Comments
[wimlib] / src / extract.c
1 /*
2  * extract.c
3  *
4  * Support for extracting WIM files.
5  *
6  * This code does NOT contain any filesystem-specific features.  In particular,
7  * security information (i.e. file permissions) and alternate data streams are
8  * ignored, except possibly to read an alternate data stream that contains
9  * symbolic link data.
10  */
11
12 /*
13  * Copyright (C) 2010 Carl Thijssen
14  * Copyright (C) 2012 Eric Biggers
15  *
16  * This file is part of wimlib, a library for working with WIM files.
17  *
18  * wimlib is free software; you can redistribute it and/or modify it under the
19  * terms of the GNU Lesser General Public License as published by the Free
20  * Software Foundation; either version 2.1 of the License, or (at your option)
21  * any later version.
22  *
23  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
24  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
25  * A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
26  * details.
27  *
28  * You should have received a copy of the GNU Lesser General Public License
29  * along with wimlib; if not, see http://www.gnu.org/licenses/.
30  */
31
32
33 #include <dirent.h>
34 #include <errno.h>
35 #include <fcntl.h>
36 #include <string.h>
37 #include <sys/stat.h>
38 #include <sys/time.h>
39 #include <unistd.h>
40
41 #include "config.h"
42 #include "dentry.h"
43 #include "lookup_table.h"
44 #include "timestamp.h"
45 #include "wimlib_internal.h"
46 #include "xml.h"
47
48 /* Internal */
49 #define WIMLIB_EXTRACT_FLAG_MULTI_IMAGE 0x80000000
50
51 static int extract_regular_file_linked(const struct dentry *dentry, 
52                                        const char *output_dir,
53                                        const char *output_path,
54                                        int extract_flags,
55                                        struct lookup_table_entry *lte)
56 {
57         /* This mode overrides the normal hard-link extraction and
58          * instead either symlinks or hardlinks *all* identical files in
59          * the WIM, even if they are in a different image (in the case
60          * of a multi-image extraction) */
61         wimlib_assert(lte->extracted_file);
62
63         if (extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) {
64                 if (link(lte->extracted_file, output_path) != 0) {
65                         ERROR_WITH_ERRNO("Failed to hard link "
66                                          "`%s' to `%s'",
67                                          output_path, lte->extracted_file);
68                         return WIMLIB_ERR_LINK;
69                 }
70         } else {
71                 int num_path_components;
72                 int num_output_dir_path_components;
73                 size_t extracted_file_len;
74                 char *p;
75                 const char *p2;
76                 size_t i;
77
78                 wimlib_assert(extract_flags & WIMLIB_EXTRACT_FLAG_SYMLINK);
79
80                 num_path_components = 
81                         get_num_path_components(dentry->full_path_utf8) - 1;
82                 num_output_dir_path_components =
83                         get_num_path_components(output_dir);
84
85                 if (extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
86                         num_path_components++;
87                         num_output_dir_path_components--;
88                 }
89                 extracted_file_len = strlen(lte->extracted_file);
90
91                 char buf[extracted_file_len + 3 * num_path_components + 1];
92                 p = &buf[0];
93
94                 for (i = 0; i < num_path_components; i++) {
95                         *p++ = '.';
96                         *p++ = '.';
97                         *p++ = '/';
98                 }
99                 p2 = lte->extracted_file;
100                 while (*p2 == '/')
101                         p2++;
102                 while (num_output_dir_path_components--)
103                         p2 = path_next_part(p2, NULL);
104                 strcpy(p, p2);
105                 if (symlink(buf, output_path) != 0) {
106                         ERROR_WITH_ERRNO("Failed to symlink `%s' to "
107                                          "`%s'",
108                                          buf, lte->extracted_file);
109                         return WIMLIB_ERR_LINK;
110                 }
111
112         }
113         return 0;
114 }
115
116 static int extract_regular_file_unlinked(WIMStruct *w,
117                                          struct dentry *dentry, 
118                                          const char *output_path,
119                                          int extract_flags,
120                                          struct lookup_table_entry *lte)
121 {
122         /* Normal mode of extraction.  Regular files and hard links are
123          * extracted in the way that they appear in the WIM. */
124
125         int out_fd;
126         const struct resource_entry *res_entry;
127         int ret;
128         const struct list_head *head = &dentry->link_group_list;
129
130         if (head->next != head &&
131              !(extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE))
132         {
133                 /* This dentry is one of a hard link set of at least 2 dentries.
134                  * If one of the other dentries has already been extracted, make
135                  * a hard link to the file corresponding to this
136                  * already-extracted directory.  Otherwise, extract the
137                  * file, and set the dentry->extracted_file field so that other
138                  * dentries in the hard link group can link to it. */
139                 struct dentry *other;
140                 list_for_each_entry(other, head, link_group_list) {
141                         if (other->extracted_file) {
142                                 DEBUG("Extracting hard link `%s' => `%s'",
143                                       output_path, other->extracted_file);
144                                 if (link(other->extracted_file, output_path) != 0) {
145                                         ERROR_WITH_ERRNO("Failed to hard link "
146                                                          "`%s' to `%s'",
147                                                          output_path,
148                                                          other->extracted_file);
149                                         return WIMLIB_ERR_LINK;
150                                 }
151                                 return 0;
152                         }
153                 }
154                 FREE(dentry->extracted_file);
155                 dentry->extracted_file = STRDUP(output_path);
156                 if (!dentry->extracted_file) {
157                         ERROR("Failed to allocate memory for filename");
158                         return WIMLIB_ERR_NOMEM;
159                 }
160         }
161
162         /* Extract the contents of the file to @output_path. */
163
164         out_fd = open(output_path, O_WRONLY | O_CREAT | O_TRUNC, 0644);
165         if (out_fd == -1) {
166                 ERROR_WITH_ERRNO("Failed to open the file `%s' for writing",
167                                  output_path);
168                 return WIMLIB_ERR_OPEN;
169         }
170
171         if (!lte) {
172                 /* Empty file with no lookup table entry */
173                 DEBUG("Empty file `%s'.", output_path);
174                 ret = 0;
175                 goto done;
176         }
177
178         ret = extract_full_wim_resource_to_fd(lte, out_fd);
179         if (ret != 0) {
180                 ERROR("Failed to extract resource to `%s'", output_path);
181                 goto done;
182         }
183
184 done:
185         if (close(out_fd) != 0) {
186                 ERROR_WITH_ERRNO("Failed to close file `%s'", output_path);
187                 ret = WIMLIB_ERR_WRITE;
188         }
189         return ret;
190 }
191
192 /* 
193  * Extracts a regular file from the WIM archive. 
194  */
195 static int extract_regular_file(WIMStruct *w, 
196                                 struct dentry *dentry, 
197                                 const char *output_dir,
198                                 const char *output_path,
199                                 int extract_flags)
200 {
201         struct lookup_table_entry *lte;
202
203         lte = dentry_first_lte(dentry, w->lookup_table);
204
205         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
206                               WIMLIB_EXTRACT_FLAG_HARDLINK)) && lte) {
207                 if (++lte->out_refcnt != 1)
208                         return extract_regular_file_linked(dentry, output_dir,
209                                                            output_path,
210                                                            extract_flags, lte);
211                 lte->extracted_file = STRDUP(output_path);
212                 if (!lte->extracted_file)
213                         return WIMLIB_ERR_NOMEM;
214         }
215
216         return extract_regular_file_unlinked(w, dentry, output_path,
217                                              extract_flags, lte);
218
219 }
220
221 static int extract_symlink(const struct dentry *dentry, const char *output_path,
222                            const WIMStruct *w)
223 {
224         char target[4096];
225         ssize_t ret = dentry_readlink(dentry, target, sizeof(target), w);
226         if (ret <= 0) {
227                 ERROR("Could not read the symbolic link from dentry `%s'",
228                       dentry->full_path_utf8);
229                 return WIMLIB_ERR_INVALID_DENTRY;
230         }
231         ret = symlink(target, output_path);
232         if (ret != 0) {
233                 ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
234                                  output_path, target);
235                 return WIMLIB_ERR_LINK;
236         }
237         return 0;
238 }
239
240 /* 
241  * Extracts a directory from the WIM archive. 
242  *
243  * @dentry:             The directory entry for the directory.
244  * @output_path:        The path to which the directory is to be extracted to.
245  * @return:             True on success, false on failure. 
246  */
247 static int extract_directory(const char *output_path, bool is_root)
248 {
249         int ret;
250         struct stat stbuf;
251         ret = stat(output_path, &stbuf);
252         if (ret == 0) {
253                 if (S_ISDIR(stbuf.st_mode)) {
254                         if (!is_root)
255                                 WARNING("`%s' already exists", output_path);
256                         return 0;
257                 } else {
258                         ERROR("`%s' is not a directory", output_path);
259                         return WIMLIB_ERR_MKDIR;
260                 }
261         } else {
262                 if (errno != ENOENT) {
263                         ERROR_WITH_ERRNO("Failed to stat `%s'", output_path);
264                         return WIMLIB_ERR_STAT;
265                 }
266         }
267         /* Compute the output path directory to the directory. */
268         if (mkdir(output_path, S_IRWXU | S_IRGRP | S_IXGRP |
269                                S_IROTH | S_IXOTH) != 0) {
270                 ERROR_WITH_ERRNO("Cannot create directory `%s'",
271                                  output_path);
272                 return WIMLIB_ERR_MKDIR;
273         }
274         return 0;
275 }
276
277 struct extract_args {
278         WIMStruct *w;
279         int extract_flags;
280         const char *output_dir;
281 #ifdef WITH_NTFS_3G
282         struct SECURITY_API *scapi;
283 #endif
284 };
285
286 /* 
287  * Extracts a file, directory, or symbolic link from the WIM archive.  For use
288  * in for_dentry_in_tree().
289  */
290 static int extract_dentry(struct dentry *dentry, void *arg)
291 {
292         struct extract_args *args = arg;
293         WIMStruct *w = args->w;
294         int extract_flags = args->extract_flags;
295         size_t len = strlen(args->output_dir);
296         char output_path[len + dentry->full_path_utf8_len + 1];
297         int ret;
298
299         if (extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) {
300                 wimlib_assert(dentry->full_path_utf8);
301                 puts(dentry->full_path_utf8);
302         }
303
304         memcpy(output_path, args->output_dir, len);
305         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
306         output_path[len + dentry->full_path_utf8_len] = '\0';
307
308         if (dentry_is_symlink(dentry))
309                 ret = extract_symlink(dentry, output_path, w);
310         else if (dentry_is_directory(dentry))
311                 ret = extract_directory(output_path, dentry_is_root(dentry));
312         else
313                 ret = extract_regular_file(w, dentry, args->output_dir,
314                                             output_path, extract_flags);
315         if (ret != 0)
316                 return ret;
317
318         return 0;
319 }
320
321 /* Apply timestamp to extracted file */
322 static int apply_dentry_timestamps(struct dentry *dentry, void *arg)
323 {
324         struct extract_args *args = arg;
325         size_t len = strlen(args->output_dir);
326         char output_path[len + dentry->full_path_utf8_len + 1];
327
328         memcpy(output_path, args->output_dir, len);
329         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
330         output_path[len + dentry->full_path_utf8_len] = '\0';
331
332         struct timeval tv[2];
333         wim_timestamp_to_timeval(dentry->last_access_time, &tv[0]);
334         wim_timestamp_to_timeval(dentry->last_write_time, &tv[1]);
335         if (lutimes(output_path, tv) != 0) {
336                 WARNING("Failed to set timestamp on file `%s': %s",
337                         output_path, strerror(errno));
338         }
339         return 0;
340 }
341
342
343 static int extract_single_image(WIMStruct *w, int image,
344                                 const char *output_dir, int extract_flags)
345 {
346         DEBUG("Extracting image %d", image);
347
348         int ret;
349         ret = wimlib_select_image(w, image);
350         if (ret != 0)
351                 return ret;
352
353         struct extract_args args = {
354                 .w = w,
355                 .extract_flags = extract_flags,
356                 .output_dir = output_dir,
357         #ifdef WITH_NTFS_3G
358                 .scapi = NULL
359         #endif
360         };
361
362         ret = for_dentry_in_tree(wim_root_dentry(w), extract_dentry, &args);
363         if (ret != 0)
364                 return ret;
365         return for_dentry_in_tree_depth(wim_root_dentry(w),
366                                         apply_dentry_timestamps, &args);
367
368 }
369
370
371 /* Extracts all images from the WIM to @output_dir, with the images placed in
372  * subdirectories named by their image names. */
373 static int extract_all_images(WIMStruct *w, const char *output_dir,
374                               int extract_flags)
375 {
376         size_t image_name_max_len = max(xml_get_max_image_name_len(w), 20);
377         size_t output_path_len = strlen(output_dir);
378         char buf[output_path_len + 1 + image_name_max_len + 1];
379         int ret;
380         int image;
381         const char *image_name;
382
383         DEBUG("Attempting to extract all images from `%s'", w->filename);
384
385         ret = extract_directory(output_dir, true);
386         if (ret != 0)
387                 return ret;
388
389         memcpy(buf, output_dir, output_path_len);
390         buf[output_path_len] = '/';
391         for (image = 1; image <= w->hdr.image_count; image++) {
392                 
393                 image_name = wimlib_get_image_name(w, image);
394                 if (*image_name) {
395                         strcpy(buf + output_path_len + 1, image_name);
396                 } else {
397                         /* Image name is empty. Use image number instead */
398                         sprintf(buf + output_path_len + 1, "%d", image);
399                 }
400                 ret = extract_single_image(w, image, buf, extract_flags);
401                 if (ret != 0)
402                         goto done;
403         }
404 done:
405         /* Restore original output directory */
406         buf[output_path_len + 1] = '\0';
407         return 0;
408 }
409
410 /* Extracts a single image or all images from a WIM file. */
411 WIMLIBAPI int wimlib_extract_image(WIMStruct *w, int image,
412                                    const char *output_dir, int flags)
413 {
414         if (!output_dir)
415                 return WIMLIB_ERR_INVALID_PARAM;
416
417         if ((flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
418                         == (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
419                 return WIMLIB_ERR_INVALID_PARAM;
420
421         for_lookup_table_entry(w->lookup_table, zero_out_refcnts, NULL);
422
423         if (image == WIM_ALL_IMAGES) {
424                 flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
425                 return extract_all_images(w, output_dir, flags);
426         } else {
427                 flags &= ~WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
428                 return extract_single_image(w, image, output_dir, flags);
429         }
430
431 }