unzip.c revision 1.24 1 1.24 joerg /* $NetBSD: unzip.c,v 1.24 2018/07/19 18:04:25 joerg Exp $ */
2 1.1 joerg
3 1.1 joerg /*-
4 1.13 joerg * Copyright (c) 2009, 2010 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 1.24 joerg __RCSID("$NetBSD: unzip.c,v 1.24 2018/07/19 18:04:25 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 1.7 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 1.17 christos static const char * y_str = ""; /* 4 digit year */
74 1.1 joerg
75 1.1 joerg /* time when unzip started */
76 1.1 joerg static time_t now;
77 1.1 joerg
78 1.1 joerg /* debug flag */
79 1.1 joerg static int unzip_debug;
80 1.1 joerg
81 1.1 joerg /* running on tty? */
82 1.1 joerg static int tty;
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 1.18 joerg __dead __printflike(1, 2) 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 1.18 joerg __dead __printflike(1, 2) 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 /* non-fatal error message + errno */
135 1.18 joerg __printflike(1, 2) 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 1.20 christos
150 1.1 joerg /* non-fatal error message, no errno */
151 1.18 joerg __printflike(1, 2) static void
152 1.1 joerg warningx(const char *fmt, ...)
153 1.1 joerg {
154 1.1 joerg va_list ap;
155 1.1 joerg
156 1.1 joerg if (noeol)
157 1.1 joerg fprintf(stdout, "\n");
158 1.1 joerg fflush(stdout);
159 1.1 joerg fprintf(stderr, "unzip: ");
160 1.1 joerg va_start(ap, fmt);
161 1.1 joerg vfprintf(stderr, fmt, ap);
162 1.1 joerg va_end(ap);
163 1.1 joerg fprintf(stderr, "\n");
164 1.1 joerg }
165 1.1 joerg
166 1.1 joerg /* informational message (if not -q) */
167 1.18 joerg __printflike(1, 2) static void
168 1.1 joerg info(const char *fmt, ...)
169 1.1 joerg {
170 1.1 joerg va_list ap;
171 1.1 joerg
172 1.1 joerg if (q_opt && !unzip_debug)
173 1.1 joerg return;
174 1.1 joerg va_start(ap, fmt);
175 1.1 joerg vfprintf(stdout, fmt, ap);
176 1.1 joerg va_end(ap);
177 1.1 joerg fflush(stdout);
178 1.1 joerg
179 1.1 joerg if (*fmt == '\0')
180 1.1 joerg noeol = 1;
181 1.1 joerg else
182 1.1 joerg noeol = fmt[strlen(fmt) - 1] != '\n';
183 1.1 joerg }
184 1.1 joerg
185 1.1 joerg /* debug message (if unzip_debug) */
186 1.18 joerg __printflike(1, 2) static void
187 1.1 joerg debug(const char *fmt, ...)
188 1.1 joerg {
189 1.1 joerg va_list ap;
190 1.1 joerg
191 1.1 joerg if (!unzip_debug)
192 1.1 joerg return;
193 1.1 joerg va_start(ap, fmt);
194 1.1 joerg vfprintf(stderr, fmt, ap);
195 1.1 joerg va_end(ap);
196 1.1 joerg fflush(stderr);
197 1.1 joerg
198 1.1 joerg if (*fmt == '\0')
199 1.1 joerg noeol = 1;
200 1.1 joerg else
201 1.1 joerg noeol = fmt[strlen(fmt) - 1] != '\n';
202 1.1 joerg }
203 1.1 joerg
204 1.1 joerg /* duplicate a path name, possibly converting to lower case */
205 1.1 joerg static char *
206 1.1 joerg pathdup(const char *path)
207 1.1 joerg {
208 1.1 joerg char *str;
209 1.1 joerg size_t i, len;
210 1.1 joerg
211 1.1 joerg len = strlen(path);
212 1.1 joerg while (len && path[len - 1] == '/')
213 1.1 joerg len--;
214 1.1 joerg if ((str = malloc(len + 1)) == NULL) {
215 1.1 joerg errno = ENOMEM;
216 1.1 joerg error("malloc()");
217 1.1 joerg }
218 1.1 joerg if (L_opt) {
219 1.1 joerg for (i = 0; i < len; ++i)
220 1.1 joerg str[i] = tolower((unsigned char)path[i]);
221 1.1 joerg } else {
222 1.1 joerg memcpy(str, path, len);
223 1.1 joerg }
224 1.1 joerg str[len] = '\0';
225 1.1 joerg
226 1.1 joerg return (str);
227 1.1 joerg }
228 1.1 joerg
229 1.1 joerg /* concatenate two path names */
230 1.1 joerg static char *
231 1.1 joerg pathcat(const char *prefix, const char *path)
232 1.1 joerg {
233 1.1 joerg char *str;
234 1.1 joerg size_t prelen, len;
235 1.1 joerg
236 1.1 joerg prelen = prefix ? strlen(prefix) + 1 : 0;
237 1.1 joerg len = strlen(path) + 1;
238 1.1 joerg if ((str = malloc(prelen + len)) == NULL) {
239 1.1 joerg errno = ENOMEM;
240 1.1 joerg error("malloc()");
241 1.1 joerg }
242 1.1 joerg if (prefix) {
243 1.1 joerg memcpy(str, prefix, prelen); /* includes zero */
244 1.1 joerg str[prelen - 1] = '/'; /* splat zero */
245 1.1 joerg }
246 1.1 joerg memcpy(str + prelen, path, len); /* includes zero */
247 1.1 joerg
248 1.1 joerg return (str);
249 1.1 joerg }
250 1.1 joerg
251 1.1 joerg /*
252 1.1 joerg * Pattern lists for include / exclude processing
253 1.1 joerg */
254 1.1 joerg struct pattern {
255 1.1 joerg STAILQ_ENTRY(pattern) link;
256 1.1 joerg char pattern[];
257 1.1 joerg };
258 1.1 joerg
259 1.1 joerg STAILQ_HEAD(pattern_list, pattern);
260 1.1 joerg static struct pattern_list include = STAILQ_HEAD_INITIALIZER(include);
261 1.1 joerg static struct pattern_list exclude = STAILQ_HEAD_INITIALIZER(exclude);
262 1.1 joerg
263 1.1 joerg /*
264 1.1 joerg * Add an entry to a pattern list
265 1.1 joerg */
266 1.1 joerg static void
267 1.1 joerg add_pattern(struct pattern_list *list, const char *pattern)
268 1.1 joerg {
269 1.1 joerg struct pattern *entry;
270 1.1 joerg size_t len;
271 1.1 joerg
272 1.1 joerg debug("adding pattern '%s'\n", pattern);
273 1.1 joerg len = strlen(pattern);
274 1.1 joerg if ((entry = malloc(sizeof *entry + len + 1)) == NULL) {
275 1.1 joerg errno = ENOMEM;
276 1.1 joerg error("malloc()");
277 1.1 joerg }
278 1.1 joerg memcpy(entry->pattern, pattern, len + 1);
279 1.1 joerg STAILQ_INSERT_TAIL(list, entry, link);
280 1.1 joerg }
281 1.1 joerg
282 1.1 joerg /*
283 1.1 joerg * Match a string against a list of patterns
284 1.1 joerg */
285 1.1 joerg static int
286 1.1 joerg match_pattern(struct pattern_list *list, const char *str)
287 1.1 joerg {
288 1.1 joerg struct pattern *entry;
289 1.1 joerg
290 1.1 joerg STAILQ_FOREACH(entry, list, link) {
291 1.7 wiz if (fnmatch(entry->pattern, str, C_opt ? FNM_CASEFOLD : 0) == 0)
292 1.1 joerg return (1);
293 1.1 joerg }
294 1.1 joerg return (0);
295 1.1 joerg }
296 1.1 joerg
297 1.1 joerg /*
298 1.1 joerg * Verify that a given pathname is in the include list and not in the
299 1.1 joerg * exclude list.
300 1.1 joerg */
301 1.1 joerg static int
302 1.1 joerg accept_pathname(const char *pathname)
303 1.1 joerg {
304 1.1 joerg
305 1.1 joerg if (!STAILQ_EMPTY(&include) && !match_pattern(&include, pathname))
306 1.1 joerg return (0);
307 1.1 joerg if (!STAILQ_EMPTY(&exclude) && match_pattern(&exclude, pathname))
308 1.1 joerg return (0);
309 1.1 joerg return (1);
310 1.1 joerg }
311 1.1 joerg
312 1.1 joerg /*
313 1.1 joerg * Create the specified directory with the specified mode, taking certain
314 1.1 joerg * precautions on they way.
315 1.1 joerg */
316 1.1 joerg static void
317 1.1 joerg make_dir(const char *path, int mode)
318 1.1 joerg {
319 1.1 joerg struct stat sb;
320 1.1 joerg
321 1.1 joerg if (lstat(path, &sb) == 0) {
322 1.1 joerg if (S_ISDIR(sb.st_mode))
323 1.1 joerg return;
324 1.1 joerg /*
325 1.1 joerg * Normally, we should either ask the user about removing
326 1.1 joerg * the non-directory of the same name as a directory we
327 1.1 joerg * wish to create, or respect the -n or -o command-line
328 1.1 joerg * options. However, this may lead to a later failure or
329 1.1 joerg * even compromise (if this non-directory happens to be a
330 1.1 joerg * symlink to somewhere unsafe), so we don't.
331 1.1 joerg */
332 1.1 joerg
333 1.1 joerg /*
334 1.1 joerg * Don't check unlink() result; failure will cause mkdir()
335 1.1 joerg * to fail later, which we will catch.
336 1.1 joerg */
337 1.1 joerg (void)unlink(path);
338 1.1 joerg }
339 1.1 joerg if (mkdir(path, mode) != 0 && errno != EEXIST)
340 1.1 joerg error("mkdir('%s')", path);
341 1.1 joerg }
342 1.1 joerg
343 1.1 joerg /*
344 1.1 joerg * Ensure that all directories leading up to (but not including) the
345 1.1 joerg * specified path exist.
346 1.1 joerg *
347 1.1 joerg * XXX inefficient + modifies the file in-place
348 1.1 joerg */
349 1.1 joerg static void
350 1.1 joerg make_parent(char *path)
351 1.1 joerg {
352 1.1 joerg struct stat sb;
353 1.1 joerg char *sep;
354 1.1 joerg
355 1.1 joerg sep = strrchr(path, '/');
356 1.1 joerg if (sep == NULL || sep == path)
357 1.1 joerg return;
358 1.1 joerg *sep = '\0';
359 1.1 joerg if (lstat(path, &sb) == 0) {
360 1.1 joerg if (S_ISDIR(sb.st_mode)) {
361 1.1 joerg *sep = '/';
362 1.1 joerg return;
363 1.1 joerg }
364 1.1 joerg unlink(path);
365 1.1 joerg }
366 1.1 joerg make_parent(path);
367 1.1 joerg mkdir(path, 0755);
368 1.1 joerg *sep = '/';
369 1.1 joerg
370 1.1 joerg #if 0
371 1.1 joerg for (sep = path; (sep = strchr(sep, '/')) != NULL; sep++) {
372 1.1 joerg /* root in case of absolute d_arg */
373 1.1 joerg if (sep == path)
374 1.1 joerg continue;
375 1.1 joerg *sep = '\0';
376 1.1 joerg make_dir(path, 0755);
377 1.1 joerg *sep = '/';
378 1.1 joerg }
379 1.1 joerg #endif
380 1.1 joerg }
381 1.1 joerg
382 1.1 joerg /*
383 1.1 joerg * Extract a directory.
384 1.1 joerg */
385 1.1 joerg static void
386 1.1 joerg extract_dir(struct archive *a, struct archive_entry *e, const char *path)
387 1.1 joerg {
388 1.1 joerg int mode;
389 1.1 joerg
390 1.24 joerg /*
391 1.24 joerg * Dropbox likes to create '/' directory entries, just ignore
392 1.24 joerg * such junk.
393 1.24 joerg */
394 1.24 joerg if (*path == '\0')
395 1.24 joerg return;
396 1.24 joerg
397 1.11 mbalmer mode = archive_entry_mode(e) & 0777;
398 1.1 joerg if (mode == 0)
399 1.1 joerg mode = 0755;
400 1.1 joerg
401 1.1 joerg /*
402 1.1 joerg * Some zipfiles contain directories with weird permissions such
403 1.1 joerg * as 0644 or 0444. This can cause strange issues such as being
404 1.1 joerg * unable to extract files into the directory we just created, or
405 1.1 joerg * the user being unable to remove the directory later without
406 1.1 joerg * first manually changing its permissions. Therefore, we whack
407 1.1 joerg * the permissions into shape, assuming that the user wants full
408 1.1 joerg * access and that anyone who gets read access also gets execute
409 1.1 joerg * access.
410 1.1 joerg */
411 1.1 joerg mode |= 0700;
412 1.1 joerg if (mode & 0040)
413 1.1 joerg mode |= 0010;
414 1.1 joerg if (mode & 0004)
415 1.1 joerg mode |= 0001;
416 1.1 joerg
417 1.10 wiz info(" creating: %s/\n", path);
418 1.1 joerg make_dir(path, mode);
419 1.1 joerg ac(archive_read_data_skip(a));
420 1.1 joerg }
421 1.1 joerg
422 1.1 joerg static unsigned char buffer[8192];
423 1.1 joerg static char spinner[] = { '|', '/', '-', '\\' };
424 1.1 joerg
425 1.8 joerg static int
426 1.8 joerg handle_existing_file(char **path)
427 1.8 joerg {
428 1.8 joerg size_t alen;
429 1.8 joerg ssize_t len;
430 1.8 joerg char buf[4];
431 1.8 joerg
432 1.8 joerg for (;;) {
433 1.8 joerg fprintf(stderr,
434 1.8 joerg "replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ",
435 1.8 joerg *path);
436 1.8 joerg fgets(buf, 4, stdin);
437 1.8 joerg switch (*buf) {
438 1.8 joerg case 'A':
439 1.8 joerg o_opt = 1;
440 1.8 joerg /* FALL THROUGH */
441 1.8 joerg case 'y':
442 1.8 joerg case 'Y':
443 1.8 joerg (void)unlink(*path);
444 1.8 joerg return 1;
445 1.8 joerg case 'N':
446 1.22 christos n_opt = 1;
447 1.8 joerg /* FALL THROUGH */
448 1.8 joerg case 'n':
449 1.8 joerg return -1;
450 1.8 joerg case 'r':
451 1.8 joerg case 'R':
452 1.8 joerg printf("New name: ");
453 1.8 joerg fflush(stdout);
454 1.8 joerg free(*path);
455 1.8 joerg *path = NULL;
456 1.8 joerg alen = 0;
457 1.8 joerg len = getline(path, &alen, stdin);
458 1.15 wiz if ((*path)[len - 1] == '\n')
459 1.8 joerg (*path)[len - 1] = '\0';
460 1.8 joerg return 0;
461 1.8 joerg default:
462 1.8 joerg break;
463 1.8 joerg }
464 1.8 joerg }
465 1.8 joerg }
466 1.8 joerg
467 1.1 joerg /*
468 1.13 joerg * Detect binary files by a combination of character white list and
469 1.13 joerg * black list. NUL bytes and other control codes without use in text files
470 1.13 joerg * result directly in switching the file to binary mode. Otherwise, at least
471 1.13 joerg * one white-listed byte has to be found.
472 1.13 joerg *
473 1.13 joerg * Black-listed: 0..6, 14..25, 28..31
474 1.13 joerg * White-listed: 9..10, 13, >= 32
475 1.13 joerg *
476 1.13 joerg * See the proginfo/txtvsbin.txt in the zip sources for a detailed discussion.
477 1.13 joerg */
478 1.13 joerg #define BYTE_IS_BINARY(x) ((x) < 32 && (0xf3ffc07fU & (1U << (x))))
479 1.13 joerg #define BYTE_IS_TEXT(x) ((x) >= 32 || (0x00002600U & (1U << (x))))
480 1.13 joerg
481 1.13 joerg static int
482 1.13 joerg check_binary(const unsigned char *buf, size_t len)
483 1.13 joerg {
484 1.13 joerg int rv;
485 1.13 joerg for (rv = 1; len--; ++buf) {
486 1.13 joerg if (BYTE_IS_BINARY(*buf))
487 1.13 joerg return 1;
488 1.13 joerg if (BYTE_IS_TEXT(*buf))
489 1.13 joerg rv = 0;
490 1.13 joerg }
491 1.13 joerg
492 1.13 joerg return rv;
493 1.13 joerg }
494 1.13 joerg
495 1.13 joerg /*
496 1.22 christos * Extract to a file descriptor
497 1.22 christos */
498 1.22 christos static int
499 1.22 christos extract2fd(struct archive *a, char *pathname, int fd)
500 1.22 christos {
501 1.22 christos int cr, text, warn;
502 1.22 christos ssize_t len;
503 1.22 christos unsigned char *p, *q, *end;
504 1.22 christos
505 1.22 christos text = a_opt;
506 1.22 christos warn = 0;
507 1.22 christos cr = 0;
508 1.22 christos
509 1.22 christos /* loop over file contents and write to fd */
510 1.22 christos for (int n = 0; ; n++) {
511 1.22 christos if (fd != STDOUT_FILENO)
512 1.22 christos if (tty && (n % 4) == 0)
513 1.22 christos info(" %c\b\b", spinner[(n / 4) % sizeof spinner]);
514 1.22 christos
515 1.22 christos len = archive_read_data(a, buffer, sizeof buffer);
516 1.22 christos
517 1.22 christos if (len < 0)
518 1.22 christos ac(len);
519 1.22 christos
520 1.22 christos /* left over CR from previous buffer */
521 1.22 christos if (a_opt && cr) {
522 1.22 christos if (len == 0 || buffer[0] != '\n')
523 1.22 christos if (write(fd, "\r", 1) != 1)
524 1.22 christos error("write('%s')", pathname);
525 1.22 christos cr = 0;
526 1.22 christos }
527 1.22 christos
528 1.22 christos /* EOF */
529 1.22 christos if (len == 0)
530 1.22 christos break;
531 1.22 christos end = buffer + len;
532 1.22 christos
533 1.22 christos /*
534 1.22 christos * Detect whether this is a text file. The correct way to
535 1.22 christos * do this is to check the least significant bit of the
536 1.22 christos * "internal file attributes" field of the corresponding
537 1.22 christos * file header in the central directory, but libarchive
538 1.22 christos * does not provide access to this field, so we have to
539 1.22 christos * guess by looking for non-ASCII characters in the
540 1.22 christos * buffer. Hopefully we won't guess wrong. If we do
541 1.22 christos * guess wrong, we print a warning message later.
542 1.22 christos */
543 1.22 christos if (a_opt && n == 0) {
544 1.22 christos if (check_binary(buffer, len))
545 1.22 christos text = 0;
546 1.22 christos }
547 1.22 christos
548 1.22 christos /* simple case */
549 1.22 christos if (!a_opt || !text) {
550 1.22 christos if (write(fd, buffer, len) != len)
551 1.22 christos error("write('%s')", pathname);
552 1.22 christos continue;
553 1.22 christos }
554 1.22 christos
555 1.22 christos /* hard case: convert \r\n to \n (sigh...) */
556 1.22 christos for (p = buffer; p < end; p = q + 1) {
557 1.22 christos for (q = p; q < end; q++) {
558 1.22 christos if (!warn && BYTE_IS_BINARY(*q)) {
559 1.22 christos warningx("%s may be corrupted due"
560 1.22 christos " to weak text file detection"
561 1.22 christos " heuristic", pathname);
562 1.22 christos warn = 1;
563 1.22 christos }
564 1.22 christos if (q[0] != '\r')
565 1.22 christos continue;
566 1.22 christos if (&q[1] == end) {
567 1.22 christos cr = 1;
568 1.22 christos break;
569 1.22 christos }
570 1.22 christos if (q[1] == '\n')
571 1.22 christos break;
572 1.22 christos }
573 1.22 christos if (write(fd, p, q - p) != q - p)
574 1.22 christos error("write('%s')", pathname);
575 1.22 christos }
576 1.22 christos }
577 1.22 christos
578 1.22 christos return text;
579 1.22 christos }
580 1.22 christos
581 1.22 christos /*
582 1.1 joerg * Extract a regular file.
583 1.1 joerg */
584 1.1 joerg static void
585 1.8 joerg extract_file(struct archive *a, struct archive_entry *e, char **path)
586 1.1 joerg {
587 1.1 joerg int mode;
588 1.1 joerg time_t mtime;
589 1.1 joerg struct stat sb;
590 1.1 joerg struct timeval tv[2];
591 1.22 christos int fd, check, text;
592 1.20 christos const char *linkname;
593 1.1 joerg
594 1.11 mbalmer mode = archive_entry_mode(e) & 0777;
595 1.1 joerg if (mode == 0)
596 1.1 joerg mode = 0644;
597 1.1 joerg mtime = archive_entry_mtime(e);
598 1.1 joerg
599 1.1 joerg /* look for existing file of same name */
600 1.8 joerg recheck:
601 1.8 joerg if (lstat(*path, &sb) == 0) {
602 1.1 joerg if (u_opt || f_opt) {
603 1.1 joerg /* check if up-to-date */
604 1.1 joerg if (S_ISREG(sb.st_mode) && sb.st_mtime >= mtime)
605 1.1 joerg return;
606 1.8 joerg (void)unlink(*path);
607 1.1 joerg } else if (o_opt) {
608 1.1 joerg /* overwrite */
609 1.8 joerg (void)unlink(*path);
610 1.1 joerg } else if (n_opt) {
611 1.1 joerg /* do not overwrite */
612 1.1 joerg return;
613 1.1 joerg } else {
614 1.8 joerg check = handle_existing_file(path);
615 1.8 joerg if (check == 0)
616 1.8 joerg goto recheck;
617 1.8 joerg if (check == -1)
618 1.8 joerg return; /* do not overwrite */
619 1.1 joerg }
620 1.1 joerg } else {
621 1.1 joerg if (f_opt)
622 1.1 joerg return;
623 1.1 joerg }
624 1.1 joerg
625 1.21 christos tv[0].tv_sec = now;
626 1.21 christos tv[0].tv_usec = 0;
627 1.21 christos tv[1].tv_sec = mtime;
628 1.21 christos tv[1].tv_usec = 0;
629 1.21 christos
630 1.20 christos /* process symlinks */
631 1.20 christos linkname = archive_entry_symlink(e);
632 1.20 christos if (linkname != NULL) {
633 1.20 christos if (symlink(linkname, *path) == -1)
634 1.20 christos error("symlink('%s', '%s')", linkname, *path);
635 1.20 christos info(" extracting: %s -> %s\n", *path, linkname);
636 1.20 christos if (lchmod(*path, mode) == -1)
637 1.20 christos warning("Cannot set mode for '%s'", *path);
638 1.20 christos if (lutimes(*path, tv) == -1)
639 1.20 christos warning("utimes('%s')", *path);
640 1.20 christos return;
641 1.20 christos }
642 1.20 christos
643 1.20 christos /* process hardlinks */
644 1.20 christos linkname = archive_entry_hardlink(e);
645 1.20 christos if (linkname != NULL) {
646 1.20 christos if (link(linkname, *path) == -1)
647 1.20 christos error("link('%s', '%s')", linkname, *path);
648 1.20 christos info(" extracting: %s link to %s\n", *path, linkname);
649 1.20 christos return;
650 1.20 christos }
651 1.20 christos
652 1.8 joerg if ((fd = open(*path, O_RDWR|O_CREAT|O_TRUNC, mode)) < 0)
653 1.8 joerg error("open('%s')", *path);
654 1.1 joerg
655 1.8 joerg info(" extracting: %s", *path);
656 1.1 joerg
657 1.22 christos text = extract2fd(a, *path, fd);
658 1.1 joerg
659 1.1 joerg if (tty)
660 1.1 joerg info(" \b\b");
661 1.1 joerg if (text)
662 1.1 joerg info(" (text)");
663 1.1 joerg info("\n");
664 1.1 joerg
665 1.1 joerg /* set access and modification time */
666 1.1 joerg if (futimes(fd, tv) != 0)
667 1.8 joerg error("utimes('%s')", *path);
668 1.1 joerg if (close(fd) != 0)
669 1.8 joerg error("close('%s')", *path);
670 1.1 joerg }
671 1.1 joerg
672 1.1 joerg /*
673 1.1 joerg * Extract a zipfile entry: first perform some sanity checks to ensure
674 1.1 joerg * that it is either a directory or a regular file and that the path is
675 1.1 joerg * not absolute and does not try to break out of the current directory;
676 1.1 joerg * then call either extract_dir() or extract_file() as appropriate.
677 1.1 joerg *
678 1.1 joerg * This is complicated a bit by the various ways in which we need to
679 1.1 joerg * manipulate the path name. Case conversion (if requested by the -L
680 1.1 joerg * option) happens first, but the include / exclude patterns are applied
681 1.1 joerg * to the full converted path name, before the directory part of the path
682 1.1 joerg * is removed in accordance with the -j option. Sanity checks are
683 1.1 joerg * intentionally done earlier than they need to be, so the user will get a
684 1.1 joerg * warning about insecure paths even for files or directories which
685 1.1 joerg * wouldn't be extracted anyway.
686 1.1 joerg */
687 1.1 joerg static void
688 1.1 joerg extract(struct archive *a, struct archive_entry *e)
689 1.1 joerg {
690 1.1 joerg char *pathname, *realpathname;
691 1.1 joerg mode_t filetype;
692 1.1 joerg char *p, *q;
693 1.1 joerg
694 1.1 joerg pathname = pathdup(archive_entry_pathname(e));
695 1.1 joerg filetype = archive_entry_filetype(e);
696 1.1 joerg
697 1.1 joerg /* sanity checks */
698 1.1 joerg if (pathname[0] == '/' ||
699 1.1 joerg strncmp(pathname, "../", 3) == 0 ||
700 1.1 joerg strstr(pathname, "/../") != NULL) {
701 1.1 joerg warningx("skipping insecure entry '%s'", pathname);
702 1.1 joerg ac(archive_read_data_skip(a));
703 1.1 joerg free(pathname);
704 1.1 joerg return;
705 1.1 joerg }
706 1.1 joerg
707 1.1 joerg /* I don't think this can happen in a zipfile.. */
708 1.20 christos if (!S_ISDIR(filetype) && !S_ISREG(filetype) && !S_ISLNK(filetype)) {
709 1.1 joerg warningx("skipping non-regular entry '%s'", pathname);
710 1.1 joerg ac(archive_read_data_skip(a));
711 1.1 joerg free(pathname);
712 1.1 joerg return;
713 1.1 joerg }
714 1.1 joerg
715 1.1 joerg /* skip directories in -j case */
716 1.1 joerg if (S_ISDIR(filetype) && j_opt) {
717 1.1 joerg ac(archive_read_data_skip(a));
718 1.1 joerg free(pathname);
719 1.1 joerg return;
720 1.1 joerg }
721 1.1 joerg
722 1.1 joerg /* apply include / exclude patterns */
723 1.1 joerg if (!accept_pathname(pathname)) {
724 1.1 joerg ac(archive_read_data_skip(a));
725 1.1 joerg free(pathname);
726 1.1 joerg return;
727 1.1 joerg }
728 1.1 joerg
729 1.1 joerg /* apply -j and -d */
730 1.1 joerg if (j_opt) {
731 1.1 joerg for (p = q = pathname; *p; ++p)
732 1.1 joerg if (*p == '/')
733 1.1 joerg q = p + 1;
734 1.1 joerg realpathname = pathcat(d_arg, q);
735 1.1 joerg } else {
736 1.1 joerg realpathname = pathcat(d_arg, pathname);
737 1.1 joerg }
738 1.1 joerg
739 1.1 joerg /* ensure that parent directory exists */
740 1.1 joerg make_parent(realpathname);
741 1.1 joerg
742 1.1 joerg if (S_ISDIR(filetype))
743 1.1 joerg extract_dir(a, e, realpathname);
744 1.1 joerg else
745 1.8 joerg extract_file(a, e, &realpathname);
746 1.1 joerg
747 1.1 joerg free(realpathname);
748 1.1 joerg free(pathname);
749 1.1 joerg }
750 1.1 joerg
751 1.2 joerg static void
752 1.2 joerg extract_stdout(struct archive *a, struct archive_entry *e)
753 1.2 joerg {
754 1.2 joerg char *pathname;
755 1.2 joerg mode_t filetype;
756 1.2 joerg
757 1.2 joerg pathname = pathdup(archive_entry_pathname(e));
758 1.2 joerg filetype = archive_entry_filetype(e);
759 1.2 joerg
760 1.2 joerg /* I don't think this can happen in a zipfile.. */
761 1.20 christos if (!S_ISDIR(filetype) && !S_ISREG(filetype) && !S_ISLNK(filetype)) {
762 1.2 joerg warningx("skipping non-regular entry '%s'", pathname);
763 1.2 joerg ac(archive_read_data_skip(a));
764 1.2 joerg free(pathname);
765 1.2 joerg return;
766 1.2 joerg }
767 1.2 joerg
768 1.2 joerg /* skip directories in -j case */
769 1.2 joerg if (S_ISDIR(filetype)) {
770 1.2 joerg ac(archive_read_data_skip(a));
771 1.2 joerg free(pathname);
772 1.2 joerg return;
773 1.2 joerg }
774 1.2 joerg
775 1.2 joerg /* apply include / exclude patterns */
776 1.2 joerg if (!accept_pathname(pathname)) {
777 1.2 joerg ac(archive_read_data_skip(a));
778 1.2 joerg free(pathname);
779 1.2 joerg return;
780 1.2 joerg }
781 1.2 joerg
782 1.3 joerg if (c_opt)
783 1.3 joerg info("x %s\n", pathname);
784 1.3 joerg
785 1.22 christos (void)extract2fd(a, pathname, STDOUT_FILENO);
786 1.2 joerg
787 1.2 joerg free(pathname);
788 1.2 joerg }
789 1.2 joerg
790 1.1 joerg /*
791 1.1 joerg * Print the name of an entry to stdout.
792 1.1 joerg */
793 1.1 joerg static void
794 1.1 joerg list(struct archive *a, struct archive_entry *e)
795 1.1 joerg {
796 1.2 joerg char buf[20];
797 1.2 joerg time_t mtime;
798 1.17 christos struct tm *tm;
799 1.1 joerg
800 1.3 joerg mtime = archive_entry_mtime(e);
801 1.17 christos tm = localtime(&mtime);
802 1.17 christos if (*y_str)
803 1.17 christos strftime(buf, sizeof(buf), "%m-%d-%G %R", tm);
804 1.17 christos else
805 1.17 christos strftime(buf, sizeof(buf), "%m-%d-%g %R", tm);
806 1.2 joerg
807 1.3 joerg if (v_opt == 1) {
808 1.3 joerg printf(" %8ju %s %s\n",
809 1.3 joerg (uintmax_t)archive_entry_size(e),
810 1.3 joerg buf, archive_entry_pathname(e));
811 1.3 joerg } else if (v_opt == 2) {
812 1.6 wiz printf("%8ju Stored %7ju 0%% %s %08x %s\n",
813 1.2 joerg (uintmax_t)archive_entry_size(e),
814 1.2 joerg (uintmax_t)archive_entry_size(e),
815 1.2 joerg buf,
816 1.2 joerg 0U,
817 1.2 joerg archive_entry_pathname(e));
818 1.2 joerg }
819 1.1 joerg ac(archive_read_data_skip(a));
820 1.1 joerg }
821 1.1 joerg
822 1.1 joerg /*
823 1.1 joerg * Extract to memory to check CRC
824 1.1 joerg */
825 1.7 wiz static int
826 1.1 joerg test(struct archive *a, struct archive_entry *e)
827 1.1 joerg {
828 1.1 joerg ssize_t len;
829 1.7 wiz int error_count;
830 1.1 joerg
831 1.7 wiz error_count = 0;
832 1.1 joerg if (S_ISDIR(archive_entry_filetype(e)))
833 1.7 wiz return 0;
834 1.1 joerg
835 1.7 wiz info(" testing: %s\t", archive_entry_pathname(e));
836 1.1 joerg while ((len = archive_read_data(a, buffer, sizeof buffer)) > 0)
837 1.1 joerg /* nothing */;
838 1.1 joerg if (len < 0) {
839 1.7 wiz info(" %s\n", archive_error_string(a));
840 1.7 wiz ++error_count;
841 1.1 joerg } else {
842 1.7 wiz info(" OK\n");
843 1.1 joerg }
844 1.1 joerg
845 1.1 joerg /* shouldn't be necessary, but it doesn't hurt */
846 1.1 joerg ac(archive_read_data_skip(a));
847 1.7 wiz
848 1.7 wiz return error_count;
849 1.1 joerg }
850 1.1 joerg
851 1.1 joerg /*
852 1.1 joerg * Main loop: open the zipfile, iterate over its contents and decide what
853 1.1 joerg * to do with each entry.
854 1.1 joerg */
855 1.1 joerg static void
856 1.1 joerg unzip(const char *fn)
857 1.1 joerg {
858 1.1 joerg struct archive *a;
859 1.1 joerg struct archive_entry *e;
860 1.22 christos int ret;
861 1.7 wiz uintmax_t total_size, file_count, error_count;
862 1.1 joerg
863 1.22 christos if ((a = archive_read_new()) == NULL)
864 1.22 christos error("archive_read_new failed");
865 1.1 joerg
866 1.1 joerg ac(archive_read_support_format_zip(a));
867 1.22 christos ac(archive_read_open_filename(a, fn, 8192));
868 1.1 joerg
869 1.12 joerg if (!q_opt && !p_opt)
870 1.9 wiz printf("Archive: %s\n", fn);
871 1.9 wiz
872 1.3 joerg if (v_opt == 1) {
873 1.17 christos printf(" Length %sDate Time Name\n", y_str);
874 1.17 christos printf(" -------- %s---- ---- ----\n", y_str);
875 1.3 joerg } else if (v_opt == 2) {
876 1.17 christos printf(" Length Method Size Ratio %sDate Time CRC-32 Name\n", y_str);
877 1.17 christos printf("-------- ------ ------- ----- %s---- ---- ------ ----\n", y_str);
878 1.3 joerg }
879 1.3 joerg
880 1.3 joerg total_size = 0;
881 1.3 joerg file_count = 0;
882 1.7 wiz error_count = 0;
883 1.1 joerg for (;;) {
884 1.1 joerg ret = archive_read_next_header(a, &e);
885 1.1 joerg if (ret == ARCHIVE_EOF)
886 1.1 joerg break;
887 1.1 joerg ac(ret);
888 1.1 joerg if (t_opt)
889 1.7 wiz error_count += test(a, e);
890 1.2 joerg else if (v_opt)
891 1.1 joerg list(a, e);
892 1.3 joerg else if (p_opt || c_opt)
893 1.2 joerg extract_stdout(a, e);
894 1.1 joerg else
895 1.1 joerg extract(a, e);
896 1.3 joerg
897 1.3 joerg total_size += archive_entry_size(e);
898 1.3 joerg ++file_count;
899 1.3 joerg }
900 1.3 joerg
901 1.3 joerg if (v_opt == 1) {
902 1.17 christos printf(" -------- %s-------\n", y_str);
903 1.17 christos printf(" %8ju %s%ju file%s\n",
904 1.17 christos total_size, y_str, file_count, file_count != 1 ? "s" : "");
905 1.3 joerg } else if (v_opt == 2) {
906 1.17 christos printf("-------- ------- --- %s-------\n", y_str);
907 1.17 christos printf("%8ju %7ju 0%% %s%ju file%s\n",
908 1.17 christos total_size, total_size, y_str, file_count,
909 1.3 joerg file_count != 1 ? "s" : "");
910 1.1 joerg }
911 1.1 joerg
912 1.23 joerg ac(archive_read_free(a));
913 1.7 wiz
914 1.7 wiz if (t_opt) {
915 1.7 wiz if (error_count > 0) {
916 1.19 joerg errorx("%ju checksum error(s) found.", error_count);
917 1.7 wiz }
918 1.7 wiz else {
919 1.7 wiz printf("No errors detected in compressed data of %s.\n",
920 1.7 wiz fn);
921 1.7 wiz }
922 1.7 wiz }
923 1.1 joerg }
924 1.1 joerg
925 1.17 christos static void __dead
926 1.1 joerg usage(void)
927 1.1 joerg {
928 1.1 joerg
929 1.17 christos fprintf(stderr, "Usage: %s [-aCcfjLlnopqtuvy] [-d dir] [-x pattern] "
930 1.17 christos "zipfile\n", getprogname());
931 1.1 joerg exit(1);
932 1.1 joerg }
933 1.1 joerg
934 1.1 joerg static int
935 1.1 joerg getopts(int argc, char *argv[])
936 1.1 joerg {
937 1.1 joerg int opt;
938 1.1 joerg
939 1.1 joerg optreset = optind = 1;
940 1.17 christos while ((opt = getopt(argc, argv, "aCcd:fjLlnopqtuvyx:")) != -1)
941 1.1 joerg switch (opt) {
942 1.1 joerg case 'a':
943 1.1 joerg a_opt = 1;
944 1.1 joerg break;
945 1.7 wiz case 'C':
946 1.7 wiz C_opt = 1;
947 1.7 wiz break;
948 1.3 joerg case 'c':
949 1.3 joerg c_opt = 1;
950 1.3 joerg break;
951 1.1 joerg case 'd':
952 1.1 joerg d_arg = optarg;
953 1.1 joerg break;
954 1.1 joerg case 'f':
955 1.1 joerg f_opt = 1;
956 1.1 joerg break;
957 1.1 joerg case 'j':
958 1.1 joerg j_opt = 1;
959 1.1 joerg break;
960 1.1 joerg case 'L':
961 1.1 joerg L_opt = 1;
962 1.1 joerg break;
963 1.1 joerg case 'l':
964 1.2 joerg if (v_opt == 0)
965 1.2 joerg v_opt = 1;
966 1.1 joerg break;
967 1.1 joerg case 'n':
968 1.1 joerg n_opt = 1;
969 1.1 joerg break;
970 1.1 joerg case 'o':
971 1.1 joerg o_opt = 1;
972 1.2 joerg q_opt = 1;
973 1.2 joerg break;
974 1.2 joerg case 'p':
975 1.2 joerg p_opt = 1;
976 1.1 joerg break;
977 1.1 joerg case 'q':
978 1.1 joerg q_opt = 1;
979 1.1 joerg break;
980 1.1 joerg case 't':
981 1.1 joerg t_opt = 1;
982 1.1 joerg break;
983 1.1 joerg case 'u':
984 1.1 joerg u_opt = 1;
985 1.1 joerg break;
986 1.2 joerg case 'v':
987 1.2 joerg v_opt = 2;
988 1.2 joerg break;
989 1.1 joerg case 'x':
990 1.1 joerg add_pattern(&exclude, optarg);
991 1.1 joerg break;
992 1.17 christos case 'y':
993 1.17 christos y_str = " ";
994 1.17 christos break;
995 1.1 joerg default:
996 1.1 joerg usage();
997 1.1 joerg }
998 1.1 joerg
999 1.1 joerg return (optind);
1000 1.1 joerg }
1001 1.1 joerg
1002 1.1 joerg int
1003 1.1 joerg main(int argc, char *argv[])
1004 1.1 joerg {
1005 1.1 joerg const char *zipfile;
1006 1.1 joerg int nopts;
1007 1.1 joerg
1008 1.1 joerg if (isatty(STDOUT_FILENO))
1009 1.1 joerg tty = 1;
1010 1.1 joerg
1011 1.1 joerg if (getenv("UNZIP_DEBUG") != NULL)
1012 1.1 joerg unzip_debug = 1;
1013 1.1 joerg for (int i = 0; i < argc; ++i)
1014 1.1 joerg debug("%s%c", argv[i], (i < argc - 1) ? ' ' : '\n');
1015 1.1 joerg
1016 1.1 joerg /*
1017 1.1 joerg * Info-ZIP's unzip(1) expects certain options to come before the
1018 1.1 joerg * zipfile name, and others to come after - though it does not
1019 1.1 joerg * enforce this. For simplicity, we accept *all* options both
1020 1.1 joerg * before and after the zipfile name.
1021 1.1 joerg */
1022 1.1 joerg nopts = getopts(argc, argv);
1023 1.1 joerg
1024 1.1 joerg if (argc <= nopts)
1025 1.1 joerg usage();
1026 1.1 joerg zipfile = argv[nopts++];
1027 1.1 joerg
1028 1.22 christos if (strcmp(zipfile, "-") == 0)
1029 1.22 christos zipfile = NULL; /* STDIN */
1030 1.22 christos
1031 1.1 joerg while (nopts < argc && *argv[nopts] != '-')
1032 1.1 joerg add_pattern(&include, argv[nopts++]);
1033 1.1 joerg
1034 1.1 joerg nopts--; /* fake argv[0] */
1035 1.1 joerg nopts += getopts(argc - nopts, argv + nopts);
1036 1.1 joerg
1037 1.1 joerg if (n_opt + o_opt + u_opt > 1)
1038 1.1 joerg errorx("-n, -o and -u are contradictory");
1039 1.1 joerg
1040 1.1 joerg time(&now);
1041 1.1 joerg
1042 1.1 joerg unzip(zipfile);
1043 1.1 joerg
1044 1.1 joerg exit(0);
1045 1.1 joerg }
1046