]> wimlib.net Git - wimlib/blob - src/extract.c
1e31cdb07909fbf309f5f015cf1858c715817dc5
[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                             && !other_dentry->extraction_skipped)
1152                         {
1153                                 DEBUG("Creating %"TS" before %"TS" "
1154                                       "to guarantee correct DOS name extraction",
1155                                       dentry_full_path(other_dentry),
1156                                       dentry_full_path(dentry));
1157                                 orig_dentry = dentry;
1158                                 dentry = other_dentry;
1159                                 break;
1160                         }
1161                 }
1162         }
1163 again:
1164         if (!build_extraction_path(path, dentry, ctx))
1165                 return 0;
1166         ret = do_dentry_extract_skeleton(path, dentry, ctx);
1167         if (ret)
1168                 return ret;
1169
1170         dentry->skeleton_extracted = 1;
1171
1172         if (orig_dentry) {
1173                 dentry = orig_dentry;
1174                 orig_dentry = NULL;
1175                 goto again;
1176         }
1177         dentry->d_inode->i_dos_name_extracted = 1;
1178         return 0;
1179 }
1180
1181 static int
1182 dentry_extract_dir_skeleton(struct wim_dentry *dentry, void *_ctx)
1183 {
1184         if (dentry->d_inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1185                 return dentry_extract_skeleton(dentry, _ctx);
1186         return 0;
1187 }
1188
1189 /* Create a file or directory, then immediately extract all streams.  The WIM
1190  * may not be read sequentially by this function.  */
1191 static int
1192 dentry_extract(struct wim_dentry *dentry, void *_ctx)
1193 {
1194         struct apply_ctx *ctx = _ctx;
1195         tchar path[ctx->ops->path_max];
1196         int ret;
1197
1198         ret = dentry_extract_skeleton(dentry, ctx);
1199         if (ret)
1200                 return ret;
1201
1202         if (!build_extraction_path(path, dentry, ctx))
1203                 return 0;
1204
1205         return extract_streams(path, ctx, dentry, NULL, NULL);
1206 }
1207
1208 /* Creates a temporary file opened for writing.  The open file descriptor is
1209  * returned in @fd_ret and its name is returned in @name_ret (dynamically
1210  * allocated).  */
1211 static int
1212 create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
1213 {
1214         tchar *name;
1215         int raw_fd;
1216
1217 retry:
1218         name = ttempnam(NULL, T("wimlib"));
1219         if (name == NULL) {
1220                 ERROR_WITH_ERRNO("Failed to create temporary filename");
1221                 return WIMLIB_ERR_NOMEM;
1222         }
1223
1224         raw_fd = topen(name, O_WRONLY | O_CREAT | O_EXCL | O_BINARY, 0600);
1225
1226         if (raw_fd < 0) {
1227                 if (errno == EEXIST) {
1228                         FREE(name);
1229                         goto retry;
1230                 }
1231                 ERROR_WITH_ERRNO("Failed to open temporary file \"%"TS"\"", name);
1232                 FREE(name);
1233                 return WIMLIB_ERR_OPEN;
1234         }
1235
1236         filedes_init(fd_ret, raw_fd);
1237         *name_ret = name;
1238         return 0;
1239 }
1240
1241 /* Extract all instances of the stream @lte that are being extracted in this
1242  * call of extract_tree(), but actually read the stream data from @lte_override.
1243  */
1244 static int
1245 extract_stream_instances(struct wim_lookup_table_entry *lte,
1246                          struct wim_lookup_table_entry *lte_override,
1247                          struct apply_ctx *ctx)
1248 {
1249         struct wim_dentry **lte_dentries;
1250         tchar path[ctx->ops->path_max];
1251         size_t i;
1252         int ret;
1253
1254         if (lte->out_refcnt <= ARRAY_LEN(lte->inline_lte_dentries))
1255                 lte_dentries = lte->inline_lte_dentries;
1256         else
1257                 lte_dentries = lte->lte_dentries;
1258
1259         for (i = 0; i < lte->out_refcnt; i++) {
1260                 struct wim_dentry *dentry = lte_dentries[i];
1261
1262                 if (dentry->tmp_flag)
1263                         continue;
1264                 if (!build_extraction_path(path, dentry, ctx))
1265                         continue;
1266                 ret = extract_streams(path, ctx, dentry, lte, lte_override);
1267                 if (ret)
1268                         goto out_clear_tmp_flags;
1269                 dentry->tmp_flag = 1;
1270         }
1271         ret = 0;
1272 out_clear_tmp_flags:
1273         for (i = 0; i < lte->out_refcnt; i++)
1274                 lte_dentries[i]->tmp_flag = 0;
1275         return ret;
1276 }
1277
1278 /* Determine whether the specified stream needs to be extracted to a temporary
1279  * file or not.
1280  *
1281  * @lte->out_refcnt specifies the number of instances of this stream that must
1282  * be extracted.
1283  *
1284  * @is_partial_res is %true if this stream is just one of multiple in a single
1285  * WIM resource being extracted.  */
1286 static bool
1287 need_tmpfile_to_extract(struct wim_lookup_table_entry *lte,
1288                         bool is_partial_res)
1289 {
1290         /* Temporary file is always required when reading a partial resource,
1291          * since in that case we retrieve all the contained streams in one pass.
1292          * */
1293         if (is_partial_res)
1294                 return true;
1295
1296         /* Otherwise we don't need a temporary file if only a single instance of
1297          * the stream is needed.  */
1298         if (lte->out_refcnt == 1)
1299                 return false;
1300
1301         wimlib_assert(lte->out_refcnt >= 2);
1302
1303         /* We also don't need a temporary file if random access to the stream is
1304          * allowed.  */
1305         if (lte->resource_location != RESOURCE_IN_WIM ||
1306             filedes_is_seekable(&lte->rspec->wim->in_fd))
1307                 return false;
1308
1309         return true;
1310 }
1311
1312 static int
1313 begin_extract_stream_to_tmpfile(struct wim_lookup_table_entry *lte,
1314                                 bool is_partial_res,
1315                                 void *_ctx)
1316 {
1317         struct apply_ctx *ctx = _ctx;
1318         int ret;
1319
1320         if (!need_tmpfile_to_extract(lte, is_partial_res)) {
1321                 DEBUG("Temporary file not needed "
1322                       "for stream (size=%"PRIu64")", lte->size);
1323                 ret = extract_stream_instances(lte, lte, ctx);
1324                 if (ret)
1325                         return ret;
1326
1327                 return BEGIN_STREAM_STATUS_SKIP_STREAM;
1328         }
1329
1330         DEBUG("Temporary file needed for stream (size=%"PRIu64")", lte->size);
1331         return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
1332 }
1333
1334 static int
1335 end_extract_stream_to_tmpfile(struct wim_lookup_table_entry *lte,
1336                               int status, void *_ctx)
1337 {
1338         struct apply_ctx *ctx = _ctx;
1339         struct wim_lookup_table_entry lte_override;
1340         int ret;
1341         int errno_save = errno;
1342
1343         ret = filedes_close(&ctx->tmpfile_fd);
1344
1345         if (status) {
1346                 ret = status;
1347                 errno = errno_save;
1348                 goto out_delete_tmpfile;
1349         }
1350
1351         if (ret) {
1352                 ERROR_WITH_ERRNO("Error writing temporary file %"TS, ctx->tmpfile_name);
1353                 ret = WIMLIB_ERR_WRITE;
1354                 goto out_delete_tmpfile;
1355         }
1356
1357         /* Now that a full stream has been extracted to a temporary file,
1358          * extract all instances of it to the actual target.  */
1359
1360         memcpy(&lte_override, lte, sizeof(struct wim_lookup_table_entry));
1361         lte_override.resource_location = RESOURCE_IN_FILE_ON_DISK;
1362         lte_override.file_on_disk = ctx->tmpfile_name;
1363
1364         ret = extract_stream_instances(lte, &lte_override, ctx);
1365
1366 out_delete_tmpfile:
1367         errno_save = errno;
1368         tunlink(ctx->tmpfile_name);
1369         FREE(ctx->tmpfile_name);
1370         errno = errno_save;
1371         return ret;
1372 }
1373
1374 /* Extracts a list of streams (ctx.stream_list), assuming that the directory
1375  * structure and empty files were already created.  This relies on the
1376  * per-`struct wim_lookup_table_entry' list of dentries that reference each
1377  * stream that was constructed earlier.  */
1378 static int
1379 extract_stream_list(struct apply_ctx *ctx)
1380 {
1381         if (!(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
1382                 /* Sequential extraction: read the streams in the order in which
1383                  * they appear in the WIM file.  */
1384                 struct read_stream_list_callbacks cbs = {
1385                         .begin_stream           = begin_extract_stream_to_tmpfile,
1386                         .begin_stream_ctx       = ctx,
1387                         .consume_chunk          = extract_chunk_to_fd,
1388                         .consume_chunk_ctx      = &ctx->tmpfile_fd,
1389                         .end_stream             = end_extract_stream_to_tmpfile,
1390                         .end_stream_ctx         = ctx,
1391                 };
1392                 return read_stream_list(&ctx->stream_list,
1393                                         offsetof(struct wim_lookup_table_entry, extraction_list),
1394                                         &cbs, VERIFY_STREAM_HASHES);
1395         } else {
1396                 /* Extract the streams in unsorted order.  */
1397                 struct wim_lookup_table_entry *lte;
1398                 int ret;
1399
1400                 list_for_each_entry(lte, &ctx->stream_list, extraction_list) {
1401                         ret = extract_stream_instances(lte, lte, ctx);
1402                         if (ret)
1403                                 return ret;
1404                 }
1405                 return 0;
1406         }
1407 }
1408
1409 #define PWM_ALLOW_WIM_HDR 0x00001
1410 #define PWM_SILENT_EOF    0x00002
1411
1412 /* Read the header from a stream in a pipable WIM.  */
1413 static int
1414 read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte,
1415                        struct wim_resource_spec *rspec,
1416                        int flags, struct wim_header_disk *hdr_ret)
1417 {
1418         union {
1419                 struct pwm_stream_hdr stream_hdr;
1420                 struct wim_header_disk pwm_hdr;
1421         } buf;
1422         struct wim_reshdr reshdr;
1423         int ret;
1424
1425         ret = full_read(&pwm->in_fd, &buf.stream_hdr, sizeof(buf.stream_hdr));
1426         if (ret)
1427                 goto read_error;
1428
1429         if ((flags & PWM_ALLOW_WIM_HDR) && buf.stream_hdr.magic == PWM_MAGIC) {
1430                 BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr));
1431                 ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1,
1432                                 sizeof(buf.pwm_hdr) - sizeof(buf.stream_hdr));
1433
1434                 if (ret)
1435                         goto read_error;
1436                 lte->resource_location = RESOURCE_NONEXISTENT;
1437                 memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
1438                 return 0;
1439         }
1440
1441         if (le64_to_cpu(buf.stream_hdr.magic) != PWM_STREAM_MAGIC) {
1442                 ERROR("Data read on pipe is invalid (expected stream header).");
1443                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
1444         }
1445
1446         copy_hash(lte->hash, buf.stream_hdr.hash);
1447
1448         reshdr.size_in_wim = 0;
1449         reshdr.flags = le32_to_cpu(buf.stream_hdr.flags);
1450         reshdr.offset_in_wim = pwm->in_fd.offset;
1451         reshdr.uncompressed_size = le64_to_cpu(buf.stream_hdr.uncompressed_size);
1452         wim_res_hdr_to_spec(&reshdr, pwm, rspec);
1453         lte_bind_wim_resource_spec(lte, rspec);
1454         lte->flags = rspec->flags;
1455         lte->size = rspec->uncompressed_size;
1456         lte->offset_in_res = 0;
1457         return 0;
1458
1459 read_error:
1460         if (ret != WIMLIB_ERR_UNEXPECTED_END_OF_FILE || !(flags & PWM_SILENT_EOF))
1461                 ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
1462         return ret;
1463 }
1464
1465 static int
1466 extract_streams_from_pipe(struct apply_ctx *ctx)
1467 {
1468         struct wim_lookup_table_entry *found_lte;
1469         struct wim_resource_spec *rspec;
1470         struct wim_lookup_table_entry *needed_lte;
1471         struct wim_lookup_table *lookup_table;
1472         struct wim_header_disk pwm_hdr;
1473         int ret;
1474         int pwm_flags;
1475
1476         ret = WIMLIB_ERR_NOMEM;
1477         found_lte = new_lookup_table_entry();
1478         if (found_lte == NULL)
1479                 goto out;
1480
1481         rspec = MALLOC(sizeof(struct wim_resource_spec));
1482         if (rspec == NULL)
1483                 goto out_free_found_lte;
1484
1485         lookup_table = ctx->wim->lookup_table;
1486         pwm_flags = PWM_ALLOW_WIM_HDR;
1487         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1488                 pwm_flags |= PWM_SILENT_EOF;
1489         memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GID_LEN);
1490         ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
1491         ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
1492         if (ctx->progress_func)
1493                 ctx->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1494                                    &ctx->progress);
1495         while (ctx->num_streams_remaining) {
1496                 if (found_lte->resource_location != RESOURCE_NONEXISTENT)
1497                         lte_unbind_wim_resource_spec(found_lte);
1498                 ret = read_pwm_stream_header(ctx->wim, found_lte, rspec,
1499                                              pwm_flags, &pwm_hdr);
1500                 if (ret) {
1501                         if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE &&
1502                             (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1503                         {
1504                                 goto resume_done;
1505                         }
1506                         goto out_free_found_lte;
1507                 }
1508
1509                 if ((found_lte->resource_location != RESOURCE_NONEXISTENT)
1510                     && !(found_lte->flags & WIM_RESHDR_FLAG_METADATA)
1511                     && (needed_lte = lookup_resource(lookup_table, found_lte->hash))
1512                     && (needed_lte->out_refcnt))
1513                 {
1514                         tchar *tmpfile_name = NULL;
1515                         struct wim_lookup_table_entry *lte_override;
1516                         struct wim_lookup_table_entry tmpfile_lte;
1517
1518                         needed_lte->offset_in_res = found_lte->offset_in_res;
1519                         needed_lte->flags = found_lte->flags;
1520                         needed_lte->size = found_lte->size;
1521
1522                         lte_unbind_wim_resource_spec(found_lte);
1523                         lte_bind_wim_resource_spec(needed_lte, rspec);
1524
1525                         if (needed_lte->out_refcnt > 1) {
1526
1527                                 struct filedes tmpfile_fd;
1528
1529                                 /* Extract stream to temporary file.  */
1530                                 ret = create_temporary_file(&tmpfile_fd, &tmpfile_name);
1531                                 if (ret) {
1532                                         lte_unbind_wim_resource_spec(needed_lte);
1533                                         goto out_free_found_lte;
1534                                 }
1535
1536                                 ret = extract_full_stream_to_fd(needed_lte,
1537                                                                 &tmpfile_fd);
1538                                 if (ret) {
1539                                         filedes_close(&tmpfile_fd);
1540                                         goto delete_tmpfile;
1541                                 }
1542
1543                                 if (filedes_close(&tmpfile_fd)) {
1544                                         ERROR_WITH_ERRNO("Error writing to temporary "
1545                                                          "file \"%"TS"\"", tmpfile_name);
1546                                         ret = WIMLIB_ERR_WRITE;
1547                                         goto delete_tmpfile;
1548                                 }
1549                                 memcpy(&tmpfile_lte, needed_lte,
1550                                        sizeof(struct wim_lookup_table_entry));
1551                                 tmpfile_lte.resource_location = RESOURCE_IN_FILE_ON_DISK;
1552                                 tmpfile_lte.file_on_disk = tmpfile_name;
1553                                 lte_override = &tmpfile_lte;
1554                         } else {
1555                                 lte_override = needed_lte;
1556                         }
1557
1558                         ret = extract_stream_instances(needed_lte, lte_override, ctx);
1559                 delete_tmpfile:
1560                         lte_unbind_wim_resource_spec(needed_lte);
1561                         if (tmpfile_name) {
1562                                 tunlink(tmpfile_name);
1563                                 FREE(tmpfile_name);
1564                         }
1565                         if (ret)
1566                                 goto out_free_found_lte;
1567                         ctx->num_streams_remaining--;
1568                 } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) {
1569                         ret = skip_wim_stream(found_lte);
1570                         if (ret)
1571                                 goto out_free_found_lte;
1572                 } else {
1573                         u16 part_number = le16_to_cpu(pwm_hdr.part_number);
1574                         u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
1575
1576                         if (part_number != ctx->progress.extract.part_number ||
1577                             total_parts != ctx->progress.extract.total_parts ||
1578                             memcmp(pwm_hdr.guid, ctx->progress.extract.guid,
1579                                    WIM_GID_LEN))
1580                         {
1581                                 ctx->progress.extract.part_number = part_number;
1582                                 ctx->progress.extract.total_parts = total_parts;
1583                                 memcpy(ctx->progress.extract.guid,
1584                                        pwm_hdr.guid, WIM_GID_LEN);
1585                                 if (ctx->progress_func) {
1586                                         ctx->progress_func(
1587                                                 WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1588                                                            &ctx->progress);
1589                                 }
1590
1591                         }
1592                 }
1593         }
1594         ret = 0;
1595 out_free_found_lte:
1596         if (found_lte->resource_location != RESOURCE_IN_WIM)
1597                 FREE(rspec);
1598         free_lookup_table_entry(found_lte);
1599 out:
1600         return ret;
1601
1602 resume_done:
1603         /* TODO */
1604         return 0;
1605 }
1606
1607 /* Finish extracting a file, directory, or symbolic link by setting file
1608  * security and timestamps.  */
1609 static int
1610 dentry_extract_final(struct wim_dentry *dentry, void *_ctx)
1611 {
1612         struct apply_ctx *ctx = _ctx;
1613         int ret;
1614         tchar path[ctx->ops->path_max];
1615
1616         if (!build_extraction_path(path, dentry, ctx))
1617                 return 0;
1618
1619         ret = extract_security(path, ctx, dentry);
1620         if (ret)
1621                 return ret;
1622
1623         if (ctx->ops->requires_final_set_attributes_pass) {
1624                 /* Set file attributes (if supported).  */
1625                 ret = extract_file_attributes(path, ctx, dentry, 1);
1626                 if (ret)
1627                         return ret;
1628         }
1629
1630         return extract_timestamps(path, ctx, dentry);
1631 }
1632
1633 /*
1634  * Extract a WIM dentry to standard output.
1635  *
1636  * This obviously doesn't make sense in all cases.  We return an error if the
1637  * dentry does not correspond to a regular file.  Otherwise we extract the
1638  * unnamed data stream only.
1639  */
1640 static int
1641 extract_dentry_to_stdout(struct wim_dentry *dentry)
1642 {
1643         int ret = 0;
1644         if (dentry->d_inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
1645                                              FILE_ATTRIBUTE_DIRECTORY))
1646         {
1647                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
1648                       "extracted to standard output", dentry_full_path(dentry));
1649                 ret = WIMLIB_ERR_NOT_A_REGULAR_FILE;
1650         } else {
1651                 struct wim_lookup_table_entry *lte;
1652
1653                 lte = inode_unnamed_lte_resolved(dentry->d_inode);
1654                 if (lte) {
1655                         struct filedes _stdout;
1656                         filedes_init(&_stdout, STDOUT_FILENO);
1657                         ret = extract_full_stream_to_fd(lte, &_stdout);
1658                 }
1659         }
1660         return ret;
1661 }
1662
1663 #ifdef __WIN32__
1664 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
1665 #else
1666 static const utf16lechar replacement_char = cpu_to_le16('?');
1667 #endif
1668
1669 static bool
1670 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
1671 {
1672         size_t i;
1673
1674         if (num_chars == 0)
1675                 return true;
1676         for (i = 0; i < num_chars; i++) {
1677                 switch (name[i]) {
1678         #ifdef __WIN32__
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                 case cpu_to_le16('<'):
1685                 case cpu_to_le16('>'):
1686                 case cpu_to_le16('|'):
1687         #endif
1688                 case cpu_to_le16('/'):
1689                 case cpu_to_le16('\0'):
1690                         if (fix)
1691                                 name[i] = replacement_char;
1692                         else
1693                                 return false;
1694                 }
1695         }
1696
1697 #ifdef __WIN32__
1698         if (name[num_chars - 1] == cpu_to_le16(' ') ||
1699             name[num_chars - 1] == cpu_to_le16('.'))
1700         {
1701                 if (fix)
1702                         name[num_chars - 1] = replacement_char;
1703                 else
1704                         return false;
1705         }
1706 #endif
1707         return true;
1708 }
1709
1710 static bool
1711 dentry_is_dot_or_dotdot(const struct wim_dentry *dentry)
1712 {
1713         const utf16lechar *file_name = dentry->file_name;
1714         return file_name != NULL &&
1715                 file_name[0] == cpu_to_le16('.') &&
1716                 (file_name[1] == cpu_to_le16('\0') ||
1717                  (file_name[1] == cpu_to_le16('.') &&
1718                   file_name[2] == cpu_to_le16('\0')));
1719 }
1720
1721 static int
1722 dentry_mark_skipped(struct wim_dentry *dentry, void *_ignore)
1723 {
1724         dentry->extraction_skipped = 1;
1725         return 0;
1726 }
1727
1728 /*
1729  * dentry_calculate_extraction_path-
1730  *
1731  * Calculate the actual filename component at which a WIM dentry will be
1732  * extracted, handling invalid filenames "properly".
1733  *
1734  * dentry->extraction_name usually will be set the same as dentry->file_name (on
1735  * UNIX, converted into the platform's multibyte encoding).  However, if the
1736  * file name contains characters that are not valid on the current platform or
1737  * has some other format that is not valid, leave dentry->extraction_name as
1738  * NULL and set dentry->extraction_skipped to indicate that this dentry should
1739  * not be extracted, unless the appropriate flag
1740  * WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES is set in the extract flags, in
1741  * which case a substitute filename will be created and set instead.
1742  *
1743  * Conflicts with case-insensitive names on Windows are handled similarly; see
1744  * below.
1745  */
1746 static int
1747 dentry_calculate_extraction_path(struct wim_dentry *dentry, void *_args)
1748 {
1749         struct apply_ctx *ctx = _args;
1750         int ret;
1751
1752         if (dentry == ctx->extract_root || dentry->extraction_skipped)
1753                 return 0;
1754
1755         if (!dentry_is_supported(dentry, &ctx->supported_features))
1756                 goto skip_dentry;
1757
1758         if (dentry_is_dot_or_dotdot(dentry)) {
1759                 /* WIM files shouldn't contain . or .. entries.  But if they are
1760                  * there, don't attempt to extract them. */
1761                 WARNING("Skipping extraction of unexpected . or .. file "
1762                         "\"%"TS"\"", dentry_full_path(dentry));
1763                 goto skip_dentry;
1764         }
1765
1766         if (!ctx->ops->supports_case_sensitive_filenames)
1767         {
1768                 struct wim_dentry *other;
1769                 list_for_each_entry(other, &dentry->case_insensitive_conflict_list,
1770                                     case_insensitive_conflict_list)
1771                 {
1772                         if (!other->extraction_skipped) {
1773                                 if (ctx->extract_flags &
1774                                     WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
1775                                         WARNING("\"%"TS"\" has the same "
1776                                                 "case-insensitive name as "
1777                                                 "\"%"TS"\"; extracting "
1778                                                 "dummy name instead",
1779                                                 dentry_full_path(dentry),
1780                                                 dentry_full_path(other));
1781                                         goto out_replace;
1782                                 } else {
1783                                         WARNING("Not extracting \"%"TS"\": "
1784                                                 "has same case-insensitive "
1785                                                 "name as \"%"TS"\"",
1786                                                 dentry_full_path(dentry),
1787                                                 dentry_full_path(other));
1788                                         goto skip_dentry;
1789                                 }
1790                         }
1791                 }
1792         }
1793
1794         if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
1795 #if TCHAR_IS_UTF16LE
1796                 dentry->extraction_name = dentry->file_name;
1797                 dentry->extraction_name_nchars = dentry->file_name_nbytes / 2;
1798                 return 0;
1799 #else
1800                 return utf16le_to_tstr(dentry->file_name,
1801                                        dentry->file_name_nbytes,
1802                                        &dentry->extraction_name,
1803                                        &dentry->extraction_name_nchars);
1804 #endif
1805         } else {
1806                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
1807                 {
1808                         WARNING("\"%"TS"\" has an invalid filename "
1809                                 "that is not supported on this platform; "
1810                                 "extracting dummy name instead",
1811                                 dentry_full_path(dentry));
1812                         goto out_replace;
1813                 } else {
1814                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
1815                                 "that is not supported on this platform",
1816                                 dentry_full_path(dentry));
1817                         goto skip_dentry;
1818                 }
1819         }
1820
1821 out_replace:
1822         {
1823                 utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2];
1824
1825                 memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
1826                 file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
1827
1828                 tchar *tchar_name;
1829                 size_t tchar_nchars;
1830         #if TCHAR_IS_UTF16LE
1831                 tchar_name = utf16_name_copy;
1832                 tchar_nchars = dentry->file_name_nbytes / 2;
1833         #else
1834                 ret = utf16le_to_tstr(utf16_name_copy,
1835                                       dentry->file_name_nbytes,
1836                                       &tchar_name, &tchar_nchars);
1837                 if (ret)
1838                         return ret;
1839         #endif
1840                 size_t fixed_name_num_chars = tchar_nchars;
1841                 tchar fixed_name[tchar_nchars + 50];
1842
1843                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
1844                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
1845                                                  T(" (invalid filename #%lu)"),
1846                                                  ++ctx->invalid_sequence);
1847         #if !TCHAR_IS_UTF16LE
1848                 FREE(tchar_name);
1849         #endif
1850                 dentry->extraction_name = memdup(fixed_name,
1851                                                  2 * fixed_name_num_chars + 2);
1852                 if (!dentry->extraction_name)
1853                         return WIMLIB_ERR_NOMEM;
1854                 dentry->extraction_name_nchars = fixed_name_num_chars;
1855         }
1856         return 0;
1857
1858 skip_dentry:
1859         for_dentry_in_tree(dentry, dentry_mark_skipped, NULL);
1860         return 0;
1861 }
1862
1863 /* Clean up dentry and inode structure after extraction.  */
1864 static int
1865 dentry_reset_needs_extraction(struct wim_dentry *dentry, void *_ignore)
1866 {
1867         struct wim_inode *inode = dentry->d_inode;
1868
1869         dentry->in_extraction_tree = 0;
1870         dentry->extraction_skipped = 0;
1871         dentry->was_hardlinked = 0;
1872         dentry->skeleton_extracted = 0;
1873         inode->i_visited = 0;
1874         FREE(inode->i_extracted_file);
1875         inode->i_extracted_file = NULL;
1876         inode->i_dos_name_extracted = 0;
1877         if ((void*)dentry->extraction_name != (void*)dentry->file_name)
1878                 FREE(dentry->extraction_name);
1879         dentry->extraction_name = NULL;
1880         return 0;
1881 }
1882
1883 /* Tally features necessary to extract a dentry and the corresponding inode.  */
1884 static int
1885 dentry_tally_features(struct wim_dentry *dentry, void *_features)
1886 {
1887         struct wim_features *features = _features;
1888         struct wim_inode *inode = dentry->d_inode;
1889
1890         if (dentry->extraction_skipped)
1891                 return 0;
1892
1893         if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
1894                 features->archive_files++;
1895         if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
1896                 features->hidden_files++;
1897         if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
1898                 features->system_files++;
1899         if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
1900                 features->compressed_files++;
1901         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
1902                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1903                         features->encrypted_directories++;
1904                 else
1905                         features->encrypted_files++;
1906         }
1907         if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
1908                 features->not_context_indexed_files++;
1909         if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
1910                 features->sparse_files++;
1911         if (inode_has_named_stream(inode))
1912                 features->named_data_streams++;
1913         if (inode->i_visited)
1914                 features->hard_links++;
1915         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1916                 features->reparse_points++;
1917                 if (inode_is_symlink(inode))
1918                         features->symlink_reparse_points++;
1919                 else
1920                         features->other_reparse_points++;
1921         }
1922         if (inode->i_security_id != -1)
1923                 features->security_descriptors++;
1924         if (dentry->short_name_nbytes)
1925                 features->short_names++;
1926         if (inode_has_unix_data(inode))
1927                 features->unix_data++;
1928         inode->i_visited = 1;
1929         return 0;
1930 }
1931
1932 /* Tally the features necessary to extract a dentry tree.  */
1933 static void
1934 dentry_tree_get_features(struct wim_dentry *root, struct wim_features *features)
1935 {
1936         memset(features, 0, sizeof(struct wim_features));
1937         for_dentry_in_tree(root, dentry_tally_features, features);
1938         dentry_tree_clear_inode_visited(root);
1939 }
1940
1941 static u32
1942 compute_supported_attributes_mask(const struct wim_features *supported_features)
1943 {
1944         u32 mask = ~(u32)0;
1945
1946         if (!supported_features->archive_files)
1947                 mask &= ~FILE_ATTRIBUTE_ARCHIVE;
1948
1949         if (!supported_features->hidden_files)
1950                 mask &= ~FILE_ATTRIBUTE_HIDDEN;
1951
1952         if (!supported_features->system_files)
1953                 mask &= ~FILE_ATTRIBUTE_SYSTEM;
1954
1955         if (!supported_features->not_context_indexed_files)
1956                 mask &= ~FILE_ATTRIBUTE_NOT_CONTENT_INDEXED;
1957
1958         if (!supported_features->compressed_files)
1959                 mask &= ~FILE_ATTRIBUTE_COMPRESSED;
1960
1961         if (!supported_features->sparse_files)
1962                 mask &= ~FILE_ATTRIBUTE_SPARSE_FILE;
1963
1964         if (!supported_features->reparse_points)
1965                 mask &= ~FILE_ATTRIBUTE_REPARSE_POINT;
1966
1967         return mask;
1968 }
1969
1970 static int
1971 do_feature_check(const struct wim_features *required_features,
1972                  const struct wim_features *supported_features,
1973                  int extract_flags,
1974                  const struct apply_operations *ops,
1975                  const tchar *wim_source_path,
1976                  bool warn)
1977 {
1978         const tchar *loc;
1979         const tchar *mode = T("this extraction mode");
1980
1981         if (wim_source_path[0] == '\0')
1982                 loc = T("the WIM image");
1983         else
1984                 loc = wim_source_path;
1985
1986         if (warn) {
1987                 /* We're an archive program, so theoretically we can do what we want
1988                  * with FILE_ATTRIBUTE_ARCHIVE (which is a dumb flag anyway).  Don't
1989                  * bother the user about it.  */
1990 #if 0
1991                 if (required_features->archive_files && !supported_features->archive_files)
1992                 {
1993                         WARNING(
1994                   "%lu files in %"TS" are marked as archived, but this attribute\n"
1995         "          is not supported in %"TS".",
1996                                 required_features->archive_files, loc, mode);
1997                 }
1998 #endif
1999
2000                 if (required_features->hidden_files && !supported_features->hidden_files)
2001                 {
2002                         WARNING(
2003                   "%lu files in %"TS" are marked as hidden, but this\n"
2004         "          attribute is not supported in %"TS".",
2005                                 required_features->hidden_files, loc, mode);
2006                 }
2007
2008                 if (required_features->system_files && !supported_features->system_files)
2009                 {
2010                         WARNING(
2011                   "%lu files in %"TS" are marked as system files,\n"
2012         "          but this attribute is not supported in %"TS".",
2013                                 required_features->system_files, loc, mode);
2014                 }
2015
2016                 if (required_features->compressed_files && !supported_features->compressed_files)
2017                 {
2018                         WARNING(
2019                   "%lu files in %"TS" are marked as being transparently\n"
2020         "          compressed, but transparent compression is not supported in\n"
2021         "          %"TS".  These files will be extracted as uncompressed.",
2022                                 required_features->compressed_files, loc, mode);
2023                 }
2024
2025                 if (required_features->encrypted_files && !supported_features->encrypted_files)
2026                 {
2027                         WARNING(
2028                   "%lu files in %"TS" are marked as being encrypted,\n"
2029         "           but encryption is not supported in %"TS".  These files\n"
2030         "           will not be extracted.",
2031                                 required_features->encrypted_files, loc, mode);
2032                 }
2033
2034                 if (required_features->encrypted_directories &&
2035                     !supported_features->encrypted_directories)
2036                 {
2037                         WARNING(
2038                   "%lu directories in %"TS" are marked as being encrypted,\n"
2039         "           but encryption is not supported in %"TS".\n"
2040         "           These directories will be extracted as unencrypted.",
2041                                 required_features->encrypted_directories, loc, mode);
2042                 }
2043
2044                 if (required_features->not_context_indexed_files &&
2045                     !supported_features->not_context_indexed_files)
2046                 {
2047                         WARNING(
2048                   "%lu files in %"TS" are marked as not content indexed,\n"
2049         "          but this attribute is not supported in %"TS".",
2050                                 required_features->not_context_indexed_files, loc, mode);
2051                 }
2052
2053                 if (required_features->sparse_files && !supported_features->sparse_files)
2054                 {
2055                         WARNING(
2056                   "%lu files in %"TS" are marked as sparse, but creating\n"
2057         "           sparse files is not supported in %"TS".  These files\n"
2058         "           will be extracted as non-sparse.",
2059                                 required_features->sparse_files, loc, mode);
2060                 }
2061
2062                 if (required_features->named_data_streams &&
2063                     !supported_features->named_data_streams)
2064                 {
2065                         WARNING(
2066                   "%lu files in %"TS" contain one or more alternate (named)\n"
2067         "          data streams, which are not supported in %"TS".\n"
2068         "          Alternate data streams will NOT be extracted.",
2069                                 required_features->named_data_streams, loc, mode);
2070                 }
2071
2072                 if (unlikely(extract_flags & (WIMLIB_EXTRACT_FLAG_HARDLINK |
2073                                               WIMLIB_EXTRACT_FLAG_SYMLINK)) &&
2074                     required_features->named_data_streams &&
2075                     supported_features->named_data_streams)
2076                 {
2077                         WARNING(
2078                   "%lu files in %"TS" contain one or more alternate (named)\n"
2079         "          data streams, which are not supported in linked extraction mode.\n"
2080         "          Alternate data streams will NOT be extracted.",
2081                                 required_features->named_data_streams, loc);
2082                 }
2083
2084                 if (required_features->hard_links && !supported_features->hard_links)
2085                 {
2086                         WARNING(
2087                   "%lu files in %"TS" are hard links, but hard links are\n"
2088         "          not supported in %"TS".  Hard links will be extracted as\n"
2089         "          duplicate copies of the linked files.",
2090                                 required_features->hard_links, loc, mode);
2091                 }
2092
2093                 if (required_features->reparse_points && !supported_features->reparse_points)
2094                 {
2095                         if (supported_features->symlink_reparse_points) {
2096                                 if (required_features->other_reparse_points) {
2097                                         WARNING(
2098                   "%lu files in %"TS" are reparse points that are neither\n"
2099         "          symbolic links nor junction points and are not supported in\n"
2100         "          %"TS".  These reparse points will not be extracted.",
2101                                                 required_features->other_reparse_points, loc,
2102                                                 mode);
2103                                 }
2104                         } else {
2105                                 WARNING(
2106                   "%lu files in %"TS" are reparse points, which are\n"
2107         "          not supported in %"TS" and will not be extracted.",
2108                                         required_features->reparse_points, loc, mode);
2109                         }
2110                 }
2111
2112                 if (required_features->security_descriptors &&
2113                     !supported_features->security_descriptors)
2114                 {
2115                         WARNING(
2116                   "%lu files in %"TS" have Windows NT security descriptors,\n"
2117         "          but extracting security descriptors is not supported in\n"
2118         "          %"TS".  No security descriptors will be extracted.",
2119                                 required_features->security_descriptors, loc, mode);
2120                 }
2121
2122                 if (required_features->short_names && !supported_features->short_names)
2123                 {
2124                         WARNING(
2125                   "%lu files in %"TS" have short (DOS) names, but\n"
2126         "          extracting short names is not supported in %"TS".\n"
2127         "          Short names will not be extracted.\n",
2128                                 required_features->short_names, loc, mode);
2129                 }
2130         }
2131
2132         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
2133             required_features->unix_data && !supported_features->unix_data)
2134         {
2135                 ERROR("Extracting UNIX data is not supported in %"TS, mode);
2136                 return WIMLIB_ERR_UNSUPPORTED;
2137         }
2138         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) &&
2139             required_features->short_names && !supported_features->short_names)
2140         {
2141                 ERROR("Extracting short names is not supported in %"TS"", mode);
2142                 return WIMLIB_ERR_UNSUPPORTED;
2143         }
2144         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
2145             !ops->set_timestamps)
2146         {
2147                 ERROR("Extracting timestamps is not supported in %"TS"", mode);
2148                 return WIMLIB_ERR_UNSUPPORTED;
2149         }
2150         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
2151                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
2152              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
2153             required_features->security_descriptors &&
2154             !supported_features->security_descriptors)
2155         {
2156                 ERROR("Extracting security descriptors is not supported in %"TS, mode);
2157                 return WIMLIB_ERR_UNSUPPORTED;
2158         }
2159
2160         if ((extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) &&
2161             !supported_features->hard_links)
2162         {
2163                 ERROR("Hard link extraction mode requested, but "
2164                       "%"TS" does not support hard links!", mode);
2165                 return WIMLIB_ERR_UNSUPPORTED;
2166         }
2167
2168         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
2169             required_features->symlink_reparse_points &&
2170             !(supported_features->symlink_reparse_points ||
2171               supported_features->reparse_points))
2172         {
2173                 ERROR("Extracting symbolic links is not supported in %"TS, mode);
2174                 return WIMLIB_ERR_UNSUPPORTED;
2175         }
2176
2177         if ((extract_flags & WIMLIB_EXTRACT_FLAG_SYMLINK) &&
2178             !supported_features->symlink_reparse_points)
2179         {
2180                 ERROR("Symbolic link extraction mode requested, but "
2181                       "%"TS" does not support symbolic "
2182                       "links!", mode);
2183                 return WIMLIB_ERR_UNSUPPORTED;
2184         }
2185         return 0;
2186 }
2187
2188 static void
2189 do_extract_warnings(struct apply_ctx *ctx)
2190 {
2191         if (ctx->partial_security_descriptors == 0 &&
2192             ctx->no_security_descriptors == 0)
2193                 return;
2194
2195         WARNING("Extraction to \"%"TS"\" complete, but with one or more warnings:",
2196                 ctx->target);
2197         if (ctx->partial_security_descriptors != 0) {
2198                 WARNING("- Could only partially set the security descriptor\n"
2199                         "            on %lu files or directories.",
2200                         ctx->partial_security_descriptors);
2201         }
2202         if (ctx->no_security_descriptors != 0) {
2203                 WARNING("- Could not set security descriptor at all\n"
2204                         "            on %lu files or directories.",
2205                         ctx->no_security_descriptors);
2206         }
2207 #ifdef __WIN32__
2208         WARNING("To fully restore all security descriptors, run the program\n"
2209                 "          with Administrator rights.");
2210 #endif
2211 }
2212
2213 static int
2214 dentry_set_skipped(struct wim_dentry *dentry, void *_ignore)
2215 {
2216         dentry->in_extraction_tree = 1;
2217         dentry->extraction_skipped = 1;
2218         return 0;
2219 }
2220
2221 static int
2222 dentry_set_not_skipped(struct wim_dentry *dentry, void *_ignore)
2223 {
2224         dentry->in_extraction_tree = 1;
2225         dentry->extraction_skipped = 0;
2226         return 0;
2227 }
2228
2229 static int
2230 extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
2231               const tchar *wim_source_path, const tchar *target,
2232               int extract_flags, wimlib_progress_func_t progress_func)
2233 {
2234         struct wim_features required_features;
2235         struct apply_ctx ctx;
2236         int ret;
2237         struct wim_lookup_table_entry *lte;
2238
2239         wimlib_assert(num_trees != 0);
2240         wimlib_assert(target != NULL);
2241
2242         if (num_trees > 1)
2243                 wimlib_assert((extract_flags & WIMLIB_EXTRACT_FLAG_PATHMODE));
2244
2245         /* Start initializing the apply_ctx.  */
2246         memset(&ctx, 0, sizeof(struct apply_ctx));
2247         ctx.wim = wim;
2248         ctx.extract_flags = extract_flags;
2249         ctx.target = target;
2250         ctx.target_nchars = tstrlen(target);
2251         ctx.progress_func = progress_func;
2252         if (progress_func) {
2253                 ctx.progress.extract.wimfile_name = wim->filename;
2254                 ctx.progress.extract.image = wim->current_image;
2255                 ctx.progress.extract.extract_flags = (extract_flags &
2256                                                       WIMLIB_EXTRACT_MASK_PUBLIC);
2257                 ctx.progress.extract.image_name = wimlib_get_image_name(wim,
2258                                                                         wim->current_image);
2259                 ctx.progress.extract.extract_root_wim_source_path = wim_source_path;
2260                 ctx.progress.extract.target = target;
2261         }
2262         INIT_LIST_HEAD(&ctx.stream_list);
2263
2264         if (extract_flags & WIMLIB_EXTRACT_FLAG_PATHMODE) {
2265                 /* Path mode --- there may be multiple trees, and targets are
2266                  * set relative to the root of the image.
2267                  *
2268                  * Consider all dentries to be in the extraction tree, but
2269                  * assume all to be skipped unless in one of the subtrees being
2270                  * extracted or one of the ancestors of the subtrees up to the
2271                  * image root.  */
2272                 ctx.extract_root = wim_root_dentry(wim);
2273                 for_dentry_in_tree(ctx.extract_root, dentry_set_skipped, NULL);
2274
2275                 for (size_t i = 0; i < num_trees; i++) {
2276                         struct wim_dentry *d;
2277
2278                         for_dentry_in_tree(trees[i], dentry_set_not_skipped, NULL);
2279                         d = trees[i];
2280                         while (d != ctx.extract_root) {
2281                                 d = d->parent;
2282                                 dentry_set_not_skipped(d, NULL);
2283                         }
2284                 }
2285         } else {
2286                 ctx.extract_root = trees[0];
2287                 for_dentry_in_tree(ctx.extract_root, dentry_set_not_skipped, NULL);
2288         }
2289
2290         /* Select the appropriate apply_operations based on the
2291          * platform and extract_flags.  */
2292 #ifdef __WIN32__
2293         ctx.ops = &win32_apply_ops;
2294 #else
2295         ctx.ops = &unix_apply_ops;
2296 #endif
2297
2298 #ifdef WITH_NTFS_3G
2299         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
2300                 ctx.ops = &ntfs_3g_apply_ops;
2301 #endif
2302
2303         /* Call the start_extract() callback.  This gives the apply_operations
2304          * implementation a chance to do any setup needed to access the volume.
2305          * Furthermore, it's expected to set the supported features of this
2306          * extraction mode (ctx.supported_features), which are determined at
2307          * runtime as they may vary depending on the actual volume.  These
2308          * features are then compared with the actual features extracting this
2309          * dentry tree requires.  Some mismatches will merely produce warnings
2310          * and the unsupported data will be ignored; others will produce errors.
2311          */
2312         ret = ctx.ops->start_extract(target, &ctx);
2313         if (ret)
2314                 goto out_dentry_reset_needs_extraction;
2315
2316         /* Get and check the features required to extract the dentry tree.  */
2317         dentry_tree_get_features(ctx.extract_root, &required_features);
2318         ret = do_feature_check(&required_features, &ctx.supported_features,
2319                                extract_flags, ctx.ops,
2320                                wim_source_path,
2321                                !(extract_flags & WIMLIB_EXTRACT_FLAG_PATHMODE));
2322         if (ret)
2323                 goto out_finish_or_abort_extract;
2324
2325         ctx.supported_attributes_mask =
2326                 compute_supported_attributes_mask(&ctx.supported_features);
2327
2328         /* Figure out whether the root dentry is being extracted to the root of
2329          * a volume and therefore needs to be treated "specially", for example
2330          * not being explicitly created and not having attributes set.  */
2331         if (ctx.ops->target_is_root && ctx.ops->root_directory_is_special)
2332                 ctx.root_dentry_is_special = ctx.ops->target_is_root(target);
2333
2334         /* Calculate the actual filename component of each extracted dentry.  In
2335          * the process, set the dentry->extraction_skipped flag on dentries that
2336          * are being skipped because of filename or supported features problems.  */
2337         ret = for_dentry_in_tree(ctx.extract_root,
2338                                  dentry_calculate_extraction_path, &ctx);
2339         if (ret)
2340                 goto out_dentry_reset_needs_extraction;
2341
2342         /* Build the list of the streams that need to be extracted and
2343          * initialize ctx.progress.extract with stream information.  */
2344         ret = for_dentry_in_tree(ctx.extract_root,
2345                                  dentry_resolve_and_zero_lte_refcnt, &ctx);
2346         if (ret)
2347                 goto out_dentry_reset_needs_extraction;
2348
2349         ret = for_dentry_in_tree(ctx.extract_root,
2350                                  dentry_add_streams_to_extract, &ctx);
2351         if (ret)
2352                 goto out_teardown_stream_list;
2353
2354         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2355                 /* When extracting from a pipe, the number of bytes of data to
2356                  * extract can't be determined in the normal way (examining the
2357                  * lookup table), since at this point all we have is a set of
2358                  * SHA1 message digests of streams that need to be extracted.
2359                  * However, we can get a reasonably accurate estimate by taking
2360                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
2361                  * data.  This does assume that a full image is being extracted,
2362                  * but currently there is no API for doing otherwise.  (Also,
2363                  * subtract <HARDLINKBYTES> from this if hard links are
2364                  * supported by the extraction mode.)  */
2365                 ctx.progress.extract.total_bytes =
2366                         wim_info_get_image_total_bytes(wim->wim_info,
2367                                                        wim->current_image);
2368                 if (ctx.supported_features.hard_links) {
2369                         ctx.progress.extract.total_bytes -=
2370                                 wim_info_get_image_hard_link_bytes(wim->wim_info,
2371                                                                    wim->current_image);
2372                 }
2373         }
2374
2375         /* Handle the special case of extracting a file to standard
2376          * output.  In that case, "root" should be a single file, not a
2377          * directory tree.  (If not, extract_dentry_to_stdout() will
2378          * return an error.)  */
2379         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
2380                 ret = 0;
2381                 for (size_t i = 0; i < num_trees; i++) {
2382                         ret = extract_dentry_to_stdout(trees[i]);
2383                         if (ret)
2384                                 break;
2385                 }
2386                 goto out_teardown_stream_list;
2387         }
2388
2389         if (ctx.ops->realpath_works_on_nonexisting_files &&
2390             ((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) ||
2391              ctx.ops->requires_realtarget_in_paths))
2392         {
2393                 ctx.realtarget = realpath(target, NULL);
2394                 if (!ctx.realtarget) {
2395                         ret = WIMLIB_ERR_NOMEM;
2396                         goto out_teardown_stream_list;
2397                 }
2398                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2399         }
2400
2401         if (progress_func) {
2402                 int msg;
2403                 if (*wim_source_path || (extract_flags & WIMLIB_EXTRACT_FLAG_PATHMODE))
2404                         msg = WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN;
2405                 else
2406                         msg = WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
2407                 progress_func(msg, &ctx.progress);
2408         }
2409
2410         if (!ctx.root_dentry_is_special)
2411         {
2412                 tchar path[ctx.ops->path_max];
2413                 if (build_extraction_path(path, ctx.extract_root, &ctx))
2414                 {
2415                         ret = extract_inode(path, &ctx, ctx.extract_root->d_inode);
2416                         if (ret)
2417                                 goto out_free_realtarget;
2418                 }
2419         }
2420
2421         /* If we need to fix up the targets of absolute symbolic links
2422          * (WIMLIB_EXTRACT_FLAG_RPFIX) or the extraction mode requires paths to
2423          * be absolute, use realpath() (or its replacement on Windows) to get
2424          * the absolute path to the extraction target.  Note that this requires
2425          * the target directory to exist, unless
2426          * realpath_works_on_nonexisting_files is set in the apply_operations.
2427          * */
2428         if (!ctx.realtarget &&
2429             (((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
2430               required_features.symlink_reparse_points) ||
2431              ctx.ops->requires_realtarget_in_paths))
2432         {
2433                 ctx.realtarget = realpath(target, NULL);
2434                 if (!ctx.realtarget) {
2435                         ret = WIMLIB_ERR_NOMEM;
2436                         goto out_free_realtarget;
2437                 }
2438                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2439         }
2440
2441         if (ctx.ops->requires_short_name_reordering) {
2442                 ret = for_dentry_in_tree(ctx.extract_root, dentry_extract_dir_skeleton,
2443                                          &ctx);
2444                 if (ret)
2445                         goto out_free_realtarget;
2446         }
2447
2448         /* Finally, the important part: extract the tree of files.  */
2449         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
2450                 /* Sequential extraction requested, so two passes are needed
2451                  * (one for directory structure, one for streams.)  */
2452                 if (progress_func)
2453                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2454                                       &ctx.progress);
2455
2456                 if (!(extract_flags & WIMLIB_EXTRACT_FLAG_RESUME)) {
2457                         ret = for_dentry_in_tree(ctx.extract_root, dentry_extract_skeleton, &ctx);
2458                         if (ret)
2459                                 goto out_free_realtarget;
2460                 }
2461                 if (progress_func)
2462                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2463                                       &ctx.progress);
2464                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
2465                         ret = extract_streams_from_pipe(&ctx);
2466                 else
2467                         ret = extract_stream_list(&ctx);
2468                 if (ret)
2469                         goto out_free_realtarget;
2470         } else {
2471                 /* Sequential extraction was not requested, so we can make do
2472                  * with one pass where we both create the files and extract
2473                  * streams.   */
2474                 if (progress_func)
2475                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2476                                       &ctx.progress);
2477                 ret = for_dentry_in_tree(ctx.extract_root, dentry_extract, &ctx);
2478                 if (ret)
2479                         goto out_free_realtarget;
2480                 if (progress_func)
2481                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2482                                       &ctx.progress);
2483         }
2484
2485         /* If the total number of bytes to extract was miscalculated, just jump
2486          * to the calculated number in order to avoid confusing the progress
2487          * function.  This should only occur when extracting from a pipe.  */
2488         if (ctx.progress.extract.completed_bytes != ctx.progress.extract.total_bytes)
2489         {
2490                 DEBUG("Calculated %"PRIu64" bytes to extract, but actually "
2491                       "extracted %"PRIu64,
2492                       ctx.progress.extract.total_bytes,
2493                       ctx.progress.extract.completed_bytes);
2494         }
2495         if (progress_func &&
2496             ctx.progress.extract.completed_bytes < ctx.progress.extract.total_bytes)
2497         {
2498                 ctx.progress.extract.completed_bytes = ctx.progress.extract.total_bytes;
2499                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, &ctx.progress);
2500         }
2501
2502         /* Apply security descriptors and timestamps.  This is done at the end,
2503          * and in a depth-first manner, to prevent timestamps from getting
2504          * changed by subsequent extract operations and to minimize the chance
2505          * of the restored security descriptors getting in our way.  */
2506         if (progress_func)
2507                 progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
2508                               &ctx.progress);
2509         ret = for_dentry_in_tree_depth(ctx.extract_root, dentry_extract_final, &ctx);
2510         if (ret)
2511                 goto out_free_realtarget;
2512
2513         if (progress_func) {
2514                 int msg;
2515                 if (*wim_source_path || (extract_flags & WIMLIB_EXTRACT_FLAG_PATHMODE))
2516                         msg = WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END;
2517                 else
2518                         msg = WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END;
2519                 progress_func(msg, &ctx.progress);
2520         }
2521
2522         do_extract_warnings(&ctx);
2523
2524         ret = 0;
2525 out_free_realtarget:
2526         FREE(ctx.realtarget);
2527 out_teardown_stream_list:
2528         /* Free memory allocated as part of the mapping from each
2529          * wim_lookup_table_entry to the dentries that reference it.  */
2530         if (!(ctx.extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER))
2531                 list_for_each_entry(lte, &ctx.stream_list, extraction_list)
2532                         if (lte->out_refcnt > ARRAY_LEN(lte->inline_lte_dentries))
2533                                 FREE(lte->lte_dentries);
2534 out_finish_or_abort_extract:
2535         if (ret) {
2536                 if (ctx.ops->abort_extract)
2537                         ctx.ops->abort_extract(&ctx);
2538         } else {
2539                 if (ctx.ops->finish_extract)
2540                         ret = ctx.ops->finish_extract(&ctx);
2541         }
2542 out_dentry_reset_needs_extraction:
2543         for_dentry_in_tree(ctx.extract_root, dentry_reset_needs_extraction, NULL);
2544         return ret;
2545 }
2546
2547 /*
2548  * extract_tree - Extract a file or directory tree from the currently selected
2549  *                WIM image.
2550  *
2551  * @wim:        WIMStruct for the WIM file, with the desired image selected
2552  *              (as wim->current_image).
2553  *
2554  * @wim_source_path:
2555  *              "Canonical" (i.e. no leading or trailing slashes, path
2556  *              separators WIM_PATH_SEPARATOR) path inside the WIM image to
2557  *              extract.  An empty string means the full image.
2558  *
2559  * @target:
2560  *              Filesystem path to extract the file or directory tree to.
2561  *              (Or, with WIMLIB_EXTRACT_FLAG_NTFS: the name of a NTFS volume.)
2562  *
2563  * @extract_flags:
2564  *              WIMLIB_EXTRACT_FLAG_*.  Also, the private flag
2565  *              WIMLIB_EXTRACT_FLAG_MULTI_IMAGE will be set if this is being
2566  *              called through wimlib_extract_image() with WIMLIB_ALL_IMAGES as
2567  *              the image.
2568  *
2569  * @progress_func:
2570  *              If non-NULL, progress function for the extraction.  The messages
2571  *              that may be sent in this function are:
2572  *
2573  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN or
2574  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
2575  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN;
2576  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END;
2577  *              WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY;
2578  *              WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS;
2579  *              WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS;
2580  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END or
2581  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END.
2582  *
2583  * Returns 0 on success; a positive WIMLIB_ERR_* code on failure.
2584  */
2585 static int
2586 extract_tree(WIMStruct *wim, const tchar *wim_source_path,
2587              const tchar *target, int extract_flags,
2588              wimlib_progress_func_t progress_func)
2589 {
2590         struct wim_dentry *root;
2591
2592         root = get_dentry(wim, wim_source_path, WIMLIB_CASE_PLATFORM_DEFAULT);
2593
2594         if (root == NULL) {
2595                   ERROR("Path \"%"TS"\" does not exist in WIM image %d",
2596                         wim_source_path, wim->current_image);
2597                   return WIMLIB_ERR_PATH_DOES_NOT_EXIST;
2598         }
2599         return extract_trees(wim, &root, 1, wim_source_path,
2600                              target, extract_flags, progress_func);
2601 }
2602
2603 /* Make sure the extraction flags make sense, and update them if needed.  */
2604 static int
2605 check_extract_flags(int extract_flags,
2606                     const u32 wim_header_flags,
2607                     int *updated_extract_flags_ret)
2608 {
2609         /* Check for invalid flag combinations  */
2610         if ((extract_flags &
2611              (WIMLIB_EXTRACT_FLAG_SYMLINK |
2612               WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK |
2613                                                  WIMLIB_EXTRACT_FLAG_HARDLINK))
2614                 return WIMLIB_ERR_INVALID_PARAM;
2615
2616         if ((extract_flags &
2617              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2618               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2619                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
2620                 return WIMLIB_ERR_INVALID_PARAM;
2621
2622         if ((extract_flags &
2623              (WIMLIB_EXTRACT_FLAG_RPFIX |
2624               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
2625                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
2626                 return WIMLIB_ERR_INVALID_PARAM;
2627
2628         if ((extract_flags &
2629              (WIMLIB_EXTRACT_FLAG_RESUME |
2630               WIMLIB_EXTRACT_FLAG_FROM_PIPE)) == WIMLIB_EXTRACT_FLAG_RESUME)
2631                 return WIMLIB_ERR_INVALID_PARAM;
2632
2633 #ifndef WITH_NTFS_3G
2634         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2635                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
2636                       "        we cannot apply a WIM image directly to a NTFS volume.");
2637                 return WIMLIB_ERR_UNSUPPORTED;
2638         }
2639 #endif
2640
2641         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
2642                               WIMLIB_EXTRACT_FLAG_NORPFIX)) == 0)
2643         {
2644                 /* Do reparse point fixups by default if the WIM header says
2645                  * they are enabled and we are extracting a full image. */
2646                 if (wim_header_flags & WIM_HDR_FLAG_RP_FIX)
2647                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
2648         }
2649
2650         /* TODO: Since UNIX data entries are stored in the file resources, in a
2651          * completely sequential extraction they may come up before the
2652          * corresponding file or symbolic link data.  This needs to be handled
2653          * better.  */
2654         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_UNIX_DATA |
2655                               WIMLIB_EXTRACT_FLAG_FILE_ORDER))
2656                                     == WIMLIB_EXTRACT_FLAG_UNIX_DATA)
2657         {
2658                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2659                         WARNING("Setting UNIX file/owner group may "
2660                                 "be impossible on some\n"
2661                                 "          symbolic links "
2662                                 "when applying from a pipe.");
2663                 } else {
2664                         extract_flags |= WIMLIB_EXTRACT_FLAG_FILE_ORDER;
2665                         WARNING("Disabling sequential extraction for "
2666                                 "UNIX data mode");
2667                 }
2668         }
2669
2670         if (updated_extract_flags_ret)
2671                 *updated_extract_flags_ret = extract_flags;
2672         return 0;
2673 }
2674
2675
2676 /* Internal function to execute extraction commands for a WIM image.  The paths
2677  * in the extract commands are expected to be already "canonicalized".  */
2678 static int
2679 do_wimlib_extract_files(WIMStruct *wim,
2680                         int image,
2681                         struct wimlib_extract_command *cmds,
2682                         size_t num_cmds,
2683                         wimlib_progress_func_t progress_func)
2684 {
2685         int ret;
2686         bool found_link_cmd = false;
2687         bool found_nolink_cmd = false;
2688
2689         /* Select the image from which we are extracting files */
2690         ret = select_wim_image(wim, image);
2691         if (ret)
2692                 return ret;
2693
2694         /* Make sure there are no streams in the WIM that have not been
2695          * checksummed yet.  */
2696         ret = wim_checksum_unhashed_streams(wim);
2697         if (ret)
2698                 return ret;
2699
2700         /* Check for problems with the extraction commands */
2701         for (size_t i = 0; i < num_cmds; i++) {
2702
2703                 if (cmds[i].fs_dest_path[0] == T('\0'))
2704                         return WIMLIB_ERR_INVALID_PARAM;
2705
2706                 if (cmds[i].extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS)
2707                         return WIMLIB_ERR_INVALID_PARAM;
2708
2709                 ret = check_extract_flags(cmds[i].extract_flags,
2710                                           wim->hdr.flags,
2711                                           &cmds[i].extract_flags);
2712                 if (ret)
2713                         return ret;
2714                 if (cmds[i].extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2715                                              WIMLIB_EXTRACT_FLAG_HARDLINK)) {
2716                         found_link_cmd = true;
2717                 } else {
2718                         found_nolink_cmd = true;
2719                 }
2720                 if (found_link_cmd && found_nolink_cmd) {
2721                         ERROR("Symlink or hardlink extraction mode must "
2722                               "be set on all extraction commands");
2723                         return WIMLIB_ERR_INVALID_PARAM;
2724                 }
2725         }
2726
2727         /* Execute the extraction commands */
2728         for (size_t i = 0; i < num_cmds; i++) {
2729                 ret = extract_tree(wim,
2730                                    cmds[i].wim_source_path,
2731                                    cmds[i].fs_dest_path,
2732                                    cmds[i].extract_flags,
2733                                    progress_func);
2734                 if (ret)
2735                         return ret;
2736         }
2737         return 0;
2738 }
2739
2740 /* API function documented in wimlib.h  */
2741 WIMLIBAPI int
2742 wimlib_extract_files(WIMStruct *wim,
2743                      int image,
2744                      const struct wimlib_extract_command *cmds,
2745                      size_t num_cmds,
2746                      int default_extract_flags,
2747                      wimlib_progress_func_t progress_func)
2748 {
2749         int ret;
2750         struct wimlib_extract_command *cmds_copy;
2751         int all_flags = 0;
2752
2753         default_extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
2754
2755         if (num_cmds == 0)
2756                 return 0;
2757
2758         cmds_copy = CALLOC(num_cmds, sizeof(cmds[0]));
2759         if (!cmds_copy)
2760                 return WIMLIB_ERR_NOMEM;
2761
2762         for (size_t i = 0; i < num_cmds; i++) {
2763                 cmds_copy[i].extract_flags = (default_extract_flags |
2764                                                  cmds[i].extract_flags)
2765                                                 & WIMLIB_EXTRACT_MASK_PUBLIC;
2766                 all_flags |= cmds_copy[i].extract_flags;
2767
2768                 cmds_copy[i].wim_source_path = canonicalize_wim_path(cmds[i].wim_source_path);
2769                 if (!cmds_copy[i].wim_source_path) {
2770                         ret = WIMLIB_ERR_NOMEM;
2771                         goto out_free_cmds_copy;
2772                 }
2773
2774                 cmds_copy[i].fs_dest_path = canonicalize_fs_path(cmds[i].fs_dest_path);
2775                 if (!cmds_copy[i].fs_dest_path) {
2776                         ret = WIMLIB_ERR_NOMEM;
2777                         goto out_free_cmds_copy;
2778                 }
2779
2780         }
2781         ret = do_wimlib_extract_files(wim, image,
2782                                       cmds_copy, num_cmds,
2783                                       progress_func);
2784
2785         if (all_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2786                          WIMLIB_EXTRACT_FLAG_HARDLINK))
2787         {
2788                 for_lookup_table_entry(wim->lookup_table,
2789                                        lte_free_extracted_file, NULL);
2790         }
2791 out_free_cmds_copy:
2792         for (size_t i = 0; i < num_cmds; i++) {
2793                 FREE(cmds_copy[i].wim_source_path);
2794                 FREE(cmds_copy[i].fs_dest_path);
2795         }
2796         FREE(cmds_copy);
2797         return ret;
2798 }
2799
2800 /*
2801  * Extracts an image from a WIM file.
2802  *
2803  * @wim:                WIMStruct for the WIM file.
2804  *
2805  * @image:              Number of the single image to extract.
2806  *
2807  * @target:             Directory or NTFS volume to extract the image to.
2808  *
2809  * @extract_flags:      Bitwise or of WIMLIB_EXTRACT_FLAG_*.
2810  *
2811  * @progress_func:      If non-NULL, a progress function to be called
2812  *                      periodically.
2813  *
2814  * Returns 0 on success; nonzero on failure.
2815  */
2816 static int
2817 extract_single_image(WIMStruct *wim, int image,
2818                      const tchar *target, int extract_flags,
2819                      wimlib_progress_func_t progress_func)
2820 {
2821         int ret;
2822         tchar *target_copy = canonicalize_fs_path(target);
2823         if (target_copy == NULL)
2824                 return WIMLIB_ERR_NOMEM;
2825         struct wimlib_extract_command cmd = {
2826                 .wim_source_path = T(""),
2827                 .fs_dest_path = target_copy,
2828                 .extract_flags = extract_flags,
2829         };
2830         ret = do_wimlib_extract_files(wim, image, &cmd, 1, progress_func);
2831         FREE(target_copy);
2832         return ret;
2833 }
2834
2835 static const tchar * const filename_forbidden_chars =
2836 T(
2837 #ifdef __WIN32__
2838 "<>:\"/\\|?*"
2839 #else
2840 "/"
2841 #endif
2842 );
2843
2844 /* This function checks if it is okay to use a WIM image's name as a directory
2845  * name.  */
2846 static bool
2847 image_name_ok_as_dir(const tchar *image_name)
2848 {
2849         return image_name && *image_name &&
2850                 !tstrpbrk(image_name, filename_forbidden_chars) &&
2851                 tstrcmp(image_name, T(".")) &&
2852                 tstrcmp(image_name, T(".."));
2853 }
2854
2855 /* Extracts all images from the WIM to the directory @target, with the images
2856  * placed in subdirectories named by their image names. */
2857 static int
2858 extract_all_images(WIMStruct *wim,
2859                    const tchar *target,
2860                    int extract_flags,
2861                    wimlib_progress_func_t progress_func)
2862 {
2863         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
2864         size_t output_path_len = tstrlen(target);
2865         tchar buf[output_path_len + 1 + image_name_max_len + 1];
2866         int ret;
2867         int image;
2868         const tchar *image_name;
2869         struct stat stbuf;
2870
2871         extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
2872
2873         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2874                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
2875                 return WIMLIB_ERR_INVALID_PARAM;
2876         }
2877
2878         if (tstat(target, &stbuf)) {
2879                 if (errno == ENOENT) {
2880                         if (tmkdir(target, 0755)) {
2881                                 ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
2882                                 return WIMLIB_ERR_MKDIR;
2883                         }
2884                 } else {
2885                         ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
2886                         return WIMLIB_ERR_STAT;
2887                 }
2888         } else if (!S_ISDIR(stbuf.st_mode)) {
2889                 ERROR("\"%"TS"\" is not a directory", target);
2890                 return WIMLIB_ERR_NOTDIR;
2891         }
2892
2893         tmemcpy(buf, target, output_path_len);
2894         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
2895         for (image = 1; image <= wim->hdr.image_count; image++) {
2896                 image_name = wimlib_get_image_name(wim, image);
2897                 if (image_name_ok_as_dir(image_name)) {
2898                         tstrcpy(buf + output_path_len + 1, image_name);
2899                 } else {
2900                         /* Image name is empty or contains forbidden characters.
2901                          * Use image number instead. */
2902                         tsprintf(buf + output_path_len + 1, T("%d"), image);
2903                 }
2904                 ret = extract_single_image(wim, image, buf, extract_flags,
2905                                            progress_func);
2906                 if (ret)
2907                         return ret;
2908         }
2909         return 0;
2910 }
2911
2912 static int
2913 do_wimlib_extract_image(WIMStruct *wim,
2914                         int image,
2915                         const tchar *target,
2916                         int extract_flags,
2917                         wimlib_progress_func_t progress_func)
2918 {
2919         int ret;
2920
2921         if (image == WIMLIB_ALL_IMAGES) {
2922                 ret = extract_all_images(wim, target, extract_flags,
2923                                          progress_func);
2924         } else {
2925                 ret = extract_single_image(wim, image, target, extract_flags,
2926                                            progress_func);
2927         }
2928
2929         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2930                              WIMLIB_EXTRACT_FLAG_HARDLINK))
2931         {
2932                 for_lookup_table_entry(wim->lookup_table,
2933                                        lte_free_extracted_file,
2934                                        NULL);
2935         }
2936         return ret;
2937 }
2938
2939 /* API function documented in wimlib.h  */
2940 WIMLIBAPI int
2941 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
2942                                const tchar *target, int extract_flags,
2943                                wimlib_progress_func_t progress_func)
2944 {
2945         int ret;
2946         WIMStruct *pwm;
2947         struct filedes *in_fd;
2948         int image;
2949         unsigned i;
2950
2951         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
2952
2953         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT)
2954                 return WIMLIB_ERR_INVALID_PARAM;
2955
2956         /* Read the WIM header from the pipe and get a WIMStruct to represent
2957          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
2958          * wimlib_open_wim(), getting a WIMStruct in this way will result in
2959          * an empty lookup table, no XML data read, and no filename set.  */
2960         ret = open_wim_as_WIMStruct(&pipe_fd,
2961                                     WIMLIB_OPEN_FLAG_FROM_PIPE,
2962                                     &pwm, progress_func);
2963         if (ret)
2964                 return ret;
2965
2966         /* Sanity check to make sure this is a pipable WIM.  */
2967         if (pwm->hdr.magic != PWM_MAGIC) {
2968                 ERROR("The WIM being read from file descriptor %d "
2969                       "is not pipable!", pipe_fd);
2970                 ret = WIMLIB_ERR_NOT_PIPABLE;
2971                 goto out_wimlib_free;
2972         }
2973
2974         /* Sanity check to make sure the first part of a pipable split WIM is
2975          * sent over the pipe first.  */
2976         if (pwm->hdr.part_number != 1) {
2977                 ERROR("The first part of the split WIM must be "
2978                       "sent over the pipe first.");
2979                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
2980                 goto out_wimlib_free;
2981         }
2982
2983         in_fd = &pwm->in_fd;
2984         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
2985
2986         /* As mentioned, the WIMStruct we created from the pipe does not have
2987          * XML data yet.  Fix this by reading the extra copy of the XML data
2988          * that directly follows the header in pipable WIMs.  (Note: see
2989          * write_pipable_wim() for more details about the format of pipable
2990          * WIMs.)  */
2991         {
2992                 struct wim_lookup_table_entry xml_lte;
2993                 struct wim_resource_spec xml_rspec;
2994                 ret = read_pwm_stream_header(pwm, &xml_lte, &xml_rspec, 0, NULL);
2995                 if (ret)
2996                         goto out_wimlib_free;
2997
2998                 if (!(xml_lte.flags & WIM_RESHDR_FLAG_METADATA))
2999                 {
3000                         ERROR("Expected XML data, but found non-metadata "
3001                               "stream.");
3002                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
3003                         goto out_wimlib_free;
3004                 }
3005
3006                 wim_res_spec_to_hdr(&xml_rspec, &pwm->hdr.xml_data_reshdr);
3007
3008                 ret = read_wim_xml_data(pwm);
3009                 if (ret)
3010                         goto out_wimlib_free;
3011
3012                 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
3013                         ERROR("Image count in XML data is not the same as in WIM header.");
3014                         ret = WIMLIB_ERR_XML;
3015                         goto out_wimlib_free;
3016                 }
3017         }
3018
3019         /* Get image index (this may use the XML data that was just read to
3020          * resolve an image name).  */
3021         if (image_num_or_name) {
3022                 image = wimlib_resolve_image(pwm, image_num_or_name);
3023                 if (image == WIMLIB_NO_IMAGE) {
3024                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
3025                               image_num_or_name);
3026                         ret = WIMLIB_ERR_INVALID_IMAGE;
3027                         goto out_wimlib_free;
3028                 } else if (image == WIMLIB_ALL_IMAGES) {
3029                         ERROR("Applying all images from a pipe is not supported.");
3030                         ret = WIMLIB_ERR_INVALID_IMAGE;
3031                         goto out_wimlib_free;
3032                 }
3033         } else {
3034                 if (pwm->hdr.image_count != 1) {
3035                         ERROR("No image was specified, but the pipable WIM "
3036                               "did not contain exactly 1 image");
3037                         ret = WIMLIB_ERR_INVALID_IMAGE;
3038                         goto out_wimlib_free;
3039                 }
3040                 image = 1;
3041         }
3042
3043         /* Load the needed metadata resource.  */
3044         for (i = 1; i <= pwm->hdr.image_count; i++) {
3045                 struct wim_lookup_table_entry *metadata_lte;
3046                 struct wim_image_metadata *imd;
3047                 struct wim_resource_spec *metadata_rspec;
3048
3049                 metadata_lte = new_lookup_table_entry();
3050                 if (metadata_lte == NULL) {
3051                         ret = WIMLIB_ERR_NOMEM;
3052                         goto out_wimlib_free;
3053                 }
3054                 metadata_rspec = MALLOC(sizeof(struct wim_resource_spec));
3055                 if (metadata_rspec == NULL) {
3056                         ret = WIMLIB_ERR_NOMEM;
3057                         free_lookup_table_entry(metadata_lte);
3058                         goto out_wimlib_free;
3059                 }
3060
3061                 ret = read_pwm_stream_header(pwm, metadata_lte, metadata_rspec, 0, NULL);
3062                 imd = pwm->image_metadata[i - 1];
3063                 imd->metadata_lte = metadata_lte;
3064                 if (ret) {
3065                         FREE(metadata_rspec);
3066                         goto out_wimlib_free;
3067                 }
3068
3069                 if (!(metadata_lte->flags & WIM_RESHDR_FLAG_METADATA)) {
3070                         ERROR("Expected metadata resource, but found "
3071                               "non-metadata stream.");
3072                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
3073                         goto out_wimlib_free;
3074                 }
3075
3076                 if (i == image) {
3077                         /* Metadata resource is for the images being extracted.
3078                          * Parse it and save the metadata in memory.  */
3079                         ret = read_metadata_resource(pwm, imd);
3080                         if (ret)
3081                                 goto out_wimlib_free;
3082                         imd->modified = 1;
3083                 } else {
3084                         /* Metadata resource is not for the image being
3085                          * extracted.  Skip over it.  */
3086                         ret = skip_wim_stream(metadata_lte);
3087                         if (ret)
3088                                 goto out_wimlib_free;
3089                 }
3090         }
3091         /* Extract the image.  */
3092         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
3093         ret = do_wimlib_extract_image(pwm, image, target,
3094                                       extract_flags, progress_func);
3095         /* Clean up and return.  */
3096 out_wimlib_free:
3097         wimlib_free(pwm);
3098         return ret;
3099 }
3100
3101 /* API function documented in wimlib.h  */
3102 WIMLIBAPI int
3103 wimlib_extract_image(WIMStruct *wim,
3104                      int image,
3105                      const tchar *target,
3106                      int extract_flags,
3107                      wimlib_progress_func_t progress_func)
3108 {
3109         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
3110         return do_wimlib_extract_image(wim, image, target, extract_flags,
3111                                        progress_func);
3112 }
3113
3114 /* API function documented in wimlib.h  */
3115 WIMLIBAPI int
3116 wimlib_extract_pathlist(WIMStruct *wim, int image,
3117                         const tchar *target,
3118                         const tchar *path_list_file,
3119                         int extract_flags,
3120                         wimlib_progress_func_t progress_func)
3121 {
3122         int ret;
3123         tchar **paths;
3124         size_t num_paths;
3125         void *mem;
3126
3127         ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
3128         if (ret) {
3129                 ERROR("Failed to read path list file \"%"TS"\"",
3130                       path_list_file);
3131                 return ret;
3132         }
3133
3134         ret = wimlib_extract_paths(wim, image, target,
3135                                    (const tchar * const *)paths, num_paths,
3136                                    extract_flags, progress_func);
3137         FREE(paths);
3138         FREE(mem);
3139         return ret;
3140 }
3141
3142 struct append_dentry_ctx {
3143         struct wim_dentry **dentries;
3144         size_t num_dentries;
3145         size_t num_alloc_dentries;
3146 };
3147
3148 static int
3149 append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
3150 {
3151         struct append_dentry_ctx *ctx = _ctx;
3152
3153         if (ctx->num_dentries == ctx->num_alloc_dentries) {
3154                 struct wim_dentry **new_dentries;
3155                 size_t new_length;
3156
3157                 new_length = max(ctx->num_alloc_dentries + 8,
3158                                  ctx->num_alloc_dentries * 3 / 2);
3159                 new_dentries = REALLOC(ctx->dentries,
3160                                        new_length * sizeof(ctx->dentries[0]));
3161                 if (new_dentries == NULL)
3162                         return WIMLIB_ERR_NOMEM;
3163                 ctx->dentries = new_dentries;
3164                 ctx->num_alloc_dentries = new_length;
3165         }
3166         ctx->dentries[ctx->num_dentries++] = dentry;
3167         return 0;
3168 }
3169
3170 /* API function documented in wimlib.h  */
3171 WIMLIBAPI int
3172 wimlib_extract_paths(WIMStruct *wim,
3173                      int image,
3174                      const tchar *target,
3175                      const tchar * const *paths,
3176                      size_t num_paths,
3177                      int extract_flags,
3178                      wimlib_progress_func_t progress_func)
3179 {
3180         int ret;
3181         struct wim_dentry **trees;
3182         size_t num_trees;
3183
3184         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
3185
3186         if (wim == NULL || target == NULL || target[0] == T('\0') ||
3187             (num_paths != 0 && paths == NULL))
3188                 return WIMLIB_ERR_INVALID_PARAM;
3189
3190         ret = check_extract_flags(extract_flags, wim->hdr.flags,
3191                                   &extract_flags);
3192         if (ret)
3193                 return ret;
3194
3195         ret = select_wim_image(wim, image);
3196         if (ret)
3197                 return ret;
3198
3199         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
3200
3201                 struct append_dentry_ctx append_dentry_ctx = {
3202                         .dentries = NULL,
3203                         .num_dentries = 0,
3204                         .num_alloc_dentries = 0,
3205                 };
3206
3207                 u32 wildcard_flags = 0;
3208
3209                 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB)
3210                         wildcard_flags |= WILDCARD_FLAG_ERROR_IF_NO_MATCH;
3211                 else
3212                         wildcard_flags |= WILDCARD_FLAG_WARN_IF_NO_MATCH;
3213
3214                 if (default_ignore_case)
3215                         wildcard_flags |= WILDCARD_FLAG_CASE_INSENSITIVE;
3216
3217                 for (size_t i = 0; i < num_paths; i++) {
3218                         tchar *path = canonicalize_wim_path(paths[i]);
3219                         if (path == NULL) {
3220                                 ret = WIMLIB_ERR_NOMEM;
3221                                 trees = append_dentry_ctx.dentries;
3222                                 goto out_free_trees;
3223                         }
3224                         ret = expand_wildcard(wim, path,
3225                                               append_dentry_cb,
3226                                               &append_dentry_ctx,
3227                                               wildcard_flags);
3228                         FREE(path);
3229                         if (ret) {
3230                                 trees = append_dentry_ctx.dentries;
3231                                 goto out_free_trees;
3232                         }
3233                 }
3234                 trees = append_dentry_ctx.dentries;
3235                 num_trees = append_dentry_ctx.num_dentries;
3236         } else {
3237                 trees = MALLOC(num_paths * sizeof(trees[0]));
3238                 if (trees == NULL)
3239                         return WIMLIB_ERR_NOMEM;
3240
3241                 for (size_t i = 0; i < num_paths; i++) {
3242
3243                         tchar *path = canonicalize_wim_path(paths[i]);
3244                         if (path == NULL) {
3245                                 ret = WIMLIB_ERR_NOMEM;
3246                                 goto out_free_trees;
3247                         }
3248
3249                         trees[i] = get_dentry(wim, path,
3250                                               WIMLIB_CASE_PLATFORM_DEFAULT);
3251                         FREE(path);
3252                         if (trees[i] == NULL) {
3253                                   ERROR("Path \"%"TS"\" does not exist "
3254                                         "in WIM image %d",
3255                                         paths[i], wim->current_image);
3256                                   ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
3257                                   goto out_free_trees;
3258                         }
3259                 }
3260                 num_trees = num_paths;
3261         }
3262
3263         if (num_trees == 0) {
3264                 ret = 0;
3265                 goto out_free_trees;
3266         }
3267
3268         ret = extract_trees(wim, trees, num_trees,
3269                             T(""), target,
3270                             ((extract_flags &
3271                                 ~WIMLIB_EXTRACT_FLAG_GLOB_PATHS)
3272                                 | WIMLIB_EXTRACT_FLAG_PATHMODE),
3273                             progress_func);
3274
3275         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
3276                              WIMLIB_EXTRACT_FLAG_HARDLINK))
3277         {
3278                 for_lookup_table_entry(wim->lookup_table,
3279                                        lte_free_extracted_file,
3280                                        NULL);
3281         }
3282 out_free_trees:
3283         FREE(trees);
3284         return ret;
3285 }