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