e42aba4d1f6d0bb8975f62088e5554fb7f34588d
[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 General Public License as published by the Free
20  * Software Foundation; either version 3 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 General Public License for more
26  * details.
27  *
28  * You should have received a copy of the GNU 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         int ret;
127         const struct list_head *head = &dentry->link_group_list;
128
129         if (head->next != head &&
130              !(extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE))
131         {
132                 /* This dentry is one of a hard link set of at least 2 dentries.
133                  * If one of the other dentries has already been extracted, make
134                  * a hard link to the file corresponding to this
135                  * already-extracted directory.  Otherwise, extract the
136                  * file, and set the dentry->extracted_file field so that other
137                  * dentries in the hard link group can link to it. */
138                 struct dentry *other;
139                 list_for_each_entry(other, head, link_group_list) {
140                         if (other->extracted_file) {
141                                 DEBUG("Extracting hard link `%s' => `%s'",
142                                       output_path, other->extracted_file);
143                                 if (link(other->extracted_file, output_path) != 0) {
144                                         ERROR_WITH_ERRNO("Failed to hard link "
145                                                          "`%s' to `%s'",
146                                                          output_path,
147                                                          other->extracted_file);
148                                         return WIMLIB_ERR_LINK;
149                                 }
150                                 return 0;
151                         }
152                 }
153                 FREE(dentry->extracted_file);
154                 dentry->extracted_file = STRDUP(output_path);
155                 if (!dentry->extracted_file) {
156                         ERROR("Failed to allocate memory for filename");
157                         return WIMLIB_ERR_NOMEM;
158                 }
159         }
160
161         /* Extract the contents of the file to @output_path. */
162
163         out_fd = open(output_path, O_WRONLY | O_CREAT | O_TRUNC, 0644);
164         if (out_fd == -1) {
165                 ERROR_WITH_ERRNO("Failed to open the file `%s' for writing",
166                                  output_path);
167                 return WIMLIB_ERR_OPEN;
168         }
169
170         if (!lte) {
171                 /* Empty file with no lookup table entry */
172                 DEBUG("Empty file `%s'.", output_path);
173                 ret = 0;
174                 goto done;
175         }
176
177         ret = extract_full_wim_resource_to_fd(lte, out_fd);
178         if (ret != 0) {
179                 ERROR("Failed to extract resource to `%s'", output_path);
180                 goto done;
181         }
182
183 done:
184         if (close(out_fd) != 0) {
185                 ERROR_WITH_ERRNO("Failed to close file `%s'", output_path);
186                 ret = WIMLIB_ERR_WRITE;
187         }
188         return ret;
189 }
190
191 /* 
192  * Extracts a regular file from the WIM archive. 
193  */
194 static int extract_regular_file(WIMStruct *w, 
195                                 struct dentry *dentry, 
196                                 const char *output_dir,
197                                 const char *output_path,
198                                 int extract_flags)
199 {
200         struct lookup_table_entry *lte;
201
202         lte = dentry_unnamed_lte(dentry, w->lookup_table);
203
204         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
205                               WIMLIB_EXTRACT_FLAG_HARDLINK)) && lte) {
206                 if (++lte->out_refcnt != 1)
207                         return extract_regular_file_linked(dentry, output_dir,
208                                                            output_path,
209                                                            extract_flags, lte);
210                 lte->extracted_file = STRDUP(output_path);
211                 if (!lte->extracted_file)
212                         return WIMLIB_ERR_NOMEM;
213         }
214
215         return extract_regular_file_unlinked(w, dentry, output_path,
216                                              extract_flags, lte);
217
218 }
219
220 static int extract_symlink(const struct dentry *dentry, const char *output_path,
221                            const WIMStruct *w)
222 {
223         char target[4096];
224         ssize_t ret = dentry_readlink(dentry, target, sizeof(target), w);
225         if (ret <= 0) {
226                 ERROR("Could not read the symbolic link from dentry `%s'",
227                       dentry->full_path_utf8);
228                 return WIMLIB_ERR_INVALID_DENTRY;
229         }
230         ret = symlink(target, output_path);
231         if (ret != 0) {
232                 ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
233                                  output_path, target);
234                 return WIMLIB_ERR_LINK;
235         }
236         return 0;
237 }
238
239 /* 
240  * Extracts a directory from the WIM archive. 
241  *
242  * @dentry:             The directory entry for the directory.
243  * @output_path:        The path to which the directory is to be extracted to.
244  * @return:             True on success, false on failure. 
245  */
246 static int extract_directory(const char *output_path, bool is_root)
247 {
248         int ret;
249         struct stat stbuf;
250         ret = stat(output_path, &stbuf);
251         if (ret == 0) {
252                 if (S_ISDIR(stbuf.st_mode)) {
253                         if (!is_root)
254                                 WARNING("`%s' already exists", output_path);
255                         return 0;
256                 } else {
257                         ERROR("`%s' is not a directory", output_path);
258                         return WIMLIB_ERR_MKDIR;
259                 }
260         } else {
261                 if (errno != ENOENT) {
262                         ERROR_WITH_ERRNO("Failed to stat `%s'", output_path);
263                         return WIMLIB_ERR_STAT;
264                 }
265         }
266         /* Compute the output path directory to the directory. */
267         if (mkdir(output_path, S_IRWXU | S_IRGRP | S_IXGRP |
268                                S_IROTH | S_IXOTH) != 0) {
269                 ERROR_WITH_ERRNO("Cannot create directory `%s'",
270                                  output_path);
271                 return WIMLIB_ERR_MKDIR;
272         }
273         return 0;
274 }
275
276 struct extract_args {
277         WIMStruct *w;
278         int extract_flags;
279         const char *output_dir;
280 };
281
282 /* 
283  * Extracts a file, directory, or symbolic link from the WIM archive.  For use
284  * in for_dentry_in_tree().
285  */
286 static int extract_dentry(struct dentry *dentry, void *arg)
287 {
288         struct extract_args *args = arg;
289         WIMStruct *w = args->w;
290         int extract_flags = args->extract_flags;
291         size_t len = strlen(args->output_dir);
292         char output_path[len + dentry->full_path_utf8_len + 1];
293
294         if (extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) {
295                 wimlib_assert(dentry->full_path_utf8);
296                 puts(dentry->full_path_utf8);
297         }
298
299         memcpy(output_path, args->output_dir, len);
300         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
301         output_path[len + dentry->full_path_utf8_len] = '\0';
302
303         if (dentry_is_symlink(dentry))
304                 return extract_symlink(dentry, output_path, w);
305         else if (dentry_is_directory(dentry))
306                 return extract_directory(output_path, dentry_is_root(dentry));
307         else
308                 return extract_regular_file(w, dentry, args->output_dir,
309                                             output_path, extract_flags);
310 }
311
312 /* Apply timestamp to extracted file */
313 static int apply_dentry_timestamps(struct dentry *dentry, void *arg)
314 {
315         struct extract_args *args = arg;
316         size_t len = strlen(args->output_dir);
317         char output_path[len + dentry->full_path_utf8_len + 1];
318
319         memcpy(output_path, args->output_dir, len);
320         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
321         output_path[len + dentry->full_path_utf8_len] = '\0';
322
323         struct timeval tv[2];
324         wim_timestamp_to_timeval(dentry->last_access_time, &tv[0]);
325         wim_timestamp_to_timeval(dentry->last_write_time, &tv[1]);
326         if (lutimes(output_path, tv) != 0) {
327                 WARNING("Failed to set timestamp on file `%s': %s",
328                         output_path, strerror(errno));
329         }
330         return 0;
331 }
332
333
334 static int extract_single_image(WIMStruct *w, int image,
335                                 const char *output_dir, int extract_flags)
336 {
337         DEBUG("Extracting image %d", image);
338
339         int ret;
340         ret = wimlib_select_image(w, image);
341         if (ret != 0)
342                 return ret;
343
344         struct extract_args args = {
345                 .w = w,
346                 .extract_flags = extract_flags,
347                 .output_dir = output_dir,
348         };
349
350         ret = for_dentry_in_tree(wim_root_dentry(w), extract_dentry, &args);
351         if (ret != 0)
352                 return ret;
353         return for_dentry_in_tree_depth(wim_root_dentry(w),
354                                         apply_dentry_timestamps, &args);
355
356 }
357
358
359 /* Extracts all images from the WIM to @output_dir, with the images placed in
360  * subdirectories named by their image names. */
361 static int extract_all_images(WIMStruct *w, const char *output_dir,
362                               int extract_flags)
363 {
364         size_t image_name_max_len = max(xml_get_max_image_name_len(w), 20);
365         size_t output_path_len = strlen(output_dir);
366         char buf[output_path_len + 1 + image_name_max_len + 1];
367         int ret;
368         int image;
369         const char *image_name;
370
371         DEBUG("Attempting to extract all images from `%s' to `%s'",
372               w->filename, output_dir);
373
374         ret = extract_directory(output_dir, true);
375         if (ret != 0)
376                 return ret;
377
378         memcpy(buf, output_dir, output_path_len);
379         buf[output_path_len] = '/';
380         for (image = 1; image <= w->hdr.image_count; image++) {
381                 
382                 image_name = wimlib_get_image_name(w, image);
383                 if (*image_name) {
384                         strcpy(buf + output_path_len + 1, image_name);
385                 } else {
386                         /* Image name is empty. Use image number instead */
387                         sprintf(buf + output_path_len + 1, "%d", image);
388                 }
389                 ret = extract_single_image(w, image, buf, extract_flags);
390                 if (ret != 0)
391                         return ret;
392         }
393         return 0;
394 }
395
396
397 /* Extracts a single image or all images from a WIM file. */
398 WIMLIBAPI int wimlib_extract_image(WIMStruct *w, int image,
399                                    const char *output_dir, int flags,
400                                    WIMStruct **additional_swms,
401                                    unsigned num_additional_swms)
402 {
403         struct lookup_table *joined_tab, *w_tab_save;
404         int ret;
405
406         DEBUG("w->filename = %s, image = %d, output_dir = %s, flags = 0x%x, "
407               "num_additional_swms = %u",
408               w->filename, image, output_dir, flags, num_additional_swms);
409
410         if (!w || !output_dir)
411                 return WIMLIB_ERR_INVALID_PARAM;
412
413         if ((flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
414                         == (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
415                 return WIMLIB_ERR_INVALID_PARAM;
416
417         ret = verify_swm_set(w, additional_swms, num_additional_swms);
418         if (ret != 0)
419                 return ret;
420
421         if (num_additional_swms) {
422                 ret = new_joined_lookup_table(w, additional_swms,
423                                               num_additional_swms, &joined_tab);
424                 if (ret != 0)
425                         return ret;
426                 w_tab_save = w->lookup_table;
427                 w->lookup_table = joined_tab;
428         }
429
430
431         for_lookup_table_entry(w->lookup_table, zero_out_refcnts, NULL);
432
433         if (image == WIM_ALL_IMAGES) {
434                 flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
435                 ret = extract_all_images(w, output_dir, flags);
436         } else {
437                 flags &= ~WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
438                 ret = extract_single_image(w, image, output_dir, flags);
439         }
440         if (num_additional_swms) {
441                 free_lookup_table(w->lookup_table);
442                 w->lookup_table = w_tab_save;
443         }
444         return ret;
445
446 }