]> wimlib.net Git - wimlib/blob - src/extract.c
Allow configurable case sensitivity
[wimlib] / src / extract.c
1 /*
2  * extract.c
3  *
4  * Support for extracting WIM images, or files or directories contained in a WIM
5  * image.
6  */
7
8 /*
9  * Copyright (C) 2012, 2013 Eric Biggers
10  *
11  * This file is part of wimlib, a library for working with WIM files.
12  *
13  * wimlib is free software; you can redistribute it and/or modify it under the
14  * terms of the GNU General Public License as published by the Free
15  * Software Foundation; either version 3 of the License, or (at your option)
16  * any later version.
17  *
18  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
19  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
20  * A PARTICULAR PURPOSE. See the GNU General Public License for more
21  * details.
22  *
23  * You should have received a copy of the GNU General Public License
24  * along with wimlib; if not, see http://www.gnu.org/licenses/.
25  */
26
27 /*
28  * This file provides the API functions wimlib_extract_image(),
29  * wimlib_extract_files(), and wimlib_extract_image_from_pipe().  Internally,
30  * all end up calling extract_tree() zero or more times to extract a tree of
31  * files from the currently selected WIM image to the specified target directory
32  * or NTFS volume.
33  *
34  * Although wimlib supports multiple extraction modes/backends (NTFS-3g, UNIX,
35  * Win32), this file does not itself have code to extract files or directories
36  * to any specific target; instead, it handles generic functionality and relies
37  * on lower-level callback functions declared in `struct apply_operations' to do
38  * the actual extraction.
39  */
40
41 #ifdef HAVE_CONFIG_H
42 #  include "config.h"
43 #endif
44
45 #include "wimlib/apply.h"
46 #include "wimlib/dentry.h"
47 #include "wimlib/encoding.h"
48 #include "wimlib/endianness.h"
49 #include "wimlib/error.h"
50 #include "wimlib/lookup_table.h"
51 #include "wimlib/metadata.h"
52 #include "wimlib/pathlist.h"
53 #include "wimlib/paths.h"
54 #include "wimlib/reparse.h"
55 #include "wimlib/resource.h"
56 #include "wimlib/security.h"
57 #ifdef __WIN32__
58 #  include "wimlib/win32.h" /* for realpath() equivalent */
59 #endif
60 #include "wimlib/xml.h"
61 #include "wimlib/wildcard.h"
62 #include "wimlib/wim.h"
63
64 #include <errno.h>
65 #include <fcntl.h>
66 #include <stdlib.h>
67 #include <sys/stat.h>
68 #include <unistd.h>
69
70 #define WIMLIB_EXTRACT_FLAG_MULTI_IMAGE 0x80000000
71 #define WIMLIB_EXTRACT_FLAG_FROM_PIPE   0x40000000
72 #define WIMLIB_EXTRACT_MASK_PUBLIC      0x3fffffff
73
74 /* Given a WIM dentry in the tree to be extracted, resolve all streams in the
75  * corresponding inode and set 'out_refcnt' in each to 0.  */
76 static int
77 dentry_resolve_and_zero_lte_refcnt(struct wim_dentry *dentry, void *_ctx)
78 {
79         struct apply_ctx *ctx = _ctx;
80         struct wim_inode *inode = dentry->d_inode;
81         struct wim_lookup_table_entry *lte;
82         int ret;
83         bool force = false;
84
85         if (dentry->extraction_skipped)
86                 return 0;
87
88         /* Special case:  when extracting from a pipe, the WIM lookup table is
89          * initially empty, so "resolving" an inode's streams is initially not
90          * possible.  However, we still need to keep track of which streams,
91          * identified by SHA1 message digests, need to be extracted, so we
92          * "resolve" the inode's streams anyway by allocating new entries.  */
93         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
94                 force = true;
95         ret = inode_resolve_ltes(inode, ctx->wim->lookup_table, force);
96         if (ret)
97                 return ret;
98         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
99                 lte = inode_stream_lte_resolved(inode, i);
100                 if (lte)
101                         lte->out_refcnt = 0;
102         }
103         return 0;
104 }
105
106 static inline bool
107 is_linked_extraction(const struct apply_ctx *ctx)
108 {
109         return 0 != (ctx->extract_flags & (WIMLIB_EXTRACT_FLAG_HARDLINK |
110                                            WIMLIB_EXTRACT_FLAG_SYMLINK));
111 }
112
113 static inline bool
114 can_extract_named_data_streams(const struct apply_ctx *ctx)
115 {
116         return ctx->supported_features.named_data_streams &&
117                 !is_linked_extraction(ctx);
118 }
119
120 static int
121 ref_stream_to_extract(struct wim_lookup_table_entry *lte,
122                       struct wim_dentry *dentry, struct apply_ctx *ctx)
123 {
124         if (!lte)
125                 return 0;
126
127         /* Tally the size only for each extraction of the stream (not hard
128          * links).  */
129         if (!(dentry->d_inode->i_visited &&
130               ctx->supported_features.hard_links) &&
131             (!is_linked_extraction(ctx) || (lte->out_refcnt == 0 &&
132                                             lte->extracted_file == NULL)))
133         {
134                 ctx->progress.extract.total_bytes += lte->size;
135                 ctx->progress.extract.num_streams++;
136         }
137
138         /* Add stream to the extraction_list only one time, even if it's going
139          * to be extracted to multiple locations.  */
140         if (lte->out_refcnt == 0) {
141                 list_add_tail(&lte->extraction_list, &ctx->stream_list);
142                 ctx->num_streams_remaining++;
143         }
144
145         if (!(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
146                 struct wim_dentry **lte_dentries;
147
148                 /* Append dentry to this stream's array of dentries referencing
149                  * it.  Use inline array to avoid memory allocation until the
150                  * number of dentries becomes too large.  */
151                 if (lte->out_refcnt < ARRAY_LEN(lte->inline_lte_dentries)) {
152                         lte_dentries = lte->inline_lte_dentries;
153                 } else {
154                         struct wim_dentry **prev_lte_dentries;
155                         size_t alloc_lte_dentries;
156
157                         if (lte->out_refcnt == ARRAY_LEN(lte->inline_lte_dentries)) {
158                                 prev_lte_dentries = NULL;
159                                 alloc_lte_dentries = ARRAY_LEN(lte->inline_lte_dentries);
160                         } else {
161                                 prev_lte_dentries = lte->lte_dentries;
162                                 alloc_lte_dentries = lte->alloc_lte_dentries;
163                         }
164
165                         if (lte->out_refcnt == alloc_lte_dentries) {
166                                 alloc_lte_dentries *= 2;
167                                 lte_dentries = REALLOC(prev_lte_dentries,
168                                                        alloc_lte_dentries *
169                                                         sizeof(lte_dentries[0]));
170                                 if (lte_dentries == NULL)
171                                         return WIMLIB_ERR_NOMEM;
172                                 if (prev_lte_dentries == NULL) {
173                                         memcpy(lte_dentries,
174                                                lte->inline_lte_dentries,
175                                                sizeof(lte->inline_lte_dentries));
176                                 }
177                                 lte->lte_dentries = lte_dentries;
178                                 lte->alloc_lte_dentries = alloc_lte_dentries;
179                         }
180                         lte_dentries = lte->lte_dentries;
181                 }
182                 lte_dentries[lte->out_refcnt] = dentry;
183         }
184         lte->out_refcnt++;
185         return 0;
186 }
187
188 /* Given a WIM dentry in the tree to be extracted, iterate through streams that
189  * need to be extracted.  For each one, add it to the list of streams to be
190  * extracted (ctx->stream_list) if not already done so, and also update the
191  * progress information (ctx->progress) with the stream.  Furthermore, if doing
192  * a sequential extraction, build a mapping from each the stream to the dentries
193  * referencing it.
194  *
195  * This uses the i_visited member of the inodes (assumed to be 0 initially).  */
196 static int
197 dentry_add_streams_to_extract(struct wim_dentry *dentry, void *_ctx)
198 {
199         struct apply_ctx *ctx = _ctx;
200         struct wim_inode *inode = dentry->d_inode;
201         int ret;
202
203         /* Don't process dentries marked as skipped.  */
204         if (dentry->extraction_skipped)
205                 return 0;
206
207         /* The unnamed data stream will always be extracted, except in an
208          * unlikely case.  */
209         if (!inode_is_encrypted_directory(inode)) {
210                 ret = ref_stream_to_extract(inode_unnamed_lte_resolved(inode),
211                                             dentry, ctx);
212                 if (ret)
213                         return ret;
214         }
215
216         /* Named data streams will be extracted only if supported in the current
217          * extraction mode and volume, and to avoid complications, if not doing
218          * a linked extraction.  */
219         if (can_extract_named_data_streams(ctx)) {
220                 for (u16 i = 0; i < inode->i_num_ads; i++) {
221                         if (!ads_entry_is_named_stream(&inode->i_ads_entries[i]))
222                                 continue;
223                         ret = ref_stream_to_extract(inode->i_ads_entries[i].lte,
224                                                     dentry, ctx);
225                         if (ret)
226                                 return ret;
227                 }
228         }
229         inode->i_visited = 1;
230         return 0;
231 }
232
233 /* Inform library user of progress of stream extraction following the successful
234  * extraction of a copy of the stream specified by @lte.  */
235 static void
236 update_extract_progress(struct apply_ctx *ctx,
237                         const struct wim_lookup_table_entry *lte)
238 {
239         wimlib_progress_func_t progress_func = ctx->progress_func;
240         union wimlib_progress_info *progress = &ctx->progress;
241
242         progress->extract.completed_bytes += lte->size;
243         if (progress_func &&
244             progress->extract.completed_bytes >= ctx->next_progress)
245         {
246                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, progress);
247                 if (progress->extract.completed_bytes >=
248                     progress->extract.total_bytes)
249                 {
250                         ctx->next_progress = ~0ULL;
251                 } else {
252                         ctx->next_progress += progress->extract.total_bytes / 128;
253                         if (ctx->next_progress > progress->extract.total_bytes)
254                                 ctx->next_progress = progress->extract.total_bytes;
255                 }
256         }
257 }
258
259 #ifndef __WIN32__
260 /* Extract a symbolic link (not directly as reparse data), handling fixing up
261  * the target of absolute symbolic links and updating the extract progress.
262  *
263  * @inode must specify the WIM inode for a symbolic link or junction reparse
264  * point.
265  *
266  * @lte_override overrides the resource used as the reparse data for the
267  * symbolic link.  */
268 static int
269 extract_symlink(const tchar *path, struct apply_ctx *ctx,
270                 struct wim_inode *inode,
271                 struct wim_lookup_table_entry *lte_override)
272 {
273         ssize_t bufsize = ctx->ops->path_max;
274         tchar target[bufsize];
275         tchar *buf = target;
276         tchar *fixed_target;
277         ssize_t sret;
278         int ret;
279
280         /* If absolute symbolic link fixups requested, reserve space in the link
281          * target buffer for the absolute path of the target directory.  */
282         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX)
283         {
284                 buf += ctx->realtarget_nchars;
285                 bufsize -= ctx->realtarget_nchars;
286         }
287
288         /* Translate the WIM inode's reparse data into the link target.  */
289         sret = wim_inode_readlink(inode, buf, bufsize - 1, lte_override);
290         if (sret < 0) {
291                 errno = -sret;
292                 return WIMLIB_ERR_READLINK;
293         }
294         buf[sret] = '\0';
295
296         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
297             buf[0] == '/')
298         {
299                 /* Fix absolute symbolic link target to point into the
300                  * actual extraction destination.  */
301                 tmemcpy(target, ctx->realtarget, ctx->realtarget_nchars);
302                 fixed_target = target;
303         } else {
304                 /* Keep same link target.  */
305                 fixed_target = buf;
306         }
307
308         /* Call into the apply_operations to create the symbolic link.  */
309         DEBUG("Creating symlink \"%"TS"\" => \"%"TS"\"",
310               path, fixed_target);
311         ret = ctx->ops->create_symlink(fixed_target, path, ctx);
312         if (ret) {
313                 ERROR_WITH_ERRNO("Failed to create symlink "
314                                  "\"%"TS"\" => \"%"TS"\"", path, fixed_target);
315                 return ret;
316         }
317
318         /* Account for reparse data consumed.  */
319         update_extract_progress(ctx,
320                                 (lte_override ? lte_override :
321                                       inode_unnamed_lte_resolved(inode)));
322         return 0;
323 }
324 #endif /* !__WIN32__ */
325
326 /* Create a file, directory, or symbolic link.  */
327 static int
328 extract_inode(const tchar *path, struct apply_ctx *ctx, struct wim_inode *inode)
329 {
330         int ret;
331
332 #ifndef __WIN32__
333         if (ctx->supported_features.symlink_reparse_points &&
334             !ctx->supported_features.reparse_points &&
335             inode_is_symlink(inode))
336         {
337                 ret = extract_symlink(path, ctx, inode, NULL);
338         } else
339 #endif /* !__WIN32__ */
340         if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
341                 ret = ctx->ops->create_directory(path, ctx, &inode->extract_cookie);
342                 if (ret) {
343                         ERROR_WITH_ERRNO("Failed to create the directory "
344                                          "\"%"TS"\"", path);
345                 }
346         } else if ((inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) &&
347                     ctx->ops->extract_encrypted_stream_creates_file &&
348                     ctx->supported_features.encrypted_files) {
349                 ret = ctx->ops->extract_encrypted_stream(
350                                 path, inode_unnamed_lte_resolved(inode), ctx);
351                 if (ret) {
352                         ERROR_WITH_ERRNO("Failed to create and extract "
353                                          "encrypted file \"%"TS"\"", path);
354                 }
355         } else {
356                 ret = ctx->ops->create_file(path, ctx, &inode->extract_cookie);
357                 if (ret) {
358                         ERROR_WITH_ERRNO("Failed to create the file "
359                                          "\"%"TS"\"", path);
360                 }
361         }
362         return ret;
363 }
364
365 static int
366 extract_hardlink(const tchar *oldpath, const tchar *newpath,
367                  struct apply_ctx *ctx)
368 {
369         int ret;
370
371         DEBUG("Creating hardlink \"%"TS"\" => \"%"TS"\"", newpath, oldpath);
372         ret = ctx->ops->create_hardlink(oldpath, newpath, ctx);
373         if (ret) {
374                 ERROR_WITH_ERRNO("Failed to create hardlink "
375                                  "\"%"TS"\" => \"%"TS"\"",
376                                  newpath, oldpath);
377         }
378         return ret;
379 }
380
381 #ifdef __WIN32__
382 static int
383 try_extract_rpfix(u8 *rpbuf,
384                   u16 *rpbuflen_p,
385                   const wchar_t *extract_root_realpath,
386                   unsigned extract_root_realpath_nchars)
387 {
388         struct reparse_data rpdata;
389         wchar_t *target;
390         size_t target_nchars;
391         size_t stripped_nchars;
392         wchar_t *stripped_target;
393         wchar_t stripped_target_nchars;
394         int ret;
395
396         utf16lechar *new_target;
397         utf16lechar *new_print_name;
398         size_t new_target_nchars;
399         size_t new_print_name_nchars;
400         utf16lechar *p;
401
402         ret = parse_reparse_data(rpbuf, *rpbuflen_p, &rpdata);
403         if (ret)
404                 return ret;
405
406         if (extract_root_realpath[0] == L'\0' ||
407             extract_root_realpath[1] != L':' ||
408             extract_root_realpath[2] != L'\\')
409                 return WIMLIB_ERR_REPARSE_POINT_FIXUP_FAILED;
410
411         ret = parse_substitute_name(rpdata.substitute_name,
412                                     rpdata.substitute_name_nbytes,
413                                     rpdata.rptag);
414         if (ret < 0)
415                 return 0;
416         stripped_nchars = ret;
417         target = rpdata.substitute_name;
418         target_nchars = rpdata.substitute_name_nbytes / sizeof(utf16lechar);
419         stripped_target = target + stripped_nchars;
420         stripped_target_nchars = target_nchars - stripped_nchars;
421
422         new_target = alloca((6 + extract_root_realpath_nchars +
423                              stripped_target_nchars) * sizeof(utf16lechar));
424
425         p = new_target;
426         if (stripped_nchars == 6) {
427                 /* Include \??\ prefix if it was present before */
428                 p = wmempcpy(p, L"\\??\\", 4);
429         }
430
431         /* Print name excludes the \??\ if present. */
432         new_print_name = p;
433         if (stripped_nchars != 0) {
434                 /* Get drive letter from real path to extract root, if a drive
435                  * letter was present before. */
436                 *p++ = extract_root_realpath[0];
437                 *p++ = extract_root_realpath[1];
438         }
439         /* Copy the rest of the extract root */
440         p = wmempcpy(p, extract_root_realpath + 2, extract_root_realpath_nchars - 2);
441
442         /* Append the stripped target */
443         p = wmempcpy(p, stripped_target, stripped_target_nchars);
444         new_target_nchars = p - new_target;
445         new_print_name_nchars = p - new_print_name;
446
447         if (new_target_nchars * sizeof(utf16lechar) >= REPARSE_POINT_MAX_SIZE ||
448             new_print_name_nchars * sizeof(utf16lechar) >= REPARSE_POINT_MAX_SIZE)
449                 return WIMLIB_ERR_REPARSE_POINT_FIXUP_FAILED;
450
451         rpdata.substitute_name = new_target;
452         rpdata.substitute_name_nbytes = new_target_nchars * sizeof(utf16lechar);
453         rpdata.print_name = new_print_name;
454         rpdata.print_name_nbytes = new_print_name_nchars * sizeof(utf16lechar);
455         return make_reparse_buffer(&rpdata, rpbuf, rpbuflen_p);
456 }
457 #endif /* __WIN32__ */
458
459 /* Set reparse data on extracted file or directory that has
460  * FILE_ATTRIBUTE_REPARSE_POINT set.  */
461 static int
462 extract_reparse_data(const tchar *path, struct apply_ctx *ctx,
463                      struct wim_inode *inode,
464                      struct wim_lookup_table_entry *lte_override)
465 {
466         int ret;
467         u8 rpbuf[REPARSE_POINT_MAX_SIZE];
468         u16 rpbuflen;
469
470         ret = wim_inode_get_reparse_data(inode, rpbuf, &rpbuflen, lte_override);
471         if (ret)
472                 goto error;
473
474 #ifdef __WIN32__
475         /* Fix up target of absolute symbolic link or junction points so
476          * that they point into the actual extraction target.  */
477         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
478             (inode->i_reparse_tag == WIM_IO_REPARSE_TAG_SYMLINK ||
479              inode->i_reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT) &&
480             !inode->i_not_rpfixed)
481         {
482                 ret = try_extract_rpfix(rpbuf, &rpbuflen, ctx->realtarget,
483                                         ctx->realtarget_nchars);
484                 if (ret && !(ctx->extract_flags &
485                              WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS))
486                 {
487                         WARNING("Reparse point fixup of \"%"TS"\" "
488                                 "failed", path);
489                         ret = 0;
490                 }
491                 if (ret)
492                         goto error;
493         }
494 #endif
495
496         ret = ctx->ops->set_reparse_data(path, rpbuf, rpbuflen, ctx);
497
498         /* On Windows, the SeCreateSymbolicLink privilege is required to create
499          * symbolic links.  To be more friendly towards non-Administrator users,
500          * we merely warn the user if symbolic links cannot be created due to
501          * insufficient permissions or privileges, unless
502          * WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS was provided.  */
503 #ifdef __WIN32__
504         if (ret && inode_is_symlink(inode) &&
505             (errno == EACCES || errno == EPERM) &&
506             !(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS))
507         {
508                 WARNING("Can't set reparse data on \"%"TS"\": "
509                         "Access denied!\n"
510                         "          You may be trying to "
511                         "extract a symbolic link without the\n"
512                         "          SeCreateSymbolicLink privilege, "
513                         "which by default non-Administrator\n"
514                         "          accounts do not have.",
515                         path);
516                 ret = 0;
517         }
518 #endif
519         if (ret)
520                 goto error;
521
522         /* Account for reparse data consumed.  */
523         update_extract_progress(ctx,
524                                 (lte_override ? lte_override :
525                                       inode_unnamed_lte_resolved(inode)));
526         return 0;
527
528 error:
529         ERROR_WITH_ERRNO("Failed to set reparse data on \"%"TS"\"", path);
530         return ret;
531 }
532
533 /*
534  * Extract zero or more streams to a file.
535  *
536  * This function operates slightly differently depending on whether @lte_spec is
537  * NULL or not.  When @lte_spec is NULL, the behavior is to extract the default
538  * file contents (unnamed stream), and, if named data streams are supported in
539  * the extract mode and volume, any named data streams.  When @lte_spec is not
540  * NULL, the behavior is to extract only all copies of the stream @lte_spec, and
541  * in addition use @lte_spec to set the reparse data or create the symbolic link
542  * if appropriate.
543  *
544  * @path
545  *      Path to file to extract (as can be passed to apply_operations
546  *      functions).
547  * @ctx
548  *      Apply context.
549  * @dentry
550  *      WIM dentry that corresponds to the file being extracted.
551  * @lte_spec
552  *      If non-NULL, specifies the lookup table entry for a stream to extract,
553  *      and only that stream will be extracted (although there may be more than
554  *      one instance of it).
555  * @lte_override
556  *      Used only if @lte_spec != NULL; it is passed to the extraction functions
557  *      rather than @lte_spec, allowing the location of the stream to be
558  *      overridden.  (This is used when the WIM is being read from a nonseekable
559  *      file, such as a pipe, when streams need to be used more than once; each
560  *      such stream is extracted to a temporary file.)
561  */
562 static int
563 extract_streams(const tchar *path, struct apply_ctx *ctx,
564                 struct wim_dentry *dentry,
565                 struct wim_lookup_table_entry *lte_spec,
566                 struct wim_lookup_table_entry *lte_override)
567 {
568         struct wim_inode *inode = dentry->d_inode;
569         struct wim_lookup_table_entry *lte;
570         file_spec_t file_spec;
571         int ret;
572
573         if (dentry->was_hardlinked)
574                 return 0;
575
576 #ifdef ENABLE_DEBUG
577         if (lte_spec) {
578                 char sha1_str[100];
579                 char *p = sha1_str;
580                 for (unsigned i = 0; i < SHA1_HASH_SIZE; i++)
581                         p += sprintf(p, "%02x", lte_override->hash[i]);
582                 DEBUG("Extracting stream SHA1=%s to \"%"TS"\"",
583                       sha1_str, path, inode->i_ino);
584         } else {
585                 DEBUG("Extracting streams to \"%"TS"\"", path, inode->i_ino);
586         }
587 #endif
588
589         if (ctx->ops->uses_cookies)
590                 file_spec.cookie = inode->extract_cookie;
591         else
592                 file_spec.path = path;
593
594         /* Unnamed data stream.  */
595         lte = inode_unnamed_lte_resolved(inode);
596         if (lte && (!lte_spec || lte == lte_spec)) {
597                 if (lte_spec)
598                         lte = lte_override;
599                 if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
600                                              FILE_ATTRIBUTE_REPARSE_POINT)))
601                 {
602                         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED &&
603                             ctx->supported_features.encrypted_files) {
604                                 if (!ctx->ops->extract_encrypted_stream_creates_file) {
605                                         ret = ctx->ops->extract_encrypted_stream(
606                                                                 path, lte, ctx);
607                                         if (ret)
608                                                 goto error;
609                                 }
610                         } else {
611                                 ret = ctx->ops->extract_unnamed_stream(
612                                                         file_spec, lte, ctx);
613                                 if (ret)
614                                         goto error;
615                         }
616                         update_extract_progress(ctx, lte);
617                 }
618                 else if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT)
619                 {
620                         ret = 0;
621                         if (ctx->supported_features.reparse_points)
622                                 ret = extract_reparse_data(path, ctx, inode, lte);
623                 #ifndef __WIN32__
624                         else if ((inode_is_symlink(inode) &&
625                                   ctx->supported_features.symlink_reparse_points))
626                                 ret = extract_symlink(path, ctx, inode, lte);
627                 #endif
628                         if (ret)
629                                 return ret;
630                 }
631         }
632
633         /* Named data streams.  */
634         if (can_extract_named_data_streams(ctx)) {
635                 for (u16 i = 0; i < inode->i_num_ads; i++) {
636                         struct wim_ads_entry *entry = &inode->i_ads_entries[i];
637
638                         if (!ads_entry_is_named_stream(entry))
639                                 continue;
640                         lte = entry->lte;
641                         if (!lte)
642                                 continue;
643                         if (lte_spec && lte_spec != lte)
644                                 continue;
645                         if (lte_spec)
646                                 lte = lte_override;
647                         ret = ctx->ops->extract_named_stream(file_spec, entry->stream_name,
648                                                              entry->stream_name_nbytes / 2,
649                                                              lte, ctx);
650                         if (ret)
651                                 goto error;
652                         update_extract_progress(ctx, lte);
653                 }
654         }
655         return 0;
656
657 error:
658         ERROR_WITH_ERRNO("Failed to extract data of \"%"TS"\"", path);
659         return ret;
660 }
661
662 /* Set attributes on an extracted file or directory if supported by the
663  * extraction mode.  */
664 static int
665 extract_file_attributes(const tchar *path, struct apply_ctx *ctx,
666                         struct wim_dentry *dentry, unsigned pass)
667 {
668         int ret;
669
670         if (ctx->ops->set_file_attributes &&
671             !(dentry == ctx->extract_root && ctx->root_dentry_is_special)) {
672                 u32 attributes = dentry->d_inode->i_attributes;
673
674                 /* Clear unsupported attributes.  */
675                 attributes &= ctx->supported_attributes_mask;
676
677                 if ((attributes & FILE_ATTRIBUTE_DIRECTORY &&
678                      !ctx->supported_features.encrypted_directories) ||
679                     (!(attributes & FILE_ATTRIBUTE_DIRECTORY) &&
680                      !ctx->supported_features.encrypted_files))
681                 {
682                         attributes &= ~FILE_ATTRIBUTE_ENCRYPTED;
683                 }
684
685                 if (attributes == 0)
686                         attributes = FILE_ATTRIBUTE_NORMAL;
687
688                 ret = ctx->ops->set_file_attributes(path, attributes, ctx, pass);
689                 if (ret) {
690                         ERROR_WITH_ERRNO("Failed to set attributes on "
691                                          "\"%"TS"\"", path);
692                         return ret;
693                 }
694         }
695         return 0;
696 }
697
698
699 /* Set or remove the short (DOS) name on an extracted file or directory if
700  * supported by the extraction mode.  Since DOS names are unimportant and it's
701  * easy to run into problems setting them on Windows (SetFileShortName()
702  * requires SE_RESTORE privilege, which only the Administrator can request, and
703  * also requires DELETE access to the file), failure is ignored unless
704  * WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES is set.  */
705 static int
706 extract_short_name(const tchar *path, struct apply_ctx *ctx,
707                    struct wim_dentry *dentry)
708 {
709         int ret;
710
711         /* The root of the dentry tree being extracted may not be extracted to
712          * its original name, so its short name should be ignored.  */
713         if (dentry == ctx->extract_root)
714                 return 0;
715
716         if (ctx->supported_features.short_names) {
717                 ret = ctx->ops->set_short_name(path,
718                                                dentry->short_name,
719                                                dentry->short_name_nbytes / 2,
720                                                ctx);
721                 if (ret && (ctx->extract_flags &
722                             WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES))
723                 {
724                         ERROR_WITH_ERRNO("Failed to set short name of "
725                                          "\"%"TS"\"", path);
726                         return ret;
727                 }
728         }
729         return 0;
730 }
731
732 /* Set security descriptor, UNIX data, or neither on an extracted file, taking
733  * into account the current extraction mode and flags.  */
734 static int
735 extract_security(const tchar *path, struct apply_ctx *ctx,
736                  struct wim_dentry *dentry)
737 {
738         int ret;
739         struct wim_inode *inode = dentry->d_inode;
740
741         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS)
742                 return 0;
743
744         if ((ctx->extract_root == dentry) && ctx->root_dentry_is_special)
745                 return 0;
746
747 #ifndef __WIN32__
748         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
749                 struct wimlib_unix_data data;
750
751                 ret = inode_get_unix_data(inode, &data, NULL);
752                 if (ret < 0)
753                         ret = 0;
754                 else if (ret == 0)
755                         ret = ctx->ops->set_unix_data(path, &data, ctx);
756                 if (ret) {
757                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) {
758                                 ERROR_WITH_ERRNO("Failed to set UNIX owner, "
759                                                  "group, and/or mode on "
760                                                  "\"%"TS"\"", path);
761                                 return ret;
762                         } else {
763                                 WARNING_WITH_ERRNO("Failed to set UNIX owner, "
764                                                    "group, and/or/mode on "
765                                                    "\"%"TS"\"", path);
766                         }
767                 }
768         }
769         else
770 #endif /* __WIN32__ */
771         if (ctx->supported_features.security_descriptors &&
772             inode->i_security_id != -1)
773         {
774                 const struct wim_security_data *sd;
775                 const u8 *desc;
776                 size_t desc_size;
777
778                 sd = wim_const_security_data(ctx->wim);
779                 desc = sd->descriptors[inode->i_security_id];
780                 desc_size = sd->sizes[inode->i_security_id];
781
782                 ret = ctx->ops->set_security_descriptor(path, desc,
783                                                         desc_size, ctx);
784                 if (ret) {
785                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) {
786                                 ERROR_WITH_ERRNO("Failed to set security "
787                                                  "descriptor on \"%"TS"\"", path);
788                                 return ret;
789                         } else {
790                         #if 0
791                                 if (errno != EACCES) {
792                                         WARNING_WITH_ERRNO("Failed to set "
793                                                            "security descriptor "
794                                                            "on \"%"TS"\"", path);
795                                 }
796                         #endif
797                                 ctx->no_security_descriptors++;
798                         }
799                 }
800         }
801         return 0;
802 }
803
804 /* Set timestamps on an extracted file.  Failure is warning-only unless
805  * WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS is set.  */
806 static int
807 extract_timestamps(const tchar *path, struct apply_ctx *ctx,
808                    struct wim_dentry *dentry)
809 {
810         struct wim_inode *inode = dentry->d_inode;
811         int ret;
812
813         if ((ctx->extract_root == dentry) && ctx->root_dentry_is_special)
814                 return 0;
815
816         if (ctx->ops->set_timestamps) {
817                 ret = ctx->ops->set_timestamps(path,
818                                                inode->i_creation_time,
819                                                inode->i_last_write_time,
820                                                inode->i_last_access_time,
821                                                ctx);
822                 if (ret) {
823                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) {
824                                 ERROR_WITH_ERRNO("Failed to set timestamps "
825                                                  "on \"%"TS"\"", path);
826                                 return ret;
827                         } else {
828                                 WARNING_WITH_ERRNO("Failed to set timestamps "
829                                                    "on \"%"TS"\"", path);
830                         }
831                 }
832         }
833         return 0;
834 }
835
836 /* Check whether the extraction of a dentry should be skipped completely.  */
837 static bool
838 dentry_is_supported(struct wim_dentry *dentry,
839                     const struct wim_features *supported_features)
840 {
841         struct wim_inode *inode = dentry->d_inode;
842
843         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
844                 return supported_features->reparse_points ||
845                         (inode_is_symlink(inode) &&
846                          supported_features->symlink_reparse_points);
847         }
848         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
849                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
850                         return supported_features->encrypted_directories != 0;
851                 else
852                         return supported_features->encrypted_files != 0;
853         }
854         return true;
855 }
856
857 /* Given a WIM dentry to extract, build the path to which to extract it, in the
858  * format understood by the callbacks in the apply_operations being used.
859  *
860  * Write the resulting path into @path, which must have room for at least
861  * ctx->ops->max_path characters including the null-terminator.
862  *
863  * Return %true if successful; %false if this WIM dentry doesn't actually need
864  * to be extracted or if the calculated path exceeds ctx->ops->max_path
865  * characters.
866  *
867  * This function clobbers the tmp_list member of @dentry and its ancestors up
868  * until the extraction root.  */
869 static bool
870 build_extraction_path(tchar path[], struct wim_dentry *dentry,
871                       struct apply_ctx *ctx)
872 {
873         size_t path_nchars;
874         LIST_HEAD(ancestor_list);
875         tchar *p = path;
876         const tchar *target_prefix;
877         size_t target_prefix_nchars;
878         struct wim_dentry *d;
879
880         if (dentry->extraction_skipped)
881                 return false;
882
883         path_nchars = ctx->ops->path_prefix_nchars;
884
885         if (ctx->ops->requires_realtarget_in_paths) {
886                 target_prefix        = ctx->realtarget;
887                 target_prefix_nchars = ctx->realtarget_nchars;
888         } else if (ctx->ops->requires_target_in_paths) {
889                 target_prefix        = ctx->target;
890                 target_prefix_nchars = ctx->target_nchars;
891         } else {
892                 target_prefix        = NULL;
893                 target_prefix_nchars = 0;
894         }
895         path_nchars += target_prefix_nchars;
896
897         for (d = dentry; d != ctx->extract_root; d = d->parent) {
898                 path_nchars += d->extraction_name_nchars + 1;
899                 list_add(&d->tmp_list, &ancestor_list);
900         }
901
902         path_nchars++; /* null terminator */
903
904         if (path_nchars > ctx->ops->path_max) {
905                 WARNING("\"%"TS"\": Path too long to extract",
906                         dentry_full_path(dentry));
907                 return false;
908         }
909
910         p = tmempcpy(p, ctx->ops->path_prefix, ctx->ops->path_prefix_nchars);
911         p = tmempcpy(p, target_prefix, target_prefix_nchars);
912         list_for_each_entry(d, &ancestor_list, tmp_list) {
913                 *p++ = ctx->ops->path_separator;
914                 p = tmempcpy(p, d->extraction_name, d->extraction_name_nchars);
915         }
916         *p++ = T('\0');
917         wimlib_assert(p - path == path_nchars);
918         return true;
919 }
920
921 static unsigned
922 get_num_path_components(const tchar *path, tchar path_separator)
923 {
924         unsigned num_components = 0;
925 #ifdef __WIN32__
926         /* Ignore drive letter.  */
927         if (path[0] != L'\0' && path[1] == L':')
928                 path += 2;
929 #endif
930
931         while (*path) {
932                 while (*path == path_separator)
933                         path++;
934                 if (*path)
935                         num_components++;
936                 while (*path && *path != path_separator)
937                         path++;
938         }
939         return num_components;
940 }
941
942 static int
943 extract_multiimage_symlink(const tchar *oldpath, const tchar *newpath,
944                            struct apply_ctx *ctx, struct wim_dentry *dentry)
945 {
946         size_t num_raw_path_components;
947         const struct wim_dentry *d;
948         size_t num_target_path_components;
949         tchar *p;
950         const tchar *p_old;
951         int ret;
952
953         num_raw_path_components = 0;
954         for (d = dentry; d != ctx->extract_root; d = d->parent)
955                 num_raw_path_components++;
956
957         if (ctx->ops->requires_realtarget_in_paths)
958                 num_target_path_components = get_num_path_components(ctx->realtarget,
959                                                                      ctx->ops->path_separator);
960         else if (ctx->ops->requires_target_in_paths)
961                 num_target_path_components = get_num_path_components(ctx->target,
962                                                                      ctx->ops->path_separator);
963         else
964                 num_target_path_components = 0;
965
966         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
967                 wimlib_assert(num_target_path_components > 0);
968                 num_raw_path_components++;
969                 num_target_path_components--;
970         }
971
972         p_old = oldpath + ctx->ops->path_prefix_nchars;
973 #ifdef __WIN32__
974         if (p_old[0] != L'\0' && p_old[1] == ':')
975                 p_old += 2;
976 #endif
977         while (*p_old == ctx->ops->path_separator)
978                 p_old++;
979         while (--num_target_path_components) {
980                 while (*p_old != ctx->ops->path_separator)
981                         p_old++;
982                 while (*p_old == ctx->ops->path_separator)
983                         p_old++;
984         }
985
986         tchar symlink_target[tstrlen(p_old) + 3 * num_raw_path_components + 1];
987
988         p = &symlink_target[0];
989         while (num_raw_path_components--) {
990                 *p++ = '.';
991                 *p++ = '.';
992                 *p++ = ctx->ops->path_separator;
993         }
994         tstrcpy(p, p_old);
995         DEBUG("Creating symlink \"%"TS"\" => \"%"TS"\"",
996               newpath, symlink_target);
997         ret = ctx->ops->create_symlink(symlink_target, newpath, ctx);
998         if (ret) {
999                 ERROR_WITH_ERRNO("Failed to create symlink "
1000                                  "\"%"TS"\" => \"%"TS"\"",
1001                                  newpath, symlink_target);
1002         }
1003         return ret;
1004 }
1005
1006 /* Create the "skeleton" of an extracted file or directory.  Don't yet extract
1007  * data streams, reparse data (including symbolic links), timestamps, and
1008  * security descriptors.  Basically, everything that doesn't require reading
1009  * non-metadata resources from the WIM file and isn't delayed until the final
1010  * pass.  */
1011 static int
1012 do_dentry_extract_skeleton(tchar path[], struct wim_dentry *dentry,
1013                            struct apply_ctx *ctx)
1014 {
1015         struct wim_inode *inode = dentry->d_inode;
1016         int ret;
1017         const tchar *oldpath;
1018
1019         if (unlikely(is_linked_extraction(ctx))) {
1020                 struct wim_lookup_table_entry *unnamed_lte;
1021
1022                 unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode);
1023                 if (unnamed_lte && unnamed_lte->extracted_file) {
1024                         oldpath = unnamed_lte->extracted_file;
1025                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK)
1026                                 goto hardlink;
1027                         else
1028                                 goto symlink;
1029                 }
1030         }
1031
1032         /* Create hard link if this dentry corresponds to an already-extracted
1033          * inode.  */
1034         if (inode->i_extracted_file) {
1035                 oldpath = inode->i_extracted_file;
1036                 goto hardlink;
1037         }
1038
1039         /* Skip symlinks unless they can be extracted as reparse points rather
1040          * than created directly.  */
1041         if (inode_is_symlink(inode) && !ctx->supported_features.reparse_points)
1042                 return 0;
1043
1044         /* Create this file or directory unless it's the extraction root, which
1045          * was already created if necessary.  */
1046         if (dentry != ctx->extract_root) {
1047                 ret = extract_inode(path, ctx, inode);
1048                 if (ret)
1049                         return ret;
1050         }
1051
1052         /* Create empty named data streams.  */
1053         if (can_extract_named_data_streams(ctx)) {
1054                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1055                         file_spec_t file_spec;
1056                         struct wim_ads_entry *entry = &inode->i_ads_entries[i];
1057
1058                         if (!ads_entry_is_named_stream(entry))
1059                                 continue;
1060                         if (entry->lte)
1061                                 continue;
1062                         if (ctx->ops->uses_cookies)
1063                                 file_spec.cookie = inode->extract_cookie;
1064                         else
1065                                 file_spec.path = path;
1066                         ret = ctx->ops->extract_named_stream(file_spec,
1067                                                              entry->stream_name,
1068                                                              entry->stream_name_nbytes / 2,
1069                                                              entry->lte, ctx);
1070                         if (ret) {
1071                                 ERROR_WITH_ERRNO("\"%"TS"\": failed to create "
1072                                                  "empty named data stream",
1073                                                  path);
1074                                 return ret;
1075                         }
1076                 }
1077         }
1078
1079         /* Set file attributes (if supported).  */
1080         ret = extract_file_attributes(path, ctx, dentry, 0);
1081         if (ret)
1082                 return ret;
1083
1084         /* Set or remove file short name (if supported).  */
1085         ret = extract_short_name(path, ctx, dentry);
1086         if (ret)
1087                 return ret;
1088
1089         /* If inode has multiple links and hard links are supported in this
1090          * extraction mode and volume, save the path to the extracted file in
1091          * case it's needed to create a hard link.  */
1092         if (unlikely(is_linked_extraction(ctx))) {
1093                 struct wim_lookup_table_entry *unnamed_lte;
1094
1095                 unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode);
1096                 if (unnamed_lte) {
1097                         unnamed_lte->extracted_file = TSTRDUP(path);
1098                         if (!unnamed_lte->extracted_file)
1099                                 return WIMLIB_ERR_NOMEM;
1100                 }
1101         } else if (inode->i_nlink > 1 && ctx->supported_features.hard_links) {
1102                 inode->i_extracted_file = TSTRDUP(path);
1103                 if (!inode->i_extracted_file)
1104                         return WIMLIB_ERR_NOMEM;
1105         }
1106         return 0;
1107
1108 symlink:
1109         ret = extract_multiimage_symlink(oldpath, path, ctx, dentry);
1110         if (ret)
1111                 return ret;
1112         dentry->was_hardlinked = 1;
1113         return 0;
1114
1115 hardlink:
1116         ret = extract_hardlink(oldpath, path, ctx);
1117         if (ret)
1118                 return ret;
1119         dentry->was_hardlinked = 1;
1120         return 0;
1121 }
1122
1123 /* This is a wrapper around do_dentry_extract_skeleton() that handles building
1124  * the path, doing short name reordering.  This is also idempotent; dentries
1125  * already processed have skeleton_extracted set and no action is taken.  See
1126  * apply_operations.requires_short_name_reordering for more details about short
1127  * name reordering.  */
1128 static int
1129 dentry_extract_skeleton(struct wim_dentry *dentry, void *_ctx)
1130 {
1131         struct apply_ctx *ctx = _ctx;
1132         tchar path[ctx->ops->path_max];
1133         struct wim_dentry *orig_dentry;
1134         struct wim_dentry *other_dentry;
1135         int ret;
1136
1137         if (dentry->skeleton_extracted)
1138                 return 0;
1139
1140         orig_dentry = NULL;
1141         if (ctx->supported_features.short_names
1142             && ctx->ops->requires_short_name_reordering
1143             && !dentry_has_short_name(dentry)
1144             && !dentry->d_inode->i_dos_name_extracted)
1145         {
1146                 inode_for_each_dentry(other_dentry, dentry->d_inode) {
1147                         if (dentry_has_short_name(other_dentry)
1148                             && !other_dentry->skeleton_extracted
1149                             && other_dentry->in_extraction_tree)
1150                         {
1151                                 DEBUG("Creating %"TS" before %"TS" "
1152                                       "to guarantee correct DOS name extraction",
1153                                       dentry_full_path(other_dentry),
1154                                       dentry_full_path(dentry));
1155                                 orig_dentry = dentry;
1156                                 dentry = other_dentry;
1157                                 break;
1158                         }
1159                 }
1160         }
1161 again:
1162         if (!build_extraction_path(path, dentry, ctx))
1163                 return 0;
1164         ret = do_dentry_extract_skeleton(path, dentry, ctx);
1165         if (ret)
1166                 return ret;
1167
1168         dentry->skeleton_extracted = 1;
1169
1170         if (orig_dentry) {
1171                 dentry = orig_dentry;
1172                 orig_dentry = NULL;
1173                 goto again;
1174         }
1175         dentry->d_inode->i_dos_name_extracted = 1;
1176         return 0;
1177 }
1178
1179 static int
1180 dentry_extract_dir_skeleton(struct wim_dentry *dentry, void *_ctx)
1181 {
1182         if (dentry->d_inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1183                 return dentry_extract_skeleton(dentry, _ctx);
1184         return 0;
1185 }
1186
1187 /* Create a file or directory, then immediately extract all streams.  The WIM
1188  * may not be read sequentially by this function.  */
1189 static int
1190 dentry_extract(struct wim_dentry *dentry, void *_ctx)
1191 {
1192         struct apply_ctx *ctx = _ctx;
1193         tchar path[ctx->ops->path_max];
1194         int ret;
1195
1196         ret = dentry_extract_skeleton(dentry, ctx);
1197         if (ret)
1198                 return ret;
1199
1200         if (!build_extraction_path(path, dentry, ctx))
1201                 return 0;
1202
1203         return extract_streams(path, ctx, dentry, NULL, NULL);
1204 }
1205
1206 /* Creates a temporary file opened for writing.  The open file descriptor is
1207  * returned in @fd_ret and its name is returned in @name_ret (dynamically
1208  * allocated).  */
1209 static int
1210 create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
1211 {
1212         tchar *name;
1213         int raw_fd;
1214
1215 retry:
1216         name = ttempnam(NULL, T("wimlib"));
1217         if (name == NULL) {
1218                 ERROR_WITH_ERRNO("Failed to create temporary filename");
1219                 return WIMLIB_ERR_NOMEM;
1220         }
1221
1222         raw_fd = topen(name, O_WRONLY | O_CREAT | O_EXCL | O_BINARY, 0600);
1223
1224         if (raw_fd < 0) {
1225                 if (errno == EEXIST) {
1226                         FREE(name);
1227                         goto retry;
1228                 }
1229                 ERROR_WITH_ERRNO("Failed to open temporary file \"%"TS"\"", name);
1230                 FREE(name);
1231                 return WIMLIB_ERR_OPEN;
1232         }
1233
1234         filedes_init(fd_ret, raw_fd);
1235         *name_ret = name;
1236         return 0;
1237 }
1238
1239 /* Extract all instances of the stream @lte that are being extracted in this
1240  * call of extract_tree(), but actually read the stream data from @lte_override.
1241  */
1242 static int
1243 extract_stream_instances(struct wim_lookup_table_entry *lte,
1244                          struct wim_lookup_table_entry *lte_override,
1245                          struct apply_ctx *ctx)
1246 {
1247         struct wim_dentry **lte_dentries;
1248         tchar path[ctx->ops->path_max];
1249         size_t i;
1250         int ret;
1251
1252         if (lte->out_refcnt <= ARRAY_LEN(lte->inline_lte_dentries))
1253                 lte_dentries = lte->inline_lte_dentries;
1254         else
1255                 lte_dentries = lte->lte_dentries;
1256
1257         for (i = 0; i < lte->out_refcnt; i++) {
1258                 struct wim_dentry *dentry = lte_dentries[i];
1259
1260                 if (dentry->tmp_flag)
1261                         continue;
1262                 if (!build_extraction_path(path, dentry, ctx))
1263                         continue;
1264                 ret = extract_streams(path, ctx, dentry, lte, lte_override);
1265                 if (ret)
1266                         goto out_clear_tmp_flags;
1267                 dentry->tmp_flag = 1;
1268         }
1269         ret = 0;
1270 out_clear_tmp_flags:
1271         for (i = 0; i < lte->out_refcnt; i++)
1272                 lte_dentries[i]->tmp_flag = 0;
1273         return ret;
1274 }
1275
1276 /* Determine whether the specified stream needs to be extracted to a temporary
1277  * file or not.
1278  *
1279  * @lte->out_refcnt specifies the number of instances of this stream that must
1280  * be extracted.
1281  *
1282  * @is_partial_res is %true if this stream is just one of multiple in a single
1283  * WIM resource being extracted.  */
1284 static bool
1285 need_tmpfile_to_extract(struct wim_lookup_table_entry *lte,
1286                         bool is_partial_res)
1287 {
1288         /* Temporary file is always required when reading a partial resource,
1289          * since in that case we retrieve all the contained streams in one pass.
1290          * */
1291         if (is_partial_res)
1292                 return true;
1293
1294         /* Otherwise we don't need a temporary file if only a single instance of
1295          * the stream is needed.  */
1296         if (lte->out_refcnt == 1)
1297                 return false;
1298
1299         wimlib_assert(lte->out_refcnt >= 2);
1300
1301         /* We also don't need a temporary file if random access to the stream is
1302          * allowed.  */
1303         if (lte->resource_location != RESOURCE_IN_WIM ||
1304             filedes_is_seekable(&lte->rspec->wim->in_fd))
1305                 return false;
1306
1307         return true;
1308 }
1309
1310 static int
1311 begin_extract_stream_to_tmpfile(struct wim_lookup_table_entry *lte,
1312                                 bool is_partial_res,
1313                                 void *_ctx)
1314 {
1315         struct apply_ctx *ctx = _ctx;
1316         int ret;
1317
1318         if (!need_tmpfile_to_extract(lte, is_partial_res)) {
1319                 DEBUG("Temporary file not needed "
1320                       "for stream (size=%"PRIu64")", lte->size);
1321                 ret = extract_stream_instances(lte, lte, ctx);
1322                 if (ret)
1323                         return ret;
1324
1325                 return BEGIN_STREAM_STATUS_SKIP_STREAM;
1326         }
1327
1328         DEBUG("Temporary file needed for stream (size=%"PRIu64")", lte->size);
1329         return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
1330 }
1331
1332 static int
1333 end_extract_stream_to_tmpfile(struct wim_lookup_table_entry *lte,
1334                               int status, void *_ctx)
1335 {
1336         struct apply_ctx *ctx = _ctx;
1337         struct wim_lookup_table_entry lte_override;
1338         int ret;
1339         int errno_save = errno;
1340
1341         ret = filedes_close(&ctx->tmpfile_fd);
1342
1343         if (status) {
1344                 ret = status;
1345                 errno = errno_save;
1346                 goto out_delete_tmpfile;
1347         }
1348
1349         if (ret) {
1350                 ERROR_WITH_ERRNO("Error writing temporary file %"TS, ctx->tmpfile_name);
1351                 ret = WIMLIB_ERR_WRITE;
1352                 goto out_delete_tmpfile;
1353         }
1354
1355         /* Now that a full stream has been extracted to a temporary file,
1356          * extract all instances of it to the actual target.  */
1357
1358         memcpy(&lte_override, lte, sizeof(struct wim_lookup_table_entry));
1359         lte_override.resource_location = RESOURCE_IN_FILE_ON_DISK;
1360         lte_override.file_on_disk = ctx->tmpfile_name;
1361
1362         ret = extract_stream_instances(lte, &lte_override, ctx);
1363
1364 out_delete_tmpfile:
1365         errno_save = errno;
1366         tunlink(ctx->tmpfile_name);
1367         FREE(ctx->tmpfile_name);
1368         errno = errno_save;
1369         return ret;
1370 }
1371
1372 /* Extracts a list of streams (ctx.stream_list), assuming that the directory
1373  * structure and empty files were already created.  This relies on the
1374  * per-`struct wim_lookup_table_entry' list of dentries that reference each
1375  * stream that was constructed earlier.  */
1376 static int
1377 extract_stream_list(struct apply_ctx *ctx)
1378 {
1379         if (!(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
1380                 /* Sequential extraction: read the streams in the order in which
1381                  * they appear in the WIM file.  */
1382                 struct read_stream_list_callbacks cbs = {
1383                         .begin_stream           = begin_extract_stream_to_tmpfile,
1384                         .begin_stream_ctx       = ctx,
1385                         .consume_chunk          = extract_chunk_to_fd,
1386                         .consume_chunk_ctx      = &ctx->tmpfile_fd,
1387                         .end_stream             = end_extract_stream_to_tmpfile,
1388                         .end_stream_ctx         = ctx,
1389                 };
1390                 return read_stream_list(&ctx->stream_list,
1391                                         offsetof(struct wim_lookup_table_entry, extraction_list),
1392                                         &cbs, VERIFY_STREAM_HASHES);
1393         } else {
1394                 /* Extract the streams in unsorted order.  */
1395                 struct wim_lookup_table_entry *lte;
1396                 int ret;
1397
1398                 list_for_each_entry(lte, &ctx->stream_list, extraction_list) {
1399                         ret = extract_stream_instances(lte, lte, ctx);
1400                         if (ret)
1401                                 return ret;
1402                 }
1403                 return 0;
1404         }
1405 }
1406
1407 #define PWM_ALLOW_WIM_HDR 0x00001
1408 #define PWM_SILENT_EOF    0x00002
1409
1410 /* Read the header from a stream in a pipable WIM.  */
1411 static int
1412 read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte,
1413                        struct wim_resource_spec *rspec,
1414                        int flags, struct wim_header_disk *hdr_ret)
1415 {
1416         union {
1417                 struct pwm_stream_hdr stream_hdr;
1418                 struct wim_header_disk pwm_hdr;
1419         } buf;
1420         struct wim_reshdr reshdr;
1421         int ret;
1422
1423         ret = full_read(&pwm->in_fd, &buf.stream_hdr, sizeof(buf.stream_hdr));
1424         if (ret)
1425                 goto read_error;
1426
1427         if ((flags & PWM_ALLOW_WIM_HDR) && buf.stream_hdr.magic == PWM_MAGIC) {
1428                 BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr));
1429                 ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1,
1430                                 sizeof(buf.pwm_hdr) - sizeof(buf.stream_hdr));
1431
1432                 if (ret)
1433                         goto read_error;
1434                 lte->resource_location = RESOURCE_NONEXISTENT;
1435                 memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
1436                 return 0;
1437         }
1438
1439         if (le64_to_cpu(buf.stream_hdr.magic) != PWM_STREAM_MAGIC) {
1440                 ERROR("Data read on pipe is invalid (expected stream header).");
1441                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
1442         }
1443
1444         copy_hash(lte->hash, buf.stream_hdr.hash);
1445
1446         reshdr.size_in_wim = 0;
1447         reshdr.flags = le32_to_cpu(buf.stream_hdr.flags);
1448         reshdr.offset_in_wim = pwm->in_fd.offset;
1449         reshdr.uncompressed_size = le64_to_cpu(buf.stream_hdr.uncompressed_size);
1450         wim_res_hdr_to_spec(&reshdr, pwm, rspec);
1451         lte_bind_wim_resource_spec(lte, rspec);
1452         lte->flags = rspec->flags;
1453         lte->size = rspec->uncompressed_size;
1454         lte->offset_in_res = 0;
1455         return 0;
1456
1457 read_error:
1458         if (ret != WIMLIB_ERR_UNEXPECTED_END_OF_FILE || !(flags & PWM_SILENT_EOF))
1459                 ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
1460         return ret;
1461 }
1462
1463 static int
1464 extract_streams_from_pipe(struct apply_ctx *ctx)
1465 {
1466         struct wim_lookup_table_entry *found_lte;
1467         struct wim_resource_spec *rspec;
1468         struct wim_lookup_table_entry *needed_lte;
1469         struct wim_lookup_table *lookup_table;
1470         struct wim_header_disk pwm_hdr;
1471         int ret;
1472         int pwm_flags;
1473
1474         ret = WIMLIB_ERR_NOMEM;
1475         found_lte = new_lookup_table_entry();
1476         if (found_lte == NULL)
1477                 goto out;
1478
1479         rspec = MALLOC(sizeof(struct wim_resource_spec));
1480         if (rspec == NULL)
1481                 goto out_free_found_lte;
1482
1483         lookup_table = ctx->wim->lookup_table;
1484         pwm_flags = PWM_ALLOW_WIM_HDR;
1485         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1486                 pwm_flags |= PWM_SILENT_EOF;
1487         memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GID_LEN);
1488         ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
1489         ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
1490         if (ctx->progress_func)
1491                 ctx->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1492                                    &ctx->progress);
1493         while (ctx->num_streams_remaining) {
1494                 if (found_lte->resource_location != RESOURCE_NONEXISTENT)
1495                         lte_unbind_wim_resource_spec(found_lte);
1496                 ret = read_pwm_stream_header(ctx->wim, found_lte, rspec,
1497                                              pwm_flags, &pwm_hdr);
1498                 if (ret) {
1499                         if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE &&
1500                             (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1501                         {
1502                                 goto resume_done;
1503                         }
1504                         goto out_free_found_lte;
1505                 }
1506
1507                 if ((found_lte->resource_location != RESOURCE_NONEXISTENT)
1508                     && !(found_lte->flags & WIM_RESHDR_FLAG_METADATA)
1509                     && (needed_lte = lookup_resource(lookup_table, found_lte->hash))
1510                     && (needed_lte->out_refcnt))
1511                 {
1512                         tchar *tmpfile_name = NULL;
1513                         struct wim_lookup_table_entry *lte_override;
1514                         struct wim_lookup_table_entry tmpfile_lte;
1515
1516                         needed_lte->offset_in_res = found_lte->offset_in_res;
1517                         needed_lte->flags = found_lte->flags;
1518                         needed_lte->size = found_lte->size;
1519
1520                         lte_unbind_wim_resource_spec(found_lte);
1521                         lte_bind_wim_resource_spec(needed_lte, rspec);
1522
1523                         if (needed_lte->out_refcnt > 1) {
1524
1525                                 struct filedes tmpfile_fd;
1526
1527                                 /* Extract stream to temporary file.  */
1528                                 ret = create_temporary_file(&tmpfile_fd, &tmpfile_name);
1529                                 if (ret)
1530                                         goto out_free_found_lte;
1531
1532                                 ret = extract_stream_to_fd(needed_lte, &tmpfile_fd,
1533                                                            needed_lte->size);
1534                                 if (ret) {
1535                                         filedes_close(&tmpfile_fd);
1536                                         goto delete_tmpfile;
1537                                 }
1538
1539                                 if (filedes_close(&tmpfile_fd)) {
1540                                         ERROR_WITH_ERRNO("Error writing to temporary "
1541                                                          "file \"%"TS"\"", tmpfile_name);
1542                                         ret = WIMLIB_ERR_WRITE;
1543                                         goto delete_tmpfile;
1544                                 }
1545                                 memcpy(&tmpfile_lte, needed_lte,
1546                                        sizeof(struct wim_lookup_table_entry));
1547                                 tmpfile_lte.resource_location = RESOURCE_IN_FILE_ON_DISK;
1548                                 tmpfile_lte.file_on_disk = tmpfile_name;
1549                                 lte_override = &tmpfile_lte;
1550                         } else {
1551                                 lte_override = needed_lte;
1552                         }
1553
1554                         ret = extract_stream_instances(needed_lte, lte_override, ctx);
1555                 delete_tmpfile:
1556                         lte_unbind_wim_resource_spec(needed_lte);
1557                         if (tmpfile_name) {
1558                                 tunlink(tmpfile_name);
1559                                 FREE(tmpfile_name);
1560                         }
1561                         if (ret)
1562                                 goto out_free_found_lte;
1563                         ctx->num_streams_remaining--;
1564                 } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) {
1565                         ret = skip_wim_stream(found_lte);
1566                         if (ret)
1567                                 goto out_free_found_lte;
1568                 } else {
1569                         u16 part_number = le16_to_cpu(pwm_hdr.part_number);
1570                         u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
1571
1572                         if (part_number != ctx->progress.extract.part_number ||
1573                             total_parts != ctx->progress.extract.total_parts ||
1574                             memcmp(pwm_hdr.guid, ctx->progress.extract.guid,
1575                                    WIM_GID_LEN))
1576                         {
1577                                 ctx->progress.extract.part_number = part_number;
1578                                 ctx->progress.extract.total_parts = total_parts;
1579                                 memcpy(ctx->progress.extract.guid,
1580                                        pwm_hdr.guid, WIM_GID_LEN);
1581                                 if (ctx->progress_func) {
1582                                         ctx->progress_func(
1583                                                 WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1584                                                            &ctx->progress);
1585                                 }
1586
1587                         }
1588                 }
1589         }
1590         ret = 0;
1591 out_free_found_lte:
1592         if (found_lte->resource_location != RESOURCE_IN_WIM)
1593                 FREE(rspec);
1594         free_lookup_table_entry(found_lte);
1595 out:
1596         return ret;
1597
1598 resume_done:
1599         /* TODO */
1600         return 0;
1601 }
1602
1603 /* Finish extracting a file, directory, or symbolic link by setting file
1604  * security and timestamps.  */
1605 static int
1606 dentry_extract_final(struct wim_dentry *dentry, void *_ctx)
1607 {
1608         struct apply_ctx *ctx = _ctx;
1609         int ret;
1610         tchar path[ctx->ops->path_max];
1611
1612         if (!build_extraction_path(path, dentry, ctx))
1613                 return 0;
1614
1615         ret = extract_security(path, ctx, dentry);
1616         if (ret)
1617                 return ret;
1618
1619         if (ctx->ops->requires_final_set_attributes_pass) {
1620                 /* Set file attributes (if supported).  */
1621                 ret = extract_file_attributes(path, ctx, dentry, 1);
1622                 if (ret)
1623                         return ret;
1624         }
1625
1626         return extract_timestamps(path, ctx, dentry);
1627 }
1628
1629 /*
1630  * Extract a WIM dentry to standard output.
1631  *
1632  * This obviously doesn't make sense in all cases.  We return an error if the
1633  * dentry does not correspond to a regular file.  Otherwise we extract the
1634  * unnamed data stream only.
1635  */
1636 static int
1637 extract_dentry_to_stdout(struct wim_dentry *dentry)
1638 {
1639         int ret = 0;
1640         if (dentry->d_inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
1641                                              FILE_ATTRIBUTE_DIRECTORY))
1642         {
1643                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
1644                       "extracted to standard output", dentry_full_path(dentry));
1645                 ret = WIMLIB_ERR_NOT_A_REGULAR_FILE;
1646         } else {
1647                 struct wim_lookup_table_entry *lte;
1648
1649                 lte = inode_unnamed_lte_resolved(dentry->d_inode);
1650                 if (lte) {
1651                         struct filedes _stdout;
1652                         filedes_init(&_stdout, STDOUT_FILENO);
1653                         ret = extract_stream_to_fd(lte, &_stdout, lte->size);
1654                 }
1655         }
1656         return ret;
1657 }
1658
1659 #ifdef __WIN32__
1660 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
1661 #else
1662 static const utf16lechar replacement_char = cpu_to_le16('?');
1663 #endif
1664
1665 static bool
1666 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
1667 {
1668         size_t i;
1669
1670         if (num_chars == 0)
1671                 return true;
1672         for (i = 0; i < num_chars; i++) {
1673                 switch (name[i]) {
1674         #ifdef __WIN32__
1675                 case cpu_to_le16('\\'):
1676                 case cpu_to_le16(':'):
1677                 case cpu_to_le16('*'):
1678                 case cpu_to_le16('?'):
1679                 case cpu_to_le16('"'):
1680                 case cpu_to_le16('<'):
1681                 case cpu_to_le16('>'):
1682                 case cpu_to_le16('|'):
1683         #endif
1684                 case cpu_to_le16('/'):
1685                 case cpu_to_le16('\0'):
1686                         if (fix)
1687                                 name[i] = replacement_char;
1688                         else
1689                                 return false;
1690                 }
1691         }
1692
1693 #ifdef __WIN32__
1694         if (name[num_chars - 1] == cpu_to_le16(' ') ||
1695             name[num_chars - 1] == cpu_to_le16('.'))
1696         {
1697                 if (fix)
1698                         name[num_chars - 1] = replacement_char;
1699                 else
1700                         return false;
1701         }
1702 #endif
1703         return true;
1704 }
1705
1706 static bool
1707 dentry_is_dot_or_dotdot(const struct wim_dentry *dentry)
1708 {
1709         const utf16lechar *file_name = dentry->file_name;
1710         return file_name != NULL &&
1711                 file_name[0] == cpu_to_le16('.') &&
1712                 (file_name[1] == cpu_to_le16('\0') ||
1713                  (file_name[1] == cpu_to_le16('.') &&
1714                   file_name[2] == cpu_to_le16('\0')));
1715 }
1716
1717 static int
1718 dentry_mark_skipped(struct wim_dentry *dentry, void *_ignore)
1719 {
1720         dentry->extraction_skipped = 1;
1721         return 0;
1722 }
1723
1724 /*
1725  * dentry_calculate_extraction_path-
1726  *
1727  * Calculate the actual filename component at which a WIM dentry will be
1728  * extracted, handling invalid filenames "properly".
1729  *
1730  * dentry->extraction_name usually will be set the same as dentry->file_name (on
1731  * UNIX, converted into the platform's multibyte encoding).  However, if the
1732  * file name contains characters that are not valid on the current platform or
1733  * has some other format that is not valid, leave dentry->extraction_name as
1734  * NULL and set dentry->extraction_skipped to indicate that this dentry should
1735  * not be extracted, unless the appropriate flag
1736  * WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES is set in the extract flags, in
1737  * which case a substitute filename will be created and set instead.
1738  *
1739  * Conflicts with case-insensitive names on Windows are handled similarly; see
1740  * below.
1741  */
1742 static int
1743 dentry_calculate_extraction_path(struct wim_dentry *dentry, void *_args)
1744 {
1745         struct apply_ctx *ctx = _args;
1746         int ret;
1747
1748         dentry->in_extraction_tree = 1;
1749
1750         if (dentry == ctx->extract_root || dentry->extraction_skipped)
1751                 return 0;
1752
1753         if (!dentry_is_supported(dentry, &ctx->supported_features))
1754                 goto skip_dentry;
1755
1756         if (dentry_is_dot_or_dotdot(dentry)) {
1757                 /* WIM files shouldn't contain . or .. entries.  But if they are
1758                  * there, don't attempt to extract them. */
1759                 WARNING("Skipping extraction of unexpected . or .. file "
1760                         "\"%"TS"\"", dentry_full_path(dentry));
1761                 goto skip_dentry;
1762         }
1763
1764 #ifdef __WIN32__
1765         if (!ctx->ops->supports_case_sensitive_filenames)
1766         {
1767                 struct wim_dentry *other;
1768                 list_for_each_entry(other, &dentry->case_insensitive_conflict_list,
1769                                     case_insensitive_conflict_list)
1770                 {
1771                         if (ctx->extract_flags &
1772                             WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
1773                                 WARNING("\"%"TS"\" has the same "
1774                                         "case-insensitive name as "
1775                                         "\"%"TS"\"; extracting "
1776                                         "dummy name instead",
1777                                         dentry_full_path(dentry),
1778                                         dentry_full_path(other));
1779                                 goto out_replace;
1780                         } else {
1781                                 WARNING("Not extracting \"%"TS"\": "
1782                                         "has same case-insensitive "
1783                                         "name as \"%"TS"\"",
1784                                         dentry_full_path(dentry),
1785                                         dentry_full_path(other));
1786                                 goto skip_dentry;
1787                         }
1788                 }
1789         }
1790 #else   /* __WIN32__ */
1791         wimlib_assert(ctx->ops->supports_case_sensitive_filenames);
1792 #endif  /* !__WIN32__ */
1793
1794         if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
1795 #ifdef __WIN32__
1796                 dentry->extraction_name = dentry->file_name;
1797                 dentry->extraction_name_nchars = dentry->file_name_nbytes / 2;
1798                 return 0;
1799 #else
1800                 return utf16le_to_tstr(dentry->file_name,
1801                                        dentry->file_name_nbytes,
1802                                        &dentry->extraction_name,
1803                                        &dentry->extraction_name_nchars);
1804 #endif
1805         } else {
1806                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
1807                 {
1808                         WARNING("\"%"TS"\" has an invalid filename "
1809                                 "that is not supported on this platform; "
1810                                 "extracting dummy name instead",
1811                                 dentry_full_path(dentry));
1812                         goto out_replace;
1813                 } else {
1814                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
1815                                 "that is not supported on this platform",
1816                                 dentry_full_path(dentry));
1817                         goto skip_dentry;
1818                 }
1819         }
1820
1821 out_replace:
1822         {
1823                 utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2];
1824
1825                 memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
1826                 file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
1827
1828                 tchar *tchar_name;
1829                 size_t tchar_nchars;
1830         #ifdef __WIN32__
1831                 tchar_name = utf16_name_copy;
1832                 tchar_nchars = dentry->file_name_nbytes / 2;
1833         #else
1834                 ret = utf16le_to_tstr(utf16_name_copy,
1835                                       dentry->file_name_nbytes,
1836                                       &tchar_name, &tchar_nchars);
1837                 if (ret)
1838                         return ret;
1839         #endif
1840                 size_t fixed_name_num_chars = tchar_nchars;
1841                 tchar fixed_name[tchar_nchars + 50];
1842
1843                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
1844                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
1845                                                  T(" (invalid filename #%lu)"),
1846                                                  ++ctx->invalid_sequence);
1847         #ifndef __WIN32__
1848                 FREE(tchar_name);
1849         #endif
1850                 dentry->extraction_name = memdup(fixed_name,
1851                                                  2 * fixed_name_num_chars + 2);
1852                 if (!dentry->extraction_name)
1853                         return WIMLIB_ERR_NOMEM;
1854                 dentry->extraction_name_nchars = fixed_name_num_chars;
1855         }
1856         return 0;
1857
1858 skip_dentry:
1859         for_dentry_in_tree(dentry, dentry_mark_skipped, NULL);
1860         return 0;
1861 }
1862
1863 /* Clean up dentry and inode structure after extraction.  */
1864 static int
1865 dentry_reset_needs_extraction(struct wim_dentry *dentry, void *_ignore)
1866 {
1867         struct wim_inode *inode = dentry->d_inode;
1868
1869         dentry->in_extraction_tree = 0;
1870         dentry->extraction_skipped = 0;
1871         dentry->was_hardlinked = 0;
1872         dentry->skeleton_extracted = 0;
1873         inode->i_visited = 0;
1874         FREE(inode->i_extracted_file);
1875         inode->i_extracted_file = NULL;
1876         inode->i_dos_name_extracted = 0;
1877         if ((void*)dentry->extraction_name != (void*)dentry->file_name)
1878                 FREE(dentry->extraction_name);
1879         dentry->extraction_name = NULL;
1880         return 0;
1881 }
1882
1883 /* Tally features necessary to extract a dentry and the corresponding inode.  */
1884 static int
1885 dentry_tally_features(struct wim_dentry *dentry, void *_features)
1886 {
1887         struct wim_features *features = _features;
1888         struct wim_inode *inode = dentry->d_inode;
1889
1890         if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
1891                 features->archive_files++;
1892         if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
1893                 features->hidden_files++;
1894         if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
1895                 features->system_files++;
1896         if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
1897                 features->compressed_files++;
1898         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
1899                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1900                         features->encrypted_directories++;
1901                 else
1902                         features->encrypted_files++;
1903         }
1904         if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
1905                 features->not_context_indexed_files++;
1906         if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
1907                 features->sparse_files++;
1908         if (inode_has_named_stream(inode))
1909                 features->named_data_streams++;
1910         if (inode->i_visited)
1911                 features->hard_links++;
1912         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1913                 features->reparse_points++;
1914                 if (inode_is_symlink(inode))
1915                         features->symlink_reparse_points++;
1916                 else
1917                         features->other_reparse_points++;
1918         }
1919         if (inode->i_security_id != -1)
1920                 features->security_descriptors++;
1921         if (dentry->short_name_nbytes)
1922                 features->short_names++;
1923         if (inode_has_unix_data(inode))
1924                 features->unix_data++;
1925         inode->i_visited = 1;
1926         return 0;
1927 }
1928
1929 /* Tally the features necessary to extract a dentry tree.  */
1930 static void
1931 dentry_tree_get_features(struct wim_dentry *root, struct wim_features *features)
1932 {
1933         memset(features, 0, sizeof(struct wim_features));
1934         for_dentry_in_tree(root, dentry_tally_features, features);
1935         dentry_tree_clear_inode_visited(root);
1936 }
1937
1938 static u32
1939 compute_supported_attributes_mask(const struct wim_features *supported_features)
1940 {
1941         u32 mask = ~(u32)0;
1942
1943         if (!supported_features->archive_files)
1944                 mask &= ~FILE_ATTRIBUTE_ARCHIVE;
1945
1946         if (!supported_features->hidden_files)
1947                 mask &= ~FILE_ATTRIBUTE_HIDDEN;
1948
1949         if (!supported_features->system_files)
1950                 mask &= ~FILE_ATTRIBUTE_SYSTEM;
1951
1952         if (!supported_features->not_context_indexed_files)
1953                 mask &= ~FILE_ATTRIBUTE_NOT_CONTENT_INDEXED;
1954
1955         if (!supported_features->compressed_files)
1956                 mask &= ~FILE_ATTRIBUTE_COMPRESSED;
1957
1958         if (!supported_features->sparse_files)
1959                 mask &= ~FILE_ATTRIBUTE_SPARSE_FILE;
1960
1961         if (!supported_features->reparse_points)
1962                 mask &= ~FILE_ATTRIBUTE_REPARSE_POINT;
1963
1964         return mask;
1965 }
1966
1967 static int
1968 do_feature_check(const struct wim_features *required_features,
1969                  const struct wim_features *supported_features,
1970                  int extract_flags,
1971                  const struct apply_operations *ops,
1972                  const tchar *wim_source_path)
1973 {
1974         const tchar *loc;
1975         const tchar *mode = T("this extraction mode");
1976
1977         if (wim_source_path[0] == '\0')
1978                 loc = T("the WIM image");
1979         else
1980                 loc = wim_source_path;
1981
1982         /* We're an archive program, so theoretically we can do what we want
1983          * with FILE_ATTRIBUTE_ARCHIVE (which is a dumb flag anyway).  Don't
1984          * bother the user about it.  */
1985 #if 0
1986         if (required_features->archive_files && !supported_features->archive_files)
1987         {
1988                 WARNING(
1989           "%lu files in %"TS" are marked as archived, but this attribute\n"
1990 "          is not supported in %"TS".",
1991                         required_features->archive_files, loc, mode);
1992         }
1993 #endif
1994
1995         if (required_features->hidden_files && !supported_features->hidden_files)
1996         {
1997                 WARNING(
1998           "%lu files in %"TS" are marked as hidden, but this\n"
1999 "          attribute is not supported in %"TS".",
2000                         required_features->hidden_files, loc, mode);
2001         }
2002
2003         if (required_features->system_files && !supported_features->system_files)
2004         {
2005                 WARNING(
2006           "%lu files in %"TS" are marked as system files,\n"
2007 "          but this attribute is not supported in %"TS".",
2008                         required_features->system_files, loc, mode);
2009         }
2010
2011         if (required_features->compressed_files && !supported_features->compressed_files)
2012         {
2013                 WARNING(
2014           "%lu files in %"TS" are marked as being transparently\n"
2015 "          compressed, but transparent compression is not supported in\n"
2016 "          %"TS".  These files will be extracted as uncompressed.",
2017                         required_features->compressed_files, loc, mode);
2018         }
2019
2020         if (required_features->encrypted_files && !supported_features->encrypted_files)
2021         {
2022                 WARNING(
2023           "%lu files in %"TS" are marked as being encrypted,\n"
2024 "           but encryption is not supported in %"TS".  These files\n"
2025 "           will not be extracted.",
2026                         required_features->encrypted_files, loc, mode);
2027         }
2028
2029         if (required_features->encrypted_directories &&
2030             !supported_features->encrypted_directories)
2031         {
2032                 WARNING(
2033           "%lu directories in %"TS" are marked as being encrypted,\n"
2034 "           but encryption is not supported in %"TS".\n"
2035 "           These directories will be extracted as unencrypted.",
2036                         required_features->encrypted_directories, loc, mode);
2037         }
2038
2039         if (required_features->not_context_indexed_files &&
2040             !supported_features->not_context_indexed_files)
2041         {
2042                 WARNING(
2043           "%lu files in %"TS" are marked as not content indexed,\n"
2044 "          but this attribute is not supported in %"TS".",
2045                         required_features->not_context_indexed_files, loc, mode);
2046         }
2047
2048         if (required_features->sparse_files && !supported_features->sparse_files)
2049         {
2050                 WARNING(
2051           "%lu files in %"TS" are marked as sparse, but creating\n"
2052 "           sparse files is not supported in %"TS".  These files\n"
2053 "           will be extracted as non-sparse.",
2054                         required_features->sparse_files, loc, mode);
2055         }
2056
2057         if (required_features->named_data_streams &&
2058             !supported_features->named_data_streams)
2059         {
2060                 WARNING(
2061           "%lu files in %"TS" contain one or more alternate (named)\n"
2062 "          data streams, which are not supported in %"TS".\n"
2063 "          Alternate data streams will NOT be extracted.",
2064                         required_features->named_data_streams, loc, mode);
2065         }
2066
2067         if (unlikely(extract_flags & (WIMLIB_EXTRACT_FLAG_HARDLINK |
2068                                       WIMLIB_EXTRACT_FLAG_SYMLINK)) &&
2069             required_features->named_data_streams &&
2070             supported_features->named_data_streams)
2071         {
2072                 WARNING(
2073           "%lu files in %"TS" contain one or more alternate (named)\n"
2074 "          data streams, which are not supported in linked extraction mode.\n"
2075 "          Alternate data streams will NOT be extracted.",
2076                         required_features->named_data_streams, loc);
2077         }
2078
2079         if (required_features->hard_links && !supported_features->hard_links)
2080         {
2081                 WARNING(
2082           "%lu files in %"TS" are hard links, but hard links are\n"
2083 "          not supported in %"TS".  Hard links will be extracted as\n"
2084 "          duplicate copies of the linked files.",
2085                         required_features->hard_links, loc, mode);
2086         }
2087
2088         if (required_features->reparse_points && !supported_features->reparse_points)
2089         {
2090                 if (supported_features->symlink_reparse_points) {
2091                         if (required_features->other_reparse_points) {
2092                                 WARNING(
2093           "%lu files in %"TS" are reparse points that are neither\n"
2094 "          symbolic links nor junction points and are not supported in\n"
2095 "          %"TS".  These reparse points will not be extracted.",
2096                                         required_features->other_reparse_points, loc,
2097                                         mode);
2098                         }
2099                 } else {
2100                         WARNING(
2101           "%lu files in %"TS" are reparse points, which are\n"
2102 "          not supported in %"TS" and will not be extracted.",
2103                                 required_features->reparse_points, loc, mode);
2104                 }
2105         }
2106
2107         if (required_features->security_descriptors &&
2108             !supported_features->security_descriptors)
2109         {
2110                 WARNING(
2111           "%lu files in %"TS" have Windows NT security descriptors,\n"
2112 "          but extracting security descriptors is not supported in\n"
2113 "          %"TS".  No security descriptors will be extracted.",
2114                         required_features->security_descriptors, loc, mode);
2115         }
2116
2117         if (required_features->short_names && !supported_features->short_names)
2118         {
2119                 WARNING(
2120           "%lu files in %"TS" have short (DOS) names, but\n"
2121 "          extracting short names is not supported in %"TS".\n"
2122 "          Short names will not be extracted.\n",
2123                         required_features->short_names, loc, mode);
2124         }
2125
2126         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
2127             required_features->unix_data && !supported_features->unix_data)
2128         {
2129                 ERROR("Extracting UNIX data is not supported in %"TS, mode);
2130                 return WIMLIB_ERR_UNSUPPORTED;
2131         }
2132         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) &&
2133             required_features->short_names && !supported_features->short_names)
2134         {
2135                 ERROR("Extracting short names is not supported in %"TS"", mode);
2136                 return WIMLIB_ERR_UNSUPPORTED;
2137         }
2138         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
2139             !ops->set_timestamps)
2140         {
2141                 ERROR("Extracting timestamps is not supported in %"TS"", mode);
2142                 return WIMLIB_ERR_UNSUPPORTED;
2143         }
2144         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
2145                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
2146              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
2147             required_features->security_descriptors &&
2148             !supported_features->security_descriptors)
2149         {
2150                 ERROR("Extracting security descriptors is not supported in %"TS, mode);
2151                 return WIMLIB_ERR_UNSUPPORTED;
2152         }
2153
2154         if ((extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) &&
2155             !supported_features->hard_links)
2156         {
2157                 ERROR("Hard link extraction mode requested, but "
2158                       "%"TS" does not support hard links!", mode);
2159                 return WIMLIB_ERR_UNSUPPORTED;
2160         }
2161
2162         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
2163             required_features->symlink_reparse_points &&
2164             !(supported_features->symlink_reparse_points ||
2165               supported_features->reparse_points))
2166         {
2167                 ERROR("Extracting symbolic links is not supported in %"TS, mode);
2168                 return WIMLIB_ERR_UNSUPPORTED;
2169         }
2170
2171         if ((extract_flags & WIMLIB_EXTRACT_FLAG_SYMLINK) &&
2172             !supported_features->symlink_reparse_points)
2173         {
2174                 ERROR("Symbolic link extraction mode requested, but "
2175                       "%"TS" does not support symbolic "
2176                       "links!", mode);
2177                 return WIMLIB_ERR_UNSUPPORTED;
2178         }
2179         return 0;
2180 }
2181
2182 static void
2183 do_extract_warnings(struct apply_ctx *ctx)
2184 {
2185         if (ctx->partial_security_descriptors == 0 &&
2186             ctx->no_security_descriptors == 0)
2187                 return;
2188
2189         WARNING("Extraction to \"%"TS"\" complete, but with one or more warnings:",
2190                 ctx->target);
2191         if (ctx->partial_security_descriptors != 0) {
2192                 WARNING("- Could only partially set the security descriptor\n"
2193                         "            on %lu files or directories.",
2194                         ctx->partial_security_descriptors);
2195         }
2196         if (ctx->no_security_descriptors != 0) {
2197                 WARNING("- Could not set security descriptor at all\n"
2198                         "            on %lu files or directories.",
2199                         ctx->no_security_descriptors);
2200         }
2201 #ifdef __WIN32__
2202         WARNING("To fully restore all security descriptors, run the program\n"
2203                 "          with Administrator rights.");
2204 #endif
2205 }
2206
2207 /*
2208  * extract_tree - Extract a file or directory tree from the currently selected
2209  *                WIM image.
2210  *
2211  * @wim:        WIMStruct for the WIM file, with the desired image selected
2212  *              (as wim->current_image).
2213  *
2214  * @wim_source_path:
2215  *              "Canonical" (i.e. no leading or trailing slashes, path
2216  *              separators WIM_PATH_SEPARATOR) path inside the WIM image to
2217  *              extract.  An empty string means the full image.
2218  *
2219  * @target:
2220  *              Filesystem path to extract the file or directory tree to.
2221  *              (Or, with WIMLIB_EXTRACT_FLAG_NTFS: the name of a NTFS volume.)
2222  *
2223  * @extract_flags:
2224  *              WIMLIB_EXTRACT_FLAG_*.  Also, the private flag
2225  *              WIMLIB_EXTRACT_FLAG_MULTI_IMAGE will be set if this is being
2226  *              called through wimlib_extract_image() with WIMLIB_ALL_IMAGES as
2227  *              the image.
2228  *
2229  * @progress_func:
2230  *              If non-NULL, progress function for the extraction.  The messages
2231  *              that may be sent in this function are:
2232  *
2233  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN or
2234  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
2235  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN;
2236  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END;
2237  *              WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY;
2238  *              WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS;
2239  *              WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS;
2240  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END or
2241  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END.
2242  *
2243  * Returns 0 on success; a positive WIMLIB_ERR_* code on failure.
2244  */
2245 static int
2246 extract_tree(WIMStruct *wim, const tchar *wim_source_path, const tchar *target,
2247              int extract_flags, wimlib_progress_func_t progress_func)
2248 {
2249         struct wim_dentry *root;
2250         struct wim_features required_features;
2251         struct apply_ctx ctx;
2252         int ret;
2253         struct wim_lookup_table_entry *lte;
2254
2255         /* Start initializing the apply_ctx.  */
2256         memset(&ctx, 0, sizeof(struct apply_ctx));
2257         ctx.wim = wim;
2258         ctx.extract_flags = extract_flags;
2259         ctx.target = target;
2260         ctx.target_nchars = tstrlen(target);
2261         ctx.progress_func = progress_func;
2262         if (progress_func) {
2263                 ctx.progress.extract.wimfile_name = wim->filename;
2264                 ctx.progress.extract.image = wim->current_image;
2265                 ctx.progress.extract.extract_flags = (extract_flags &
2266                                                       WIMLIB_EXTRACT_MASK_PUBLIC);
2267                 ctx.progress.extract.image_name = wimlib_get_image_name(wim,
2268                                                                         wim->current_image);
2269                 ctx.progress.extract.extract_root_wim_source_path = wim_source_path;
2270                 ctx.progress.extract.target = target;
2271         }
2272         INIT_LIST_HEAD(&ctx.stream_list);
2273
2274         /* Translate the path to extract into the corresponding
2275          * `struct wim_dentry', which will be the root of the
2276          * "dentry tree" to extract.  */
2277         root = get_dentry(wim, wim_source_path, WIMLIB_CASE_PLATFORM_DEFAULT);
2278         if (!root) {
2279                 ERROR("Path \"%"TS"\" does not exist in WIM image %d",
2280                       wim_source_path, wim->current_image);
2281                 ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
2282                 goto out;
2283         }
2284
2285         ctx.extract_root = root;
2286
2287         /* Select the appropriate apply_operations based on the
2288          * platform and extract_flags.  */
2289 #ifdef __WIN32__
2290         ctx.ops = &win32_apply_ops;
2291 #else
2292         ctx.ops = &unix_apply_ops;
2293 #endif
2294
2295 #ifdef WITH_NTFS_3G
2296         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
2297                 ctx.ops = &ntfs_3g_apply_ops;
2298 #endif
2299
2300         /* Call the start_extract() callback.  This gives the apply_operations
2301          * implementation a chance to do any setup needed to access the volume.
2302          * Furthermore, it's expected to set the supported features of this
2303          * extraction mode (ctx.supported_features), which are determined at
2304          * runtime as they may vary depending on the actual volume.  These
2305          * features are then compared with the actual features extracting this
2306          * dentry tree requires.  Some mismatches will merely produce warnings
2307          * and the unsupported data will be ignored; others will produce errors.
2308          */
2309         ret = ctx.ops->start_extract(target, &ctx);
2310         if (ret)
2311                 goto out;
2312
2313         dentry_tree_get_features(root, &required_features);
2314         ret = do_feature_check(&required_features, &ctx.supported_features,
2315                                extract_flags, ctx.ops, wim_source_path);
2316         if (ret)
2317                 goto out_finish_or_abort_extract;
2318
2319         ctx.supported_attributes_mask =
2320                 compute_supported_attributes_mask(&ctx.supported_features);
2321
2322         /* Figure out whether the root dentry is being extracted to the root of
2323          * a volume and therefore needs to be treated "specially", for example
2324          * not being explicitly created and not having attributes set.  */
2325         if (ctx.ops->target_is_root && ctx.ops->root_directory_is_special)
2326                 ctx.root_dentry_is_special = ctx.ops->target_is_root(target);
2327
2328         /* Calculate the actual filename component of each extracted dentry.  In
2329          * the process, set the dentry->extraction_skipped flag on dentries that
2330          * are being skipped for some reason (e.g. invalid filename).  */
2331         ret = for_dentry_in_tree(root, dentry_calculate_extraction_path, &ctx);
2332         if (ret)
2333                 goto out_dentry_reset_needs_extraction;
2334
2335         /* Build the list of the streams that need to be extracted and
2336          * initialize ctx.progress.extract with stream information.  */
2337         ret = for_dentry_in_tree(ctx.extract_root,
2338                                  dentry_resolve_and_zero_lte_refcnt, &ctx);
2339         if (ret)
2340                 goto out_dentry_reset_needs_extraction;
2341
2342         ret = for_dentry_in_tree(ctx.extract_root,
2343                                  dentry_add_streams_to_extract, &ctx);
2344         if (ret)
2345                 goto out_teardown_stream_list;
2346
2347         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2348                 /* When extracting from a pipe, the number of bytes of data to
2349                  * extract can't be determined in the normal way (examining the
2350                  * lookup table), since at this point all we have is a set of
2351                  * SHA1 message digests of streams that need to be extracted.
2352                  * However, we can get a reasonably accurate estimate by taking
2353                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
2354                  * data.  This does assume that a full image is being extracted,
2355                  * but currently there is no API for doing otherwise.  (Also,
2356                  * subtract <HARDLINKBYTES> from this if hard links are
2357                  * supported by the extraction mode.)  */
2358                 ctx.progress.extract.total_bytes =
2359                         wim_info_get_image_total_bytes(wim->wim_info,
2360                                                        wim->current_image);
2361                 if (ctx.supported_features.hard_links) {
2362                         ctx.progress.extract.total_bytes -=
2363                                 wim_info_get_image_hard_link_bytes(wim->wim_info,
2364                                                                    wim->current_image);
2365                 }
2366         }
2367
2368         /* Handle the special case of extracting a file to standard
2369          * output.  In that case, "root" should be a single file, not a
2370          * directory tree.  (If not, extract_dentry_to_stdout() will
2371          * return an error.)  */
2372         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
2373                 ret = extract_dentry_to_stdout(root);
2374                 goto out_teardown_stream_list;
2375         }
2376
2377         if (ctx.ops->realpath_works_on_nonexisting_files &&
2378             ((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) ||
2379              ctx.ops->requires_realtarget_in_paths))
2380         {
2381                 ctx.realtarget = realpath(target, NULL);
2382                 if (!ctx.realtarget) {
2383                         ret = WIMLIB_ERR_NOMEM;
2384                         goto out_teardown_stream_list;
2385                 }
2386                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2387         }
2388
2389         if (progress_func) {
2390                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN :
2391                                                  WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN,
2392                               &ctx.progress);
2393         }
2394
2395         if (!ctx.root_dentry_is_special)
2396         {
2397                 tchar path[ctx.ops->path_max];
2398                 if (build_extraction_path(path, root, &ctx))
2399                 {
2400                         ret = extract_inode(path, &ctx, root->d_inode);
2401                         if (ret)
2402                                 goto out_free_realtarget;
2403                 }
2404         }
2405
2406         /* If we need to fix up the targets of absolute symbolic links
2407          * (WIMLIB_EXTRACT_FLAG_RPFIX) or the extraction mode requires paths to
2408          * be absolute, use realpath() (or its replacement on Windows) to get
2409          * the absolute path to the extraction target.  Note that this requires
2410          * the target directory to exist, unless
2411          * realpath_works_on_nonexisting_files is set in the apply_operations.
2412          * */
2413         if (!ctx.realtarget &&
2414             (((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
2415               required_features.symlink_reparse_points) ||
2416              ctx.ops->requires_realtarget_in_paths))
2417         {
2418                 ctx.realtarget = realpath(target, NULL);
2419                 if (!ctx.realtarget) {
2420                         ret = WIMLIB_ERR_NOMEM;
2421                         goto out_free_realtarget;
2422                 }
2423                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2424         }
2425
2426         if (ctx.ops->requires_short_name_reordering) {
2427                 ret = for_dentry_in_tree(root, dentry_extract_dir_skeleton,
2428                                          &ctx);
2429                 if (ret)
2430                         goto out_free_realtarget;
2431         }
2432
2433         /* Finally, the important part: extract the tree of files.  */
2434         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
2435                 /* Sequential extraction requested, so two passes are needed
2436                  * (one for directory structure, one for streams.)  */
2437                 if (progress_func)
2438                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2439                                       &ctx.progress);
2440
2441                 if (!(extract_flags & WIMLIB_EXTRACT_FLAG_RESUME)) {
2442                         ret = for_dentry_in_tree(root, dentry_extract_skeleton, &ctx);
2443                         if (ret)
2444                                 goto out_free_realtarget;
2445                 }
2446                 if (progress_func)
2447                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2448                                       &ctx.progress);
2449                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
2450                         ret = extract_streams_from_pipe(&ctx);
2451                 else
2452                         ret = extract_stream_list(&ctx);
2453                 if (ret)
2454                         goto out_free_realtarget;
2455         } else {
2456                 /* Sequential extraction was not requested, so we can make do
2457                  * with one pass where we both create the files and extract
2458                  * streams.   */
2459                 if (progress_func)
2460                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2461                                       &ctx.progress);
2462                 ret = for_dentry_in_tree(root, dentry_extract, &ctx);
2463                 if (ret)
2464                         goto out_free_realtarget;
2465                 if (progress_func)
2466                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2467                                       &ctx.progress);
2468         }
2469
2470         /* If the total number of bytes to extract was miscalculated, just jump
2471          * to the calculated number in order to avoid confusing the progress
2472          * function.  This should only occur when extracting from a pipe.  */
2473         if (ctx.progress.extract.completed_bytes != ctx.progress.extract.total_bytes)
2474         {
2475                 DEBUG("Calculated %"PRIu64" bytes to extract, but actually "
2476                       "extracted %"PRIu64,
2477                       ctx.progress.extract.total_bytes,
2478                       ctx.progress.extract.completed_bytes);
2479         }
2480         if (progress_func &&
2481             ctx.progress.extract.completed_bytes < ctx.progress.extract.total_bytes)
2482         {
2483                 ctx.progress.extract.completed_bytes = ctx.progress.extract.total_bytes;
2484                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, &ctx.progress);
2485         }
2486
2487         /* Apply security descriptors and timestamps.  This is done at the end,
2488          * and in a depth-first manner, to prevent timestamps from getting
2489          * changed by subsequent extract operations and to minimize the chance
2490          * of the restored security descriptors getting in our way.  */
2491         if (progress_func)
2492                 progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
2493                               &ctx.progress);
2494         ret = for_dentry_in_tree_depth(root, dentry_extract_final, &ctx);
2495         if (ret)
2496                 goto out_free_realtarget;
2497
2498         if (progress_func) {
2499                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END :
2500                               WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END,
2501                               &ctx.progress);
2502         }
2503
2504         do_extract_warnings(&ctx);
2505
2506         ret = 0;
2507 out_free_realtarget:
2508         FREE(ctx.realtarget);
2509 out_teardown_stream_list:
2510         /* Free memory allocated as part of the mapping from each
2511          * wim_lookup_table_entry to the dentries that reference it.  */
2512         if (!(ctx.extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER))
2513                 list_for_each_entry(lte, &ctx.stream_list, extraction_list)
2514                         if (lte->out_refcnt > ARRAY_LEN(lte->inline_lte_dentries))
2515                                 FREE(lte->lte_dentries);
2516 out_dentry_reset_needs_extraction:
2517         for_dentry_in_tree(root, dentry_reset_needs_extraction, NULL);
2518 out_finish_or_abort_extract:
2519         if (ret) {
2520                 if (ctx.ops->abort_extract)
2521                         ctx.ops->abort_extract(&ctx);
2522         } else {
2523                 if (ctx.ops->finish_extract)
2524                         ret = ctx.ops->finish_extract(&ctx);
2525         }
2526 out:
2527         return ret;
2528 }
2529
2530 /* Validates a single wimlib_extract_command, mostly checking to make sure the
2531  * extract flags make sense. */
2532 static int
2533 check_extract_command(struct wimlib_extract_command *cmd, int wim_header_flags)
2534 {
2535         int extract_flags;
2536
2537         /* Empty destination path? */
2538         if (cmd->fs_dest_path[0] == T('\0'))
2539                 return WIMLIB_ERR_INVALID_PARAM;
2540
2541         extract_flags = cmd->extract_flags;
2542
2543         /* Check for invalid flag combinations  */
2544         if ((extract_flags &
2545              (WIMLIB_EXTRACT_FLAG_SYMLINK |
2546               WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK |
2547                                                  WIMLIB_EXTRACT_FLAG_HARDLINK))
2548                 return WIMLIB_ERR_INVALID_PARAM;
2549
2550         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS)
2551                 return WIMLIB_ERR_INVALID_PARAM;
2552
2553         if ((extract_flags &
2554              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2555               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2556                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
2557                 return WIMLIB_ERR_INVALID_PARAM;
2558
2559         if ((extract_flags &
2560              (WIMLIB_EXTRACT_FLAG_RPFIX |
2561               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
2562                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
2563                 return WIMLIB_ERR_INVALID_PARAM;
2564
2565         if ((extract_flags &
2566              (WIMLIB_EXTRACT_FLAG_RESUME |
2567               WIMLIB_EXTRACT_FLAG_FROM_PIPE)) == WIMLIB_EXTRACT_FLAG_RESUME)
2568                 return WIMLIB_ERR_INVALID_PARAM;
2569
2570         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2571 #ifndef WITH_NTFS_3G
2572                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
2573                       "        we cannot apply a WIM image directly to a NTFS volume.");
2574                 return WIMLIB_ERR_UNSUPPORTED;
2575 #endif
2576         }
2577
2578         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
2579                               WIMLIB_EXTRACT_FLAG_NORPFIX)) == 0)
2580         {
2581                 /* Do reparse point fixups by default if the WIM header says
2582                  * they are enabled and we are extracting a full image. */
2583                 if (wim_header_flags & WIM_HDR_FLAG_RP_FIX)
2584                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
2585         }
2586
2587         /* TODO: Since UNIX data entries are stored in the file resources, in a
2588          * completely sequential extraction they may come up before the
2589          * corresponding file or symbolic link data.  This needs to be handled
2590          * better.  */
2591         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_UNIX_DATA |
2592                               WIMLIB_EXTRACT_FLAG_FILE_ORDER))
2593                                     == WIMLIB_EXTRACT_FLAG_UNIX_DATA)
2594         {
2595                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2596                         WARNING("Setting UNIX file/owner group may "
2597                                 "be impossible on some\n"
2598                                 "          symbolic links "
2599                                 "when applying from a pipe.");
2600                 } else {
2601                         extract_flags |= WIMLIB_EXTRACT_FLAG_FILE_ORDER;
2602                         WARNING("Disabling sequential extraction for "
2603                                 "UNIX data mode");
2604                 }
2605         }
2606
2607         cmd->extract_flags = extract_flags;
2608         return 0;
2609 }
2610
2611
2612 /* Internal function to execute extraction commands for a WIM image.  The paths
2613  * in the extract commands are expected to be already "canonicalized".  */
2614 static int
2615 do_wimlib_extract_files(WIMStruct *wim,
2616                         int image,
2617                         struct wimlib_extract_command *cmds,
2618                         size_t num_cmds,
2619                         wimlib_progress_func_t progress_func)
2620 {
2621         int ret;
2622         bool found_link_cmd = false;
2623         bool found_nolink_cmd = false;
2624
2625         /* Select the image from which we are extracting files */
2626         ret = select_wim_image(wim, image);
2627         if (ret)
2628                 return ret;
2629
2630         /* Make sure there are no streams in the WIM that have not been
2631          * checksummed yet.  */
2632         ret = wim_checksum_unhashed_streams(wim);
2633         if (ret)
2634                 return ret;
2635
2636         /* Check for problems with the extraction commands */
2637         for (size_t i = 0; i < num_cmds; i++) {
2638                 ret = check_extract_command(&cmds[i], wim->hdr.flags);
2639                 if (ret)
2640                         return ret;
2641                 if (cmds[i].extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2642                                              WIMLIB_EXTRACT_FLAG_HARDLINK)) {
2643                         found_link_cmd = true;
2644                 } else {
2645                         found_nolink_cmd = true;
2646                 }
2647                 if (found_link_cmd && found_nolink_cmd) {
2648                         ERROR("Symlink or hardlink extraction mode must "
2649                               "be set on all extraction commands");
2650                         return WIMLIB_ERR_INVALID_PARAM;
2651                 }
2652         }
2653
2654         /* Execute the extraction commands */
2655         for (size_t i = 0; i < num_cmds; i++) {
2656                 ret = extract_tree(wim,
2657                                    cmds[i].wim_source_path,
2658                                    cmds[i].fs_dest_path,
2659                                    cmds[i].extract_flags,
2660                                    progress_func);
2661                 if (ret)
2662                         return ret;
2663         }
2664         return 0;
2665 }
2666
2667 /* API function documented in wimlib.h  */
2668 WIMLIBAPI int
2669 wimlib_extract_files(WIMStruct *wim,
2670                      int image,
2671                      const struct wimlib_extract_command *cmds,
2672                      size_t num_cmds,
2673                      int default_extract_flags,
2674                      wimlib_progress_func_t progress_func)
2675 {
2676         int ret;
2677         struct wimlib_extract_command *cmds_copy;
2678         int all_flags = 0;
2679
2680         default_extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
2681
2682         if (num_cmds == 0)
2683                 return 0;
2684
2685         cmds_copy = CALLOC(num_cmds, sizeof(cmds[0]));
2686         if (!cmds_copy)
2687                 return WIMLIB_ERR_NOMEM;
2688
2689         for (size_t i = 0; i < num_cmds; i++) {
2690                 cmds_copy[i].extract_flags = (default_extract_flags |
2691                                                  cmds[i].extract_flags)
2692                                                 & WIMLIB_EXTRACT_MASK_PUBLIC;
2693                 all_flags |= cmds_copy[i].extract_flags;
2694
2695                 cmds_copy[i].wim_source_path = canonicalize_wim_path(cmds[i].wim_source_path);
2696                 if (!cmds_copy[i].wim_source_path) {
2697                         ret = WIMLIB_ERR_NOMEM;
2698                         goto out_free_cmds_copy;
2699                 }
2700
2701                 cmds_copy[i].fs_dest_path = canonicalize_fs_path(cmds[i].fs_dest_path);
2702                 if (!cmds_copy[i].fs_dest_path) {
2703                         ret = WIMLIB_ERR_NOMEM;
2704                         goto out_free_cmds_copy;
2705                 }
2706
2707         }
2708         ret = do_wimlib_extract_files(wim, image,
2709                                       cmds_copy, num_cmds,
2710                                       progress_func);
2711
2712         if (all_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2713                          WIMLIB_EXTRACT_FLAG_HARDLINK))
2714         {
2715                 for_lookup_table_entry(wim->lookup_table,
2716                                        lte_free_extracted_file, NULL);
2717         }
2718 out_free_cmds_copy:
2719         for (size_t i = 0; i < num_cmds; i++) {
2720                 FREE(cmds_copy[i].wim_source_path);
2721                 FREE(cmds_copy[i].fs_dest_path);
2722         }
2723         FREE(cmds_copy);
2724         return ret;
2725 }
2726
2727 /*
2728  * Extracts an image from a WIM file.
2729  *
2730  * @wim:                WIMStruct for the WIM file.
2731  *
2732  * @image:              Number of the single image to extract.
2733  *
2734  * @target:             Directory or NTFS volume to extract the image to.
2735  *
2736  * @extract_flags:      Bitwise or of WIMLIB_EXTRACT_FLAG_*.
2737  *
2738  * @progress_func:      If non-NULL, a progress function to be called
2739  *                      periodically.
2740  *
2741  * Returns 0 on success; nonzero on failure.
2742  */
2743 static int
2744 extract_single_image(WIMStruct *wim, int image,
2745                      const tchar *target, int extract_flags,
2746                      wimlib_progress_func_t progress_func)
2747 {
2748         int ret;
2749         tchar *target_copy = canonicalize_fs_path(target);
2750         if (target_copy == NULL)
2751                 return WIMLIB_ERR_NOMEM;
2752         struct wimlib_extract_command cmd = {
2753                 .wim_source_path = T(""),
2754                 .fs_dest_path = target_copy,
2755                 .extract_flags = extract_flags,
2756         };
2757         ret = do_wimlib_extract_files(wim, image, &cmd, 1, progress_func);
2758         FREE(target_copy);
2759         return ret;
2760 }
2761
2762 static const tchar * const filename_forbidden_chars =
2763 T(
2764 #ifdef __WIN32__
2765 "<>:\"/\\|?*"
2766 #else
2767 "/"
2768 #endif
2769 );
2770
2771 /* This function checks if it is okay to use a WIM image's name as a directory
2772  * name.  */
2773 static bool
2774 image_name_ok_as_dir(const tchar *image_name)
2775 {
2776         return image_name && *image_name &&
2777                 !tstrpbrk(image_name, filename_forbidden_chars) &&
2778                 tstrcmp(image_name, T(".")) &&
2779                 tstrcmp(image_name, T(".."));
2780 }
2781
2782 /* Extracts all images from the WIM to the directory @target, with the images
2783  * placed in subdirectories named by their image names. */
2784 static int
2785 extract_all_images(WIMStruct *wim,
2786                    const tchar *target,
2787                    int extract_flags,
2788                    wimlib_progress_func_t progress_func)
2789 {
2790         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
2791         size_t output_path_len = tstrlen(target);
2792         tchar buf[output_path_len + 1 + image_name_max_len + 1];
2793         int ret;
2794         int image;
2795         const tchar *image_name;
2796         struct stat stbuf;
2797
2798         extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
2799
2800         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2801                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
2802                 return WIMLIB_ERR_INVALID_PARAM;
2803         }
2804
2805         if (tstat(target, &stbuf)) {
2806                 if (errno == ENOENT) {
2807                         if (tmkdir(target, 0755)) {
2808                                 ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
2809                                 return WIMLIB_ERR_MKDIR;
2810                         }
2811                 } else {
2812                         ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
2813                         return WIMLIB_ERR_STAT;
2814                 }
2815         } else if (!S_ISDIR(stbuf.st_mode)) {
2816                 ERROR("\"%"TS"\" is not a directory", target);
2817                 return WIMLIB_ERR_NOTDIR;
2818         }
2819
2820         tmemcpy(buf, target, output_path_len);
2821         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
2822         for (image = 1; image <= wim->hdr.image_count; image++) {
2823                 image_name = wimlib_get_image_name(wim, image);
2824                 if (image_name_ok_as_dir(image_name)) {
2825                         tstrcpy(buf + output_path_len + 1, image_name);
2826                 } else {
2827                         /* Image name is empty or contains forbidden characters.
2828                          * Use image number instead. */
2829                         tsprintf(buf + output_path_len + 1, T("%d"), image);
2830                 }
2831                 ret = extract_single_image(wim, image, buf, extract_flags,
2832                                            progress_func);
2833                 if (ret)
2834                         return ret;
2835         }
2836         return 0;
2837 }
2838
2839 static int
2840 do_wimlib_extract_image(WIMStruct *wim,
2841                         int image,
2842                         const tchar *target,
2843                         int extract_flags,
2844                         wimlib_progress_func_t progress_func)
2845 {
2846         int ret;
2847
2848         if (image == WIMLIB_ALL_IMAGES) {
2849                 ret = extract_all_images(wim, target, extract_flags,
2850                                          progress_func);
2851         } else {
2852                 ret = extract_single_image(wim, image, target, extract_flags,
2853                                            progress_func);
2854         }
2855
2856         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2857                              WIMLIB_EXTRACT_FLAG_HARDLINK))
2858         {
2859                 for_lookup_table_entry(wim->lookup_table,
2860                                        lte_free_extracted_file,
2861                                        NULL);
2862         }
2863         return ret;
2864 }
2865
2866 /* API function documented in wimlib.h  */
2867 WIMLIBAPI int
2868 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
2869                                const tchar *target, int extract_flags,
2870                                wimlib_progress_func_t progress_func)
2871 {
2872         int ret;
2873         WIMStruct *pwm;
2874         struct filedes *in_fd;
2875         int image;
2876         unsigned i;
2877
2878         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
2879
2880         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT)
2881                 return WIMLIB_ERR_INVALID_PARAM;
2882
2883         /* Read the WIM header from the pipe and get a WIMStruct to represent
2884          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
2885          * wimlib_open_wim(), getting a WIMStruct in this way will result in
2886          * an empty lookup table, no XML data read, and no filename set.  */
2887         ret = open_wim_as_WIMStruct(&pipe_fd,
2888                                     WIMLIB_OPEN_FLAG_FROM_PIPE,
2889                                     &pwm, progress_func);
2890         if (ret)
2891                 return ret;
2892
2893         /* Sanity check to make sure this is a pipable WIM.  */
2894         if (pwm->hdr.magic != PWM_MAGIC) {
2895                 ERROR("The WIM being read from file descriptor %d "
2896                       "is not pipable!", pipe_fd);
2897                 ret = WIMLIB_ERR_NOT_PIPABLE;
2898                 goto out_wimlib_free;
2899         }
2900
2901         /* Sanity check to make sure the first part of a pipable split WIM is
2902          * sent over the pipe first.  */
2903         if (pwm->hdr.part_number != 1) {
2904                 ERROR("The first part of the split WIM must be "
2905                       "sent over the pipe first.");
2906                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
2907                 goto out_wimlib_free;
2908         }
2909
2910         in_fd = &pwm->in_fd;
2911         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
2912
2913         /* As mentioned, the WIMStruct we created from the pipe does not have
2914          * XML data yet.  Fix this by reading the extra copy of the XML data
2915          * that directly follows the header in pipable WIMs.  (Note: see
2916          * write_pipable_wim() for more details about the format of pipable
2917          * WIMs.)  */
2918         {
2919                 struct wim_lookup_table_entry xml_lte;
2920                 struct wim_resource_spec xml_rspec;
2921                 ret = read_pwm_stream_header(pwm, &xml_lte, &xml_rspec, 0, NULL);
2922                 if (ret)
2923                         goto out_wimlib_free;
2924
2925                 if (!(xml_lte.flags & WIM_RESHDR_FLAG_METADATA))
2926                 {
2927                         ERROR("Expected XML data, but found non-metadata "
2928                               "stream.");
2929                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
2930                         goto out_wimlib_free;
2931                 }
2932
2933                 wim_res_spec_to_hdr(&xml_rspec, &pwm->hdr.xml_data_reshdr);
2934
2935                 ret = read_wim_xml_data(pwm);
2936                 if (ret)
2937                         goto out_wimlib_free;
2938
2939                 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
2940                         ERROR("Image count in XML data is not the same as in WIM header.");
2941                         ret = WIMLIB_ERR_XML;
2942                         goto out_wimlib_free;
2943                 }
2944         }
2945
2946         /* Get image index (this may use the XML data that was just read to
2947          * resolve an image name).  */
2948         if (image_num_or_name) {
2949                 image = wimlib_resolve_image(pwm, image_num_or_name);
2950                 if (image == WIMLIB_NO_IMAGE) {
2951                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
2952                               image_num_or_name);
2953                         ret = WIMLIB_ERR_INVALID_IMAGE;
2954                         goto out_wimlib_free;
2955                 } else if (image == WIMLIB_ALL_IMAGES) {
2956                         ERROR("Applying all images from a pipe is not supported.");
2957                         ret = WIMLIB_ERR_INVALID_IMAGE;
2958                         goto out_wimlib_free;
2959                 }
2960         } else {
2961                 if (pwm->hdr.image_count != 1) {
2962                         ERROR("No image was specified, but the pipable WIM "
2963                               "did not contain exactly 1 image");
2964                         ret = WIMLIB_ERR_INVALID_IMAGE;
2965                         goto out_wimlib_free;
2966                 }
2967                 image = 1;
2968         }
2969
2970         /* Load the needed metadata resource.  */
2971         for (i = 1; i <= pwm->hdr.image_count; i++) {
2972                 struct wim_lookup_table_entry *metadata_lte;
2973                 struct wim_image_metadata *imd;
2974                 struct wim_resource_spec *metadata_rspec;
2975
2976                 metadata_lte = new_lookup_table_entry();
2977                 if (metadata_lte == NULL) {
2978                         ret = WIMLIB_ERR_NOMEM;
2979                         goto out_wimlib_free;
2980                 }
2981                 metadata_rspec = MALLOC(sizeof(struct wim_resource_spec));
2982                 if (metadata_rspec == NULL) {
2983                         ret = WIMLIB_ERR_NOMEM;
2984                         free_lookup_table_entry(metadata_lte);
2985                         goto out_wimlib_free;
2986                 }
2987
2988                 ret = read_pwm_stream_header(pwm, metadata_lte, metadata_rspec, 0, NULL);
2989                 imd = pwm->image_metadata[i - 1];
2990                 imd->metadata_lte = metadata_lte;
2991                 if (ret) {
2992                         FREE(metadata_rspec);
2993                         goto out_wimlib_free;
2994                 }
2995
2996                 if (!(metadata_lte->flags & WIM_RESHDR_FLAG_METADATA)) {
2997                         ERROR("Expected metadata resource, but found "
2998                               "non-metadata stream.");
2999                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
3000                         goto out_wimlib_free;
3001                 }
3002
3003                 if (i == image) {
3004                         /* Metadata resource is for the images being extracted.
3005                          * Parse it and save the metadata in memory.  */
3006                         ret = read_metadata_resource(pwm, imd);
3007                         if (ret)
3008                                 goto out_wimlib_free;
3009                         imd->modified = 1;
3010                 } else {
3011                         /* Metadata resource is not for the image being
3012                          * extracted.  Skip over it.  */
3013                         ret = skip_wim_stream(metadata_lte);
3014                         if (ret)
3015                                 goto out_wimlib_free;
3016                 }
3017         }
3018         /* Extract the image.  */
3019         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
3020         ret = do_wimlib_extract_image(pwm, image, target,
3021                                       extract_flags, progress_func);
3022         /* Clean up and return.  */
3023 out_wimlib_free:
3024         wimlib_free(pwm);
3025         return ret;
3026 }
3027
3028 /* API function documented in wimlib.h  */
3029 WIMLIBAPI int
3030 wimlib_extract_image(WIMStruct *wim,
3031                      int image,
3032                      const tchar *target,
3033                      int extract_flags,
3034                      wimlib_progress_func_t progress_func)
3035 {
3036         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
3037         return do_wimlib_extract_image(wim, image, target, extract_flags,
3038                                        progress_func);
3039 }
3040
3041 /* API function documented in wimlib.h  */
3042 WIMLIBAPI int
3043 wimlib_extract_pathlist(WIMStruct *wim, int image,
3044                         const tchar *target,
3045                         const tchar *path_list_file,
3046                         int extract_flags,
3047                         wimlib_progress_func_t progress_func)
3048 {
3049         int ret;
3050         tchar **paths;
3051         size_t num_paths;
3052         void *mem;
3053
3054         ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
3055         if (ret)
3056                 return ret;
3057
3058         ret = wimlib_extract_paths(wim, image, target,
3059                                    (const tchar * const *)paths, num_paths,
3060                                    extract_flags, progress_func);
3061         FREE(paths);
3062         FREE(mem);
3063         return ret;
3064 }
3065
3066 /* API function documented in wimlib.h  */
3067 WIMLIBAPI int
3068 wimlib_extract_paths(WIMStruct *wim,
3069                      int image,
3070                      const tchar *target,
3071                      const tchar * const *paths,
3072                      size_t num_paths,
3073                      int extract_flags,
3074                      wimlib_progress_func_t progress_func)
3075 {
3076         int ret;
3077         tchar **expanded_paths;
3078         size_t num_expanded_paths;
3079         struct wimlib_extract_command *cmds;
3080
3081         ret = select_wim_image(wim, image);
3082         if (ret)
3083                 return ret;
3084
3085         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
3086                 int wildcard_flags = 0;
3087
3088                 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB)
3089                         wildcard_flags |= WILDCARD_FLAG_ERROR_IF_NO_MATCH;
3090                 else
3091                         wildcard_flags |= WILDCARD_FLAG_WARN_IF_NO_MATCH;
3092
3093                 if (default_ignore_case)
3094                         wildcard_flags |= WILDCARD_FLAG_CASE_INSENSITIVE;
3095
3096                 ret = expand_wildcard_wim_paths(wim, paths, num_paths,
3097                                                 &expanded_paths,
3098                                                 &num_expanded_paths,
3099                                                 wildcard_flags);
3100                 if (ret)
3101                         return ret;
3102         } else {
3103                 expanded_paths = (tchar**)paths;
3104                 num_expanded_paths = num_paths;
3105         }
3106
3107         cmds = CALLOC(num_expanded_paths, sizeof(cmds[0]));
3108         if (cmds == NULL) {
3109                 ret = WIMLIB_ERR_NOMEM;
3110                 goto out_free_expanded_paths;
3111         }
3112
3113         for (size_t i = 0; i < num_expanded_paths; i++) {
3114                 cmds[i].wim_source_path = expanded_paths[i];
3115                 cmds[i].extract_flags = 0;
3116
3117                 tchar *dest_path;
3118                 size_t dest_len = 0;
3119                 dest_len += tstrlen(target);
3120                 dest_len += 1;
3121                 dest_len += tstrlen(expanded_paths[i]);
3122                 dest_len += 1;
3123
3124                 dest_path = MALLOC(dest_len * sizeof(tchar));
3125                 if (dest_path == NULL) {
3126                         ret = WIMLIB_ERR_NOMEM;
3127                         goto out_free_extraction_cmds;
3128                 }
3129                 tchar *p = dest_path;
3130                 p = tmempcpy(p, target, tstrlen(target));
3131                 *p++ = OS_PREFERRED_PATH_SEPARATOR;
3132                 for (tchar *path_p = expanded_paths[i]; *path_p != '\0'; path_p++) {
3133                         if (is_any_path_separator(*path_p))
3134                                 *p++ = OS_PREFERRED_PATH_SEPARATOR;
3135                         else
3136                                 *p++ = *path_p;
3137                 }
3138                 *p++ = T('\0');
3139                 wimlib_assert(p - dest_path == dest_len);
3140                 cmds[i].fs_dest_path = dest_path;
3141         }
3142
3143         ret = wimlib_extract_files(wim, image,
3144                                    cmds, num_expanded_paths,
3145                                    extract_flags & ~WIMLIB_EXTRACT_FLAG_GLOB_PATHS,
3146                                    progress_func);
3147 out_free_extraction_cmds:
3148         for (size_t i = 0; i < num_expanded_paths; i++)
3149                 FREE(cmds[i].fs_dest_path);
3150         FREE(cmds);
3151 out_free_expanded_paths:
3152         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
3153                 for (size_t i = 0; i < num_expanded_paths; i++)
3154                         FREE(expanded_paths[i]);
3155                 FREE(expanded_paths);
3156         }
3157         return ret;
3158 }