]> wimlib.net Git - wimlib/blob - src/extract.c
abce21f030f0c7776d49f2f4e1ff27aba5e9ceb6
[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_FLAG_PATHMODE    0x20000000
73 #define WIMLIB_EXTRACT_MASK_PUBLIC      0x1fffffff
74
75 /* Given a WIM dentry in the tree to be extracted, resolve all streams in the
76  * corresponding inode and set 'out_refcnt' in each to 0.  */
77 static int
78 dentry_resolve_and_zero_lte_refcnt(struct wim_dentry *dentry, void *_ctx)
79 {
80         struct apply_ctx *ctx = _ctx;
81         struct wim_inode *inode = dentry->d_inode;
82         struct wim_lookup_table_entry *lte;
83         int ret;
84         bool force = false;
85
86         if (dentry->extraction_skipped)
87                 return 0;
88
89         /* Special case:  when extracting from a pipe, the WIM lookup table is
90          * initially empty, so "resolving" an inode's streams is initially not
91          * possible.  However, we still need to keep track of which streams,
92          * identified by SHA1 message digests, need to be extracted, so we
93          * "resolve" the inode's streams anyway by allocating new entries.  */
94         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
95                 force = true;
96         ret = inode_resolve_ltes(inode, ctx->wim->lookup_table, force);
97         if (ret)
98                 return ret;
99         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
100                 lte = inode_stream_lte_resolved(inode, i);
101                 if (lte)
102                         lte->out_refcnt = 0;
103         }
104         return 0;
105 }
106
107 static inline bool
108 is_linked_extraction(const struct apply_ctx *ctx)
109 {
110         return 0 != (ctx->extract_flags & (WIMLIB_EXTRACT_FLAG_HARDLINK |
111                                            WIMLIB_EXTRACT_FLAG_SYMLINK));
112 }
113
114 static inline bool
115 can_extract_named_data_streams(const struct apply_ctx *ctx)
116 {
117         return ctx->supported_features.named_data_streams &&
118                 !is_linked_extraction(ctx);
119 }
120
121 static int
122 ref_stream_to_extract(struct wim_lookup_table_entry *lte,
123                       struct wim_dentry *dentry, struct apply_ctx *ctx)
124 {
125         if (!lte)
126                 return 0;
127
128         /* Tally the size only for each extraction of the stream (not hard
129          * links).  */
130         if (!(dentry->d_inode->i_visited &&
131               ctx->supported_features.hard_links) &&
132             (!is_linked_extraction(ctx) || (lte->out_refcnt == 0 &&
133                                             lte->extracted_file == NULL)))
134         {
135                 ctx->progress.extract.total_bytes += lte->size;
136                 ctx->progress.extract.num_streams++;
137         }
138
139         /* Add stream to the extraction_list only one time, even if it's going
140          * to be extracted to multiple locations.  */
141         if (lte->out_refcnt == 0) {
142                 list_add_tail(&lte->extraction_list, &ctx->stream_list);
143                 ctx->num_streams_remaining++;
144         }
145
146         if (!(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
147                 struct wim_dentry **lte_dentries;
148
149                 /* Append dentry to this stream's array of dentries referencing
150                  * it.  Use inline array to avoid memory allocation until the
151                  * number of dentries becomes too large.  */
152                 if (lte->out_refcnt < ARRAY_LEN(lte->inline_lte_dentries)) {
153                         lte_dentries = lte->inline_lte_dentries;
154                 } else {
155                         struct wim_dentry **prev_lte_dentries;
156                         size_t alloc_lte_dentries;
157
158                         if (lte->out_refcnt == ARRAY_LEN(lte->inline_lte_dentries)) {
159                                 prev_lte_dentries = NULL;
160                                 alloc_lte_dentries = ARRAY_LEN(lte->inline_lte_dentries);
161                         } else {
162                                 prev_lte_dentries = lte->lte_dentries;
163                                 alloc_lte_dentries = lte->alloc_lte_dentries;
164                         }
165
166                         if (lte->out_refcnt == alloc_lte_dentries) {
167                                 alloc_lte_dentries *= 2;
168                                 lte_dentries = REALLOC(prev_lte_dentries,
169                                                        alloc_lte_dentries *
170                                                         sizeof(lte_dentries[0]));
171                                 if (lte_dentries == NULL)
172                                         return WIMLIB_ERR_NOMEM;
173                                 if (prev_lte_dentries == NULL) {
174                                         memcpy(lte_dentries,
175                                                lte->inline_lte_dentries,
176                                                sizeof(lte->inline_lte_dentries));
177                                 }
178                                 lte->lte_dentries = lte_dentries;
179                                 lte->alloc_lte_dentries = alloc_lte_dentries;
180                         }
181                         lte_dentries = lte->lte_dentries;
182                 }
183                 lte_dentries[lte->out_refcnt] = dentry;
184         }
185         lte->out_refcnt++;
186         return 0;
187 }
188
189 /* Given a WIM dentry in the tree to be extracted, iterate through streams that
190  * need to be extracted.  For each one, add it to the list of streams to be
191  * extracted (ctx->stream_list) if not already done so, and also update the
192  * progress information (ctx->progress) with the stream.  Furthermore, if doing
193  * a sequential extraction, build a mapping from each the stream to the dentries
194  * referencing it.
195  *
196  * This uses the i_visited member of the inodes (assumed to be 0 initially).  */
197 static int
198 dentry_add_streams_to_extract(struct wim_dentry *dentry, void *_ctx)
199 {
200         struct apply_ctx *ctx = _ctx;
201         struct wim_inode *inode = dentry->d_inode;
202         int ret;
203
204         /* Don't process dentries marked as skipped.  */
205         if (dentry->extraction_skipped)
206                 return 0;
207
208         /* The unnamed data stream will always be extracted, except in an
209          * unlikely case.  */
210         if (!inode_is_encrypted_directory(inode)) {
211                 ret = ref_stream_to_extract(inode_unnamed_lte_resolved(inode),
212                                             dentry, ctx);
213                 if (ret)
214                         return ret;
215         }
216
217         /* Named data streams will be extracted only if supported in the current
218          * extraction mode and volume, and to avoid complications, if not doing
219          * a linked extraction.  */
220         if (can_extract_named_data_streams(ctx)) {
221                 for (u16 i = 0; i < inode->i_num_ads; i++) {
222                         if (!ads_entry_is_named_stream(&inode->i_ads_entries[i]))
223                                 continue;
224                         ret = ref_stream_to_extract(inode->i_ads_entries[i].lte,
225                                                     dentry, ctx);
226                         if (ret)
227                                 return ret;
228                 }
229         }
230         inode->i_visited = 1;
231         return 0;
232 }
233
234 /* Inform library user of progress of stream extraction following the successful
235  * extraction of a copy of the stream specified by @lte.  */
236 static void
237 update_extract_progress(struct apply_ctx *ctx,
238                         const struct wim_lookup_table_entry *lte)
239 {
240         wimlib_progress_func_t progress_func = ctx->progress_func;
241         union wimlib_progress_info *progress = &ctx->progress;
242
243         progress->extract.completed_bytes += lte->size;
244         if (progress_func &&
245             progress->extract.completed_bytes >= ctx->next_progress)
246         {
247                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, progress);
248                 if (progress->extract.completed_bytes >=
249                     progress->extract.total_bytes)
250                 {
251                         ctx->next_progress = ~0ULL;
252                 } else {
253                         ctx->next_progress += progress->extract.total_bytes / 128;
254                         if (ctx->next_progress > progress->extract.total_bytes)
255                                 ctx->next_progress = progress->extract.total_bytes;
256                 }
257         }
258 }
259
260 #ifndef __WIN32__
261 /* Extract a symbolic link (not directly as reparse data), handling fixing up
262  * the target of absolute symbolic links and updating the extract progress.
263  *
264  * @inode must specify the WIM inode for a symbolic link or junction reparse
265  * point.
266  *
267  * @lte_override overrides the resource used as the reparse data for the
268  * symbolic link.  */
269 static int
270 extract_symlink(const tchar *path, struct apply_ctx *ctx,
271                 struct wim_inode *inode,
272                 struct wim_lookup_table_entry *lte_override)
273 {
274         ssize_t bufsize = ctx->ops->path_max;
275         tchar target[bufsize];
276         tchar *buf = target;
277         tchar *fixed_target;
278         ssize_t sret;
279         int ret;
280
281         /* If absolute symbolic link fixups requested, reserve space in the link
282          * target buffer for the absolute path of the target directory.  */
283         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX)
284         {
285                 buf += ctx->realtarget_nchars;
286                 bufsize -= ctx->realtarget_nchars;
287         }
288
289         /* Translate the WIM inode's reparse data into the link target.  */
290         sret = wim_inode_readlink(inode, buf, bufsize - 1, lte_override);
291         if (sret < 0) {
292                 errno = -sret;
293                 return WIMLIB_ERR_READLINK;
294         }
295         buf[sret] = '\0';
296
297         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
298             buf[0] == '/')
299         {
300                 /* Fix absolute symbolic link target to point into the
301                  * actual extraction destination.  */
302                 tmemcpy(target, ctx->realtarget, ctx->realtarget_nchars);
303                 fixed_target = target;
304         } else {
305                 /* Keep same link target.  */
306                 fixed_target = buf;
307         }
308
309         /* Call into the apply_operations to create the symbolic link.  */
310         DEBUG("Creating symlink \"%"TS"\" => \"%"TS"\"",
311               path, fixed_target);
312         ret = ctx->ops->create_symlink(fixed_target, path, ctx);
313         if (ret) {
314                 ERROR_WITH_ERRNO("Failed to create symlink "
315                                  "\"%"TS"\" => \"%"TS"\"", path, fixed_target);
316                 return ret;
317         }
318
319         /* Account for reparse data consumed.  */
320         update_extract_progress(ctx,
321                                 (lte_override ? lte_override :
322                                       inode_unnamed_lte_resolved(inode)));
323         return 0;
324 }
325 #endif /* !__WIN32__ */
326
327 /* Create a file, directory, or symbolic link.  */
328 static int
329 extract_inode(const tchar *path, struct apply_ctx *ctx, struct wim_inode *inode)
330 {
331         int ret;
332
333 #ifndef __WIN32__
334         if (ctx->supported_features.symlink_reparse_points &&
335             !ctx->supported_features.reparse_points &&
336             inode_is_symlink(inode))
337         {
338                 ret = extract_symlink(path, ctx, inode, NULL);
339         } else
340 #endif /* !__WIN32__ */
341         if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
342                 ret = ctx->ops->create_directory(path, ctx, &inode->extract_cookie);
343                 if (ret) {
344                         ERROR_WITH_ERRNO("Failed to create the directory "
345                                          "\"%"TS"\"", path);
346                 }
347         } else if ((inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) &&
348                     ctx->ops->extract_encrypted_stream_creates_file &&
349                     ctx->supported_features.encrypted_files) {
350                 ret = ctx->ops->extract_encrypted_stream(
351                                 path, inode_unnamed_lte_resolved(inode), ctx);
352                 if (ret) {
353                         ERROR_WITH_ERRNO("Failed to create and extract "
354                                          "encrypted file \"%"TS"\"", path);
355                 }
356         } else {
357                 ret = ctx->ops->create_file(path, ctx, &inode->extract_cookie);
358                 if (ret) {
359                         ERROR_WITH_ERRNO("Failed to create the file "
360                                          "\"%"TS"\"", path);
361                 }
362         }
363         return ret;
364 }
365
366 static int
367 extract_hardlink(const tchar *oldpath, const tchar *newpath,
368                  struct apply_ctx *ctx)
369 {
370         int ret;
371
372         DEBUG("Creating hardlink \"%"TS"\" => \"%"TS"\"", newpath, oldpath);
373         ret = ctx->ops->create_hardlink(oldpath, newpath, ctx);
374         if (ret) {
375                 ERROR_WITH_ERRNO("Failed to create hardlink "
376                                  "\"%"TS"\" => \"%"TS"\"",
377                                  newpath, oldpath);
378         }
379         return ret;
380 }
381
382 #ifdef __WIN32__
383 static int
384 try_extract_rpfix(u8 *rpbuf,
385                   u16 *rpbuflen_p,
386                   const wchar_t *extract_root_realpath,
387                   unsigned extract_root_realpath_nchars)
388 {
389         struct reparse_data rpdata;
390         wchar_t *target;
391         size_t target_nchars;
392         size_t stripped_nchars;
393         wchar_t *stripped_target;
394         wchar_t stripped_target_nchars;
395         int ret;
396
397         utf16lechar *new_target;
398         utf16lechar *new_print_name;
399         size_t new_target_nchars;
400         size_t new_print_name_nchars;
401         utf16lechar *p;
402
403         ret = parse_reparse_data(rpbuf, *rpbuflen_p, &rpdata);
404         if (ret)
405                 return ret;
406
407         if (extract_root_realpath[0] == L'\0' ||
408             extract_root_realpath[1] != L':' ||
409             extract_root_realpath[2] != L'\\')
410                 return WIMLIB_ERR_REPARSE_POINT_FIXUP_FAILED;
411
412         ret = parse_substitute_name(rpdata.substitute_name,
413                                     rpdata.substitute_name_nbytes,
414                                     rpdata.rptag);
415         if (ret < 0)
416                 return 0;
417         stripped_nchars = ret;
418         target = rpdata.substitute_name;
419         target_nchars = rpdata.substitute_name_nbytes / sizeof(utf16lechar);
420         stripped_target = target + stripped_nchars;
421         stripped_target_nchars = target_nchars - stripped_nchars;
422
423         new_target = alloca((6 + extract_root_realpath_nchars +
424                              stripped_target_nchars) * sizeof(utf16lechar));
425
426         p = new_target;
427         if (stripped_nchars == 6) {
428                 /* Include \??\ prefix if it was present before */
429                 p = wmempcpy(p, L"\\??\\", 4);
430         }
431
432         /* Print name excludes the \??\ if present. */
433         new_print_name = p;
434         if (stripped_nchars != 0) {
435                 /* Get drive letter from real path to extract root, if a drive
436                  * letter was present before. */
437                 *p++ = extract_root_realpath[0];
438                 *p++ = extract_root_realpath[1];
439         }
440         /* Copy the rest of the extract root */
441         p = wmempcpy(p, extract_root_realpath + 2, extract_root_realpath_nchars - 2);
442
443         /* Append the stripped target */
444         p = wmempcpy(p, stripped_target, stripped_target_nchars);
445         new_target_nchars = p - new_target;
446         new_print_name_nchars = p - new_print_name;
447
448         if (new_target_nchars * sizeof(utf16lechar) >= REPARSE_POINT_MAX_SIZE ||
449             new_print_name_nchars * sizeof(utf16lechar) >= REPARSE_POINT_MAX_SIZE)
450                 return WIMLIB_ERR_REPARSE_POINT_FIXUP_FAILED;
451
452         rpdata.substitute_name = new_target;
453         rpdata.substitute_name_nbytes = new_target_nchars * sizeof(utf16lechar);
454         rpdata.print_name = new_print_name;
455         rpdata.print_name_nbytes = new_print_name_nchars * sizeof(utf16lechar);
456         return make_reparse_buffer(&rpdata, rpbuf, rpbuflen_p);
457 }
458 #endif /* __WIN32__ */
459
460 /* Set reparse data on extracted file or directory that has
461  * FILE_ATTRIBUTE_REPARSE_POINT set.  */
462 static int
463 extract_reparse_data(const tchar *path, struct apply_ctx *ctx,
464                      struct wim_inode *inode,
465                      struct wim_lookup_table_entry *lte_override)
466 {
467         int ret;
468         u8 rpbuf[REPARSE_POINT_MAX_SIZE];
469         u16 rpbuflen;
470
471         ret = wim_inode_get_reparse_data(inode, rpbuf, &rpbuflen, lte_override);
472         if (ret)
473                 goto error;
474
475 #ifdef __WIN32__
476         /* Fix up target of absolute symbolic link or junction points so
477          * that they point into the actual extraction target.  */
478         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
479             (inode->i_reparse_tag == WIM_IO_REPARSE_TAG_SYMLINK ||
480              inode->i_reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT) &&
481             !inode->i_not_rpfixed)
482         {
483                 ret = try_extract_rpfix(rpbuf, &rpbuflen, ctx->realtarget,
484                                         ctx->realtarget_nchars);
485                 if (ret && !(ctx->extract_flags &
486                              WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS))
487                 {
488                         WARNING("Reparse point fixup of \"%"TS"\" "
489                                 "failed", path);
490                         ret = 0;
491                 }
492                 if (ret)
493                         goto error;
494         }
495 #endif
496
497         ret = ctx->ops->set_reparse_data(path, rpbuf, rpbuflen, ctx);
498
499         /* On Windows, the SeCreateSymbolicLink privilege is required to create
500          * symbolic links.  To be more friendly towards non-Administrator users,
501          * we merely warn the user if symbolic links cannot be created due to
502          * insufficient permissions or privileges, unless
503          * WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS was provided.  */
504 #ifdef __WIN32__
505         if (ret && inode_is_symlink(inode) &&
506             (errno == EACCES || errno == EPERM) &&
507             !(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS))
508         {
509                 WARNING("Can't set reparse data on \"%"TS"\": "
510                         "Access denied!\n"
511                         "          You may be trying to "
512                         "extract a symbolic link without the\n"
513                         "          SeCreateSymbolicLink privilege, "
514                         "which by default non-Administrator\n"
515                         "          accounts do not have.",
516                         path);
517                 ret = 0;
518         }
519 #endif
520         if (ret)
521                 goto error;
522
523         /* Account for reparse data consumed.  */
524         update_extract_progress(ctx,
525                                 (lte_override ? lte_override :
526                                       inode_unnamed_lte_resolved(inode)));
527         return 0;
528
529 error:
530         ERROR_WITH_ERRNO("Failed to set reparse data on \"%"TS"\"", path);
531         return ret;
532 }
533
534 /*
535  * Extract zero or more streams to a file.
536  *
537  * This function operates slightly differently depending on whether @lte_spec is
538  * NULL or not.  When @lte_spec is NULL, the behavior is to extract the default
539  * file contents (unnamed stream), and, if named data streams are supported in
540  * the extract mode and volume, any named data streams.  When @lte_spec is not
541  * NULL, the behavior is to extract only all copies of the stream @lte_spec, and
542  * in addition use @lte_spec to set the reparse data or create the symbolic link
543  * if appropriate.
544  *
545  * @path
546  *      Path to file to extract (as can be passed to apply_operations
547  *      functions).
548  * @ctx
549  *      Apply context.
550  * @dentry
551  *      WIM dentry that corresponds to the file being extracted.
552  * @lte_spec
553  *      If non-NULL, specifies the lookup table entry for a stream to extract,
554  *      and only that stream will be extracted (although there may be more than
555  *      one instance of it).
556  * @lte_override
557  *      Used only if @lte_spec != NULL; it is passed to the extraction functions
558  *      rather than @lte_spec, allowing the location of the stream to be
559  *      overridden.  (This is used when the WIM is being read from a nonseekable
560  *      file, such as a pipe, when streams need to be used more than once; each
561  *      such stream is extracted to a temporary file.)
562  */
563 static int
564 extract_streams(const tchar *path, struct apply_ctx *ctx,
565                 struct wim_dentry *dentry,
566                 struct wim_lookup_table_entry *lte_spec,
567                 struct wim_lookup_table_entry *lte_override)
568 {
569         struct wim_inode *inode = dentry->d_inode;
570         struct wim_lookup_table_entry *lte;
571         file_spec_t file_spec;
572         int ret;
573
574         if (dentry->was_hardlinked)
575                 return 0;
576
577 #ifdef ENABLE_DEBUG
578         if (lte_spec) {
579                 char sha1_str[100];
580                 char *p = sha1_str;
581                 for (unsigned i = 0; i < SHA1_HASH_SIZE; i++)
582                         p += sprintf(p, "%02x", lte_override->hash[i]);
583                 DEBUG("Extracting stream SHA1=%s to \"%"TS"\"",
584                       sha1_str, path, inode->i_ino);
585         } else {
586                 DEBUG("Extracting streams to \"%"TS"\"", path, inode->i_ino);
587         }
588 #endif
589
590         if (ctx->ops->uses_cookies)
591                 file_spec.cookie = inode->extract_cookie;
592         else
593                 file_spec.path = path;
594
595         /* Unnamed data stream.  */
596         lte = inode_unnamed_lte_resolved(inode);
597         if (lte && (!lte_spec || lte == lte_spec)) {
598                 if (lte_spec)
599                         lte = lte_override;
600                 if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
601                                              FILE_ATTRIBUTE_REPARSE_POINT)))
602                 {
603                         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED &&
604                             ctx->supported_features.encrypted_files) {
605                                 if (!ctx->ops->extract_encrypted_stream_creates_file) {
606                                         ret = ctx->ops->extract_encrypted_stream(
607                                                                 path, lte, ctx);
608                                         if (ret)
609                                                 goto error;
610                                 }
611                         } else {
612                                 ret = ctx->ops->extract_unnamed_stream(
613                                                         file_spec, lte, ctx);
614                                 if (ret)
615                                         goto error;
616                         }
617                         update_extract_progress(ctx, lte);
618                 }
619                 else if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT)
620                 {
621                         ret = 0;
622                         if (ctx->supported_features.reparse_points)
623                                 ret = extract_reparse_data(path, ctx, inode, lte);
624                 #ifndef __WIN32__
625                         else if ((inode_is_symlink(inode) &&
626                                   ctx->supported_features.symlink_reparse_points))
627                                 ret = extract_symlink(path, ctx, inode, lte);
628                 #endif
629                         if (ret)
630                                 return ret;
631                 }
632         }
633
634         /* Named data streams.  */
635         if (can_extract_named_data_streams(ctx)) {
636                 for (u16 i = 0; i < inode->i_num_ads; i++) {
637                         struct wim_ads_entry *entry = &inode->i_ads_entries[i];
638
639                         if (!ads_entry_is_named_stream(entry))
640                                 continue;
641                         lte = entry->lte;
642                         if (!lte)
643                                 continue;
644                         if (lte_spec && lte_spec != lte)
645                                 continue;
646                         if (lte_spec)
647                                 lte = lte_override;
648                         ret = ctx->ops->extract_named_stream(file_spec, entry->stream_name,
649                                                              entry->stream_name_nbytes / 2,
650                                                              lte, ctx);
651                         if (ret)
652                                 goto error;
653                         update_extract_progress(ctx, lte);
654                 }
655         }
656         return 0;
657
658 error:
659         ERROR_WITH_ERRNO("Failed to extract data of \"%"TS"\"", path);
660         return ret;
661 }
662
663 /* Set attributes on an extracted file or directory if supported by the
664  * extraction mode.  */
665 static int
666 extract_file_attributes(const tchar *path, struct apply_ctx *ctx,
667                         struct wim_dentry *dentry, unsigned pass)
668 {
669         int ret;
670
671         if (ctx->ops->set_file_attributes &&
672             !(dentry == ctx->extract_root && ctx->root_dentry_is_special)) {
673                 u32 attributes = dentry->d_inode->i_attributes;
674
675                 /* Clear unsupported attributes.  */
676                 attributes &= ctx->supported_attributes_mask;
677
678                 if ((attributes & FILE_ATTRIBUTE_DIRECTORY &&
679                      !ctx->supported_features.encrypted_directories) ||
680                     (!(attributes & FILE_ATTRIBUTE_DIRECTORY) &&
681                      !ctx->supported_features.encrypted_files))
682                 {
683                         attributes &= ~FILE_ATTRIBUTE_ENCRYPTED;
684                 }
685
686                 if (attributes == 0)
687                         attributes = FILE_ATTRIBUTE_NORMAL;
688
689                 ret = ctx->ops->set_file_attributes(path, attributes, ctx, pass);
690                 if (ret) {
691                         ERROR_WITH_ERRNO("Failed to set attributes on "
692                                          "\"%"TS"\"", path);
693                         return ret;
694                 }
695         }
696         return 0;
697 }
698
699
700 /* Set or remove the short (DOS) name on an extracted file or directory if
701  * supported by the extraction mode.  Since DOS names are unimportant and it's
702  * easy to run into problems setting them on Windows (SetFileShortName()
703  * requires SE_RESTORE privilege, which only the Administrator can request, and
704  * also requires DELETE access to the file), failure is ignored unless
705  * WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES is set.  */
706 static int
707 extract_short_name(const tchar *path, struct apply_ctx *ctx,
708                    struct wim_dentry *dentry)
709 {
710         int ret;
711
712         /* The root of the dentry tree being extracted may not be extracted to
713          * its original name, so its short name should be ignored.  */
714         if (dentry == ctx->extract_root)
715                 return 0;
716
717         if (ctx->supported_features.short_names) {
718                 ret = ctx->ops->set_short_name(path,
719                                                dentry->short_name,
720                                                dentry->short_name_nbytes / 2,
721                                                ctx);
722                 if (ret && (ctx->extract_flags &
723                             WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES))
724                 {
725                         ERROR_WITH_ERRNO("Failed to set short name of "
726                                          "\"%"TS"\"", path);
727                         return ret;
728                 }
729         }
730         return 0;
731 }
732
733 /* Set security descriptor, UNIX data, or neither on an extracted file, taking
734  * into account the current extraction mode and flags.  */
735 static int
736 extract_security(const tchar *path, struct apply_ctx *ctx,
737                  struct wim_dentry *dentry)
738 {
739         int ret;
740         struct wim_inode *inode = dentry->d_inode;
741
742         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS)
743                 return 0;
744
745         if ((ctx->extract_root == dentry) && ctx->root_dentry_is_special)
746                 return 0;
747
748 #ifndef __WIN32__
749         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
750                 struct wimlib_unix_data data;
751
752                 ret = inode_get_unix_data(inode, &data, NULL);
753                 if (ret < 0)
754                         ret = 0;
755                 else if (ret == 0)
756                         ret = ctx->ops->set_unix_data(path, &data, ctx);
757                 if (ret) {
758                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) {
759                                 ERROR_WITH_ERRNO("Failed to set UNIX owner, "
760                                                  "group, and/or mode on "
761                                                  "\"%"TS"\"", path);
762                                 return ret;
763                         } else {
764                                 WARNING_WITH_ERRNO("Failed to set UNIX owner, "
765                                                    "group, and/or/mode on "
766                                                    "\"%"TS"\"", path);
767                         }
768                 }
769         }
770         else
771 #endif /* __WIN32__ */
772         if (ctx->supported_features.security_descriptors &&
773             inode->i_security_id != -1)
774         {
775                 const struct wim_security_data *sd;
776                 const u8 *desc;
777                 size_t desc_size;
778
779                 sd = wim_const_security_data(ctx->wim);
780                 desc = sd->descriptors[inode->i_security_id];
781                 desc_size = sd->sizes[inode->i_security_id];
782
783                 ret = ctx->ops->set_security_descriptor(path, desc,
784                                                         desc_size, ctx);
785                 if (ret) {
786                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) {
787                                 ERROR_WITH_ERRNO("Failed to set security "
788                                                  "descriptor on \"%"TS"\"", path);
789                                 return ret;
790                         } else {
791                         #if 0
792                                 if (errno != EACCES) {
793                                         WARNING_WITH_ERRNO("Failed to set "
794                                                            "security descriptor "
795                                                            "on \"%"TS"\"", path);
796                                 }
797                         #endif
798                                 ctx->no_security_descriptors++;
799                         }
800                 }
801         }
802         return 0;
803 }
804
805 /* Set timestamps on an extracted file.  Failure is warning-only unless
806  * WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS is set.  */
807 static int
808 extract_timestamps(const tchar *path, struct apply_ctx *ctx,
809                    struct wim_dentry *dentry)
810 {
811         struct wim_inode *inode = dentry->d_inode;
812         int ret;
813
814         if ((ctx->extract_root == dentry) && ctx->root_dentry_is_special)
815                 return 0;
816
817         if (ctx->ops->set_timestamps) {
818                 ret = ctx->ops->set_timestamps(path,
819                                                inode->i_creation_time,
820                                                inode->i_last_write_time,
821                                                inode->i_last_access_time,
822                                                ctx);
823                 if (ret) {
824                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) {
825                                 ERROR_WITH_ERRNO("Failed to set timestamps "
826                                                  "on \"%"TS"\"", path);
827                                 return ret;
828                         } else {
829                                 WARNING_WITH_ERRNO("Failed to set timestamps "
830                                                    "on \"%"TS"\"", path);
831                         }
832                 }
833         }
834         return 0;
835 }
836
837 /* Check whether the extraction of a dentry should be skipped completely.  */
838 static bool
839 dentry_is_supported(struct wim_dentry *dentry,
840                     const struct wim_features *supported_features)
841 {
842         struct wim_inode *inode = dentry->d_inode;
843
844         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
845                 return supported_features->reparse_points ||
846                         (inode_is_symlink(inode) &&
847                          supported_features->symlink_reparse_points);
848         }
849         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
850                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
851                         return supported_features->encrypted_directories != 0;
852                 else
853                         return supported_features->encrypted_files != 0;
854         }
855         return true;
856 }
857
858 /* Given a WIM dentry to extract, build the path to which to extract it, in the
859  * format understood by the callbacks in the apply_operations being used.
860  *
861  * Write the resulting path into @path, which must have room for at least
862  * ctx->ops->max_path characters including the null-terminator.
863  *
864  * Return %true if successful; %false if this WIM dentry doesn't actually need
865  * to be extracted or if the calculated path exceeds ctx->ops->max_path
866  * characters.
867  *
868  * This function clobbers the tmp_list member of @dentry and its ancestors up
869  * until the extraction root.  */
870 static bool
871 build_extraction_path(tchar path[], struct wim_dentry *dentry,
872                       struct apply_ctx *ctx)
873 {
874         size_t path_nchars;
875         LIST_HEAD(ancestor_list);
876         tchar *p = path;
877         const tchar *target_prefix;
878         size_t target_prefix_nchars;
879         struct wim_dentry *d;
880
881         if (dentry->extraction_skipped)
882                 return false;
883
884         path_nchars = ctx->ops->path_prefix_nchars;
885
886         if (ctx->ops->requires_realtarget_in_paths) {
887                 target_prefix        = ctx->realtarget;
888                 target_prefix_nchars = ctx->realtarget_nchars;
889         } else if (ctx->ops->requires_target_in_paths) {
890                 target_prefix        = ctx->target;
891                 target_prefix_nchars = ctx->target_nchars;
892         } else {
893                 target_prefix        = NULL;
894                 target_prefix_nchars = 0;
895         }
896         path_nchars += target_prefix_nchars;
897
898         for (d = dentry; d != ctx->extract_root; d = d->parent) {
899                 path_nchars += d->extraction_name_nchars + 1;
900                 list_add(&d->tmp_list, &ancestor_list);
901         }
902
903         path_nchars++; /* null terminator */
904
905         if (path_nchars > ctx->ops->path_max) {
906                 WARNING("\"%"TS"\": Path too long to extract",
907                         dentry_full_path(dentry));
908                 return false;
909         }
910
911         p = tmempcpy(p, ctx->ops->path_prefix, ctx->ops->path_prefix_nchars);
912         p = tmempcpy(p, target_prefix, target_prefix_nchars);
913         list_for_each_entry(d, &ancestor_list, tmp_list) {
914                 *p++ = ctx->ops->path_separator;
915                 p = tmempcpy(p, d->extraction_name, d->extraction_name_nchars);
916         }
917         *p++ = T('\0');
918         wimlib_assert(p - path == path_nchars);
919         return true;
920 }
921
922 static unsigned
923 get_num_path_components(const tchar *path, tchar path_separator)
924 {
925         unsigned num_components = 0;
926 #ifdef __WIN32__
927         /* Ignore drive letter.  */
928         if (path[0] != L'\0' && path[1] == L':')
929                 path += 2;
930 #endif
931
932         while (*path) {
933                 while (*path == path_separator)
934                         path++;
935                 if (*path)
936                         num_components++;
937                 while (*path && *path != path_separator)
938                         path++;
939         }
940         return num_components;
941 }
942
943 static int
944 extract_multiimage_symlink(const tchar *oldpath, const tchar *newpath,
945                            struct apply_ctx *ctx, struct wim_dentry *dentry)
946 {
947         size_t num_raw_path_components;
948         const struct wim_dentry *d;
949         size_t num_target_path_components;
950         tchar *p;
951         const tchar *p_old;
952         int ret;
953
954         num_raw_path_components = 0;
955         for (d = dentry; d != ctx->extract_root; d = d->parent)
956                 num_raw_path_components++;
957
958         if (ctx->ops->requires_realtarget_in_paths)
959                 num_target_path_components = get_num_path_components(ctx->realtarget,
960                                                                      ctx->ops->path_separator);
961         else if (ctx->ops->requires_target_in_paths)
962                 num_target_path_components = get_num_path_components(ctx->target,
963                                                                      ctx->ops->path_separator);
964         else
965                 num_target_path_components = 0;
966
967         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
968                 wimlib_assert(num_target_path_components > 0);
969                 num_raw_path_components++;
970                 num_target_path_components--;
971         }
972
973         p_old = oldpath + ctx->ops->path_prefix_nchars;
974 #ifdef __WIN32__
975         if (p_old[0] != L'\0' && p_old[1] == ':')
976                 p_old += 2;
977 #endif
978         while (*p_old == ctx->ops->path_separator)
979                 p_old++;
980         while (--num_target_path_components) {
981                 while (*p_old != ctx->ops->path_separator)
982                         p_old++;
983                 while (*p_old == ctx->ops->path_separator)
984                         p_old++;
985         }
986
987         tchar symlink_target[tstrlen(p_old) + 3 * num_raw_path_components + 1];
988
989         p = &symlink_target[0];
990         while (num_raw_path_components--) {
991                 *p++ = '.';
992                 *p++ = '.';
993                 *p++ = ctx->ops->path_separator;
994         }
995         tstrcpy(p, p_old);
996         DEBUG("Creating symlink \"%"TS"\" => \"%"TS"\"",
997               newpath, symlink_target);
998         ret = ctx->ops->create_symlink(symlink_target, newpath, ctx);
999         if (ret) {
1000                 ERROR_WITH_ERRNO("Failed to create symlink "
1001                                  "\"%"TS"\" => \"%"TS"\"",
1002                                  newpath, symlink_target);
1003         }
1004         return ret;
1005 }
1006
1007 /* Create the "skeleton" of an extracted file or directory.  Don't yet extract
1008  * data streams, reparse data (including symbolic links), timestamps, and
1009  * security descriptors.  Basically, everything that doesn't require reading
1010  * non-metadata resources from the WIM file and isn't delayed until the final
1011  * pass.  */
1012 static int
1013 do_dentry_extract_skeleton(tchar path[], struct wim_dentry *dentry,
1014                            struct apply_ctx *ctx)
1015 {
1016         struct wim_inode *inode = dentry->d_inode;
1017         int ret;
1018         const tchar *oldpath;
1019
1020         if (unlikely(is_linked_extraction(ctx))) {
1021                 struct wim_lookup_table_entry *unnamed_lte;
1022
1023                 unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode);
1024                 if (unnamed_lte && unnamed_lte->extracted_file) {
1025                         oldpath = unnamed_lte->extracted_file;
1026                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK)
1027                                 goto hardlink;
1028                         else
1029                                 goto symlink;
1030                 }
1031         }
1032
1033         /* Create hard link if this dentry corresponds to an already-extracted
1034          * inode.  */
1035         if (inode->i_extracted_file) {
1036                 oldpath = inode->i_extracted_file;
1037                 goto hardlink;
1038         }
1039
1040         /* Skip symlinks unless they can be extracted as reparse points rather
1041          * than created directly.  */
1042         if (inode_is_symlink(inode) && !ctx->supported_features.reparse_points)
1043                 return 0;
1044
1045         /* Create this file or directory unless it's the extraction root, which
1046          * was already created if necessary.  */
1047         if (dentry != ctx->extract_root) {
1048                 ret = extract_inode(path, ctx, inode);
1049                 if (ret)
1050                         return ret;
1051         }
1052
1053         /* Create empty named data streams.  */
1054         if (can_extract_named_data_streams(ctx)) {
1055                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1056                         file_spec_t file_spec;
1057                         struct wim_ads_entry *entry = &inode->i_ads_entries[i];
1058
1059                         if (!ads_entry_is_named_stream(entry))
1060                                 continue;
1061                         if (entry->lte)
1062                                 continue;
1063                         if (ctx->ops->uses_cookies)
1064                                 file_spec.cookie = inode->extract_cookie;
1065                         else
1066                                 file_spec.path = path;
1067                         ret = ctx->ops->extract_named_stream(file_spec,
1068                                                              entry->stream_name,
1069                                                              entry->stream_name_nbytes / 2,
1070                                                              entry->lte, ctx);
1071                         if (ret) {
1072                                 ERROR_WITH_ERRNO("\"%"TS"\": failed to create "
1073                                                  "empty named data stream",
1074                                                  path);
1075                                 return ret;
1076                         }
1077                 }
1078         }
1079
1080         /* Set file attributes (if supported).  */
1081         ret = extract_file_attributes(path, ctx, dentry, 0);
1082         if (ret)
1083                 return ret;
1084
1085         /* Set or remove file short name (if supported).  */
1086         ret = extract_short_name(path, ctx, dentry);
1087         if (ret)
1088                 return ret;
1089
1090         /* If inode has multiple links and hard links are supported in this
1091          * extraction mode and volume, save the path to the extracted file in
1092          * case it's needed to create a hard link.  */
1093         if (unlikely(is_linked_extraction(ctx))) {
1094                 struct wim_lookup_table_entry *unnamed_lte;
1095
1096                 unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode);
1097                 if (unnamed_lte) {
1098                         unnamed_lte->extracted_file = TSTRDUP(path);
1099                         if (!unnamed_lte->extracted_file)
1100                                 return WIMLIB_ERR_NOMEM;
1101                 }
1102         } else if (inode->i_nlink > 1 && ctx->supported_features.hard_links) {
1103                 inode->i_extracted_file = TSTRDUP(path);
1104                 if (!inode->i_extracted_file)
1105                         return WIMLIB_ERR_NOMEM;
1106         }
1107         return 0;
1108
1109 symlink:
1110         ret = extract_multiimage_symlink(oldpath, path, ctx, dentry);
1111         if (ret)
1112                 return ret;
1113         dentry->was_hardlinked = 1;
1114         return 0;
1115
1116 hardlink:
1117         ret = extract_hardlink(oldpath, path, ctx);
1118         if (ret)
1119                 return ret;
1120         dentry->was_hardlinked = 1;
1121         return 0;
1122 }
1123
1124 /* This is a wrapper around do_dentry_extract_skeleton() that handles building
1125  * the path, doing short name reordering.  This is also idempotent; dentries
1126  * already processed have skeleton_extracted set and no action is taken.  See
1127  * apply_operations.requires_short_name_reordering for more details about short
1128  * name reordering.  */
1129 static int
1130 dentry_extract_skeleton(struct wim_dentry *dentry, void *_ctx)
1131 {
1132         struct apply_ctx *ctx = _ctx;
1133         tchar path[ctx->ops->path_max];
1134         struct wim_dentry *orig_dentry;
1135         struct wim_dentry *other_dentry;
1136         int ret;
1137
1138         if (dentry->skeleton_extracted)
1139                 return 0;
1140
1141         orig_dentry = NULL;
1142         if (ctx->supported_features.short_names
1143             && ctx->ops->requires_short_name_reordering
1144             && !dentry_has_short_name(dentry)
1145             && !dentry->d_inode->i_dos_name_extracted)
1146         {
1147                 inode_for_each_dentry(other_dentry, dentry->d_inode) {
1148                         if (dentry_has_short_name(other_dentry)
1149                             && !other_dentry->skeleton_extracted
1150                             && other_dentry->in_extraction_tree)
1151                         {
1152                                 DEBUG("Creating %"TS" before %"TS" "
1153                                       "to guarantee correct DOS name extraction",
1154                                       dentry_full_path(other_dentry),
1155                                       dentry_full_path(dentry));
1156                                 orig_dentry = dentry;
1157                                 dentry = other_dentry;
1158                                 break;
1159                         }
1160                 }
1161         }
1162 again:
1163         if (!build_extraction_path(path, dentry, ctx))
1164                 return 0;
1165         ret = do_dentry_extract_skeleton(path, dentry, ctx);
1166         if (ret)
1167                 return ret;
1168
1169         dentry->skeleton_extracted = 1;
1170
1171         if (orig_dentry) {
1172                 dentry = orig_dentry;
1173                 orig_dentry = NULL;
1174                 goto again;
1175         }
1176         dentry->d_inode->i_dos_name_extracted = 1;
1177         return 0;
1178 }
1179
1180 static int
1181 dentry_extract_dir_skeleton(struct wim_dentry *dentry, void *_ctx)
1182 {
1183         if (dentry->d_inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1184                 return dentry_extract_skeleton(dentry, _ctx);
1185         return 0;
1186 }
1187
1188 /* Create a file or directory, then immediately extract all streams.  The WIM
1189  * may not be read sequentially by this function.  */
1190 static int
1191 dentry_extract(struct wim_dentry *dentry, void *_ctx)
1192 {
1193         struct apply_ctx *ctx = _ctx;
1194         tchar path[ctx->ops->path_max];
1195         int ret;
1196
1197         ret = dentry_extract_skeleton(dentry, ctx);
1198         if (ret)
1199                 return ret;
1200
1201         if (!build_extraction_path(path, dentry, ctx))
1202                 return 0;
1203
1204         return extract_streams(path, ctx, dentry, NULL, NULL);
1205 }
1206
1207 /* Creates a temporary file opened for writing.  The open file descriptor is
1208  * returned in @fd_ret and its name is returned in @name_ret (dynamically
1209  * allocated).  */
1210 static int
1211 create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
1212 {
1213         tchar *name;
1214         int raw_fd;
1215
1216 retry:
1217         name = ttempnam(NULL, T("wimlib"));
1218         if (name == NULL) {
1219                 ERROR_WITH_ERRNO("Failed to create temporary filename");
1220                 return WIMLIB_ERR_NOMEM;
1221         }
1222
1223         raw_fd = topen(name, O_WRONLY | O_CREAT | O_EXCL | O_BINARY, 0600);
1224
1225         if (raw_fd < 0) {
1226                 if (errno == EEXIST) {
1227                         FREE(name);
1228                         goto retry;
1229                 }
1230                 ERROR_WITH_ERRNO("Failed to open temporary file \"%"TS"\"", name);
1231                 FREE(name);
1232                 return WIMLIB_ERR_OPEN;
1233         }
1234
1235         filedes_init(fd_ret, raw_fd);
1236         *name_ret = name;
1237         return 0;
1238 }
1239
1240 /* Extract all instances of the stream @lte that are being extracted in this
1241  * call of extract_tree(), but actually read the stream data from @lte_override.
1242  */
1243 static int
1244 extract_stream_instances(struct wim_lookup_table_entry *lte,
1245                          struct wim_lookup_table_entry *lte_override,
1246                          struct apply_ctx *ctx)
1247 {
1248         struct wim_dentry **lte_dentries;
1249         tchar path[ctx->ops->path_max];
1250         size_t i;
1251         int ret;
1252
1253         if (lte->out_refcnt <= ARRAY_LEN(lte->inline_lte_dentries))
1254                 lte_dentries = lte->inline_lte_dentries;
1255         else
1256                 lte_dentries = lte->lte_dentries;
1257
1258         for (i = 0; i < lte->out_refcnt; i++) {
1259                 struct wim_dentry *dentry = lte_dentries[i];
1260
1261                 if (dentry->tmp_flag)
1262                         continue;
1263                 if (!build_extraction_path(path, dentry, ctx))
1264                         continue;
1265                 ret = extract_streams(path, ctx, dentry, lte, lte_override);
1266                 if (ret)
1267                         goto out_clear_tmp_flags;
1268                 dentry->tmp_flag = 1;
1269         }
1270         ret = 0;
1271 out_clear_tmp_flags:
1272         for (i = 0; i < lte->out_refcnt; i++)
1273                 lte_dentries[i]->tmp_flag = 0;
1274         return ret;
1275 }
1276
1277 /* Determine whether the specified stream needs to be extracted to a temporary
1278  * file or not.
1279  *
1280  * @lte->out_refcnt specifies the number of instances of this stream that must
1281  * be extracted.
1282  *
1283  * @is_partial_res is %true if this stream is just one of multiple in a single
1284  * WIM resource being extracted.  */
1285 static bool
1286 need_tmpfile_to_extract(struct wim_lookup_table_entry *lte,
1287                         bool is_partial_res)
1288 {
1289         /* Temporary file is always required when reading a partial resource,
1290          * since in that case we retrieve all the contained streams in one pass.
1291          * */
1292         if (is_partial_res)
1293                 return true;
1294
1295         /* Otherwise we don't need a temporary file if only a single instance of
1296          * the stream is needed.  */
1297         if (lte->out_refcnt == 1)
1298                 return false;
1299
1300         wimlib_assert(lte->out_refcnt >= 2);
1301
1302         /* We also don't need a temporary file if random access to the stream is
1303          * allowed.  */
1304         if (lte->resource_location != RESOURCE_IN_WIM ||
1305             filedes_is_seekable(&lte->rspec->wim->in_fd))
1306                 return false;
1307
1308         return true;
1309 }
1310
1311 static int
1312 begin_extract_stream_to_tmpfile(struct wim_lookup_table_entry *lte,
1313                                 bool is_partial_res,
1314                                 void *_ctx)
1315 {
1316         struct apply_ctx *ctx = _ctx;
1317         int ret;
1318
1319         if (!need_tmpfile_to_extract(lte, is_partial_res)) {
1320                 DEBUG("Temporary file not needed "
1321                       "for stream (size=%"PRIu64")", lte->size);
1322                 ret = extract_stream_instances(lte, lte, ctx);
1323                 if (ret)
1324                         return ret;
1325
1326                 return BEGIN_STREAM_STATUS_SKIP_STREAM;
1327         }
1328
1329         DEBUG("Temporary file needed for stream (size=%"PRIu64")", lte->size);
1330         return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
1331 }
1332
1333 static int
1334 end_extract_stream_to_tmpfile(struct wim_lookup_table_entry *lte,
1335                               int status, void *_ctx)
1336 {
1337         struct apply_ctx *ctx = _ctx;
1338         struct wim_lookup_table_entry lte_override;
1339         int ret;
1340         int errno_save = errno;
1341
1342         ret = filedes_close(&ctx->tmpfile_fd);
1343
1344         if (status) {
1345                 ret = status;
1346                 errno = errno_save;
1347                 goto out_delete_tmpfile;
1348         }
1349
1350         if (ret) {
1351                 ERROR_WITH_ERRNO("Error writing temporary file %"TS, ctx->tmpfile_name);
1352                 ret = WIMLIB_ERR_WRITE;
1353                 goto out_delete_tmpfile;
1354         }
1355
1356         /* Now that a full stream has been extracted to a temporary file,
1357          * extract all instances of it to the actual target.  */
1358
1359         memcpy(&lte_override, lte, sizeof(struct wim_lookup_table_entry));
1360         lte_override.resource_location = RESOURCE_IN_FILE_ON_DISK;
1361         lte_override.file_on_disk = ctx->tmpfile_name;
1362
1363         ret = extract_stream_instances(lte, &lte_override, ctx);
1364
1365 out_delete_tmpfile:
1366         errno_save = errno;
1367         tunlink(ctx->tmpfile_name);
1368         FREE(ctx->tmpfile_name);
1369         errno = errno_save;
1370         return ret;
1371 }
1372
1373 /* Extracts a list of streams (ctx.stream_list), assuming that the directory
1374  * structure and empty files were already created.  This relies on the
1375  * per-`struct wim_lookup_table_entry' list of dentries that reference each
1376  * stream that was constructed earlier.  */
1377 static int
1378 extract_stream_list(struct apply_ctx *ctx)
1379 {
1380         if (!(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
1381                 /* Sequential extraction: read the streams in the order in which
1382                  * they appear in the WIM file.  */
1383                 struct read_stream_list_callbacks cbs = {
1384                         .begin_stream           = begin_extract_stream_to_tmpfile,
1385                         .begin_stream_ctx       = ctx,
1386                         .consume_chunk          = extract_chunk_to_fd,
1387                         .consume_chunk_ctx      = &ctx->tmpfile_fd,
1388                         .end_stream             = end_extract_stream_to_tmpfile,
1389                         .end_stream_ctx         = ctx,
1390                 };
1391                 return read_stream_list(&ctx->stream_list,
1392                                         offsetof(struct wim_lookup_table_entry, extraction_list),
1393                                         &cbs, VERIFY_STREAM_HASHES);
1394         } else {
1395                 /* Extract the streams in unsorted order.  */
1396                 struct wim_lookup_table_entry *lte;
1397                 int ret;
1398
1399                 list_for_each_entry(lte, &ctx->stream_list, extraction_list) {
1400                         ret = extract_stream_instances(lte, lte, ctx);
1401                         if (ret)
1402                                 return ret;
1403                 }
1404                 return 0;
1405         }
1406 }
1407
1408 #define PWM_ALLOW_WIM_HDR 0x00001
1409 #define PWM_SILENT_EOF    0x00002
1410
1411 /* Read the header from a stream in a pipable WIM.  */
1412 static int
1413 read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte,
1414                        struct wim_resource_spec *rspec,
1415                        int flags, struct wim_header_disk *hdr_ret)
1416 {
1417         union {
1418                 struct pwm_stream_hdr stream_hdr;
1419                 struct wim_header_disk pwm_hdr;
1420         } buf;
1421         struct wim_reshdr reshdr;
1422         int ret;
1423
1424         ret = full_read(&pwm->in_fd, &buf.stream_hdr, sizeof(buf.stream_hdr));
1425         if (ret)
1426                 goto read_error;
1427
1428         if ((flags & PWM_ALLOW_WIM_HDR) && buf.stream_hdr.magic == PWM_MAGIC) {
1429                 BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr));
1430                 ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1,
1431                                 sizeof(buf.pwm_hdr) - sizeof(buf.stream_hdr));
1432
1433                 if (ret)
1434                         goto read_error;
1435                 lte->resource_location = RESOURCE_NONEXISTENT;
1436                 memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
1437                 return 0;
1438         }
1439
1440         if (le64_to_cpu(buf.stream_hdr.magic) != PWM_STREAM_MAGIC) {
1441                 ERROR("Data read on pipe is invalid (expected stream header).");
1442                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
1443         }
1444
1445         copy_hash(lte->hash, buf.stream_hdr.hash);
1446
1447         reshdr.size_in_wim = 0;
1448         reshdr.flags = le32_to_cpu(buf.stream_hdr.flags);
1449         reshdr.offset_in_wim = pwm->in_fd.offset;
1450         reshdr.uncompressed_size = le64_to_cpu(buf.stream_hdr.uncompressed_size);
1451         wim_res_hdr_to_spec(&reshdr, pwm, rspec);
1452         lte_bind_wim_resource_spec(lte, rspec);
1453         lte->flags = rspec->flags;
1454         lte->size = rspec->uncompressed_size;
1455         lte->offset_in_res = 0;
1456         return 0;
1457
1458 read_error:
1459         if (ret != WIMLIB_ERR_UNEXPECTED_END_OF_FILE || !(flags & PWM_SILENT_EOF))
1460                 ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
1461         return ret;
1462 }
1463
1464 static int
1465 extract_streams_from_pipe(struct apply_ctx *ctx)
1466 {
1467         struct wim_lookup_table_entry *found_lte;
1468         struct wim_resource_spec *rspec;
1469         struct wim_lookup_table_entry *needed_lte;
1470         struct wim_lookup_table *lookup_table;
1471         struct wim_header_disk pwm_hdr;
1472         int ret;
1473         int pwm_flags;
1474
1475         ret = WIMLIB_ERR_NOMEM;
1476         found_lte = new_lookup_table_entry();
1477         if (found_lte == NULL)
1478                 goto out;
1479
1480         rspec = MALLOC(sizeof(struct wim_resource_spec));
1481         if (rspec == NULL)
1482                 goto out_free_found_lte;
1483
1484         lookup_table = ctx->wim->lookup_table;
1485         pwm_flags = PWM_ALLOW_WIM_HDR;
1486         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1487                 pwm_flags |= PWM_SILENT_EOF;
1488         memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GID_LEN);
1489         ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
1490         ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
1491         if (ctx->progress_func)
1492                 ctx->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1493                                    &ctx->progress);
1494         while (ctx->num_streams_remaining) {
1495                 if (found_lte->resource_location != RESOURCE_NONEXISTENT)
1496                         lte_unbind_wim_resource_spec(found_lte);
1497                 ret = read_pwm_stream_header(ctx->wim, found_lte, rspec,
1498                                              pwm_flags, &pwm_hdr);
1499                 if (ret) {
1500                         if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE &&
1501                             (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1502                         {
1503                                 goto resume_done;
1504                         }
1505                         goto out_free_found_lte;
1506                 }
1507
1508                 if ((found_lte->resource_location != RESOURCE_NONEXISTENT)
1509                     && !(found_lte->flags & WIM_RESHDR_FLAG_METADATA)
1510                     && (needed_lte = lookup_resource(lookup_table, found_lte->hash))
1511                     && (needed_lte->out_refcnt))
1512                 {
1513                         tchar *tmpfile_name = NULL;
1514                         struct wim_lookup_table_entry *lte_override;
1515                         struct wim_lookup_table_entry tmpfile_lte;
1516
1517                         needed_lte->offset_in_res = found_lte->offset_in_res;
1518                         needed_lte->flags = found_lte->flags;
1519                         needed_lte->size = found_lte->size;
1520
1521                         lte_unbind_wim_resource_spec(found_lte);
1522                         lte_bind_wim_resource_spec(needed_lte, rspec);
1523
1524                         if (needed_lte->out_refcnt > 1) {
1525
1526                                 struct filedes tmpfile_fd;
1527
1528                                 /* Extract stream to temporary file.  */
1529                                 ret = create_temporary_file(&tmpfile_fd, &tmpfile_name);
1530                                 if (ret)
1531                                         goto out_free_found_lte;
1532
1533                                 ret = extract_stream_to_fd(needed_lte, &tmpfile_fd,
1534                                                            needed_lte->size);
1535                                 if (ret) {
1536                                         filedes_close(&tmpfile_fd);
1537                                         goto delete_tmpfile;
1538                                 }
1539
1540                                 if (filedes_close(&tmpfile_fd)) {
1541                                         ERROR_WITH_ERRNO("Error writing to temporary "
1542                                                          "file \"%"TS"\"", tmpfile_name);
1543                                         ret = WIMLIB_ERR_WRITE;
1544                                         goto delete_tmpfile;
1545                                 }
1546                                 memcpy(&tmpfile_lte, needed_lte,
1547                                        sizeof(struct wim_lookup_table_entry));
1548                                 tmpfile_lte.resource_location = RESOURCE_IN_FILE_ON_DISK;
1549                                 tmpfile_lte.file_on_disk = tmpfile_name;
1550                                 lte_override = &tmpfile_lte;
1551                         } else {
1552                                 lte_override = needed_lte;
1553                         }
1554
1555                         ret = extract_stream_instances(needed_lte, lte_override, ctx);
1556                 delete_tmpfile:
1557                         lte_unbind_wim_resource_spec(needed_lte);
1558                         if (tmpfile_name) {
1559                                 tunlink(tmpfile_name);
1560                                 FREE(tmpfile_name);
1561                         }
1562                         if (ret)
1563                                 goto out_free_found_lte;
1564                         ctx->num_streams_remaining--;
1565                 } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) {
1566                         ret = skip_wim_stream(found_lte);
1567                         if (ret)
1568                                 goto out_free_found_lte;
1569                 } else {
1570                         u16 part_number = le16_to_cpu(pwm_hdr.part_number);
1571                         u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
1572
1573                         if (part_number != ctx->progress.extract.part_number ||
1574                             total_parts != ctx->progress.extract.total_parts ||
1575                             memcmp(pwm_hdr.guid, ctx->progress.extract.guid,
1576                                    WIM_GID_LEN))
1577                         {
1578                                 ctx->progress.extract.part_number = part_number;
1579                                 ctx->progress.extract.total_parts = total_parts;
1580                                 memcpy(ctx->progress.extract.guid,
1581                                        pwm_hdr.guid, WIM_GID_LEN);
1582                                 if (ctx->progress_func) {
1583                                         ctx->progress_func(
1584                                                 WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1585                                                            &ctx->progress);
1586                                 }
1587
1588                         }
1589                 }
1590         }
1591         ret = 0;
1592 out_free_found_lte:
1593         if (found_lte->resource_location != RESOURCE_IN_WIM)
1594                 FREE(rspec);
1595         free_lookup_table_entry(found_lte);
1596 out:
1597         return ret;
1598
1599 resume_done:
1600         /* TODO */
1601         return 0;
1602 }
1603
1604 /* Finish extracting a file, directory, or symbolic link by setting file
1605  * security and timestamps.  */
1606 static int
1607 dentry_extract_final(struct wim_dentry *dentry, void *_ctx)
1608 {
1609         struct apply_ctx *ctx = _ctx;
1610         int ret;
1611         tchar path[ctx->ops->path_max];
1612
1613         if (!build_extraction_path(path, dentry, ctx))
1614                 return 0;
1615
1616         ret = extract_security(path, ctx, dentry);
1617         if (ret)
1618                 return ret;
1619
1620         if (ctx->ops->requires_final_set_attributes_pass) {
1621                 /* Set file attributes (if supported).  */
1622                 ret = extract_file_attributes(path, ctx, dentry, 1);
1623                 if (ret)
1624                         return ret;
1625         }
1626
1627         return extract_timestamps(path, ctx, dentry);
1628 }
1629
1630 /*
1631  * Extract a WIM dentry to standard output.
1632  *
1633  * This obviously doesn't make sense in all cases.  We return an error if the
1634  * dentry does not correspond to a regular file.  Otherwise we extract the
1635  * unnamed data stream only.
1636  */
1637 static int
1638 extract_dentry_to_stdout(struct wim_dentry *dentry)
1639 {
1640         int ret = 0;
1641         if (dentry->d_inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
1642                                              FILE_ATTRIBUTE_DIRECTORY))
1643         {
1644                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
1645                       "extracted to standard output", dentry_full_path(dentry));
1646                 ret = WIMLIB_ERR_NOT_A_REGULAR_FILE;
1647         } else {
1648                 struct wim_lookup_table_entry *lte;
1649
1650                 lte = inode_unnamed_lte_resolved(dentry->d_inode);
1651                 if (lte) {
1652                         struct filedes _stdout;
1653                         filedes_init(&_stdout, STDOUT_FILENO);
1654                         ret = extract_stream_to_fd(lte, &_stdout, lte->size);
1655                 }
1656         }
1657         return ret;
1658 }
1659
1660 #ifdef __WIN32__
1661 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
1662 #else
1663 static const utf16lechar replacement_char = cpu_to_le16('?');
1664 #endif
1665
1666 static bool
1667 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
1668 {
1669         size_t i;
1670
1671         if (num_chars == 0)
1672                 return true;
1673         for (i = 0; i < num_chars; i++) {
1674                 switch (name[i]) {
1675         #ifdef __WIN32__
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                 case cpu_to_le16('|'):
1684         #endif
1685                 case cpu_to_le16('/'):
1686                 case cpu_to_le16('\0'):
1687                         if (fix)
1688                                 name[i] = replacement_char;
1689                         else
1690                                 return false;
1691                 }
1692         }
1693
1694 #ifdef __WIN32__
1695         if (name[num_chars - 1] == cpu_to_le16(' ') ||
1696             name[num_chars - 1] == cpu_to_le16('.'))
1697         {
1698                 if (fix)
1699                         name[num_chars - 1] = replacement_char;
1700                 else
1701                         return false;
1702         }
1703 #endif
1704         return true;
1705 }
1706
1707 static bool
1708 dentry_is_dot_or_dotdot(const struct wim_dentry *dentry)
1709 {
1710         const utf16lechar *file_name = dentry->file_name;
1711         return file_name != NULL &&
1712                 file_name[0] == cpu_to_le16('.') &&
1713                 (file_name[1] == cpu_to_le16('\0') ||
1714                  (file_name[1] == cpu_to_le16('.') &&
1715                   file_name[2] == cpu_to_le16('\0')));
1716 }
1717
1718 static int
1719 dentry_mark_skipped(struct wim_dentry *dentry, void *_ignore)
1720 {
1721         dentry->extraction_skipped = 1;
1722         return 0;
1723 }
1724
1725 /*
1726  * dentry_calculate_extraction_path-
1727  *
1728  * Calculate the actual filename component at which a WIM dentry will be
1729  * extracted, handling invalid filenames "properly".
1730  *
1731  * dentry->extraction_name usually will be set the same as dentry->file_name (on
1732  * UNIX, converted into the platform's multibyte encoding).  However, if the
1733  * file name contains characters that are not valid on the current platform or
1734  * has some other format that is not valid, leave dentry->extraction_name as
1735  * NULL and set dentry->extraction_skipped to indicate that this dentry should
1736  * not be extracted, unless the appropriate flag
1737  * WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES is set in the extract flags, in
1738  * which case a substitute filename will be created and set instead.
1739  *
1740  * Conflicts with case-insensitive names on Windows are handled similarly; see
1741  * below.
1742  */
1743 static int
1744 dentry_calculate_extraction_path(struct wim_dentry *dentry, void *_args)
1745 {
1746         struct apply_ctx *ctx = _args;
1747         int ret;
1748
1749         if (dentry == ctx->extract_root || dentry->extraction_skipped)
1750                 return 0;
1751
1752         if (!dentry_is_supported(dentry, &ctx->supported_features))
1753                 goto skip_dentry;
1754
1755         if (dentry_is_dot_or_dotdot(dentry)) {
1756                 /* WIM files shouldn't contain . or .. entries.  But if they are
1757                  * there, don't attempt to extract them. */
1758                 WARNING("Skipping extraction of unexpected . or .. file "
1759                         "\"%"TS"\"", dentry_full_path(dentry));
1760                 goto skip_dentry;
1761         }
1762
1763         if (!ctx->ops->supports_case_sensitive_filenames)
1764         {
1765                 struct wim_dentry *other;
1766                 list_for_each_entry(other, &dentry->case_insensitive_conflict_list,
1767                                     case_insensitive_conflict_list)
1768                 {
1769                         if (!other->extraction_skipped) {
1770                                 if (ctx->extract_flags &
1771                                     WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
1772                                         WARNING("\"%"TS"\" has the same "
1773                                                 "case-insensitive name as "
1774                                                 "\"%"TS"\"; extracting "
1775                                                 "dummy name instead",
1776                                                 dentry_full_path(dentry),
1777                                                 dentry_full_path(other));
1778                                         goto out_replace;
1779                                 } else {
1780                                         WARNING("Not extracting \"%"TS"\": "
1781                                                 "has same case-insensitive "
1782                                                 "name as \"%"TS"\"",
1783                                                 dentry_full_path(dentry),
1784                                                 dentry_full_path(other));
1785                                         goto skip_dentry;
1786                                 }
1787                         }
1788                 }
1789         }
1790
1791         if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
1792 #if TCHAR_IS_UTF16LE
1793                 dentry->extraction_name = dentry->file_name;
1794                 dentry->extraction_name_nchars = dentry->file_name_nbytes / 2;
1795                 return 0;
1796 #else
1797                 return utf16le_to_tstr(dentry->file_name,
1798                                        dentry->file_name_nbytes,
1799                                        &dentry->extraction_name,
1800                                        &dentry->extraction_name_nchars);
1801 #endif
1802         } else {
1803                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
1804                 {
1805                         WARNING("\"%"TS"\" has an invalid filename "
1806                                 "that is not supported on this platform; "
1807                                 "extracting dummy name instead",
1808                                 dentry_full_path(dentry));
1809                         goto out_replace;
1810                 } else {
1811                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
1812                                 "that is not supported on this platform",
1813                                 dentry_full_path(dentry));
1814                         goto skip_dentry;
1815                 }
1816         }
1817
1818 out_replace:
1819         {
1820                 utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2];
1821
1822                 memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
1823                 file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
1824
1825                 tchar *tchar_name;
1826                 size_t tchar_nchars;
1827         #if TCHAR_IS_UTF16LE
1828                 tchar_name = utf16_name_copy;
1829                 tchar_nchars = dentry->file_name_nbytes / 2;
1830         #else
1831                 ret = utf16le_to_tstr(utf16_name_copy,
1832                                       dentry->file_name_nbytes,
1833                                       &tchar_name, &tchar_nchars);
1834                 if (ret)
1835                         return ret;
1836         #endif
1837                 size_t fixed_name_num_chars = tchar_nchars;
1838                 tchar fixed_name[tchar_nchars + 50];
1839
1840                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
1841                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
1842                                                  T(" (invalid filename #%lu)"),
1843                                                  ++ctx->invalid_sequence);
1844         #if !TCHAR_IS_UTF16LE
1845                 FREE(tchar_name);
1846         #endif
1847                 dentry->extraction_name = memdup(fixed_name,
1848                                                  2 * fixed_name_num_chars + 2);
1849                 if (!dentry->extraction_name)
1850                         return WIMLIB_ERR_NOMEM;
1851                 dentry->extraction_name_nchars = fixed_name_num_chars;
1852         }
1853         return 0;
1854
1855 skip_dentry:
1856         for_dentry_in_tree(dentry, dentry_mark_skipped, NULL);
1857         return 0;
1858 }
1859
1860 /* Clean up dentry and inode structure after extraction.  */
1861 static int
1862 dentry_reset_needs_extraction(struct wim_dentry *dentry, void *_ignore)
1863 {
1864         struct wim_inode *inode = dentry->d_inode;
1865
1866         dentry->in_extraction_tree = 0;
1867         dentry->extraction_skipped = 0;
1868         dentry->was_hardlinked = 0;
1869         dentry->skeleton_extracted = 0;
1870         inode->i_visited = 0;
1871         FREE(inode->i_extracted_file);
1872         inode->i_extracted_file = NULL;
1873         inode->i_dos_name_extracted = 0;
1874         if ((void*)dentry->extraction_name != (void*)dentry->file_name)
1875                 FREE(dentry->extraction_name);
1876         dentry->extraction_name = NULL;
1877         return 0;
1878 }
1879
1880 /* Tally features necessary to extract a dentry and the corresponding inode.  */
1881 static int
1882 dentry_tally_features(struct wim_dentry *dentry, void *_features)
1883 {
1884         struct wim_features *features = _features;
1885         struct wim_inode *inode = dentry->d_inode;
1886
1887         if (dentry->extraction_skipped)
1888                 return 0;
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                  bool warn)
1974 {
1975         const tchar *loc;
1976         const tchar *mode = T("this extraction mode");
1977
1978         if (wim_source_path[0] == '\0')
1979                 loc = T("the WIM image");
1980         else
1981                 loc = wim_source_path;
1982
1983         if (warn) {
1984                 /* We're an archive program, so theoretically we can do what we want
1985                  * with FILE_ATTRIBUTE_ARCHIVE (which is a dumb flag anyway).  Don't
1986                  * bother the user about it.  */
1987 #if 0
1988                 if (required_features->archive_files && !supported_features->archive_files)
1989                 {
1990                         WARNING(
1991                   "%lu files in %"TS" are marked as archived, but this attribute\n"
1992         "          is not supported in %"TS".",
1993                                 required_features->archive_files, loc, mode);
1994                 }
1995 #endif
1996
1997                 if (required_features->hidden_files && !supported_features->hidden_files)
1998                 {
1999                         WARNING(
2000                   "%lu files in %"TS" are marked as hidden, but this\n"
2001         "          attribute is not supported in %"TS".",
2002                                 required_features->hidden_files, loc, mode);
2003                 }
2004
2005                 if (required_features->system_files && !supported_features->system_files)
2006                 {
2007                         WARNING(
2008                   "%lu files in %"TS" are marked as system files,\n"
2009         "          but this attribute is not supported in %"TS".",
2010                                 required_features->system_files, loc, mode);
2011                 }
2012
2013                 if (required_features->compressed_files && !supported_features->compressed_files)
2014                 {
2015                         WARNING(
2016                   "%lu files in %"TS" are marked as being transparently\n"
2017         "          compressed, but transparent compression is not supported in\n"
2018         "          %"TS".  These files will be extracted as uncompressed.",
2019                                 required_features->compressed_files, loc, mode);
2020                 }
2021
2022                 if (required_features->encrypted_files && !supported_features->encrypted_files)
2023                 {
2024                         WARNING(
2025                   "%lu files in %"TS" are marked as being encrypted,\n"
2026         "           but encryption is not supported in %"TS".  These files\n"
2027         "           will not be extracted.",
2028                                 required_features->encrypted_files, loc, mode);
2029                 }
2030
2031                 if (required_features->encrypted_directories &&
2032                     !supported_features->encrypted_directories)
2033                 {
2034                         WARNING(
2035                   "%lu directories in %"TS" are marked as being encrypted,\n"
2036         "           but encryption is not supported in %"TS".\n"
2037         "           These directories will be extracted as unencrypted.",
2038                                 required_features->encrypted_directories, loc, mode);
2039                 }
2040
2041                 if (required_features->not_context_indexed_files &&
2042                     !supported_features->not_context_indexed_files)
2043                 {
2044                         WARNING(
2045                   "%lu files in %"TS" are marked as not content indexed,\n"
2046         "          but this attribute is not supported in %"TS".",
2047                                 required_features->not_context_indexed_files, loc, mode);
2048                 }
2049
2050                 if (required_features->sparse_files && !supported_features->sparse_files)
2051                 {
2052                         WARNING(
2053                   "%lu files in %"TS" are marked as sparse, but creating\n"
2054         "           sparse files is not supported in %"TS".  These files\n"
2055         "           will be extracted as non-sparse.",
2056                                 required_features->sparse_files, loc, mode);
2057                 }
2058
2059                 if (required_features->named_data_streams &&
2060                     !supported_features->named_data_streams)
2061                 {
2062                         WARNING(
2063                   "%lu files in %"TS" contain one or more alternate (named)\n"
2064         "          data streams, which are not supported in %"TS".\n"
2065         "          Alternate data streams will NOT be extracted.",
2066                                 required_features->named_data_streams, loc, mode);
2067                 }
2068
2069                 if (unlikely(extract_flags & (WIMLIB_EXTRACT_FLAG_HARDLINK |
2070                                               WIMLIB_EXTRACT_FLAG_SYMLINK)) &&
2071                     required_features->named_data_streams &&
2072                     supported_features->named_data_streams)
2073                 {
2074                         WARNING(
2075                   "%lu files in %"TS" contain one or more alternate (named)\n"
2076         "          data streams, which are not supported in linked extraction mode.\n"
2077         "          Alternate data streams will NOT be extracted.",
2078                                 required_features->named_data_streams, loc);
2079                 }
2080
2081                 if (required_features->hard_links && !supported_features->hard_links)
2082                 {
2083                         WARNING(
2084                   "%lu files in %"TS" are hard links, but hard links are\n"
2085         "          not supported in %"TS".  Hard links will be extracted as\n"
2086         "          duplicate copies of the linked files.",
2087                                 required_features->hard_links, loc, mode);
2088                 }
2089
2090                 if (required_features->reparse_points && !supported_features->reparse_points)
2091                 {
2092                         if (supported_features->symlink_reparse_points) {
2093                                 if (required_features->other_reparse_points) {
2094                                         WARNING(
2095                   "%lu files in %"TS" are reparse points that are neither\n"
2096         "          symbolic links nor junction points and are not supported in\n"
2097         "          %"TS".  These reparse points will not be extracted.",
2098                                                 required_features->other_reparse_points, loc,
2099                                                 mode);
2100                                 }
2101                         } else {
2102                                 WARNING(
2103                   "%lu files in %"TS" are reparse points, which are\n"
2104         "          not supported in %"TS" and will not be extracted.",
2105                                         required_features->reparse_points, loc, mode);
2106                         }
2107                 }
2108
2109                 if (required_features->security_descriptors &&
2110                     !supported_features->security_descriptors)
2111                 {
2112                         WARNING(
2113                   "%lu files in %"TS" have Windows NT security descriptors,\n"
2114         "          but extracting security descriptors is not supported in\n"
2115         "          %"TS".  No security descriptors will be extracted.",
2116                                 required_features->security_descriptors, loc, mode);
2117                 }
2118
2119                 if (required_features->short_names && !supported_features->short_names)
2120                 {
2121                         WARNING(
2122                   "%lu files in %"TS" have short (DOS) names, but\n"
2123         "          extracting short names is not supported in %"TS".\n"
2124         "          Short names will not be extracted.\n",
2125                                 required_features->short_names, loc, mode);
2126                 }
2127         }
2128
2129         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
2130             required_features->unix_data && !supported_features->unix_data)
2131         {
2132                 ERROR("Extracting UNIX data is not supported in %"TS, mode);
2133                 return WIMLIB_ERR_UNSUPPORTED;
2134         }
2135         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) &&
2136             required_features->short_names && !supported_features->short_names)
2137         {
2138                 ERROR("Extracting short names is not supported in %"TS"", mode);
2139                 return WIMLIB_ERR_UNSUPPORTED;
2140         }
2141         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
2142             !ops->set_timestamps)
2143         {
2144                 ERROR("Extracting timestamps is not supported in %"TS"", mode);
2145                 return WIMLIB_ERR_UNSUPPORTED;
2146         }
2147         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
2148                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
2149              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
2150             required_features->security_descriptors &&
2151             !supported_features->security_descriptors)
2152         {
2153                 ERROR("Extracting security descriptors is not supported in %"TS, mode);
2154                 return WIMLIB_ERR_UNSUPPORTED;
2155         }
2156
2157         if ((extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) &&
2158             !supported_features->hard_links)
2159         {
2160                 ERROR("Hard link extraction mode requested, but "
2161                       "%"TS" does not support hard links!", mode);
2162                 return WIMLIB_ERR_UNSUPPORTED;
2163         }
2164
2165         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
2166             required_features->symlink_reparse_points &&
2167             !(supported_features->symlink_reparse_points ||
2168               supported_features->reparse_points))
2169         {
2170                 ERROR("Extracting symbolic links is not supported in %"TS, mode);
2171                 return WIMLIB_ERR_UNSUPPORTED;
2172         }
2173
2174         if ((extract_flags & WIMLIB_EXTRACT_FLAG_SYMLINK) &&
2175             !supported_features->symlink_reparse_points)
2176         {
2177                 ERROR("Symbolic link extraction mode requested, but "
2178                       "%"TS" does not support symbolic "
2179                       "links!", mode);
2180                 return WIMLIB_ERR_UNSUPPORTED;
2181         }
2182         return 0;
2183 }
2184
2185 static void
2186 do_extract_warnings(struct apply_ctx *ctx)
2187 {
2188         if (ctx->partial_security_descriptors == 0 &&
2189             ctx->no_security_descriptors == 0)
2190                 return;
2191
2192         WARNING("Extraction to \"%"TS"\" complete, but with one or more warnings:",
2193                 ctx->target);
2194         if (ctx->partial_security_descriptors != 0) {
2195                 WARNING("- Could only partially set the security descriptor\n"
2196                         "            on %lu files or directories.",
2197                         ctx->partial_security_descriptors);
2198         }
2199         if (ctx->no_security_descriptors != 0) {
2200                 WARNING("- Could not set security descriptor at all\n"
2201                         "            on %lu files or directories.",
2202                         ctx->no_security_descriptors);
2203         }
2204 #ifdef __WIN32__
2205         WARNING("To fully restore all security descriptors, run the program\n"
2206                 "          with Administrator rights.");
2207 #endif
2208 }
2209
2210 static int
2211 dentry_set_skipped(struct wim_dentry *dentry, void *_ignore)
2212 {
2213         dentry->in_extraction_tree = 1;
2214         dentry->extraction_skipped = 1;
2215         return 0;
2216 }
2217
2218 static int
2219 dentry_set_not_skipped(struct wim_dentry *dentry, void *_ignore)
2220 {
2221         dentry->in_extraction_tree = 1;
2222         dentry->extraction_skipped = 0;
2223         return 0;
2224 }
2225
2226 static int
2227 extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
2228               const tchar *wim_source_path, const tchar *target,
2229               int extract_flags, wimlib_progress_func_t progress_func)
2230 {
2231         struct wim_features required_features;
2232         struct apply_ctx ctx;
2233         int ret;
2234         struct wim_lookup_table_entry *lte;
2235
2236         wimlib_assert(num_trees != 0);
2237         wimlib_assert(target != NULL);
2238
2239         if (num_trees > 1)
2240                 wimlib_assert((extract_flags & WIMLIB_EXTRACT_FLAG_PATHMODE));
2241
2242         /* Start initializing the apply_ctx.  */
2243         memset(&ctx, 0, sizeof(struct apply_ctx));
2244         ctx.wim = wim;
2245         ctx.extract_flags = extract_flags;
2246         ctx.target = target;
2247         ctx.target_nchars = tstrlen(target);
2248         ctx.progress_func = progress_func;
2249         if (progress_func) {
2250                 ctx.progress.extract.wimfile_name = wim->filename;
2251                 ctx.progress.extract.image = wim->current_image;
2252                 ctx.progress.extract.extract_flags = (extract_flags &
2253                                                       WIMLIB_EXTRACT_MASK_PUBLIC);
2254                 ctx.progress.extract.image_name = wimlib_get_image_name(wim,
2255                                                                         wim->current_image);
2256                 ctx.progress.extract.extract_root_wim_source_path = wim_source_path;
2257                 ctx.progress.extract.target = target;
2258         }
2259         INIT_LIST_HEAD(&ctx.stream_list);
2260
2261         if (extract_flags & WIMLIB_EXTRACT_FLAG_PATHMODE) {
2262                 /* Path mode --- there may be multiple trees, and targets are
2263                  * set relative to the root of the image.
2264                  *
2265                  * Consider all dentries to be in the extraction tree, but
2266                  * assume all to be skipped unless in one of the subtrees being
2267                  * extracted or one of the ancestors of the subtrees up to the
2268                  * image root.  */
2269                 ctx.extract_root = wim_root_dentry(wim);
2270                 for_dentry_in_tree(ctx.extract_root, dentry_set_skipped, NULL);
2271
2272                 for (size_t i = 0; i < num_trees; i++) {
2273                         struct wim_dentry *d;
2274
2275                         for_dentry_in_tree(trees[i], dentry_set_not_skipped, NULL);
2276                         d = trees[i];
2277                         while (d != ctx.extract_root) {
2278                                 d = d->parent;
2279                                 dentry_set_not_skipped(d, NULL);
2280                         }
2281                 }
2282         } else {
2283                 ctx.extract_root = trees[0];
2284                 for_dentry_in_tree(ctx.extract_root, dentry_set_not_skipped, NULL);
2285         }
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_dentry_reset_needs_extraction;
2312
2313         /* Get and check the features required to extract the dentry tree.  */
2314         dentry_tree_get_features(ctx.extract_root, &required_features);
2315         ret = do_feature_check(&required_features, &ctx.supported_features,
2316                                extract_flags, ctx.ops,
2317                                wim_source_path,
2318                                !(extract_flags & WIMLIB_EXTRACT_FLAG_PATHMODE));
2319         if (ret)
2320                 goto out_finish_or_abort_extract;
2321
2322         ctx.supported_attributes_mask =
2323                 compute_supported_attributes_mask(&ctx.supported_features);
2324
2325         /* Figure out whether the root dentry is being extracted to the root of
2326          * a volume and therefore needs to be treated "specially", for example
2327          * not being explicitly created and not having attributes set.  */
2328         if (ctx.ops->target_is_root && ctx.ops->root_directory_is_special)
2329                 ctx.root_dentry_is_special = ctx.ops->target_is_root(target);
2330
2331         /* Calculate the actual filename component of each extracted dentry.  In
2332          * the process, set the dentry->extraction_skipped flag on dentries that
2333          * are being skipped because of filename or supported features problems.  */
2334         ret = for_dentry_in_tree(ctx.extract_root,
2335                                  dentry_calculate_extraction_path, &ctx);
2336         if (ret)
2337                 goto out_dentry_reset_needs_extraction;
2338
2339         /* Build the list of the streams that need to be extracted and
2340          * initialize ctx.progress.extract with stream information.  */
2341         ret = for_dentry_in_tree(ctx.extract_root,
2342                                  dentry_resolve_and_zero_lte_refcnt, &ctx);
2343         if (ret)
2344                 goto out_dentry_reset_needs_extraction;
2345
2346         ret = for_dentry_in_tree(ctx.extract_root,
2347                                  dentry_add_streams_to_extract, &ctx);
2348         if (ret)
2349                 goto out_teardown_stream_list;
2350
2351         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2352                 /* When extracting from a pipe, the number of bytes of data to
2353                  * extract can't be determined in the normal way (examining the
2354                  * lookup table), since at this point all we have is a set of
2355                  * SHA1 message digests of streams that need to be extracted.
2356                  * However, we can get a reasonably accurate estimate by taking
2357                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
2358                  * data.  This does assume that a full image is being extracted,
2359                  * but currently there is no API for doing otherwise.  (Also,
2360                  * subtract <HARDLINKBYTES> from this if hard links are
2361                  * supported by the extraction mode.)  */
2362                 ctx.progress.extract.total_bytes =
2363                         wim_info_get_image_total_bytes(wim->wim_info,
2364                                                        wim->current_image);
2365                 if (ctx.supported_features.hard_links) {
2366                         ctx.progress.extract.total_bytes -=
2367                                 wim_info_get_image_hard_link_bytes(wim->wim_info,
2368                                                                    wim->current_image);
2369                 }
2370         }
2371
2372         /* Handle the special case of extracting a file to standard
2373          * output.  In that case, "root" should be a single file, not a
2374          * directory tree.  (If not, extract_dentry_to_stdout() will
2375          * return an error.)  */
2376         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
2377                 ret = extract_dentry_to_stdout(ctx.extract_root);
2378                 goto out_teardown_stream_list;
2379         }
2380
2381         if (ctx.ops->realpath_works_on_nonexisting_files &&
2382             ((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) ||
2383              ctx.ops->requires_realtarget_in_paths))
2384         {
2385                 ctx.realtarget = realpath(target, NULL);
2386                 if (!ctx.realtarget) {
2387                         ret = WIMLIB_ERR_NOMEM;
2388                         goto out_teardown_stream_list;
2389                 }
2390                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2391         }
2392
2393         if (progress_func) {
2394                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN :
2395                                                  WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN,
2396                               &ctx.progress);
2397         }
2398
2399         if (!ctx.root_dentry_is_special)
2400         {
2401                 tchar path[ctx.ops->path_max];
2402                 if (build_extraction_path(path, ctx.extract_root, &ctx))
2403                 {
2404                         ret = extract_inode(path, &ctx, ctx.extract_root->d_inode);
2405                         if (ret)
2406                                 goto out_free_realtarget;
2407                 }
2408         }
2409
2410         /* If we need to fix up the targets of absolute symbolic links
2411          * (WIMLIB_EXTRACT_FLAG_RPFIX) or the extraction mode requires paths to
2412          * be absolute, use realpath() (or its replacement on Windows) to get
2413          * the absolute path to the extraction target.  Note that this requires
2414          * the target directory to exist, unless
2415          * realpath_works_on_nonexisting_files is set in the apply_operations.
2416          * */
2417         if (!ctx.realtarget &&
2418             (((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
2419               required_features.symlink_reparse_points) ||
2420              ctx.ops->requires_realtarget_in_paths))
2421         {
2422                 ctx.realtarget = realpath(target, NULL);
2423                 if (!ctx.realtarget) {
2424                         ret = WIMLIB_ERR_NOMEM;
2425                         goto out_free_realtarget;
2426                 }
2427                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2428         }
2429
2430         if (ctx.ops->requires_short_name_reordering) {
2431                 ret = for_dentry_in_tree(ctx.extract_root, dentry_extract_dir_skeleton,
2432                                          &ctx);
2433                 if (ret)
2434                         goto out_free_realtarget;
2435         }
2436
2437         /* Finally, the important part: extract the tree of files.  */
2438         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
2439                 /* Sequential extraction requested, so two passes are needed
2440                  * (one for directory structure, one for streams.)  */
2441                 if (progress_func)
2442                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2443                                       &ctx.progress);
2444
2445                 if (!(extract_flags & WIMLIB_EXTRACT_FLAG_RESUME)) {
2446                         ret = for_dentry_in_tree(ctx.extract_root, dentry_extract_skeleton, &ctx);
2447                         if (ret)
2448                                 goto out_free_realtarget;
2449                 }
2450                 if (progress_func)
2451                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2452                                       &ctx.progress);
2453                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
2454                         ret = extract_streams_from_pipe(&ctx);
2455                 else
2456                         ret = extract_stream_list(&ctx);
2457                 if (ret)
2458                         goto out_free_realtarget;
2459         } else {
2460                 /* Sequential extraction was not requested, so we can make do
2461                  * with one pass where we both create the files and extract
2462                  * streams.   */
2463                 if (progress_func)
2464                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2465                                       &ctx.progress);
2466                 ret = for_dentry_in_tree(ctx.extract_root, dentry_extract, &ctx);
2467                 if (ret)
2468                         goto out_free_realtarget;
2469                 if (progress_func)
2470                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2471                                       &ctx.progress);
2472         }
2473
2474         /* If the total number of bytes to extract was miscalculated, just jump
2475          * to the calculated number in order to avoid confusing the progress
2476          * function.  This should only occur when extracting from a pipe.  */
2477         if (ctx.progress.extract.completed_bytes != ctx.progress.extract.total_bytes)
2478         {
2479                 DEBUG("Calculated %"PRIu64" bytes to extract, but actually "
2480                       "extracted %"PRIu64,
2481                       ctx.progress.extract.total_bytes,
2482                       ctx.progress.extract.completed_bytes);
2483         }
2484         if (progress_func &&
2485             ctx.progress.extract.completed_bytes < ctx.progress.extract.total_bytes)
2486         {
2487                 ctx.progress.extract.completed_bytes = ctx.progress.extract.total_bytes;
2488                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, &ctx.progress);
2489         }
2490
2491         /* Apply security descriptors and timestamps.  This is done at the end,
2492          * and in a depth-first manner, to prevent timestamps from getting
2493          * changed by subsequent extract operations and to minimize the chance
2494          * of the restored security descriptors getting in our way.  */
2495         if (progress_func)
2496                 progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
2497                               &ctx.progress);
2498         ret = for_dentry_in_tree_depth(ctx.extract_root, dentry_extract_final, &ctx);
2499         if (ret)
2500                 goto out_free_realtarget;
2501
2502         if (progress_func) {
2503                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END :
2504                               WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END,
2505                               &ctx.progress);
2506         }
2507
2508         do_extract_warnings(&ctx);
2509
2510         ret = 0;
2511 out_free_realtarget:
2512         FREE(ctx.realtarget);
2513 out_teardown_stream_list:
2514         /* Free memory allocated as part of the mapping from each
2515          * wim_lookup_table_entry to the dentries that reference it.  */
2516         if (!(ctx.extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER))
2517                 list_for_each_entry(lte, &ctx.stream_list, extraction_list)
2518                         if (lte->out_refcnt > ARRAY_LEN(lte->inline_lte_dentries))
2519                                 FREE(lte->lte_dentries);
2520 out_finish_or_abort_extract:
2521         if (ret) {
2522                 if (ctx.ops->abort_extract)
2523                         ctx.ops->abort_extract(&ctx);
2524         } else {
2525                 if (ctx.ops->finish_extract)
2526                         ret = ctx.ops->finish_extract(&ctx);
2527         }
2528 out_dentry_reset_needs_extraction:
2529         for_dentry_in_tree(ctx.extract_root, dentry_reset_needs_extraction, NULL);
2530         return ret;
2531 }
2532
2533 /*
2534  * extract_tree - Extract a file or directory tree from the currently selected
2535  *                WIM image.
2536  *
2537  * @wim:        WIMStruct for the WIM file, with the desired image selected
2538  *              (as wim->current_image).
2539  *
2540  * @wim_source_path:
2541  *              "Canonical" (i.e. no leading or trailing slashes, path
2542  *              separators WIM_PATH_SEPARATOR) path inside the WIM image to
2543  *              extract.  An empty string means the full image.
2544  *
2545  * @target:
2546  *              Filesystem path to extract the file or directory tree to.
2547  *              (Or, with WIMLIB_EXTRACT_FLAG_NTFS: the name of a NTFS volume.)
2548  *
2549  * @extract_flags:
2550  *              WIMLIB_EXTRACT_FLAG_*.  Also, the private flag
2551  *              WIMLIB_EXTRACT_FLAG_MULTI_IMAGE will be set if this is being
2552  *              called through wimlib_extract_image() with WIMLIB_ALL_IMAGES as
2553  *              the image.
2554  *
2555  * @progress_func:
2556  *              If non-NULL, progress function for the extraction.  The messages
2557  *              that may be sent in this function are:
2558  *
2559  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN or
2560  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
2561  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN;
2562  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END;
2563  *              WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY;
2564  *              WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS;
2565  *              WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS;
2566  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END or
2567  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END.
2568  *
2569  * Returns 0 on success; a positive WIMLIB_ERR_* code on failure.
2570  */
2571 static int
2572 extract_tree(WIMStruct *wim, const tchar *wim_source_path,
2573              const tchar *target, int extract_flags,
2574              wimlib_progress_func_t progress_func)
2575 {
2576         struct wim_dentry *root;
2577
2578         root = get_dentry(wim, wim_source_path, WIMLIB_CASE_PLATFORM_DEFAULT);
2579
2580         if (root == NULL) {
2581                   ERROR("Path \"%"TS"\" does not exist in WIM image %d",
2582                         wim_source_path, wim->current_image);
2583                   return WIMLIB_ERR_PATH_DOES_NOT_EXIST;
2584         }
2585         return extract_trees(wim, &root, 1, wim_source_path,
2586                              target, extract_flags, progress_func);
2587 }
2588
2589 /* Validates a single wimlib_extract_command, mostly checking to make sure the
2590  * extract flags make sense. */
2591 static int
2592 check_extract_command(struct wimlib_extract_command *cmd, int wim_header_flags)
2593 {
2594         int extract_flags;
2595
2596         /* Empty destination path? */
2597         if (cmd->fs_dest_path[0] == T('\0'))
2598                 return WIMLIB_ERR_INVALID_PARAM;
2599
2600         extract_flags = cmd->extract_flags;
2601
2602         /* Check for invalid flag combinations  */
2603         if ((extract_flags &
2604              (WIMLIB_EXTRACT_FLAG_SYMLINK |
2605               WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK |
2606                                                  WIMLIB_EXTRACT_FLAG_HARDLINK))
2607                 return WIMLIB_ERR_INVALID_PARAM;
2608
2609         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS)
2610                 return WIMLIB_ERR_INVALID_PARAM;
2611
2612         if ((extract_flags &
2613              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2614               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2615                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
2616                 return WIMLIB_ERR_INVALID_PARAM;
2617
2618         if ((extract_flags &
2619              (WIMLIB_EXTRACT_FLAG_RPFIX |
2620               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
2621                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
2622                 return WIMLIB_ERR_INVALID_PARAM;
2623
2624         if ((extract_flags &
2625              (WIMLIB_EXTRACT_FLAG_RESUME |
2626               WIMLIB_EXTRACT_FLAG_FROM_PIPE)) == WIMLIB_EXTRACT_FLAG_RESUME)
2627                 return WIMLIB_ERR_INVALID_PARAM;
2628
2629         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2630 #ifndef WITH_NTFS_3G
2631                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
2632                       "        we cannot apply a WIM image directly to a NTFS volume.");
2633                 return WIMLIB_ERR_UNSUPPORTED;
2634 #endif
2635         }
2636
2637         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
2638                               WIMLIB_EXTRACT_FLAG_NORPFIX)) == 0)
2639         {
2640                 /* Do reparse point fixups by default if the WIM header says
2641                  * they are enabled and we are extracting a full image. */
2642                 if (wim_header_flags & WIM_HDR_FLAG_RP_FIX)
2643                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
2644         }
2645
2646         /* TODO: Since UNIX data entries are stored in the file resources, in a
2647          * completely sequential extraction they may come up before the
2648          * corresponding file or symbolic link data.  This needs to be handled
2649          * better.  */
2650         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_UNIX_DATA |
2651                               WIMLIB_EXTRACT_FLAG_FILE_ORDER))
2652                                     == WIMLIB_EXTRACT_FLAG_UNIX_DATA)
2653         {
2654                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2655                         WARNING("Setting UNIX file/owner group may "
2656                                 "be impossible on some\n"
2657                                 "          symbolic links "
2658                                 "when applying from a pipe.");
2659                 } else {
2660                         extract_flags |= WIMLIB_EXTRACT_FLAG_FILE_ORDER;
2661                         WARNING("Disabling sequential extraction for "
2662                                 "UNIX data mode");
2663                 }
2664         }
2665
2666         cmd->extract_flags = extract_flags;
2667         return 0;
2668 }
2669
2670
2671 /* Internal function to execute extraction commands for a WIM image.  The paths
2672  * in the extract commands are expected to be already "canonicalized".  */
2673 static int
2674 do_wimlib_extract_files(WIMStruct *wim,
2675                         int image,
2676                         struct wimlib_extract_command *cmds,
2677                         size_t num_cmds,
2678                         wimlib_progress_func_t progress_func)
2679 {
2680         int ret;
2681         bool found_link_cmd = false;
2682         bool found_nolink_cmd = false;
2683
2684         /* Select the image from which we are extracting files */
2685         ret = select_wim_image(wim, image);
2686         if (ret)
2687                 return ret;
2688
2689         /* Make sure there are no streams in the WIM that have not been
2690          * checksummed yet.  */
2691         ret = wim_checksum_unhashed_streams(wim);
2692         if (ret)
2693                 return ret;
2694
2695         /* Check for problems with the extraction commands */
2696         for (size_t i = 0; i < num_cmds; i++) {
2697                 ret = check_extract_command(&cmds[i], wim->hdr.flags);
2698                 if (ret)
2699                         return ret;
2700                 if (cmds[i].extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2701                                              WIMLIB_EXTRACT_FLAG_HARDLINK)) {
2702                         found_link_cmd = true;
2703                 } else {
2704                         found_nolink_cmd = true;
2705                 }
2706                 if (found_link_cmd && found_nolink_cmd) {
2707                         ERROR("Symlink or hardlink extraction mode must "
2708                               "be set on all extraction commands");
2709                         return WIMLIB_ERR_INVALID_PARAM;
2710                 }
2711         }
2712
2713         /* Execute the extraction commands */
2714         for (size_t i = 0; i < num_cmds; i++) {
2715                 ret = extract_tree(wim,
2716                                    cmds[i].wim_source_path,
2717                                    cmds[i].fs_dest_path,
2718                                    cmds[i].extract_flags,
2719                                    progress_func);
2720                 if (ret)
2721                         return ret;
2722         }
2723         return 0;
2724 }
2725
2726 /* API function documented in wimlib.h  */
2727 WIMLIBAPI int
2728 wimlib_extract_files(WIMStruct *wim,
2729                      int image,
2730                      const struct wimlib_extract_command *cmds,
2731                      size_t num_cmds,
2732                      int default_extract_flags,
2733                      wimlib_progress_func_t progress_func)
2734 {
2735         int ret;
2736         struct wimlib_extract_command *cmds_copy;
2737         int all_flags = 0;
2738
2739         default_extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
2740
2741         if (num_cmds == 0)
2742                 return 0;
2743
2744         cmds_copy = CALLOC(num_cmds, sizeof(cmds[0]));
2745         if (!cmds_copy)
2746                 return WIMLIB_ERR_NOMEM;
2747
2748         for (size_t i = 0; i < num_cmds; i++) {
2749                 cmds_copy[i].extract_flags = (default_extract_flags |
2750                                                  cmds[i].extract_flags)
2751                                                 & WIMLIB_EXTRACT_MASK_PUBLIC;
2752                 all_flags |= cmds_copy[i].extract_flags;
2753
2754                 cmds_copy[i].wim_source_path = canonicalize_wim_path(cmds[i].wim_source_path);
2755                 if (!cmds_copy[i].wim_source_path) {
2756                         ret = WIMLIB_ERR_NOMEM;
2757                         goto out_free_cmds_copy;
2758                 }
2759
2760                 cmds_copy[i].fs_dest_path = canonicalize_fs_path(cmds[i].fs_dest_path);
2761                 if (!cmds_copy[i].fs_dest_path) {
2762                         ret = WIMLIB_ERR_NOMEM;
2763                         goto out_free_cmds_copy;
2764                 }
2765
2766         }
2767         ret = do_wimlib_extract_files(wim, image,
2768                                       cmds_copy, num_cmds,
2769                                       progress_func);
2770
2771         if (all_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2772                          WIMLIB_EXTRACT_FLAG_HARDLINK))
2773         {
2774                 for_lookup_table_entry(wim->lookup_table,
2775                                        lte_free_extracted_file, NULL);
2776         }
2777 out_free_cmds_copy:
2778         for (size_t i = 0; i < num_cmds; i++) {
2779                 FREE(cmds_copy[i].wim_source_path);
2780                 FREE(cmds_copy[i].fs_dest_path);
2781         }
2782         FREE(cmds_copy);
2783         return ret;
2784 }
2785
2786 /*
2787  * Extracts an image from a WIM file.
2788  *
2789  * @wim:                WIMStruct for the WIM file.
2790  *
2791  * @image:              Number of the single image to extract.
2792  *
2793  * @target:             Directory or NTFS volume to extract the image to.
2794  *
2795  * @extract_flags:      Bitwise or of WIMLIB_EXTRACT_FLAG_*.
2796  *
2797  * @progress_func:      If non-NULL, a progress function to be called
2798  *                      periodically.
2799  *
2800  * Returns 0 on success; nonzero on failure.
2801  */
2802 static int
2803 extract_single_image(WIMStruct *wim, int image,
2804                      const tchar *target, int extract_flags,
2805                      wimlib_progress_func_t progress_func)
2806 {
2807         int ret;
2808         tchar *target_copy = canonicalize_fs_path(target);
2809         if (target_copy == NULL)
2810                 return WIMLIB_ERR_NOMEM;
2811         struct wimlib_extract_command cmd = {
2812                 .wim_source_path = T(""),
2813                 .fs_dest_path = target_copy,
2814                 .extract_flags = extract_flags,
2815         };
2816         ret = do_wimlib_extract_files(wim, image, &cmd, 1, progress_func);
2817         FREE(target_copy);
2818         return ret;
2819 }
2820
2821 static const tchar * const filename_forbidden_chars =
2822 T(
2823 #ifdef __WIN32__
2824 "<>:\"/\\|?*"
2825 #else
2826 "/"
2827 #endif
2828 );
2829
2830 /* This function checks if it is okay to use a WIM image's name as a directory
2831  * name.  */
2832 static bool
2833 image_name_ok_as_dir(const tchar *image_name)
2834 {
2835         return image_name && *image_name &&
2836                 !tstrpbrk(image_name, filename_forbidden_chars) &&
2837                 tstrcmp(image_name, T(".")) &&
2838                 tstrcmp(image_name, T(".."));
2839 }
2840
2841 /* Extracts all images from the WIM to the directory @target, with the images
2842  * placed in subdirectories named by their image names. */
2843 static int
2844 extract_all_images(WIMStruct *wim,
2845                    const tchar *target,
2846                    int extract_flags,
2847                    wimlib_progress_func_t progress_func)
2848 {
2849         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
2850         size_t output_path_len = tstrlen(target);
2851         tchar buf[output_path_len + 1 + image_name_max_len + 1];
2852         int ret;
2853         int image;
2854         const tchar *image_name;
2855         struct stat stbuf;
2856
2857         extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
2858
2859         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2860                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
2861                 return WIMLIB_ERR_INVALID_PARAM;
2862         }
2863
2864         if (tstat(target, &stbuf)) {
2865                 if (errno == ENOENT) {
2866                         if (tmkdir(target, 0755)) {
2867                                 ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
2868                                 return WIMLIB_ERR_MKDIR;
2869                         }
2870                 } else {
2871                         ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
2872                         return WIMLIB_ERR_STAT;
2873                 }
2874         } else if (!S_ISDIR(stbuf.st_mode)) {
2875                 ERROR("\"%"TS"\" is not a directory", target);
2876                 return WIMLIB_ERR_NOTDIR;
2877         }
2878
2879         tmemcpy(buf, target, output_path_len);
2880         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
2881         for (image = 1; image <= wim->hdr.image_count; image++) {
2882                 image_name = wimlib_get_image_name(wim, image);
2883                 if (image_name_ok_as_dir(image_name)) {
2884                         tstrcpy(buf + output_path_len + 1, image_name);
2885                 } else {
2886                         /* Image name is empty or contains forbidden characters.
2887                          * Use image number instead. */
2888                         tsprintf(buf + output_path_len + 1, T("%d"), image);
2889                 }
2890                 ret = extract_single_image(wim, image, buf, extract_flags,
2891                                            progress_func);
2892                 if (ret)
2893                         return ret;
2894         }
2895         return 0;
2896 }
2897
2898 static int
2899 do_wimlib_extract_image(WIMStruct *wim,
2900                         int image,
2901                         const tchar *target,
2902                         int extract_flags,
2903                         wimlib_progress_func_t progress_func)
2904 {
2905         int ret;
2906
2907         if (image == WIMLIB_ALL_IMAGES) {
2908                 ret = extract_all_images(wim, target, extract_flags,
2909                                          progress_func);
2910         } else {
2911                 ret = extract_single_image(wim, image, target, extract_flags,
2912                                            progress_func);
2913         }
2914
2915         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2916                              WIMLIB_EXTRACT_FLAG_HARDLINK))
2917         {
2918                 for_lookup_table_entry(wim->lookup_table,
2919                                        lte_free_extracted_file,
2920                                        NULL);
2921         }
2922         return ret;
2923 }
2924
2925 /* API function documented in wimlib.h  */
2926 WIMLIBAPI int
2927 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
2928                                const tchar *target, int extract_flags,
2929                                wimlib_progress_func_t progress_func)
2930 {
2931         int ret;
2932         WIMStruct *pwm;
2933         struct filedes *in_fd;
2934         int image;
2935         unsigned i;
2936
2937         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
2938
2939         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT)
2940                 return WIMLIB_ERR_INVALID_PARAM;
2941
2942         /* Read the WIM header from the pipe and get a WIMStruct to represent
2943          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
2944          * wimlib_open_wim(), getting a WIMStruct in this way will result in
2945          * an empty lookup table, no XML data read, and no filename set.  */
2946         ret = open_wim_as_WIMStruct(&pipe_fd,
2947                                     WIMLIB_OPEN_FLAG_FROM_PIPE,
2948                                     &pwm, progress_func);
2949         if (ret)
2950                 return ret;
2951
2952         /* Sanity check to make sure this is a pipable WIM.  */
2953         if (pwm->hdr.magic != PWM_MAGIC) {
2954                 ERROR("The WIM being read from file descriptor %d "
2955                       "is not pipable!", pipe_fd);
2956                 ret = WIMLIB_ERR_NOT_PIPABLE;
2957                 goto out_wimlib_free;
2958         }
2959
2960         /* Sanity check to make sure the first part of a pipable split WIM is
2961          * sent over the pipe first.  */
2962         if (pwm->hdr.part_number != 1) {
2963                 ERROR("The first part of the split WIM must be "
2964                       "sent over the pipe first.");
2965                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
2966                 goto out_wimlib_free;
2967         }
2968
2969         in_fd = &pwm->in_fd;
2970         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
2971
2972         /* As mentioned, the WIMStruct we created from the pipe does not have
2973          * XML data yet.  Fix this by reading the extra copy of the XML data
2974          * that directly follows the header in pipable WIMs.  (Note: see
2975          * write_pipable_wim() for more details about the format of pipable
2976          * WIMs.)  */
2977         {
2978                 struct wim_lookup_table_entry xml_lte;
2979                 struct wim_resource_spec xml_rspec;
2980                 ret = read_pwm_stream_header(pwm, &xml_lte, &xml_rspec, 0, NULL);
2981                 if (ret)
2982                         goto out_wimlib_free;
2983
2984                 if (!(xml_lte.flags & WIM_RESHDR_FLAG_METADATA))
2985                 {
2986                         ERROR("Expected XML data, but found non-metadata "
2987                               "stream.");
2988                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
2989                         goto out_wimlib_free;
2990                 }
2991
2992                 wim_res_spec_to_hdr(&xml_rspec, &pwm->hdr.xml_data_reshdr);
2993
2994                 ret = read_wim_xml_data(pwm);
2995                 if (ret)
2996                         goto out_wimlib_free;
2997
2998                 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
2999                         ERROR("Image count in XML data is not the same as in WIM header.");
3000                         ret = WIMLIB_ERR_XML;
3001                         goto out_wimlib_free;
3002                 }
3003         }
3004
3005         /* Get image index (this may use the XML data that was just read to
3006          * resolve an image name).  */
3007         if (image_num_or_name) {
3008                 image = wimlib_resolve_image(pwm, image_num_or_name);
3009                 if (image == WIMLIB_NO_IMAGE) {
3010                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
3011                               image_num_or_name);
3012                         ret = WIMLIB_ERR_INVALID_IMAGE;
3013                         goto out_wimlib_free;
3014                 } else if (image == WIMLIB_ALL_IMAGES) {
3015                         ERROR("Applying all images from a pipe is not supported.");
3016                         ret = WIMLIB_ERR_INVALID_IMAGE;
3017                         goto out_wimlib_free;
3018                 }
3019         } else {
3020                 if (pwm->hdr.image_count != 1) {
3021                         ERROR("No image was specified, but the pipable WIM "
3022                               "did not contain exactly 1 image");
3023                         ret = WIMLIB_ERR_INVALID_IMAGE;
3024                         goto out_wimlib_free;
3025                 }
3026                 image = 1;
3027         }
3028
3029         /* Load the needed metadata resource.  */
3030         for (i = 1; i <= pwm->hdr.image_count; i++) {
3031                 struct wim_lookup_table_entry *metadata_lte;
3032                 struct wim_image_metadata *imd;
3033                 struct wim_resource_spec *metadata_rspec;
3034
3035                 metadata_lte = new_lookup_table_entry();
3036                 if (metadata_lte == NULL) {
3037                         ret = WIMLIB_ERR_NOMEM;
3038                         goto out_wimlib_free;
3039                 }
3040                 metadata_rspec = MALLOC(sizeof(struct wim_resource_spec));
3041                 if (metadata_rspec == NULL) {
3042                         ret = WIMLIB_ERR_NOMEM;
3043                         free_lookup_table_entry(metadata_lte);
3044                         goto out_wimlib_free;
3045                 }
3046
3047                 ret = read_pwm_stream_header(pwm, metadata_lte, metadata_rspec, 0, NULL);
3048                 imd = pwm->image_metadata[i - 1];
3049                 imd->metadata_lte = metadata_lte;
3050                 if (ret) {
3051                         FREE(metadata_rspec);
3052                         goto out_wimlib_free;
3053                 }
3054
3055                 if (!(metadata_lte->flags & WIM_RESHDR_FLAG_METADATA)) {
3056                         ERROR("Expected metadata resource, but found "
3057                               "non-metadata stream.");
3058                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
3059                         goto out_wimlib_free;
3060                 }
3061
3062                 if (i == image) {
3063                         /* Metadata resource is for the images being extracted.
3064                          * Parse it and save the metadata in memory.  */
3065                         ret = read_metadata_resource(pwm, imd);
3066                         if (ret)
3067                                 goto out_wimlib_free;
3068                         imd->modified = 1;
3069                 } else {
3070                         /* Metadata resource is not for the image being
3071                          * extracted.  Skip over it.  */
3072                         ret = skip_wim_stream(metadata_lte);
3073                         if (ret)
3074                                 goto out_wimlib_free;
3075                 }
3076         }
3077         /* Extract the image.  */
3078         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
3079         ret = do_wimlib_extract_image(pwm, image, target,
3080                                       extract_flags, progress_func);
3081         /* Clean up and return.  */
3082 out_wimlib_free:
3083         wimlib_free(pwm);
3084         return ret;
3085 }
3086
3087 /* API function documented in wimlib.h  */
3088 WIMLIBAPI int
3089 wimlib_extract_image(WIMStruct *wim,
3090                      int image,
3091                      const tchar *target,
3092                      int extract_flags,
3093                      wimlib_progress_func_t progress_func)
3094 {
3095         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
3096         return do_wimlib_extract_image(wim, image, target, extract_flags,
3097                                        progress_func);
3098 }
3099
3100 /* API function documented in wimlib.h  */
3101 WIMLIBAPI int
3102 wimlib_extract_pathlist(WIMStruct *wim, int image,
3103                         const tchar *target,
3104                         const tchar *path_list_file,
3105                         int extract_flags,
3106                         wimlib_progress_func_t progress_func)
3107 {
3108         int ret;
3109         tchar **paths;
3110         size_t num_paths;
3111         void *mem;
3112
3113         ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
3114         if (ret)
3115                 return ret;
3116
3117         ret = wimlib_extract_paths(wim, image, target,
3118                                    (const tchar * const *)paths, num_paths,
3119                                    extract_flags, progress_func);
3120         FREE(paths);
3121         FREE(mem);
3122         return ret;
3123 }
3124
3125 struct append_dentry_ctx {
3126         struct wim_dentry **dentries;
3127         size_t num_dentries;
3128         size_t num_alloc_dentries;
3129 };
3130
3131 static int
3132 append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
3133 {
3134         struct append_dentry_ctx *ctx = _ctx;
3135
3136         if (ctx->num_dentries == ctx->num_alloc_dentries) {
3137                 struct wim_dentry **new_dentries;
3138                 size_t new_length;
3139
3140                 new_length = max(ctx->num_alloc_dentries + 8,
3141                                  ctx->num_alloc_dentries * 3 / 2);
3142                 new_dentries = REALLOC(ctx->dentries,
3143                                        new_length * sizeof(ctx->dentries[0]));
3144                 if (new_dentries == NULL)
3145                         return WIMLIB_ERR_NOMEM;
3146                 ctx->dentries = new_dentries;
3147                 ctx->num_alloc_dentries = new_length;
3148         }
3149         ctx->dentries[ctx->num_dentries++] = dentry;
3150         return 0;
3151 }
3152
3153 /* API function documented in wimlib.h  */
3154 WIMLIBAPI int
3155 wimlib_extract_paths(WIMStruct *wim,
3156                      int image,
3157                      const tchar *target,
3158                      const tchar * const *paths,
3159                      size_t num_paths,
3160                      int extract_flags,
3161                      wimlib_progress_func_t progress_func)
3162 {
3163         int ret;
3164         struct append_dentry_ctx append_dentry_ctx = {
3165                 .dentries = NULL,
3166                 .num_dentries = 0,
3167                 .num_alloc_dentries = 0,
3168         };
3169         struct wim_dentry **trees;
3170         size_t num_trees;
3171
3172         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
3173
3174         if (target == NULL || (num_paths != 0 && paths == NULL))
3175                 return WIMLIB_ERR_INVALID_PARAM;
3176
3177         ret = select_wim_image(wim, image);
3178         if (ret)
3179                 return ret;
3180
3181         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
3182                 u32 wildcard_flags = 0;
3183
3184                 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB)
3185                         wildcard_flags |= WILDCARD_FLAG_ERROR_IF_NO_MATCH;
3186                 else
3187                         wildcard_flags |= WILDCARD_FLAG_WARN_IF_NO_MATCH;
3188
3189                 if (default_ignore_case)
3190                         wildcard_flags |= WILDCARD_FLAG_CASE_INSENSITIVE;
3191
3192                 for (size_t i = 0; i < num_paths; i++) {
3193                         ret = expand_wildcard(wim, paths[i],
3194                                               append_dentry_cb,
3195                                               &append_dentry_ctx,
3196                                               wildcard_flags);
3197                         if (ret) {
3198                                 trees = append_dentry_ctx.dentries;
3199                                 goto out_free_trees;
3200                         }
3201                 }
3202                 trees = append_dentry_ctx.dentries;
3203                 num_trees = append_dentry_ctx.num_dentries;
3204         } else {
3205                 trees = MALLOC(num_paths * sizeof(trees[0]));
3206                 if (trees == NULL)
3207                         return WIMLIB_ERR_NOMEM;
3208
3209                 for (size_t i = 0; i < num_paths; i++) {
3210                         trees[i] = get_dentry(wim, paths[i],
3211                                               WIMLIB_CASE_PLATFORM_DEFAULT);
3212                         if (trees[i] == NULL) {
3213                                   ERROR("Path \"%"TS"\" does not exist "
3214                                         "in WIM image %d",
3215                                         paths[i], wim->current_image);
3216                                   ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
3217                                   goto out_free_trees;
3218                         }
3219                 }
3220                 num_trees = num_paths;
3221         }
3222
3223         if (num_trees == 0) {
3224                 ret = 0;
3225                 goto out_free_trees;
3226         }
3227
3228         ret = extract_trees(wim, trees, num_trees,
3229                             T(""), target,
3230                             ((extract_flags &
3231                                 ~WIMLIB_EXTRACT_FLAG_GLOB_PATHS)
3232                                 | WIMLIB_EXTRACT_FLAG_PATHMODE),
3233                             progress_func);
3234 out_free_trees:
3235         FREE(trees);
3236         return ret;
3237 }