unzip.c revision 1.11 1 1.11 mbalmer /* $NetBSD: unzip.c,v 1.11 2010/01/09 09:27:42 mbalmer 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 1.11 mbalmer __RCSID("$NetBSD: unzip.c,v 1.11 2010/01/09 09:27:42 mbalmer 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 1.7 wiz static int C_opt; /* match case-insensitively */
61 1.4 wiz static int c_opt; /* extract to stdout */
62 1.1 joerg static const char *d_arg; /* directory */
63 1.1 joerg static int f_opt; /* update existing files only */
64 1.1 joerg static int j_opt; /* junk directories */
65 1.1 joerg static int L_opt; /* lowercase names */
66 1.1 joerg static int n_opt; /* never overwrite */
67 1.1 joerg static int o_opt; /* always overwrite */
68 1.3 joerg static int p_opt; /* extract to stdout, quiet */
69 1.1 joerg static int q_opt; /* quiet */
70 1.1 joerg static int t_opt; /* test */
71 1.1 joerg static int u_opt; /* update */
72 1.3 joerg static int v_opt; /* verbose/list */
73 1.1 joerg
74 1.1 joerg /* time when unzip started */
75 1.1 joerg static time_t now;
76 1.1 joerg
77 1.1 joerg /* debug flag */
78 1.1 joerg static int unzip_debug;
79 1.1 joerg
80 1.1 joerg /* running on tty? */
81 1.1 joerg static int tty;
82 1.1 joerg
83 1.1 joerg /* convenience macro */
84 1.1 joerg /* XXX should differentiate between ARCHIVE_{WARN,FAIL,RETRY} */
85 1.1 joerg #define ac(call) \
86 1.1 joerg do { \
87 1.1 joerg int acret = (call); \
88 1.1 joerg if (acret != ARCHIVE_OK) \
89 1.1 joerg errorx("%s", archive_error_string(a)); \
90 1.1 joerg } while (0)
91 1.1 joerg
92 1.1 joerg /*
93 1.1 joerg * Indicates that last info() did not end with EOL. This helps error() et
94 1.1 joerg * al. avoid printing an error message on the same line as an incomplete
95 1.1 joerg * informational message.
96 1.1 joerg */
97 1.1 joerg static int noeol;
98 1.1 joerg
99 1.1 joerg /* fatal error message + errno */
100 1.1 joerg static void
101 1.1 joerg error(const char *fmt, ...)
102 1.1 joerg {
103 1.1 joerg va_list ap;
104 1.1 joerg
105 1.1 joerg if (noeol)
106 1.1 joerg fprintf(stdout, "\n");
107 1.1 joerg fflush(stdout);
108 1.1 joerg fprintf(stderr, "unzip: ");
109 1.1 joerg va_start(ap, fmt);
110 1.1 joerg vfprintf(stderr, fmt, ap);
111 1.1 joerg va_end(ap);
112 1.1 joerg fprintf(stderr, ": %s\n", strerror(errno));
113 1.1 joerg exit(1);
114 1.1 joerg }
115 1.1 joerg
116 1.1 joerg /* fatal error message, no errno */
117 1.1 joerg static void
118 1.1 joerg errorx(const char *fmt, ...)
119 1.1 joerg {
120 1.1 joerg va_list ap;
121 1.1 joerg
122 1.1 joerg if (noeol)
123 1.1 joerg fprintf(stdout, "\n");
124 1.1 joerg fflush(stdout);
125 1.1 joerg fprintf(stderr, "unzip: ");
126 1.1 joerg va_start(ap, fmt);
127 1.1 joerg vfprintf(stderr, fmt, ap);
128 1.1 joerg va_end(ap);
129 1.1 joerg fprintf(stderr, "\n");
130 1.1 joerg exit(1);
131 1.1 joerg }
132 1.1 joerg
133 1.1 joerg #if 0
134 1.1 joerg /* non-fatal error message + errno */
135 1.1 joerg static void
136 1.1 joerg warning(const char *fmt, ...)
137 1.1 joerg {
138 1.1 joerg va_list ap;
139 1.1 joerg
140 1.1 joerg if (noeol)
141 1.1 joerg fprintf(stdout, "\n");
142 1.1 joerg fflush(stdout);
143 1.1 joerg fprintf(stderr, "unzip: ");
144 1.1 joerg va_start(ap, fmt);
145 1.1 joerg vfprintf(stderr, fmt, ap);
146 1.1 joerg va_end(ap);
147 1.1 joerg fprintf(stderr, ": %s\n", strerror(errno));
148 1.1 joerg }
149 1.1 joerg #endif
150 1.1 joerg
151 1.1 joerg /* non-fatal error message, no errno */
152 1.1 joerg static void
153 1.1 joerg warningx(const char *fmt, ...)
154 1.1 joerg {
155 1.1 joerg va_list ap;
156 1.1 joerg
157 1.1 joerg if (noeol)
158 1.1 joerg fprintf(stdout, "\n");
159 1.1 joerg fflush(stdout);
160 1.1 joerg fprintf(stderr, "unzip: ");
161 1.1 joerg va_start(ap, fmt);
162 1.1 joerg vfprintf(stderr, fmt, ap);
163 1.1 joerg va_end(ap);
164 1.1 joerg fprintf(stderr, "\n");
165 1.1 joerg }
166 1.1 joerg
167 1.1 joerg /* informational message (if not -q) */
168 1.1 joerg static void
169 1.1 joerg info(const char *fmt, ...)
170 1.1 joerg {
171 1.1 joerg va_list ap;
172 1.1 joerg
173 1.1 joerg if (q_opt && !unzip_debug)
174 1.1 joerg return;
175 1.1 joerg va_start(ap, fmt);
176 1.1 joerg vfprintf(stdout, fmt, ap);
177 1.1 joerg va_end(ap);
178 1.1 joerg fflush(stdout);
179 1.1 joerg
180 1.1 joerg if (*fmt == '\0')
181 1.1 joerg noeol = 1;
182 1.1 joerg else
183 1.1 joerg noeol = fmt[strlen(fmt) - 1] != '\n';
184 1.1 joerg }
185 1.1 joerg
186 1.1 joerg /* debug message (if unzip_debug) */
187 1.1 joerg static void
188 1.1 joerg debug(const char *fmt, ...)
189 1.1 joerg {
190 1.1 joerg va_list ap;
191 1.1 joerg
192 1.1 joerg if (!unzip_debug)
193 1.1 joerg return;
194 1.1 joerg va_start(ap, fmt);
195 1.1 joerg vfprintf(stderr, fmt, ap);
196 1.1 joerg va_end(ap);
197 1.1 joerg fflush(stderr);
198 1.1 joerg
199 1.1 joerg if (*fmt == '\0')
200 1.1 joerg noeol = 1;
201 1.1 joerg else
202 1.1 joerg noeol = fmt[strlen(fmt) - 1] != '\n';
203 1.1 joerg }
204 1.1 joerg
205 1.1 joerg /* duplicate a path name, possibly converting to lower case */
206 1.1 joerg static char *
207 1.1 joerg pathdup(const char *path)
208 1.1 joerg {
209 1.1 joerg char *str;
210 1.1 joerg size_t i, len;
211 1.1 joerg
212 1.1 joerg len = strlen(path);
213 1.1 joerg while (len && path[len - 1] == '/')
214 1.1 joerg len--;
215 1.1 joerg if ((str = malloc(len + 1)) == NULL) {
216 1.1 joerg errno = ENOMEM;
217 1.1 joerg error("malloc()");
218 1.1 joerg }
219 1.1 joerg if (L_opt) {
220 1.1 joerg for (i = 0; i < len; ++i)
221 1.1 joerg str[i] = tolower((unsigned char)path[i]);
222 1.1 joerg } else {
223 1.1 joerg memcpy(str, path, len);
224 1.1 joerg }
225 1.1 joerg str[len] = '\0';
226 1.1 joerg
227 1.1 joerg return (str);
228 1.1 joerg }
229 1.1 joerg
230 1.1 joerg /* concatenate two path names */
231 1.1 joerg static char *
232 1.1 joerg pathcat(const char *prefix, const char *path)
233 1.1 joerg {
234 1.1 joerg char *str;
235 1.1 joerg size_t prelen, len;
236 1.1 joerg
237 1.1 joerg prelen = prefix ? strlen(prefix) + 1 : 0;
238 1.1 joerg len = strlen(path) + 1;
239 1.1 joerg if ((str = malloc(prelen + len)) == NULL) {
240 1.1 joerg errno = ENOMEM;
241 1.1 joerg error("malloc()");
242 1.1 joerg }
243 1.1 joerg if (prefix) {
244 1.1 joerg memcpy(str, prefix, prelen); /* includes zero */
245 1.1 joerg str[prelen - 1] = '/'; /* splat zero */
246 1.1 joerg }
247 1.1 joerg memcpy(str + prelen, path, len); /* includes zero */
248 1.1 joerg
249 1.1 joerg return (str);
250 1.1 joerg }
251 1.1 joerg
252 1.1 joerg /*
253 1.1 joerg * Pattern lists for include / exclude processing
254 1.1 joerg */
255 1.1 joerg struct pattern {
256 1.1 joerg STAILQ_ENTRY(pattern) link;
257 1.1 joerg char pattern[];
258 1.1 joerg };
259 1.1 joerg
260 1.1 joerg STAILQ_HEAD(pattern_list, pattern);
261 1.1 joerg static struct pattern_list include = STAILQ_HEAD_INITIALIZER(include);
262 1.1 joerg static struct pattern_list exclude = STAILQ_HEAD_INITIALIZER(exclude);
263 1.1 joerg
264 1.1 joerg /*
265 1.1 joerg * Add an entry to a pattern list
266 1.1 joerg */
267 1.1 joerg static void
268 1.1 joerg add_pattern(struct pattern_list *list, const char *pattern)
269 1.1 joerg {
270 1.1 joerg struct pattern *entry;
271 1.1 joerg size_t len;
272 1.1 joerg
273 1.1 joerg debug("adding pattern '%s'\n", pattern);
274 1.1 joerg len = strlen(pattern);
275 1.1 joerg if ((entry = malloc(sizeof *entry + len + 1)) == NULL) {
276 1.1 joerg errno = ENOMEM;
277 1.1 joerg error("malloc()");
278 1.1 joerg }
279 1.1 joerg memcpy(entry->pattern, pattern, len + 1);
280 1.1 joerg STAILQ_INSERT_TAIL(list, entry, link);
281 1.1 joerg }
282 1.1 joerg
283 1.1 joerg /*
284 1.1 joerg * Match a string against a list of patterns
285 1.1 joerg */
286 1.1 joerg static int
287 1.1 joerg match_pattern(struct pattern_list *list, const char *str)
288 1.1 joerg {
289 1.1 joerg struct pattern *entry;
290 1.1 joerg
291 1.1 joerg STAILQ_FOREACH(entry, list, link) {
292 1.7 wiz if (fnmatch(entry->pattern, str, C_opt ? FNM_CASEFOLD : 0) == 0)
293 1.1 joerg return (1);
294 1.1 joerg }
295 1.1 joerg return (0);
296 1.1 joerg }
297 1.1 joerg
298 1.1 joerg /*
299 1.1 joerg * Verify that a given pathname is in the include list and not in the
300 1.1 joerg * exclude list.
301 1.1 joerg */
302 1.1 joerg static int
303 1.1 joerg accept_pathname(const char *pathname)
304 1.1 joerg {
305 1.1 joerg
306 1.1 joerg if (!STAILQ_EMPTY(&include) && !match_pattern(&include, pathname))
307 1.1 joerg return (0);
308 1.1 joerg if (!STAILQ_EMPTY(&exclude) && match_pattern(&exclude, pathname))
309 1.1 joerg return (0);
310 1.1 joerg return (1);
311 1.1 joerg }
312 1.1 joerg
313 1.1 joerg /*
314 1.1 joerg * Create the specified directory with the specified mode, taking certain
315 1.1 joerg * precautions on they way.
316 1.1 joerg */
317 1.1 joerg static void
318 1.1 joerg make_dir(const char *path, int mode)
319 1.1 joerg {
320 1.1 joerg struct stat sb;
321 1.1 joerg
322 1.1 joerg if (lstat(path, &sb) == 0) {
323 1.1 joerg if (S_ISDIR(sb.st_mode))
324 1.1 joerg return;
325 1.1 joerg /*
326 1.1 joerg * Normally, we should either ask the user about removing
327 1.1 joerg * the non-directory of the same name as a directory we
328 1.1 joerg * wish to create, or respect the -n or -o command-line
329 1.1 joerg * options. However, this may lead to a later failure or
330 1.1 joerg * even compromise (if this non-directory happens to be a
331 1.1 joerg * symlink to somewhere unsafe), so we don't.
332 1.1 joerg */
333 1.1 joerg
334 1.1 joerg /*
335 1.1 joerg * Don't check unlink() result; failure will cause mkdir()
336 1.1 joerg * to fail later, which we will catch.
337 1.1 joerg */
338 1.1 joerg (void)unlink(path);
339 1.1 joerg }
340 1.1 joerg if (mkdir(path, mode) != 0 && errno != EEXIST)
341 1.1 joerg error("mkdir('%s')", path);
342 1.1 joerg }
343 1.1 joerg
344 1.1 joerg /*
345 1.1 joerg * Ensure that all directories leading up to (but not including) the
346 1.1 joerg * specified path exist.
347 1.1 joerg *
348 1.1 joerg * XXX inefficient + modifies the file in-place
349 1.1 joerg */
350 1.1 joerg static void
351 1.1 joerg make_parent(char *path)
352 1.1 joerg {
353 1.1 joerg struct stat sb;
354 1.1 joerg char *sep;
355 1.1 joerg
356 1.1 joerg sep = strrchr(path, '/');
357 1.1 joerg if (sep == NULL || sep == path)
358 1.1 joerg return;
359 1.1 joerg *sep = '\0';
360 1.1 joerg if (lstat(path, &sb) == 0) {
361 1.1 joerg if (S_ISDIR(sb.st_mode)) {
362 1.1 joerg *sep = '/';
363 1.1 joerg return;
364 1.1 joerg }
365 1.1 joerg unlink(path);
366 1.1 joerg }
367 1.1 joerg make_parent(path);
368 1.1 joerg mkdir(path, 0755);
369 1.1 joerg *sep = '/';
370 1.1 joerg
371 1.1 joerg #if 0
372 1.1 joerg for (sep = path; (sep = strchr(sep, '/')) != NULL; sep++) {
373 1.1 joerg /* root in case of absolute d_arg */
374 1.1 joerg if (sep == path)
375 1.1 joerg continue;
376 1.1 joerg *sep = '\0';
377 1.1 joerg make_dir(path, 0755);
378 1.1 joerg *sep = '/';
379 1.1 joerg }
380 1.1 joerg #endif
381 1.1 joerg }
382 1.1 joerg
383 1.1 joerg /*
384 1.1 joerg * Extract a directory.
385 1.1 joerg */
386 1.1 joerg static void
387 1.1 joerg extract_dir(struct archive *a, struct archive_entry *e, const char *path)
388 1.1 joerg {
389 1.1 joerg int mode;
390 1.1 joerg
391 1.11 mbalmer mode = archive_entry_mode(e) & 0777;
392 1.1 joerg if (mode == 0)
393 1.1 joerg mode = 0755;
394 1.1 joerg
395 1.1 joerg /*
396 1.1 joerg * Some zipfiles contain directories with weird permissions such
397 1.1 joerg * as 0644 or 0444. This can cause strange issues such as being
398 1.1 joerg * unable to extract files into the directory we just created, or
399 1.1 joerg * the user being unable to remove the directory later without
400 1.1 joerg * first manually changing its permissions. Therefore, we whack
401 1.1 joerg * the permissions into shape, assuming that the user wants full
402 1.1 joerg * access and that anyone who gets read access also gets execute
403 1.1 joerg * access.
404 1.1 joerg */
405 1.1 joerg mode |= 0700;
406 1.1 joerg if (mode & 0040)
407 1.1 joerg mode |= 0010;
408 1.1 joerg if (mode & 0004)
409 1.1 joerg mode |= 0001;
410 1.1 joerg
411 1.10 wiz info(" creating: %s/\n", path);
412 1.1 joerg make_dir(path, mode);
413 1.1 joerg ac(archive_read_data_skip(a));
414 1.1 joerg }
415 1.1 joerg
416 1.1 joerg static unsigned char buffer[8192];
417 1.1 joerg static char spinner[] = { '|', '/', '-', '\\' };
418 1.1 joerg
419 1.8 joerg static int
420 1.8 joerg handle_existing_file(char **path)
421 1.8 joerg {
422 1.8 joerg size_t alen;
423 1.8 joerg ssize_t len;
424 1.8 joerg char buf[4];
425 1.8 joerg
426 1.8 joerg for (;;) {
427 1.8 joerg fprintf(stderr,
428 1.8 joerg "replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ",
429 1.8 joerg *path);
430 1.8 joerg fgets(buf, 4, stdin);
431 1.8 joerg switch (*buf) {
432 1.8 joerg case 'A':
433 1.8 joerg o_opt = 1;
434 1.8 joerg /* FALL THROUGH */
435 1.8 joerg case 'y':
436 1.8 joerg case 'Y':
437 1.8 joerg (void)unlink(*path);
438 1.8 joerg return 1;
439 1.8 joerg case 'N':
440 1.8 joerg n_opt = 1;
441 1.8 joerg /* FALL THROUGH */
442 1.8 joerg case 'n':
443 1.8 joerg return -1;
444 1.8 joerg case 'r':
445 1.8 joerg case 'R':
446 1.8 joerg printf("New name: ");
447 1.8 joerg fflush(stdout);
448 1.8 joerg free(*path);
449 1.8 joerg *path = NULL;
450 1.8 joerg alen = 0;
451 1.8 joerg len = getline(path, &alen, stdin);
452 1.8 joerg if ((*path)[len - 1] != '\n')
453 1.8 joerg (*path)[len - 1] = '\0';
454 1.8 joerg return 0;
455 1.8 joerg default:
456 1.8 joerg break;
457 1.8 joerg }
458 1.8 joerg }
459 1.8 joerg }
460 1.8 joerg
461 1.1 joerg /*
462 1.1 joerg * Extract a regular file.
463 1.1 joerg */
464 1.1 joerg static void
465 1.8 joerg extract_file(struct archive *a, struct archive_entry *e, char **path)
466 1.1 joerg {
467 1.1 joerg int mode;
468 1.1 joerg time_t mtime;
469 1.1 joerg struct stat sb;
470 1.1 joerg struct timeval tv[2];
471 1.8 joerg int cr, fd, text, warn, check;
472 1.1 joerg ssize_t len;
473 1.1 joerg unsigned char *p, *q, *end;
474 1.1 joerg
475 1.11 mbalmer mode = archive_entry_mode(e) & 0777;
476 1.1 joerg if (mode == 0)
477 1.1 joerg mode = 0644;
478 1.1 joerg mtime = archive_entry_mtime(e);
479 1.1 joerg
480 1.1 joerg /* look for existing file of same name */
481 1.8 joerg recheck:
482 1.8 joerg if (lstat(*path, &sb) == 0) {
483 1.1 joerg if (u_opt || f_opt) {
484 1.1 joerg /* check if up-to-date */
485 1.1 joerg if (S_ISREG(sb.st_mode) && sb.st_mtime >= mtime)
486 1.1 joerg return;
487 1.8 joerg (void)unlink(*path);
488 1.1 joerg } else if (o_opt) {
489 1.1 joerg /* overwrite */
490 1.8 joerg (void)unlink(*path);
491 1.1 joerg } else if (n_opt) {
492 1.1 joerg /* do not overwrite */
493 1.1 joerg return;
494 1.1 joerg } else {
495 1.8 joerg check = handle_existing_file(path);
496 1.8 joerg if (check == 0)
497 1.8 joerg goto recheck;
498 1.8 joerg if (check == -1)
499 1.8 joerg return; /* do not overwrite */
500 1.1 joerg }
501 1.1 joerg } else {
502 1.1 joerg if (f_opt)
503 1.1 joerg return;
504 1.1 joerg }
505 1.1 joerg
506 1.8 joerg if ((fd = open(*path, O_RDWR|O_CREAT|O_TRUNC, mode)) < 0)
507 1.8 joerg error("open('%s')", *path);
508 1.1 joerg
509 1.1 joerg /* loop over file contents and write to disk */
510 1.8 joerg info(" extracting: %s", *path);
511 1.1 joerg text = a_opt;
512 1.1 joerg warn = 0;
513 1.1 joerg cr = 0;
514 1.1 joerg for (int n = 0; ; n++) {
515 1.1 joerg if (tty && (n % 4) == 0)
516 1.1 joerg info(" %c\b\b", spinner[(n / 4) % sizeof spinner]);
517 1.1 joerg
518 1.1 joerg len = archive_read_data(a, buffer, sizeof buffer);
519 1.1 joerg
520 1.1 joerg if (len < 0)
521 1.1 joerg ac(len);
522 1.1 joerg
523 1.1 joerg /* left over CR from previous buffer */
524 1.1 joerg if (a_opt && cr) {
525 1.1 joerg if (len == 0 || buffer[0] != '\n')
526 1.1 joerg if (write(fd, "\r", 1) != 1)
527 1.8 joerg error("write('%s')", *path);
528 1.1 joerg cr = 0;
529 1.1 joerg }
530 1.1 joerg
531 1.1 joerg /* EOF */
532 1.1 joerg if (len == 0)
533 1.1 joerg break;
534 1.1 joerg end = buffer + len;
535 1.1 joerg
536 1.1 joerg /*
537 1.1 joerg * Detect whether this is a text file. The correct way to
538 1.1 joerg * do this is to check the least significant bit of the
539 1.1 joerg * "internal file attributes" field of the corresponding
540 1.1 joerg * file header in the central directory, but libarchive
541 1.1 joerg * does not read the central directory, so we have to
542 1.1 joerg * guess by looking for non-ASCII characters in the
543 1.1 joerg * buffer. Hopefully we won't guess wrong. If we do
544 1.1 joerg * guess wrong, we print a warning message later.
545 1.1 joerg */
546 1.1 joerg if (a_opt && n == 0) {
547 1.1 joerg for (p = buffer; p < end; ++p) {
548 1.1 joerg if (!isascii((unsigned char)*p)) {
549 1.1 joerg text = 0;
550 1.1 joerg break;
551 1.1 joerg }
552 1.1 joerg }
553 1.1 joerg }
554 1.1 joerg
555 1.1 joerg /* simple case */
556 1.1 joerg if (!a_opt || !text) {
557 1.1 joerg if (write(fd, buffer, len) != len)
558 1.1 joerg error("write('%s')", path);
559 1.1 joerg continue;
560 1.1 joerg }
561 1.1 joerg
562 1.1 joerg /* hard case: convert \r\n to \n (sigh...) */
563 1.1 joerg for (p = buffer; p < end; p = q + 1) {
564 1.1 joerg for (q = p; q < end; q++) {
565 1.1 joerg if (!warn && !isascii(*q)) {
566 1.1 joerg warningx("%s may be corrupted due"
567 1.1 joerg " to weak text file detection"
568 1.8 joerg " heuristic", *path);
569 1.1 joerg warn = 1;
570 1.1 joerg }
571 1.1 joerg if (q[0] != '\r')
572 1.1 joerg continue;
573 1.1 joerg if (&q[1] == end) {
574 1.1 joerg cr = 1;
575 1.1 joerg break;
576 1.1 joerg }
577 1.1 joerg if (q[1] == '\n')
578 1.1 joerg break;
579 1.1 joerg }
580 1.1 joerg if (write(fd, p, q - p) != q - p)
581 1.8 joerg error("write('%s')", *path);
582 1.1 joerg }
583 1.1 joerg }
584 1.1 joerg if (tty)
585 1.1 joerg info(" \b\b");
586 1.1 joerg if (text)
587 1.1 joerg info(" (text)");
588 1.1 joerg info("\n");
589 1.1 joerg
590 1.1 joerg /* set access and modification time */
591 1.1 joerg tv[0].tv_sec = now;
592 1.1 joerg tv[0].tv_usec = 0;
593 1.1 joerg tv[1].tv_sec = mtime;
594 1.1 joerg tv[1].tv_usec = 0;
595 1.1 joerg if (futimes(fd, tv) != 0)
596 1.8 joerg error("utimes('%s')", *path);
597 1.1 joerg if (close(fd) != 0)
598 1.8 joerg error("close('%s')", *path);
599 1.1 joerg }
600 1.1 joerg
601 1.1 joerg /*
602 1.1 joerg * Extract a zipfile entry: first perform some sanity checks to ensure
603 1.1 joerg * that it is either a directory or a regular file and that the path is
604 1.1 joerg * not absolute and does not try to break out of the current directory;
605 1.1 joerg * then call either extract_dir() or extract_file() as appropriate.
606 1.1 joerg *
607 1.1 joerg * This is complicated a bit by the various ways in which we need to
608 1.1 joerg * manipulate the path name. Case conversion (if requested by the -L
609 1.1 joerg * option) happens first, but the include / exclude patterns are applied
610 1.1 joerg * to the full converted path name, before the directory part of the path
611 1.1 joerg * is removed in accordance with the -j option. Sanity checks are
612 1.1 joerg * intentionally done earlier than they need to be, so the user will get a
613 1.1 joerg * warning about insecure paths even for files or directories which
614 1.1 joerg * wouldn't be extracted anyway.
615 1.1 joerg */
616 1.1 joerg static void
617 1.1 joerg extract(struct archive *a, struct archive_entry *e)
618 1.1 joerg {
619 1.1 joerg char *pathname, *realpathname;
620 1.1 joerg mode_t filetype;
621 1.1 joerg char *p, *q;
622 1.1 joerg
623 1.1 joerg pathname = pathdup(archive_entry_pathname(e));
624 1.1 joerg filetype = archive_entry_filetype(e);
625 1.1 joerg
626 1.1 joerg /* sanity checks */
627 1.1 joerg if (pathname[0] == '/' ||
628 1.1 joerg strncmp(pathname, "../", 3) == 0 ||
629 1.1 joerg strstr(pathname, "/../") != NULL) {
630 1.1 joerg warningx("skipping insecure entry '%s'", pathname);
631 1.1 joerg ac(archive_read_data_skip(a));
632 1.1 joerg free(pathname);
633 1.1 joerg return;
634 1.1 joerg }
635 1.1 joerg
636 1.1 joerg /* I don't think this can happen in a zipfile.. */
637 1.1 joerg if (!S_ISDIR(filetype) && !S_ISREG(filetype)) {
638 1.1 joerg warningx("skipping non-regular entry '%s'", pathname);
639 1.1 joerg ac(archive_read_data_skip(a));
640 1.1 joerg free(pathname);
641 1.1 joerg return;
642 1.1 joerg }
643 1.1 joerg
644 1.1 joerg /* skip directories in -j case */
645 1.1 joerg if (S_ISDIR(filetype) && j_opt) {
646 1.1 joerg ac(archive_read_data_skip(a));
647 1.1 joerg free(pathname);
648 1.1 joerg return;
649 1.1 joerg }
650 1.1 joerg
651 1.1 joerg /* apply include / exclude patterns */
652 1.1 joerg if (!accept_pathname(pathname)) {
653 1.1 joerg ac(archive_read_data_skip(a));
654 1.1 joerg free(pathname);
655 1.1 joerg return;
656 1.1 joerg }
657 1.1 joerg
658 1.1 joerg /* apply -j and -d */
659 1.1 joerg if (j_opt) {
660 1.1 joerg for (p = q = pathname; *p; ++p)
661 1.1 joerg if (*p == '/')
662 1.1 joerg q = p + 1;
663 1.1 joerg realpathname = pathcat(d_arg, q);
664 1.1 joerg } else {
665 1.1 joerg realpathname = pathcat(d_arg, pathname);
666 1.1 joerg }
667 1.1 joerg
668 1.1 joerg /* ensure that parent directory exists */
669 1.1 joerg make_parent(realpathname);
670 1.1 joerg
671 1.1 joerg if (S_ISDIR(filetype))
672 1.1 joerg extract_dir(a, e, realpathname);
673 1.1 joerg else
674 1.8 joerg extract_file(a, e, &realpathname);
675 1.1 joerg
676 1.1 joerg free(realpathname);
677 1.1 joerg free(pathname);
678 1.1 joerg }
679 1.1 joerg
680 1.2 joerg static void
681 1.2 joerg extract_stdout(struct archive *a, struct archive_entry *e)
682 1.2 joerg {
683 1.2 joerg char *pathname;
684 1.2 joerg mode_t filetype;
685 1.2 joerg int cr, text, warn;
686 1.2 joerg ssize_t len;
687 1.2 joerg unsigned char *p, *q, *end;
688 1.2 joerg
689 1.2 joerg pathname = pathdup(archive_entry_pathname(e));
690 1.2 joerg filetype = archive_entry_filetype(e);
691 1.2 joerg
692 1.2 joerg /* I don't think this can happen in a zipfile.. */
693 1.2 joerg if (!S_ISDIR(filetype) && !S_ISREG(filetype)) {
694 1.2 joerg warningx("skipping non-regular entry '%s'", pathname);
695 1.2 joerg ac(archive_read_data_skip(a));
696 1.2 joerg free(pathname);
697 1.2 joerg return;
698 1.2 joerg }
699 1.2 joerg
700 1.2 joerg /* skip directories in -j case */
701 1.2 joerg if (S_ISDIR(filetype)) {
702 1.2 joerg ac(archive_read_data_skip(a));
703 1.2 joerg free(pathname);
704 1.2 joerg return;
705 1.2 joerg }
706 1.2 joerg
707 1.2 joerg /* apply include / exclude patterns */
708 1.2 joerg if (!accept_pathname(pathname)) {
709 1.2 joerg ac(archive_read_data_skip(a));
710 1.2 joerg free(pathname);
711 1.2 joerg return;
712 1.2 joerg }
713 1.2 joerg
714 1.3 joerg if (c_opt)
715 1.3 joerg info("x %s\n", pathname);
716 1.3 joerg
717 1.2 joerg text = a_opt;
718 1.2 joerg warn = 0;
719 1.2 joerg cr = 0;
720 1.2 joerg for (int n = 0; ; n++) {
721 1.2 joerg len = archive_read_data(a, buffer, sizeof buffer);
722 1.2 joerg
723 1.2 joerg if (len < 0)
724 1.2 joerg ac(len);
725 1.2 joerg
726 1.2 joerg /* left over CR from previous buffer */
727 1.2 joerg if (a_opt && cr) {
728 1.3 joerg if (len == 0 || buffer[0] != '\n') {
729 1.3 joerg if (fwrite("\r", 1, 1, stderr) != 1)
730 1.2 joerg error("write('%s')", pathname);
731 1.3 joerg }
732 1.2 joerg cr = 0;
733 1.2 joerg }
734 1.2 joerg
735 1.2 joerg /* EOF */
736 1.2 joerg if (len == 0)
737 1.2 joerg break;
738 1.2 joerg end = buffer + len;
739 1.2 joerg
740 1.2 joerg /*
741 1.2 joerg * Detect whether this is a text file. The correct way to
742 1.2 joerg * do this is to check the least significant bit of the
743 1.2 joerg * "internal file attributes" field of the corresponding
744 1.2 joerg * file header in the central directory, but libarchive
745 1.2 joerg * does not read the central directory, so we have to
746 1.2 joerg * guess by looking for non-ASCII characters in the
747 1.2 joerg * buffer. Hopefully we won't guess wrong. If we do
748 1.2 joerg * guess wrong, we print a warning message later.
749 1.2 joerg */
750 1.2 joerg if (a_opt && n == 0) {
751 1.2 joerg for (p = buffer; p < end; ++p) {
752 1.2 joerg if (!isascii((unsigned char)*p)) {
753 1.2 joerg text = 0;
754 1.2 joerg break;
755 1.2 joerg }
756 1.2 joerg }
757 1.2 joerg }
758 1.2 joerg
759 1.2 joerg /* simple case */
760 1.2 joerg if (!a_opt || !text) {
761 1.3 joerg if (fwrite(buffer, 1, len, stdout) != (size_t)len)
762 1.2 joerg error("write('%s')", pathname);
763 1.2 joerg continue;
764 1.2 joerg }
765 1.2 joerg
766 1.2 joerg /* hard case: convert \r\n to \n (sigh...) */
767 1.2 joerg for (p = buffer; p < end; p = q + 1) {
768 1.2 joerg for (q = p; q < end; q++) {
769 1.2 joerg if (!warn && !isascii(*q)) {
770 1.2 joerg warningx("%s may be corrupted due"
771 1.2 joerg " to weak text file detection"
772 1.2 joerg " heuristic", pathname);
773 1.2 joerg warn = 1;
774 1.2 joerg }
775 1.2 joerg if (q[0] != '\r')
776 1.2 joerg continue;
777 1.2 joerg if (&q[1] == end) {
778 1.2 joerg cr = 1;
779 1.2 joerg break;
780 1.2 joerg }
781 1.2 joerg if (q[1] == '\n')
782 1.2 joerg break;
783 1.2 joerg }
784 1.3 joerg if (fwrite(p, 1, q - p, stdout) != (size_t)(q - p))
785 1.2 joerg error("write('%s')", pathname);
786 1.2 joerg }
787 1.2 joerg }
788 1.2 joerg
789 1.2 joerg free(pathname);
790 1.2 joerg }
791 1.2 joerg
792 1.1 joerg /*
793 1.1 joerg * Print the name of an entry to stdout.
794 1.1 joerg */
795 1.1 joerg static void
796 1.1 joerg list(struct archive *a, struct archive_entry *e)
797 1.1 joerg {
798 1.2 joerg char buf[20];
799 1.2 joerg time_t mtime;
800 1.1 joerg
801 1.3 joerg mtime = archive_entry_mtime(e);
802 1.3 joerg strftime(buf, sizeof(buf), "%m-%d-%g %R", localtime(&mtime));
803 1.2 joerg
804 1.3 joerg if (v_opt == 1) {
805 1.3 joerg printf(" %8ju %s %s\n",
806 1.3 joerg (uintmax_t)archive_entry_size(e),
807 1.3 joerg buf, archive_entry_pathname(e));
808 1.3 joerg } else if (v_opt == 2) {
809 1.6 wiz printf("%8ju Stored %7ju 0%% %s %08x %s\n",
810 1.2 joerg (uintmax_t)archive_entry_size(e),
811 1.2 joerg (uintmax_t)archive_entry_size(e),
812 1.2 joerg buf,
813 1.2 joerg 0U,
814 1.2 joerg archive_entry_pathname(e));
815 1.2 joerg }
816 1.1 joerg ac(archive_read_data_skip(a));
817 1.1 joerg }
818 1.1 joerg
819 1.1 joerg /*
820 1.1 joerg * Extract to memory to check CRC
821 1.1 joerg */
822 1.7 wiz static int
823 1.1 joerg test(struct archive *a, struct archive_entry *e)
824 1.1 joerg {
825 1.1 joerg ssize_t len;
826 1.7 wiz int error_count;
827 1.1 joerg
828 1.7 wiz error_count = 0;
829 1.1 joerg if (S_ISDIR(archive_entry_filetype(e)))
830 1.7 wiz return 0;
831 1.1 joerg
832 1.7 wiz info(" testing: %s\t", archive_entry_pathname(e));
833 1.1 joerg while ((len = archive_read_data(a, buffer, sizeof buffer)) > 0)
834 1.1 joerg /* nothing */;
835 1.1 joerg if (len < 0) {
836 1.7 wiz info(" %s\n", archive_error_string(a));
837 1.7 wiz ++error_count;
838 1.1 joerg } else {
839 1.7 wiz info(" OK\n");
840 1.1 joerg }
841 1.1 joerg
842 1.1 joerg /* shouldn't be necessary, but it doesn't hurt */
843 1.1 joerg ac(archive_read_data_skip(a));
844 1.7 wiz
845 1.7 wiz return error_count;
846 1.1 joerg }
847 1.1 joerg
848 1.1 joerg
849 1.1 joerg /*
850 1.1 joerg * Main loop: open the zipfile, iterate over its contents and decide what
851 1.1 joerg * to do with each entry.
852 1.1 joerg */
853 1.1 joerg static void
854 1.1 joerg unzip(const char *fn)
855 1.1 joerg {
856 1.1 joerg struct archive *a;
857 1.1 joerg struct archive_entry *e;
858 1.1 joerg int fd, ret;
859 1.7 wiz uintmax_t total_size, file_count, error_count;
860 1.1 joerg
861 1.1 joerg if ((fd = open(fn, O_RDONLY)) < 0)
862 1.1 joerg error("%s", fn);
863 1.1 joerg
864 1.1 joerg a = archive_read_new();
865 1.1 joerg ac(archive_read_support_format_zip(a));
866 1.1 joerg ac(archive_read_open_fd(a, fd, 8192));
867 1.1 joerg
868 1.9 wiz if (!q_opt)
869 1.9 wiz printf("Archive: %s\n", fn);
870 1.9 wiz
871 1.3 joerg if (v_opt == 1) {
872 1.6 wiz printf(" Length Date Time Name\n");
873 1.6 wiz printf(" -------- ---- ---- ----\n");
874 1.3 joerg } else if (v_opt == 2) {
875 1.6 wiz printf(" Length Method Size Ratio Date Time CRC-32 Name\n");
876 1.6 wiz printf("-------- ------ ------- ----- ---- ---- ------ ----\n");
877 1.3 joerg }
878 1.3 joerg
879 1.3 joerg total_size = 0;
880 1.3 joerg file_count = 0;
881 1.7 wiz error_count = 0;
882 1.1 joerg for (;;) {
883 1.1 joerg ret = archive_read_next_header(a, &e);
884 1.1 joerg if (ret == ARCHIVE_EOF)
885 1.1 joerg break;
886 1.1 joerg ac(ret);
887 1.1 joerg if (t_opt)
888 1.7 wiz error_count += test(a, e);
889 1.2 joerg else if (v_opt)
890 1.1 joerg list(a, e);
891 1.3 joerg else if (p_opt || c_opt)
892 1.2 joerg extract_stdout(a, e);
893 1.1 joerg else
894 1.1 joerg extract(a, e);
895 1.3 joerg
896 1.3 joerg total_size += archive_entry_size(e);
897 1.3 joerg ++file_count;
898 1.3 joerg }
899 1.3 joerg
900 1.3 joerg if (v_opt == 1) {
901 1.3 joerg printf(" -------- -------\n");
902 1.3 joerg printf(" %8ju %ju file%s\n",
903 1.3 joerg total_size, file_count, file_count != 1 ? "s" : "");
904 1.3 joerg } else if (v_opt == 2) {
905 1.6 wiz printf("-------- ------- --- -------\n");
906 1.6 wiz printf("%8ju %7ju 0%% %ju file%s\n",
907 1.3 joerg total_size, total_size, file_count,
908 1.3 joerg file_count != 1 ? "s" : "");
909 1.1 joerg }
910 1.1 joerg
911 1.1 joerg ac(archive_read_close(a));
912 1.1 joerg (void)archive_read_finish(a);
913 1.7 wiz
914 1.1 joerg if (close(fd) != 0)
915 1.1 joerg error("%s", fn);
916 1.1 joerg
917 1.7 wiz if (t_opt) {
918 1.7 wiz if (error_count > 0) {
919 1.7 wiz errorx("%d checksum error(s) found.", error_count);
920 1.7 wiz }
921 1.7 wiz else {
922 1.7 wiz printf("No errors detected in compressed data of %s.\n",
923 1.7 wiz fn);
924 1.7 wiz }
925 1.7 wiz }
926 1.1 joerg }
927 1.1 joerg
928 1.1 joerg static void
929 1.1 joerg usage(void)
930 1.1 joerg {
931 1.1 joerg
932 1.7 wiz fprintf(stderr, "usage: unzip [-aCcfjLlnopqtuv] [-d dir] [-x pattern] zipfile\n");
933 1.1 joerg exit(1);
934 1.1 joerg }
935 1.1 joerg
936 1.1 joerg static int
937 1.1 joerg getopts(int argc, char *argv[])
938 1.1 joerg {
939 1.1 joerg int opt;
940 1.1 joerg
941 1.1 joerg optreset = optind = 1;
942 1.7 wiz while ((opt = getopt(argc, argv, "aCcd:fjLlnopqtuvx:")) != -1)
943 1.1 joerg switch (opt) {
944 1.1 joerg case 'a':
945 1.1 joerg a_opt = 1;
946 1.1 joerg break;
947 1.7 wiz case 'C':
948 1.7 wiz C_opt = 1;
949 1.7 wiz break;
950 1.3 joerg case 'c':
951 1.3 joerg c_opt = 1;
952 1.3 joerg break;
953 1.1 joerg case 'd':
954 1.1 joerg d_arg = optarg;
955 1.1 joerg break;
956 1.1 joerg case 'f':
957 1.1 joerg f_opt = 1;
958 1.1 joerg break;
959 1.1 joerg case 'j':
960 1.1 joerg j_opt = 1;
961 1.1 joerg break;
962 1.1 joerg case 'L':
963 1.1 joerg L_opt = 1;
964 1.1 joerg break;
965 1.1 joerg case 'l':
966 1.2 joerg if (v_opt == 0)
967 1.2 joerg v_opt = 1;
968 1.1 joerg break;
969 1.1 joerg case 'n':
970 1.1 joerg n_opt = 1;
971 1.1 joerg break;
972 1.1 joerg case 'o':
973 1.1 joerg o_opt = 1;
974 1.2 joerg q_opt = 1;
975 1.2 joerg break;
976 1.2 joerg case 'p':
977 1.2 joerg p_opt = 1;
978 1.1 joerg break;
979 1.1 joerg case 'q':
980 1.1 joerg q_opt = 1;
981 1.1 joerg break;
982 1.1 joerg case 't':
983 1.1 joerg t_opt = 1;
984 1.1 joerg break;
985 1.1 joerg case 'u':
986 1.1 joerg u_opt = 1;
987 1.1 joerg break;
988 1.2 joerg case 'v':
989 1.2 joerg v_opt = 2;
990 1.2 joerg break;
991 1.1 joerg case 'x':
992 1.1 joerg add_pattern(&exclude, optarg);
993 1.1 joerg break;
994 1.1 joerg default:
995 1.1 joerg usage();
996 1.1 joerg }
997 1.1 joerg
998 1.1 joerg return (optind);
999 1.1 joerg }
1000 1.1 joerg
1001 1.1 joerg int
1002 1.1 joerg main(int argc, char *argv[])
1003 1.1 joerg {
1004 1.1 joerg const char *zipfile;
1005 1.1 joerg int nopts;
1006 1.1 joerg
1007 1.1 joerg if (isatty(STDOUT_FILENO))
1008 1.1 joerg tty = 1;
1009 1.1 joerg
1010 1.1 joerg if (getenv("UNZIP_DEBUG") != NULL)
1011 1.1 joerg unzip_debug = 1;
1012 1.1 joerg for (int i = 0; i < argc; ++i)
1013 1.1 joerg debug("%s%c", argv[i], (i < argc - 1) ? ' ' : '\n');
1014 1.1 joerg
1015 1.1 joerg /*
1016 1.1 joerg * Info-ZIP's unzip(1) expects certain options to come before the
1017 1.1 joerg * zipfile name, and others to come after - though it does not
1018 1.1 joerg * enforce this. For simplicity, we accept *all* options both
1019 1.1 joerg * before and after the zipfile name.
1020 1.1 joerg */
1021 1.1 joerg nopts = getopts(argc, argv);
1022 1.1 joerg
1023 1.1 joerg if (argc <= nopts)
1024 1.1 joerg usage();
1025 1.1 joerg zipfile = argv[nopts++];
1026 1.1 joerg
1027 1.1 joerg while (nopts < argc && *argv[nopts] != '-')
1028 1.1 joerg add_pattern(&include, argv[nopts++]);
1029 1.1 joerg
1030 1.1 joerg nopts--; /* fake argv[0] */
1031 1.1 joerg nopts += getopts(argc - nopts, argv + nopts);
1032 1.1 joerg
1033 1.1 joerg if (n_opt + o_opt + u_opt > 1)
1034 1.1 joerg errorx("-n, -o and -u are contradictory");
1035 1.1 joerg
1036 1.1 joerg time(&now);
1037 1.1 joerg
1038 1.1 joerg unzip(zipfile);
1039 1.1 joerg
1040 1.1 joerg exit(0);
1041 1.1 joerg }
1042