exclude patterns */
607 1.1 joerg if (!accept_pathname(pathname)) {
608 1.1 joerg ac(archive_read_data_skip(a));
609 1.1 joerg free(pathname);
610 1.1 joerg return;
611 1.1 joerg }
612 1.1 joerg
613 1.1 joerg /* apply -j and -d */
614 1.1 joerg if (j_opt) {
615 1.1 joerg for (p = q = pathname; *p; ++p)
616 1.1 joerg if (*p == '/')
617 1.1 joerg q = p + 1;
618 1.1 joerg realpathname = pathcat(d_arg, q);
619 1.1 joerg } else {
620 1.1 joerg realpathname = pathcat(d_arg, pathname);
621 1.1 joerg }
622 1.1 joerg
623 1.1 joerg /* ensure that parent directory exists */
624 1.1 joerg make_parent(realpathname);
625 1.1 joerg
626 1.1 joerg if (S_ISDIR(filetype))
627 1.1 joerg extract_dir(a, e, realpathname);
628 1.1 joerg else
629 1.1 joerg extract_file(a, e, realpathname);
630 1.1 joerg
631 1.1 joerg free(realpathname);
632 1.1 joerg free(pathname);
633 1.1 joerg }
634 1.1 joerg
635 1.2 joerg static void
636 1.2 joerg extract_stdout(struct archive *a, struct archive_entry *e)
637 1.2 joerg {
638 1.2 joerg char *pathname;
639 1.2 joerg mode_t filetype;
640 1.2 joerg int cr, text, warn;
641 1.2 joerg ssize_t len;
642 1.2 joerg unsigned char *p, *q, *end;
643 1.2 joerg
644 1.2 joerg pathname = pathdup(archive_entry_pathname(e));
645 1.2 joerg filetype = archive_entry_filetype(e);
646 1.2 joerg
647 1.2 joerg /* I don't think this can happen in a zipfile.. */
648 1.2 joerg if (!S_ISDIR(filetype) && !S_ISREG(filetype)) {
649 1.2 joerg warningx("skipping non-regular entry '%s'", pathname);
650 1.2 joerg ac(archive_read_data_skip(a));
651 1.2 joerg free(pathname);
652 1.2 joerg return;
653 1.2 joerg }
654 1.2 joerg
655 1.2 joerg /* skip directories in -j case */
656 1.2 joerg if (S_ISDIR(filetype)) {
657 1.2 joerg ac(archive_read_data_skip(a));
658 1.2 joerg free(pathname);
659 1.2 joerg return;
660 1.2 joerg }
661 1.2 joerg
662 1.2 joerg /* apply include / exclude patterns */
663 1.2 joerg if (!accept_pathname(pathname)) {
664 1.2 joerg ac(archive_read_data_skip(a));
665 1.2 joerg free(pathname);
666 1.2 joerg return;
667 1.2 joerg }
668 1.2 joerg
669 1.2 joerg text = a_opt;
670 1.2 joerg warn = 0;
671 1.2 joerg cr = 0;
672 1.2 joerg for (int n = 0; ; n++) {
673 1.2 joerg len = archive_read_data(a, buffer, sizeof buffer);
674 1.2 joerg
675 1.2 joerg if (len < 0)
676 1.2 joerg ac(len);
677 1.2 joerg
678 1.2 joerg /* left over CR from previous buffer */
679 1.2 joerg if (a_opt && cr) {
680 1.2 joerg if (len == 0 || buffer[0] != '\n')
681 1.2 joerg if (write(STDOUT_FILENO, "\r", 1) != 1)
682 1.2 joerg error("write('%s')", pathname);
683 1.2 joerg cr = 0;
684 1.2 joerg }
685 1.2 joerg
686 1.2 joerg /* EOF */
687 1.2 joerg if (len == 0)
688 1.2 joerg break;
689 1.2 joerg end = buffer + len;
690 1.2 joerg
691 1.2 joerg /*
692 1.2 joerg * Detect whether this is a text file. The correct way to
693 1.2 joerg * do this is to check the least significant bit of the
694 1.2 joerg * "internal file attributes" field of the corresponding
695 1.2 joerg * file header in the central directory, but libarchive
696 1.2 joerg * does not read the central directory, so we have to
697 1.2 joerg * guess by looking for non-ASCII characters in the
698 1.2 joerg * buffer. Hopefully we won't guess wrong. If we do
699 1.2 joerg * guess wrong, we print a warning message later.
700 1.2 joerg */
701 1.2 joerg if (a_opt && n == 0) {
702 1.2 joerg for (p = buffer; p < end; ++p) {
703 1.2 joerg if (!isascii((unsigned char)*p)) {
704 1.2 joerg text = 0;
705 1.2 joerg break;
706 1.2 joerg }
707 1.2 joerg }
708 1.2 joerg }
709 1.2 joerg
710 1.2 joerg /* simple case */
711 1.2 joerg if (!a_opt || !text) {
712 1.2 joerg if (write(STDOUT_FILENO, buffer, len) != len)
713 1.2 joerg error("write('%s')", pathname);
714 1.2 joerg continue;
715 1.2 joerg }
716 1.2 joerg
717 1.2 joerg /* hard case: convert \r\n to \n (sigh...) */
718 1.2 joerg for (p = buffer; p < end; p = q + 1) {
719 1.2 joerg for (q = p; q < end; q++) {
720 1.2 joerg if (!warn && !isascii(*q)) {
721 1.2 joerg warningx("%s may be corrupted due"
722 1.2 joerg " to weak text file detection"
723 1.2 joerg " heuristic", pathname);
724 1.2 joerg warn = 1;
725 1.2 joerg }
726 1.2 joerg if (q[0] != '\r')
727 1.2 joerg continue;
728 1.2 joerg if (&q[1] == end) {
729 1.2 joerg cr = 1;
730 1.2 joerg break;
731 1.2 joerg }
732 1.2 joerg if (q[1] == '\n')
733 1.2 joerg break;
734 1.2 joerg }
735 1.2 joerg if (write(STDOUT_FILENO, p, q - p) != q - p)
736 1.2 joerg error("write('%s')", pathname);
737 1.2 joerg }
738 1.2 joerg }
739 1.2 joerg
740 1.2 joerg free(pathname);
741 1.2 joerg }
742 1.2 joerg
743 1.1 joerg /*
744 1.1 joerg * Print the name of an entry to stdout.
745 1.1 joerg */
746 1.1 joerg static void
747 1.1 joerg list(struct archive *a, struct archive_entry *e)
748 1.1 joerg {
749 1.2 joerg static int printed_header;
750 1.2 joerg char buf[20];
751 1.2 joerg time_t mtime;
752 1.1 joerg
753 1.2 joerg if (!printed_header && !q_opt) {
754 1.2 joerg printed_header = 1;
755 1.2 joerg printf(" Length Method Size Ratio Date Time CRC-32 Name\n");
756 1.2 joerg printf("-------- ------ ------- ----- ---- ---- ------ ----\n");
757 1.2 joerg }
758 1.2 joerg
759 1.2 joerg if (v_opt == 2) {
760 1.2 joerg mtime = archive_entry_mtime(e);
761 1.2 joerg strftime(buf, sizeof(buf), "%m-%d-%g %R", localtime(&mtime));
762 1.2 joerg printf("%8ju Stored %7ju 0%% %s %08x %s\n",
763 1.2 joerg (uintmax_t)archive_entry_size(e),
764 1.2 joerg (uintmax_t)archive_entry_size(e),
765 1.2 joerg buf,
766 1.2 joerg 0U,
767 1.2 joerg archive_entry_pathname(e));
768 1.2 joerg } else {
769 1.2 joerg printf("%s\n", archive_entry_pathname(e));
770 1.2 joerg }
771 1.1 joerg ac(archive_read_data_skip(a));
772 1.1 joerg }
773 1.1 joerg
774 1.1 joerg /*
775 1.1 joerg * Extract to memory to check CRC
776 1.1 joerg */
777 1.1 joerg static void
778 1.1 joerg test(struct archive *a, struct archive_entry *e)
779 1.1 joerg {
780 1.1 joerg ssize_t len;
781 1.1 joerg
782 1.1 joerg if (S_ISDIR(archive_entry_filetype(e)))
783 1.1 joerg return;
784 1.1 joerg
785 1.1 joerg info("%s ", archive_entry_pathname(e));
786 1.1 joerg while ((len = archive_read_data(a, buffer, sizeof buffer)) > 0)
787 1.1 joerg /* nothing */;
788 1.1 joerg if (len < 0) {
789 1.1 joerg info("%s\n", archive_error_string(a));
790 1.1 joerg ++test_failed;
791 1.1 joerg } else {
792 1.1 joerg info("OK\n");
793 1.1 joerg }
794 1.1 joerg
795 1.1 joerg /* shouldn't be necessary, but it doesn't hurt */
796 1.1 joerg ac(archive_read_data_skip(a));
797 1.1 joerg }
798 1.1 joerg
799 1.1 joerg
800 1.1 joerg /*
801 1.1 joerg * Main loop: open the zipfile, iterate over its contents and decide what
802 1.1 joerg * to do with each entry.
803 1.1 joerg */
804 1.1 joerg static void
805 1.1 joerg unzip(const char *fn)
806 1.1 joerg {
807 1.1 joerg struct archive *a;
808 1.1 joerg struct archive_entry *e;
809 1.1 joerg int fd, ret;
810 1.1 joerg
811 1.1 joerg if ((fd = open(fn, O_RDONLY)) < 0)
812 1.1 joerg error("%s", fn);
813 1.1 joerg
814 1.1 joerg a = archive_read_new();
815 1.1 joerg ac(archive_read_support_format_zip(a));
816 1.1 joerg ac(archive_read_open_fd(a, fd, 8192));
817 1.1 joerg
818 1.1 joerg for (;;) {
819 1.1 joerg ret = archive_read_next_header(a, &e);
820 1.1 joerg if (ret == ARCHIVE_EOF)
821 1.1 joerg break;
822 1.1 joerg ac(ret);
823 1.1 joerg if (t_opt)
824 1.1 joerg test(a, e);
825 1.2 joerg else if (v_opt)
826 1.1 joerg list(a, e);
827 1.2 joerg else if (p_opt)
828 1.2 joerg extract_stdout(a, e);
829 1.1 joerg else
830 1.1 joerg extract(a, e);
831 1.1 joerg }
832 1.1 joerg
833 1.1 joerg ac(archive_read_close(a));
834 1.1 joerg (void)archive_read_finish(a);
835 1.1 joerg if (close(fd) != 0)
836 1.1 joerg error("%s", fn);
837 1.1 joerg
838 1.1 joerg if (t_opt && test_failed)
839 1.1 joerg errorx("%d checksum error(s) found.", test_failed);
840 1.1 joerg }
841 1.1 joerg
842 1.1 joerg static void
843 1.1 joerg usage(void)
844 1.1 joerg {
845 1.1 joerg
846 1.1 joerg fprintf(stderr, "usage: unzip [-ajLlnoqtu] [-d dir] zipfile\n");
847 1.1 joerg exit(1);
848 1.1 joerg }
849 1.1 joerg
850 1.1 joerg static int
851 1.1 joerg getopts(int argc, char *argv[])
852 1.1 joerg {
853 1.1 joerg int opt;
854 1.1 joerg
855 1.1 joerg optreset = optind = 1;
856 1.2 joerg while ((opt = getopt(argc, argv, "ad:fjLlnopqtuvx:")) != -1)
857 1.1 joerg switch (opt) {
858 1.1 joerg case 'a':
859 1.1 joerg a_opt = 1;
860 1.1 joerg break;
861 1.1 joerg case 'd':
862 1.1 joerg d_arg = optarg;
863 1.1 joerg break;
864 1.1 joerg case 'f':
865 1.1 joerg f_opt = 1;
866 1.1 joerg break;
867 1.1 joerg case 'j':
868 1.1 joerg j_opt = 1;
869 1.1 joerg break;
870 1.1 joerg case 'L':
871 1.1 joerg L_opt = 1;
872 1.1 joerg break;
873 1.1 joerg case 'l':
874 1.2 joerg if (v_opt == 0)
875 1.2 joerg v_opt = 1;
876 1.1 joerg break;
877 1.1 joerg case 'n':
878 1.1 joerg n_opt = 1;
879 1.1 joerg break;
880 1.1 joerg case 'o':
881 1.1 joerg o_opt = 1;
882 1.2 joerg q_opt = 1;
883 1.2 joerg break;
884 1.2 joerg case 'p':
885 1.2 joerg p_opt = 1;
886 1.1 joerg break;
887 1.1 joerg case 'q':
888 1.1 joerg q_opt = 1;
889 1.1 joerg break;
890 1.1 joerg case 't':
891 1.1 joerg t_opt = 1;
892 1.1 joerg break;
893 1.1 joerg case 'u':
894 1.1 joerg u_opt = 1;
895 1.1 joerg break;
896 1.2 joerg case 'v':
897 1.2 joerg v_opt = 2;
898 1.2 joerg break;
899 1.1 joerg case 'x':
900 1.1 joerg add_pattern(&exclude, optarg);
901 1.1 joerg break;
902 1.1 joerg default:
903 1.1 joerg usage();
904 1.1 joerg }
905 1.1 joerg
906 1.1 joerg return (optind);
907 1.1 joerg }
908 1.1 joerg
909 1.1 joerg int
910 1.1 joerg main(int argc, char *argv[])
911 1.1 joerg {
912 1.1 joerg const char *zipfile;
913 1.1 joerg int nopts;
914 1.1 joerg
915 1.1 joerg if (isatty(STDOUT_FILENO))
916 1.1 joerg tty = 1;
917 1.1 joerg
918 1.1 joerg if (getenv("UNZIP_DEBUG") != NULL)
919 1.1 joerg unzip_debug = 1;
920 1.1 joerg for (int i = 0; i < argc; ++i)
921 1.1 joerg debug("%s%c", argv[i], (i < argc - 1) ? ' ' : '\n');
922 1.1 joerg
923 1.1 joerg /*
924 1.1 joerg * Info-ZIP's unzip(1) expects certain options to come before the
925 1.1 joerg * zipfile name, and others to come after - though it does not
926 1.1 joerg * enforce this. For simplicity, we accept *all* options both
927 1.1 joerg * before and after the zipfile name.
928 1.1 joerg */
929 1.1 joerg nopts = getopts(argc, argv);
930 1.1 joerg
931 1.1 joerg if (argc <= nopts)
932 1.1 joerg usage();
933 1.1 joerg zipfile = argv[nopts++];
934 1.1 joerg
935 1.1 joerg while (nopts < argc && *argv[nopts] != '-')
936 1.1 joerg add_pattern(&include, argv[nopts++]);
937 1.1 joerg
938 1.1 joerg nopts--; /* fake argv[0] */
939 1.1 joerg nopts += getopts(argc - nopts, argv + nopts);
940 1.1 joerg
941 1.1 joerg if (n_opt + o_opt + u_opt > 1)
942 1.1 joerg errorx("-n, -o and -u are contradictory");
943 1.1 joerg
944 1.1 joerg time(&now);
945 1.1 joerg
946 1.1 joerg unzip(zipfile);
947 1.1 joerg
948 1.1 joerg exit(0);
949 1.1 joerg }
950