]> wimlib.net Git - wimlib/blob - src/extract.c
88b277ac080fed49a61a40cb7a0aa52c6b487696
[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)
646 {
647         int ret;
648
649         if (ctx->ops->set_file_attributes) {
650                 if (dentry == ctx->extract_root && ctx->root_dentry_is_special)
651                         return 0;
652                 ret = ctx->ops->set_file_attributes(path,
653                                                     dentry->d_inode->i_attributes,
654                                                     ctx);
655                 if (ret) {
656                         ERROR_WITH_ERRNO("Failed to set attributes on "
657                                          "\"%"TS"\"", path);
658                         return ret;
659                 }
660         }
661         return 0;
662 }
663
664
665 /* Set or remove the short (DOS) name on an extracted file or directory if
666  * supported by the extraction mode.  Since DOS names are unimportant and it's
667  * easy to run into problems setting them on Windows (SetFileShortName()
668  * requires SE_RESTORE privilege, which only the Administrator can request, and
669  * also requires DELETE access to the file), failure is ignored unless
670  * WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES is set.  */
671 static int
672 extract_short_name(const tchar *path, struct apply_ctx *ctx,
673                    struct wim_dentry *dentry)
674 {
675         int ret;
676
677         /* The root of the dentry tree being extracted may not be extracted to
678          * its original name, so its short name should be ignored.  */
679         if (dentry == ctx->extract_root)
680                 return 0;
681
682         if (ctx->supported_features.short_names) {
683                 ret = ctx->ops->set_short_name(path,
684                                                dentry->short_name,
685                                                dentry->short_name_nbytes / 2,
686                                                ctx);
687                 if (ret && (ctx->extract_flags &
688                             WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES))
689                 {
690                         ERROR_WITH_ERRNO("Failed to set short name of "
691                                          "\"%"TS"\"", path);
692                         return ret;
693                 }
694         }
695         return 0;
696 }
697
698 /* Set security descriptor, UNIX data, or neither on an extracted file, taking
699  * into account the current extraction mode and flags.  */
700 static int
701 extract_security(const tchar *path, struct apply_ctx *ctx,
702                  struct wim_dentry *dentry)
703 {
704         int ret;
705         struct wim_inode *inode = dentry->d_inode;
706
707         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS)
708                 return 0;
709
710         if ((ctx->extract_root == dentry) && ctx->root_dentry_is_special)
711                 return 0;
712
713 #ifndef __WIN32__
714         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
715                 struct wimlib_unix_data data;
716
717                 ret = inode_get_unix_data(inode, &data, NULL);
718                 if (ret < 0)
719                         ret = 0;
720                 else if (ret == 0)
721                         ret = ctx->ops->set_unix_data(path, &data, ctx);
722                 if (ret) {
723                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) {
724                                 ERROR_WITH_ERRNO("Failed to set UNIX owner, "
725                                                  "group, and/or mode on "
726                                                  "\"%"TS"\"", path);
727                                 return ret;
728                         } else {
729                                 WARNING_WITH_ERRNO("Failed to set UNIX owner, "
730                                                    "group, and/or/mode on "
731                                                    "\"%"TS"\"", path);
732                         }
733                 }
734         }
735         else
736 #endif /* __WIN32__ */
737         if (ctx->supported_features.security_descriptors &&
738             inode->i_security_id != -1)
739         {
740                 const struct wim_security_data *sd;
741                 const u8 *desc;
742                 size_t desc_size;
743
744                 sd = wim_const_security_data(ctx->wim);
745                 desc = sd->descriptors[inode->i_security_id];
746                 desc_size = sd->sizes[inode->i_security_id];
747
748                 ret = ctx->ops->set_security_descriptor(path, desc,
749                                                         desc_size, ctx);
750                 if (ret) {
751                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) {
752                                 ERROR_WITH_ERRNO("Failed to set security "
753                                                  "descriptor on \"%"TS"\"", path);
754                                 return ret;
755                         } else {
756                                 if (errno != EACCES) {
757                                         WARNING_WITH_ERRNO("Failed to set "
758                                                            "security descriptor "
759                                                            "on \"%"TS"\"", path);
760                                 }
761                         }
762                 }
763         }
764         return 0;
765 }
766
767 /* Set timestamps on an extracted file.  Failure is warning-only unless
768  * WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS is set.  */
769 static int
770 extract_timestamps(const tchar *path, struct apply_ctx *ctx,
771                    struct wim_dentry *dentry)
772 {
773         struct wim_inode *inode = dentry->d_inode;
774         int ret;
775
776         if ((ctx->extract_root == dentry) && ctx->root_dentry_is_special)
777                 return 0;
778
779         if (ctx->ops->set_timestamps) {
780                 ret = ctx->ops->set_timestamps(path,
781                                                inode->i_creation_time,
782                                                inode->i_last_write_time,
783                                                inode->i_last_access_time,
784                                                ctx);
785                 if (ret) {
786                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) {
787                                 ERROR_WITH_ERRNO("Failed to set timestamps "
788                                                  "on \"%"TS"\"", path);
789                                 return ret;
790                         } else {
791                                 WARNING_WITH_ERRNO("Failed to set timestamps "
792                                                    "on \"%"TS"\"", path);
793                         }
794                 }
795         }
796         return 0;
797 }
798
799 /* Check whether the extraction of a dentry should be skipped completely.  */
800 static bool
801 dentry_is_supported(struct wim_dentry *dentry,
802                     const struct wim_features *supported_features)
803 {
804         struct wim_inode *inode = dentry->d_inode;
805
806         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
807                 if (supported_features->reparse_points)
808                         return true;
809                 if (supported_features->symlink_reparse_points &&
810                     inode_is_symlink(inode))
811                         return true;
812                 return false;
813         }
814         return true;
815 }
816
817 /* Given a WIM dentry to extract, build the path to which to extract it, in the
818  * format understood by the callbacks in the apply_operations being used.
819  *
820  * Write the resulting path into @path, which must have room for at least
821  * ctx->ops->max_path characters including the null-terminator.
822  *
823  * Return %true if successful; %false if this WIM dentry doesn't actually need
824  * to be extracted or if the calculated path exceeds ctx->ops->max_path
825  * characters.
826  *
827  * This function clobbers the tmp_list member of @dentry and its ancestors up
828  * until the extraction root.  */
829 static bool
830 build_extraction_path(tchar path[], struct wim_dentry *dentry,
831                       struct apply_ctx *ctx)
832 {
833         size_t path_nchars;
834         LIST_HEAD(ancestor_list);
835         tchar *p = path;
836         const tchar *target_prefix;
837         size_t target_prefix_nchars;
838         struct wim_dentry *d;
839
840         if (dentry->extraction_skipped)
841                 return false;
842
843         path_nchars = ctx->ops->path_prefix_nchars;
844
845         if (ctx->ops->requires_realtarget_in_paths) {
846                 target_prefix        = ctx->realtarget;
847                 target_prefix_nchars = ctx->realtarget_nchars;
848         } else if (ctx->ops->requires_target_in_paths) {
849                 target_prefix        = ctx->target;
850                 target_prefix_nchars = ctx->target_nchars;
851         } else {
852                 target_prefix        = NULL;
853                 target_prefix_nchars = 0;
854         }
855         path_nchars += target_prefix_nchars;
856
857         for (d = dentry; d != ctx->extract_root; d = d->parent) {
858                 path_nchars += d->extraction_name_nchars + 1;
859                 list_add(&d->tmp_list, &ancestor_list);
860         }
861
862         path_nchars++; /* null terminator */
863
864         if (path_nchars > ctx->ops->path_max) {
865                 WARNING("\"%"TS"\": Path too long to extract",
866                         dentry_full_path(dentry));
867                 return false;
868         }
869
870         p = tmempcpy(p, ctx->ops->path_prefix, ctx->ops->path_prefix_nchars);
871         p = tmempcpy(p, target_prefix, target_prefix_nchars);
872         list_for_each_entry(d, &ancestor_list, tmp_list) {
873                 *p++ = ctx->ops->path_separator;
874                 p = tmempcpy(p, d->extraction_name, d->extraction_name_nchars);
875         }
876         *p++ = T('\0');
877         wimlib_assert(p - path == path_nchars);
878         return true;
879 }
880
881 static unsigned
882 get_num_path_components(const tchar *path, tchar path_separator)
883 {
884         unsigned num_components = 0;
885
886         while (*path) {
887                 while (*path == path_separator)
888                         path++;
889                 if (*path)
890                         num_components++;
891                 while (*path && *path != path_separator)
892                         path++;
893         }
894         return num_components;
895 }
896
897 static int
898 extract_multiimage_symlink(const tchar *oldpath, const tchar *newpath,
899                            struct apply_ctx *ctx, struct wim_dentry *dentry)
900 {
901         size_t num_raw_path_components;
902         const struct wim_dentry *d;
903         size_t num_target_path_components;
904         tchar *p;
905         const tchar *p_old;
906         int ret;
907
908         num_raw_path_components = 0;
909         for (d = dentry; d != ctx->extract_root; d = d->parent)
910                 num_raw_path_components++;
911
912         if (ctx->ops->requires_realtarget_in_paths)
913                 num_target_path_components = get_num_path_components(ctx->realtarget,
914                                                                      ctx->ops->path_separator);
915         else if (ctx->ops->requires_target_in_paths)
916                 num_target_path_components = get_num_path_components(ctx->target,
917                                                                      ctx->ops->path_separator);
918         else
919                 num_target_path_components = 0;
920
921         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
922                 wimlib_assert(num_target_path_components > 0);
923                 num_raw_path_components++;
924                 num_target_path_components--;
925         }
926
927         p_old = oldpath;
928         while (*p_old == ctx->ops->path_separator)
929                 p_old++;
930         while (--num_target_path_components) {
931                 while (*p_old != ctx->ops->path_separator)
932                         p_old++;
933                 while (*p_old == ctx->ops->path_separator)
934                         p_old++;
935         }
936
937         tchar symlink_target[tstrlen(p_old) + 3 * num_raw_path_components + 1];
938
939         p = &symlink_target[0];
940         while (num_raw_path_components--) {
941                 *p++ = '.';
942                 *p++ = '.';
943                 *p++ = ctx->ops->path_separator;
944         }
945         tstrcpy(p, p_old);
946         DEBUG("Creating symlink \"%"TS"\" => \"%"TS"\"",
947               newpath, symlink_target);
948         ret = ctx->ops->create_symlink(symlink_target, newpath, ctx);
949         if (ret) {
950                 ERROR_WITH_ERRNO("Failed to create symlink "
951                                  "\"%"TS"\" => \"%"TS"\"",
952                                  newpath, symlink_target);
953         }
954         return ret;
955 }
956
957 /* Create the "skeleton" of an extracted file or directory.  Don't yet extract
958  * data streams, reparse data (including symbolic links), timestamps, and
959  * security descriptors.  Basically, everything that doesn't require reading
960  * non-metadata resources from the WIM file and isn't delayed until the final
961  * pass.  */
962 static int
963 do_dentry_extract_skeleton(tchar path[], struct wim_dentry *dentry,
964                            struct apply_ctx *ctx)
965 {
966         struct wim_inode *inode = dentry->d_inode;
967         int ret;
968         const tchar *oldpath;
969
970         if (unlikely(is_linked_extraction(ctx))) {
971                 struct wim_lookup_table_entry *unnamed_lte;
972
973                 unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode);
974                 if (unnamed_lte && unnamed_lte->extracted_file) {
975                         oldpath = unnamed_lte->extracted_file;
976                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK)
977                                 goto hardlink;
978                         else
979                                 goto symlink;
980                 }
981         }
982
983         /* Create hard link if this dentry corresponds to an already-extracted
984          * inode.  */
985         if (inode->i_extracted_file) {
986                 oldpath = inode->i_extracted_file;
987                 goto hardlink;
988         }
989
990         /* Skip symlinks unless they can be extracted as reparse points rather
991          * than created directly.  */
992         if (inode_is_symlink(inode) && !ctx->supported_features.reparse_points)
993                 return 0;
994
995         /* Create this file or directory unless it's the extraction root, which
996          * was already created if necessary.  */
997         if (dentry != ctx->extract_root) {
998                 ret = extract_inode(path, ctx, inode);
999                 if (ret)
1000                         return ret;
1001         }
1002
1003         /* Create empty named data streams.  */
1004         if (can_extract_named_data_streams(ctx)) {
1005                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1006                         file_spec_t file_spec;
1007                         struct wim_ads_entry *entry = &inode->i_ads_entries[i];
1008
1009                         if (!ads_entry_is_named_stream(entry))
1010                                 continue;
1011                         if (entry->lte)
1012                                 continue;
1013                         if (ctx->ops->uses_cookies)
1014                                 file_spec.cookie = inode->extract_cookie;
1015                         else
1016                                 file_spec.path = path;
1017                         ret = ctx->ops->extract_named_stream(file_spec,
1018                                                              entry->stream_name,
1019                                                              entry->stream_name_nbytes / 2,
1020                                                              entry->lte, ctx);
1021                         if (ret) {
1022                                 ERROR_WITH_ERRNO("\"%"TS"\": failed to create "
1023                                                  "empty named data stream",
1024                                                  path);
1025                                 return ret;
1026                         }
1027                 }
1028         }
1029
1030         /* Set file attributes (if supported).  */
1031         ret = extract_file_attributes(path, ctx, dentry);
1032         if (ret)
1033                 return ret;
1034
1035         /* Set or remove file short name (if supported).  */
1036         ret = extract_short_name(path, ctx, dentry);
1037         if (ret)
1038                 return ret;
1039
1040         /* If inode has multiple links and hard links are supported in this
1041          * extraction mode and volume, save the path to the extracted file in
1042          * case it's needed to create a hard link.  */
1043         if (unlikely(is_linked_extraction(ctx))) {
1044                 struct wim_lookup_table_entry *unnamed_lte;
1045
1046                 unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode);
1047                 if (unnamed_lte) {
1048                         unnamed_lte->extracted_file = TSTRDUP(path);
1049                         if (!unnamed_lte->extracted_file)
1050                                 return WIMLIB_ERR_NOMEM;
1051                 }
1052         } else if (inode->i_nlink > 1 && ctx->supported_features.hard_links) {
1053                 inode->i_extracted_file = TSTRDUP(path);
1054                 if (!inode->i_extracted_file)
1055                         return WIMLIB_ERR_NOMEM;
1056         }
1057         return 0;
1058
1059 symlink:
1060         ret = extract_multiimage_symlink(oldpath, path, ctx, dentry);
1061         if (ret)
1062                 return ret;
1063         dentry->was_hardlinked = 1;
1064         return 0;
1065
1066 hardlink:
1067         ret = extract_hardlink(oldpath, path, ctx);
1068         if (ret)
1069                 return ret;
1070         dentry->was_hardlinked = 1;
1071         return 0;
1072 }
1073
1074 static int
1075 dentry_extract_skeleton(struct wim_dentry *dentry, void *_ctx)
1076 {
1077         struct apply_ctx *ctx = _ctx;
1078         tchar path[ctx->ops->path_max];
1079         struct wim_dentry *orig_dentry;
1080         struct wim_dentry *other_dentry;
1081         int ret;
1082
1083         /* Here we may re-order the extraction of multiple names (hard links)
1084          * for the same file in the same directory in order to ensure the short
1085          * (DOS) name is set correctly.  A short name is always associated with
1086          * exactly one long name, and at least on NTFS, only one long name for a
1087          * file can have a short name associated with it.  (More specifically,
1088          * there can be unlimited names in the POSIX namespace, but only one
1089          * name can be in the Win32+DOS namespace, or one name in the Win32
1090          * namespace with a corresponding name in the DOS namespace.) To ensure
1091          * the short name of a file is associated with the correct long name in
1092          * a directory, we extract the long name with a corresponding short name
1093          * before any additional names.  This can affect NTFS-3g extraction
1094          * (which uses ntfs_set_ntfs_dos_name(), which doesn't allow specifying
1095          * the long name to associate with a short name) and may affect Win32
1096          * extraction as well (which uses SetFileShortName()).  */
1097
1098         if (dentry->skeleton_extracted)
1099                 return 0;
1100         orig_dentry = NULL;
1101         if (ctx->supported_features.short_names
1102             && !dentry_has_short_name(dentry)
1103             && !dentry->d_inode->i_dos_name_extracted)
1104         {
1105                 inode_for_each_dentry(other_dentry, dentry->d_inode) {
1106                         if (dentry_has_short_name(other_dentry)
1107                             && !other_dentry->skeleton_extracted
1108                             && other_dentry->parent == dentry->parent)
1109                         {
1110                                 DEBUG("Creating %"TS" before %"TS" "
1111                                       "to guarantee correct DOS name extraction",
1112                                       dentry_full_path(other_dentry),
1113                                       dentry_full_path(dentry));
1114                                 orig_dentry = dentry;
1115                                 dentry = other_dentry;
1116                                 break;
1117                         }
1118                 }
1119         }
1120 again:
1121         if (!build_extraction_path(path, dentry, ctx))
1122                 return 0;
1123         ret = do_dentry_extract_skeleton(path, dentry, ctx);
1124         if (ret)
1125                 return ret;
1126
1127         dentry->skeleton_extracted = 1;
1128
1129         if (orig_dentry) {
1130                 dentry = orig_dentry;
1131                 orig_dentry = NULL;
1132                 goto again;
1133         }
1134         dentry->d_inode->i_dos_name_extracted = 1;
1135         return 0;
1136 }
1137
1138 /* Create a file or directory, then immediately extract all streams.  This
1139  * assumes that WIMLIB_EXTRACT_FLAG_SEQUENTIAL is not specified, since the WIM
1140  * may not be read sequentially by this function.  */
1141 static int
1142 dentry_extract(struct wim_dentry *dentry, void *_ctx)
1143 {
1144         struct apply_ctx *ctx = _ctx;
1145         tchar path[ctx->ops->path_max];
1146         int ret;
1147
1148         ret = dentry_extract_skeleton(dentry, ctx);
1149         if (ret)
1150                 return ret;
1151
1152         if (!build_extraction_path(path, dentry, ctx))
1153                 return 0;
1154
1155         return extract_streams(path, ctx, dentry, NULL, NULL);
1156 }
1157
1158 /* Extract all instances of the stream @lte that are being extracted in this
1159  * call of extract_tree().  @can_seek specifies whether the WIM file descriptor
1160  * is seekable or not (e.g. is a pipe).  If not and the stream needs to be
1161  * extracted multiple times, it is extracted to a temporary file first.
1162  *
1163  * This is intended for use with sequential extraction of a WIM image
1164  * (WIMLIB_EXTRACT_FLAG_SEQUENTIAL specified).  */
1165 static int
1166 extract_stream_instances(struct wim_lookup_table_entry *lte,
1167                          struct apply_ctx *ctx, bool can_seek)
1168 {
1169         struct wim_dentry **lte_dentries;
1170         struct wim_lookup_table_entry *lte_tmp = NULL;
1171         struct wim_lookup_table_entry *lte_override;
1172         tchar *stream_tmp_filename = NULL;
1173         tchar path[ctx->ops->path_max];
1174         unsigned i;
1175         int ret;
1176
1177         if (lte->out_refcnt <= ARRAY_LEN(lte->inline_lte_dentries))
1178                 lte_dentries = lte->inline_lte_dentries;
1179         else
1180                 lte_dentries = lte->lte_dentries;
1181
1182         if (likely(can_seek || lte->out_refcnt < 2)) {
1183                 lte_override = lte;
1184         } else {
1185                 /* Need to extract stream to temporary file.  */
1186                 struct filedes fd;
1187                 int raw_fd;
1188
1189                 stream_tmp_filename = ttempnam(NULL, T("wimlib"));
1190                 if (!stream_tmp_filename) {
1191                         ERROR_WITH_ERRNO("Failed to create temporary filename");
1192                         ret = WIMLIB_ERR_OPEN;
1193                         goto out;
1194                 }
1195
1196                 lte_tmp = memdup(lte, sizeof(struct wim_lookup_table_entry));
1197                 if (!lte_tmp) {
1198                         ret = WIMLIB_ERR_NOMEM;
1199                         goto out_free_stream_tmp_filename;
1200                 }
1201                 lte_tmp->resource_location = RESOURCE_IN_FILE_ON_DISK;
1202                 lte_tmp->file_on_disk = stream_tmp_filename;
1203                 lte_override = lte_tmp;
1204
1205                 raw_fd = topen(stream_tmp_filename,
1206                                O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, 0600);
1207                 if (raw_fd < 0) {
1208                         ERROR_WITH_ERRNO("Failed to open temporary file");
1209                         ret = WIMLIB_ERR_OPEN;
1210                         goto out_free_lte_tmp;
1211                 }
1212                 filedes_init(&fd, raw_fd);
1213                 ret = extract_wim_resource_to_fd(lte, &fd,
1214                                                  wim_resource_size(lte));
1215                 if (filedes_close(&fd) && !ret)
1216                         ret = WIMLIB_ERR_WRITE;
1217                 if (ret)
1218                         goto out_unlink_stream_tmp_file;
1219         }
1220
1221         /* Extract all instances of the stream, reading either from the stream
1222          * in the WIM file or from the temporary file containing the stream.
1223          * dentry->tmp_flag is used to ensure that each dentry is processed only
1224          * once regardless of how many times this stream appears in the streams
1225          * of the corresponding inode.  */
1226         for (i = 0; i < lte->out_refcnt; i++) {
1227                 struct wim_dentry *dentry = lte_dentries[i];
1228
1229                 if (dentry->tmp_flag)
1230                         continue;
1231                 if (!build_extraction_path(path, dentry, ctx))
1232                         continue;
1233                 ret = extract_streams(path, ctx, dentry,
1234                                       lte, lte_override);
1235                 if (ret)
1236                         goto out_clear_tmp_flags;
1237                 dentry->tmp_flag = 1;
1238         }
1239         ret = 0;
1240 out_clear_tmp_flags:
1241         for (i = 0; i < lte->out_refcnt; i++)
1242                 lte_dentries[i]->tmp_flag = 0;
1243 out_unlink_stream_tmp_file:
1244         if (stream_tmp_filename)
1245                 tunlink(stream_tmp_filename);
1246 out_free_lte_tmp:
1247         FREE(lte_tmp);
1248 out_free_stream_tmp_filename:
1249         FREE(stream_tmp_filename);
1250 out:
1251         return ret;
1252 }
1253
1254 /* Extracts a list of streams (ctx.stream_list), assuming that the directory
1255  * structure and empty files were already created.  This relies on the
1256  * per-`struct wim_lookup_table_entry' list of dentries that reference each
1257  * stream that was constructed earlier.  Streams are extracted exactly in the
1258  * order of the stream list; however, unless the WIM's file descriptor is
1259  * detected to be non-seekable, streams may be read from the WIM file more than
1260  * one time if multiple copies need to be extracted.  */
1261 static int
1262 extract_stream_list(struct apply_ctx *ctx)
1263 {
1264         struct wim_lookup_table_entry *lte;
1265         bool can_seek;
1266         int ret;
1267
1268         can_seek = (lseek(ctx->wim->in_fd.fd, 0, SEEK_CUR) != -1);
1269         list_for_each_entry(lte, &ctx->stream_list, extraction_list) {
1270                 ret = extract_stream_instances(lte, ctx, can_seek);
1271                 if (ret)
1272                         return ret;
1273         }
1274         return 0;
1275 }
1276
1277 #define PWM_ALLOW_WIM_HDR 0x00001
1278 #define PWM_SILENT_EOF    0x00002
1279
1280 /* Read the header from a stream in a pipable WIM.  */
1281 static int
1282 read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte,
1283                        int flags, struct wim_header_disk *hdr_ret)
1284 {
1285         union {
1286                 struct pwm_stream_hdr stream_hdr;
1287                 struct wim_header_disk pwm_hdr;
1288         } buf;
1289         int ret;
1290
1291         ret = full_read(&pwm->in_fd, &buf.stream_hdr, sizeof(buf.stream_hdr));
1292         if (ret)
1293                 goto read_error;
1294
1295         if ((flags & PWM_ALLOW_WIM_HDR) && buf.stream_hdr.magic == PWM_MAGIC) {
1296                 BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr));
1297                 ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1,
1298                                 sizeof(buf.pwm_hdr) - sizeof(buf.stream_hdr));
1299
1300                 if (ret)
1301                         goto read_error;
1302                 lte->resource_location = RESOURCE_NONEXISTENT;
1303                 memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
1304                 return 0;
1305         }
1306
1307         if (buf.stream_hdr.magic != PWM_STREAM_MAGIC) {
1308                 ERROR("Data read on pipe is invalid (expected stream header).");
1309                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
1310         }
1311
1312         lte->resource_entry.original_size = le64_to_cpu(buf.stream_hdr.uncompressed_size);
1313         copy_hash(lte->hash, buf.stream_hdr.hash);
1314         lte->resource_entry.flags = le32_to_cpu(buf.stream_hdr.flags);
1315         lte->resource_entry.offset = pwm->in_fd.offset;
1316         lte->resource_location = RESOURCE_IN_WIM;
1317         lte->wim = pwm;
1318         if (lte->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED) {
1319                 lte->compression_type = pwm->compression_type;
1320                 lte->resource_entry.size = 0;
1321         } else {
1322                 lte->compression_type = WIMLIB_COMPRESSION_TYPE_NONE;
1323                 lte->resource_entry.size = lte->resource_entry.original_size;
1324         }
1325         lte->is_pipable = 1;
1326         return 0;
1327
1328 read_error:
1329         if (ret != WIMLIB_ERR_UNEXPECTED_END_OF_FILE || !(flags & PWM_SILENT_EOF))
1330                 ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
1331         return ret;
1332 }
1333
1334 /* Skip over an unneeded stream in a pipable WIM being read from a pipe.  */
1335 static int
1336 skip_pwm_stream(struct wim_lookup_table_entry *lte)
1337 {
1338         return read_partial_wim_resource(lte, wim_resource_size(lte),
1339                                          NULL, NULL,
1340                                          WIMLIB_READ_RESOURCE_FLAG_SEEK_ONLY,
1341                                          0);
1342 }
1343
1344 static int
1345 extract_streams_from_pipe(struct apply_ctx *ctx)
1346 {
1347         struct wim_lookup_table_entry *found_lte;
1348         struct wim_lookup_table_entry *needed_lte;
1349         struct wim_lookup_table *lookup_table;
1350         struct wim_header_disk pwm_hdr;
1351         int ret;
1352         int pwm_flags;
1353
1354         ret = WIMLIB_ERR_NOMEM;
1355         found_lte = new_lookup_table_entry();
1356         if (!found_lte)
1357                 goto out;
1358
1359         lookup_table = ctx->wim->lookup_table;
1360         pwm_flags = PWM_ALLOW_WIM_HDR;
1361         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1362                 pwm_flags |= PWM_SILENT_EOF;
1363         memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GID_LEN);
1364         ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
1365         ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
1366         if (ctx->progress_func)
1367                 ctx->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1368                                    &ctx->progress);
1369         while (ctx->num_streams_remaining) {
1370                 ret = read_pwm_stream_header(ctx->wim, found_lte, pwm_flags,
1371                                              &pwm_hdr);
1372                 if (ret) {
1373                         if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE &&
1374                             (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1375                         {
1376                                 goto resume_done;
1377                         }
1378                         goto out_free_found_lte;
1379                 }
1380
1381                 if ((found_lte->resource_location != RESOURCE_NONEXISTENT)
1382                     && !(found_lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA)
1383                     && (needed_lte = __lookup_resource(lookup_table, found_lte->hash))
1384                     && (needed_lte->out_refcnt))
1385                 {
1386                         copy_resource_entry(&needed_lte->resource_entry,
1387                                             &found_lte->resource_entry);
1388                         needed_lte->resource_location = found_lte->resource_location;
1389                         needed_lte->wim               = found_lte->wim;
1390                         needed_lte->compression_type  = found_lte->compression_type;
1391                         needed_lte->is_pipable        = found_lte->is_pipable;
1392
1393                         ret = extract_stream_instances(needed_lte, ctx, false);
1394                         if (ret)
1395                                 goto out_free_found_lte;
1396                         ctx->num_streams_remaining--;
1397                 } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) {
1398                         ret = skip_pwm_stream(found_lte);
1399                         if (ret)
1400                                 goto out_free_found_lte;
1401                 } else {
1402                         u16 part_number = le16_to_cpu(pwm_hdr.part_number);
1403                         u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
1404
1405                         if (part_number != ctx->progress.extract.part_number ||
1406                             total_parts != ctx->progress.extract.total_parts ||
1407                             memcmp(pwm_hdr.guid, ctx->progress.extract.guid,
1408                                    WIM_GID_LEN))
1409                         {
1410                                 ctx->progress.extract.part_number = part_number;
1411                                 ctx->progress.extract.total_parts = total_parts;
1412                                 memcpy(ctx->progress.extract.guid,
1413                                        pwm_hdr.guid, WIM_GID_LEN);
1414                                 if (ctx->progress_func) {
1415                                         ctx->progress_func(
1416                                                 WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1417                                                            &ctx->progress);
1418                                 }
1419
1420                         }
1421                 }
1422         }
1423         ret = 0;
1424 out_free_found_lte:
1425         free_lookup_table_entry(found_lte);
1426 out:
1427         return ret;
1428
1429 resume_done:
1430         /* TODO */
1431         return 0;
1432 }
1433
1434 /* Finish extracting a file, directory, or symbolic link by setting file
1435  * security and timestamps.  */
1436 static int
1437 dentry_extract_final(struct wim_dentry *dentry, void *_ctx)
1438 {
1439         struct apply_ctx *ctx = _ctx;
1440         int ret;
1441         tchar path[ctx->ops->path_max];
1442
1443         if (!build_extraction_path(path, dentry, ctx))
1444                 return 0;
1445
1446         ret = extract_security(path, ctx, dentry);
1447         if (ret)
1448                 return ret;
1449
1450         return extract_timestamps(path, ctx, dentry);
1451 }
1452
1453 /*
1454  * Extract a WIM dentry to standard output.
1455  *
1456  * This obviously doesn't make sense in all cases.  We return an error if the
1457  * dentry does not correspond to a regular file.  Otherwise we extract the
1458  * unnamed data stream only.
1459  */
1460 static int
1461 extract_dentry_to_stdout(struct wim_dentry *dentry)
1462 {
1463         int ret = 0;
1464         if (dentry->d_inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
1465                                              FILE_ATTRIBUTE_DIRECTORY))
1466         {
1467                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
1468                       "extracted to standard output", dentry_full_path(dentry));
1469                 ret = WIMLIB_ERR_NOT_A_REGULAR_FILE;
1470         } else {
1471                 struct wim_lookup_table_entry *lte;
1472
1473                 lte = inode_unnamed_lte_resolved(dentry->d_inode);
1474                 if (lte) {
1475                         struct filedes _stdout;
1476                         filedes_init(&_stdout, STDOUT_FILENO);
1477                         ret = extract_wim_resource_to_fd(lte, &_stdout,
1478                                                          wim_resource_size(lte));
1479                 }
1480         }
1481         return ret;
1482 }
1483
1484 #ifdef __WIN32__
1485 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
1486 #else
1487 static const utf16lechar replacement_char = cpu_to_le16('?');
1488 #endif
1489
1490 static bool
1491 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
1492 {
1493         size_t i;
1494
1495         if (num_chars == 0)
1496                 return true;
1497         for (i = 0; i < num_chars; i++) {
1498                 switch (name[i]) {
1499         #ifdef __WIN32__
1500                 case cpu_to_le16('\\'):
1501                 case cpu_to_le16(':'):
1502                 case cpu_to_le16('*'):
1503                 case cpu_to_le16('?'):
1504                 case cpu_to_le16('"'):
1505                 case cpu_to_le16('<'):
1506                 case cpu_to_le16('>'):
1507                 case cpu_to_le16('|'):
1508         #endif
1509                 case cpu_to_le16('/'):
1510                 case cpu_to_le16('\0'):
1511                         if (fix)
1512                                 name[i] = replacement_char;
1513                         else
1514                                 return false;
1515                 }
1516         }
1517
1518 #ifdef __WIN32__
1519         if (name[num_chars - 1] == cpu_to_le16(' ') ||
1520             name[num_chars - 1] == cpu_to_le16('.'))
1521         {
1522                 if (fix)
1523                         name[num_chars - 1] = replacement_char;
1524                 else
1525                         return false;
1526         }
1527 #endif
1528         return true;
1529 }
1530
1531 static bool
1532 dentry_is_dot_or_dotdot(const struct wim_dentry *dentry)
1533 {
1534         const utf16lechar *file_name = dentry->file_name;
1535         return file_name != NULL &&
1536                 file_name[0] == cpu_to_le16('.') &&
1537                 (file_name[1] == cpu_to_le16('\0') ||
1538                  (file_name[1] == cpu_to_le16('.') &&
1539                   file_name[2] == cpu_to_le16('\0')));
1540 }
1541
1542 static int
1543 dentry_mark_skipped(struct wim_dentry *dentry, void *_ignore)
1544 {
1545         dentry->extraction_skipped = 1;
1546         return 0;
1547 }
1548
1549 /*
1550  * dentry_calculate_extraction_path-
1551  *
1552  * Calculate the actual filename component at which a WIM dentry will be
1553  * extracted, handling invalid filenames "properly".
1554  *
1555  * dentry->extraction_name usually will be set the same as dentry->file_name (on
1556  * UNIX, converted into the platform's multibyte encoding).  However, if the
1557  * file name contains characters that are not valid on the current platform or
1558  * has some other format that is not valid, leave dentry->extraction_name as
1559  * NULL and set dentry->extraction_skipped to indicate that this dentry should
1560  * not be extracted, unless the appropriate flag
1561  * WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES is set in the extract flags, in
1562  * which case a substitute filename will be created and set instead.
1563  *
1564  * Conflicts with case-insensitive names on Windows are handled similarly; see
1565  * below.
1566  */
1567 static int
1568 dentry_calculate_extraction_path(struct wim_dentry *dentry, void *_args)
1569 {
1570         struct apply_ctx *ctx = _args;
1571         int ret;
1572
1573         if (dentry == ctx->extract_root || dentry->extraction_skipped)
1574                 return 0;
1575
1576         if (!dentry_is_supported(dentry, &ctx->supported_features))
1577                 goto skip_dentry;
1578
1579         if (dentry_is_dot_or_dotdot(dentry)) {
1580                 /* WIM files shouldn't contain . or .. entries.  But if they are
1581                  * there, don't attempt to extract them. */
1582                 WARNING("Skipping extraction of unexpected . or .. file "
1583                         "\"%"TS"\"", dentry_full_path(dentry));
1584                 goto skip_dentry;
1585         }
1586
1587 #ifdef __WIN32__
1588         if (!ctx->ops->supports_case_sensitive_filenames)
1589         {
1590                 struct wim_dentry *other;
1591                 list_for_each_entry(other, &dentry->case_insensitive_conflict_list,
1592                                     case_insensitive_conflict_list)
1593                 {
1594                         if (ctx->extract_flags &
1595                             WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
1596                                 WARNING("\"%"TS"\" has the same "
1597                                         "case-insensitive name as "
1598                                         "\"%"TS"\"; extracting "
1599                                         "dummy name instead",
1600                                         dentry_full_path(dentry),
1601                                         dentry_full_path(other));
1602                                 goto out_replace;
1603                         } else {
1604                                 WARNING("Not extracting \"%"TS"\": "
1605                                         "has same case-insensitive "
1606                                         "name as \"%"TS"\"",
1607                                         dentry_full_path(dentry),
1608                                         dentry_full_path(other));
1609                                 goto skip_dentry;
1610                         }
1611                 }
1612         }
1613 #else   /* __WIN32__ */
1614         wimlib_assert(ctx->ops->supports_case_sensitive_filenames);
1615 #endif  /* !__WIN32__ */
1616
1617         if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
1618 #ifdef __WIN32__
1619                 dentry->extraction_name = dentry->file_name;
1620                 dentry->extraction_name_nchars = dentry->file_name_nbytes / 2;
1621                 return 0;
1622 #else
1623                 return utf16le_to_tstr(dentry->file_name,
1624                                        dentry->file_name_nbytes,
1625                                        &dentry->extraction_name,
1626                                        &dentry->extraction_name_nchars);
1627 #endif
1628         } else {
1629                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
1630                 {
1631                         WARNING("\"%"TS"\" has an invalid filename "
1632                                 "that is not supported on this platform; "
1633                                 "extracting dummy name instead",
1634                                 dentry_full_path(dentry));
1635                         goto out_replace;
1636                 } else {
1637                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
1638                                 "that is not supported on this platform",
1639                                 dentry_full_path(dentry));
1640                         goto skip_dentry;
1641                 }
1642         }
1643
1644 out_replace:
1645         {
1646                 utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2];
1647
1648                 memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
1649                 file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
1650
1651                 tchar *tchar_name;
1652                 size_t tchar_nchars;
1653         #ifdef __WIN32__
1654                 tchar_name = utf16_name_copy;
1655                 tchar_nchars = dentry->file_name_nbytes / 2;
1656         #else
1657                 ret = utf16le_to_tstr(utf16_name_copy,
1658                                       dentry->file_name_nbytes,
1659                                       &tchar_name, &tchar_nchars);
1660                 if (ret)
1661                         return ret;
1662         #endif
1663                 size_t fixed_name_num_chars = tchar_nchars;
1664                 tchar fixed_name[tchar_nchars + 50];
1665
1666                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
1667                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
1668                                                  T(" (invalid filename #%lu)"),
1669                                                  ++ctx->invalid_sequence);
1670         #ifndef __WIN32__
1671                 FREE(tchar_name);
1672         #endif
1673                 dentry->extraction_name = memdup(fixed_name,
1674                                                  2 * fixed_name_num_chars + 2);
1675                 if (!dentry->extraction_name)
1676                         return WIMLIB_ERR_NOMEM;
1677                 dentry->extraction_name_nchars = fixed_name_num_chars;
1678         }
1679         return 0;
1680
1681 skip_dentry:
1682         for_dentry_in_tree(dentry, dentry_mark_skipped, NULL);
1683         return 0;
1684 }
1685
1686 /* Clean up dentry and inode structure after extraction.  */
1687 static int
1688 dentry_reset_needs_extraction(struct wim_dentry *dentry, void *_ignore)
1689 {
1690         struct wim_inode *inode = dentry->d_inode;
1691
1692         dentry->extraction_skipped = 0;
1693         dentry->was_hardlinked = 0;
1694         dentry->skeleton_extracted = 0;
1695         inode->i_visited = 0;
1696         FREE(inode->i_extracted_file);
1697         inode->i_extracted_file = NULL;
1698         inode->i_dos_name_extracted = 0;
1699         if ((void*)dentry->extraction_name != (void*)dentry->file_name)
1700                 FREE(dentry->extraction_name);
1701         dentry->extraction_name = NULL;
1702         return 0;
1703 }
1704
1705 /* Tally features necessary to extract a dentry and the corresponding inode.  */
1706 static int
1707 dentry_tally_features(struct wim_dentry *dentry, void *_features)
1708 {
1709         struct wim_features *features = _features;
1710         struct wim_inode *inode = dentry->d_inode;
1711
1712         if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
1713                 features->archive_files++;
1714         if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
1715                 features->hidden_files++;
1716         if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
1717                 features->system_files++;
1718         if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
1719                 features->compressed_files++;
1720         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED)
1721                 features->encrypted_files++;
1722         if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
1723                 features->not_context_indexed_files++;
1724         if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
1725                 features->sparse_files++;
1726         if (inode_has_named_stream(inode))
1727                 features->named_data_streams++;
1728         if (inode->i_visited)
1729                 features->hard_links++;
1730         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1731                 features->reparse_points++;
1732                 if (inode_is_symlink(inode))
1733                         features->symlink_reparse_points++;
1734                 else
1735                         features->other_reparse_points++;
1736         }
1737         if (inode->i_security_id != -1)
1738                 features->security_descriptors++;
1739         if (dentry->short_name_nbytes)
1740                 features->short_names++;
1741         if (inode_has_unix_data(inode))
1742                 features->unix_data++;
1743         inode->i_visited = 1;
1744         return 0;
1745 }
1746
1747 static int
1748 dentry_clear_inode_visited(struct wim_dentry *dentry, void *_ignore)
1749 {
1750         dentry->d_inode->i_visited = 0;
1751         return 0;
1752 }
1753
1754 /* Tally the features necessary to extract a dentry tree.  */
1755 static void
1756 dentry_tree_get_features(struct wim_dentry *root, struct wim_features *features)
1757 {
1758         memset(features, 0, sizeof(struct wim_features));
1759         for_dentry_in_tree(root, dentry_tally_features, features);
1760         for_dentry_in_tree(root, dentry_clear_inode_visited, NULL);
1761 }
1762
1763 static int
1764 do_feature_check(const struct wim_features *required_features,
1765                  const struct wim_features *supported_features,
1766                  int extract_flags,
1767                  const struct apply_operations *ops,
1768                  const tchar *wim_source_path)
1769 {
1770         const tchar *loc;
1771         const tchar *mode = "this extraction mode";
1772
1773         if (wim_source_path[0] == '\0')
1774                 loc = "the WIM image";
1775         else
1776                 loc = wim_source_path;
1777
1778         /* We're an archive program, so theoretically we can do what we want
1779          * with FILE_ATTRIBUTE_ARCHIVE (which is a dumb flag anyway).  Don't
1780          * bother the user about it.  */
1781 #if 0
1782         if (required_features->archive_files && !supported_features->archive_files)
1783         {
1784                 WARNING(
1785           "%lu files in %"TS" are marked as archived, but this attribute\n"
1786 "          is not supported in %"TS".",
1787                         required_features->archive_files, loc, mode);
1788         }
1789 #endif
1790
1791         if (required_features->hidden_files && !supported_features->hidden_files)
1792         {
1793                 WARNING(
1794           "%lu files in %"TS" are marked as hidden, but this\n"
1795 "          attribute is not supported in %"TS".",
1796                         required_features->hidden_files, loc, mode);
1797         }
1798
1799         if (required_features->system_files && !supported_features->system_files)
1800         {
1801                 WARNING(
1802           "%lu files in %"TS" are marked as system files,\n"
1803 "          but this attribute is not supported in %"TS".",
1804                         required_features->system_files, loc, mode);
1805         }
1806
1807         if (required_features->compressed_files && !supported_features->compressed_files)
1808         {
1809                 WARNING(
1810           "%lu files in %"TS" are marked as being transparently\n"
1811 "          compressed, but transparent compression is not supported in\n"
1812 "          %"TS".  These files will be extracted as uncompressed.",
1813                         required_features->compressed_files, loc, mode);
1814         }
1815
1816         if (required_features->encrypted_files && !supported_features->encrypted_files)
1817         {
1818                 WARNING(
1819           "%lu files in %"TS" are marked as being encrypted,\n"
1820 "           but encryption is not supported in %"TS".  These files\n"
1821 "           will be extracted as raw encrypted data instead.",
1822                         required_features->encrypted_files, loc, mode);
1823         }
1824
1825         if (required_features->not_context_indexed_files &&
1826             !supported_features->not_context_indexed_files)
1827         {
1828                 WARNING(
1829           "%lu files in %"TS" are marked as not content indexed,\n"
1830 "          but this attribute is not supported in %"TS".",
1831                         required_features->not_context_indexed_files, loc, mode);
1832         }
1833
1834         if (required_features->sparse_files && !supported_features->sparse_files)
1835         {
1836                 WARNING(
1837           "%lu files in %"TS" are marked as sparse, but creating\n"
1838 "           sparse files is not supported in %"TS".  These files\n"
1839 "           will be extracted as non-sparse.",
1840                         required_features->sparse_files, loc, mode);
1841         }
1842
1843         if (required_features->named_data_streams &&
1844             !supported_features->named_data_streams)
1845         {
1846                 WARNING(
1847           "%lu files in %"TS" contain one or more alternate (named)\n"
1848 "          data streams, which are not supported in %"TS".\n"
1849 "          Alternate data streams will NOT be extracted.",
1850                         required_features->named_data_streams, loc, mode);
1851         }
1852
1853         if (unlikely(extract_flags & (WIMLIB_EXTRACT_FLAG_HARDLINK |
1854                                       WIMLIB_EXTRACT_FLAG_SYMLINK)) &&
1855             required_features->named_data_streams &&
1856             supported_features->named_data_streams)
1857         {
1858                 WARNING(
1859           "%lu files in %"TS" contain one or more alternate (named)\n"
1860 "          data streams, which are not supported in linked extraction mode.\n"
1861 "          Alternate data streams will NOT be extracted.",
1862                         required_features->named_data_streams, loc);
1863         }
1864
1865         if (required_features->hard_links && !supported_features->hard_links)
1866         {
1867                 WARNING(
1868           "%lu files in %"TS" are hard links, but hard links are\n"
1869 "          not supported in %"TS".  Hard links will be extracted as\n"
1870 "          duplicate copies of the linked files.",
1871                         required_features->hard_links, loc, mode);
1872         }
1873
1874         if (required_features->reparse_points && !supported_features->reparse_points)
1875         {
1876                 if (supported_features->symlink_reparse_points) {
1877                         if (required_features->other_reparse_points) {
1878                                 WARNING(
1879           "%lu files in %"TS" are reparse points that are neither\n"
1880 "          symbolic links nor junction points and are not supported in\n"
1881 "          %"TS".  These reparse points will not be extracted.",
1882                                         required_features->other_reparse_points, loc,
1883                                         mode);
1884                         }
1885                 } else {
1886                         WARNING(
1887           "%lu files in %"TS" are reparse points, which are\n"
1888 "          not supported in %"TS" and will not be extracted.",
1889                                 required_features->reparse_points, loc, mode);
1890                 }
1891         }
1892
1893         if (required_features->security_descriptors &&
1894             !supported_features->security_descriptors)
1895         {
1896                 WARNING(
1897           "%lu files in %"TS" have Windows NT security descriptors,\n"
1898 "          but extracting security descriptors is not supported in\n"
1899 "          %"TS".  No security descriptors will be extracted.",
1900                         required_features->security_descriptors, loc, mode);
1901         }
1902
1903         if (required_features->short_names && !supported_features->short_names)
1904         {
1905                 WARNING(
1906           "%lu files in %"TS" have short (DOS) names, but\n"
1907 "          extracting short names is not supported in %"TS".\n"
1908 "          Short names will not be extracted.\n",
1909                         required_features->short_names, loc, mode);
1910         }
1911
1912         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
1913             required_features->unix_data && !supported_features->unix_data)
1914         {
1915                 ERROR("Extracting UNIX data is not supported in %"TS, mode);
1916                 return WIMLIB_ERR_UNSUPPORTED;
1917         }
1918         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) &&
1919             required_features->short_names && !supported_features->short_names)
1920         {
1921                 ERROR("Extracting short names is not supported in %"TS"", mode);
1922                 return WIMLIB_ERR_UNSUPPORTED;
1923         }
1924         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
1925             !ops->set_timestamps)
1926         {
1927                 ERROR("Extracting timestamps is not supported in %"TS"", mode);
1928                 return WIMLIB_ERR_UNSUPPORTED;
1929         }
1930         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
1931                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1932              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
1933             required_features->security_descriptors &&
1934             !supported_features->security_descriptors)
1935         {
1936                 ERROR("Extracting security descriptors is not supported in %"TS, mode);
1937                 return WIMLIB_ERR_UNSUPPORTED;
1938         }
1939
1940         if ((extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) &&
1941             !supported_features->hard_links)
1942         {
1943                 ERROR("Hard link extraction mode requested, but "
1944                       "%"TS" does not support hard links!", mode);
1945                 return WIMLIB_ERR_UNSUPPORTED;
1946         }
1947
1948         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
1949             required_features->symlink_reparse_points &&
1950             !(supported_features->symlink_reparse_points ||
1951               supported_features->reparse_points))
1952         {
1953                 ERROR("Extracting symbolic links is not supported in %"TS, mode);
1954                 return WIMLIB_ERR_UNSUPPORTED;
1955         }
1956
1957         if ((extract_flags & WIMLIB_EXTRACT_FLAG_SYMLINK) &&
1958             !supported_features->symlink_reparse_points)
1959         {
1960                 ERROR("Symbolic link extraction mode requested, but "
1961                       "%"TS" does not support symbolic "
1962                       "links!", mode);
1963                 return WIMLIB_ERR_UNSUPPORTED;
1964         }
1965         return 0;
1966 }
1967
1968 /*
1969  * extract_tree - Extract a file or directory tree from the currently selected
1970  *                WIM image.
1971  *
1972  * @wim:        WIMStruct for the WIM file, with the desired image selected
1973  *              (as wim->current_image).
1974  *
1975  * @wim_source_path:
1976  *              "Canonical" (i.e. no leading or trailing slashes, path
1977  *              separators WIM_PATH_SEPARATOR) path inside the WIM image to
1978  *              extract.  An empty string means the full image.
1979  *
1980  * @target:
1981  *              Filesystem path to extract the file or directory tree to.
1982  *              (Or, with WIMLIB_EXTRACT_FLAG_NTFS: the name of a NTFS volume.)
1983  *
1984  * @extract_flags:
1985  *              WIMLIB_EXTRACT_FLAG_*.  Also, the private flag
1986  *              WIMLIB_EXTRACT_FLAG_MULTI_IMAGE will be set if this is being
1987  *              called through wimlib_extract_image() with WIMLIB_ALL_IMAGES as
1988  *              the image.
1989  *
1990  * @progress_func:
1991  *              If non-NULL, progress function for the extraction.  The messages
1992  *              that may be sent in this function are:
1993  *
1994  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN or
1995  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
1996  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN;
1997  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END;
1998  *              WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY;
1999  *              WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS;
2000  *              WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS;
2001  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END or
2002  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END.
2003  *
2004  * Returns 0 on success; a positive WIMLIB_ERR_* code on failure.
2005  */
2006 static int
2007 extract_tree(WIMStruct *wim, const tchar *wim_source_path, const tchar *target,
2008              int extract_flags, wimlib_progress_func_t progress_func)
2009 {
2010         struct wim_dentry *root;
2011         struct wim_features required_features;
2012         struct apply_ctx ctx;
2013         int ret;
2014         struct wim_lookup_table_entry *lte;
2015
2016         /* Start initializing the apply_ctx.  */
2017         memset(&ctx, 0, sizeof(struct apply_ctx));
2018         ctx.wim = wim;
2019         ctx.extract_flags = extract_flags;
2020         ctx.target = target;
2021         ctx.target_nchars = tstrlen(target);
2022         ctx.progress_func = progress_func;
2023         if (progress_func) {
2024                 ctx.progress.extract.wimfile_name = wim->filename;
2025                 ctx.progress.extract.image = wim->current_image;
2026                 ctx.progress.extract.extract_flags = (extract_flags &
2027                                                       WIMLIB_EXTRACT_MASK_PUBLIC);
2028                 ctx.progress.extract.image_name = wimlib_get_image_name(wim,
2029                                                                         wim->current_image);
2030                 ctx.progress.extract.extract_root_wim_source_path = wim_source_path;
2031                 ctx.progress.extract.target = target;
2032         }
2033         INIT_LIST_HEAD(&ctx.stream_list);
2034
2035         /* Translate the path to extract into the corresponding
2036          * `struct wim_dentry', which will be the root of the
2037          * "dentry tree" to extract.  */
2038         root = get_dentry(wim, wim_source_path);
2039         if (!root) {
2040                 ERROR("Path \"%"TS"\" does not exist in WIM image %d",
2041                       wim_source_path, wim->current_image);
2042                 ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
2043                 goto out;
2044         }
2045
2046         ctx.extract_root = root;
2047
2048         /* Select the appropriate apply_operations based on the
2049          * platform and extract_flags.  */
2050 #ifdef __WIN32__
2051         ctx.ops = &win32_apply_ops;
2052 #else
2053         ctx.ops = &unix_apply_ops;
2054 #endif
2055
2056 #ifdef WITH_NTFS_3G
2057         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
2058                 ctx.ops = &ntfs_3g_apply_ops;
2059 #endif
2060
2061         /* Call the start_extract() callback.  This gives the apply_operations
2062          * implementation a chance to do any setup needed to access the volume.
2063          * Furthermore, it's expected to set the supported features of this
2064          * extraction mode (ctx.supported_features), which are determined at
2065          * runtime as they may vary depending on the actual volume.  These
2066          * features are then compared with the actual features extracting this
2067          * dentry tree requires.  Some mismatches will merely produce warnings
2068          * and the unsupported data will be ignored; others will produce errors.
2069          */
2070         ret = ctx.ops->start_extract(target, &ctx);
2071         if (ret)
2072                 goto out;
2073
2074         dentry_tree_get_features(root, &required_features);
2075         ret = do_feature_check(&required_features, &ctx.supported_features,
2076                                extract_flags, ctx.ops, wim_source_path);
2077         if (ret)
2078                 goto out_finish_or_abort_extract;
2079
2080         /* Figure out whether the root dentry is being extracted to the root of
2081          * a volume and therefore needs to be treated "specially", for example
2082          * not being explicitly created and not having attributes set.  */
2083         if (ctx.ops->target_is_root && ctx.ops->root_directory_is_special)
2084                 ctx.root_dentry_is_special = ctx.ops->target_is_root(target);
2085
2086         /* Calculate the actual filename component of each extracted dentry.  In
2087          * the process, set the dentry->extraction_skipped flag on dentries that
2088          * are being skipped for some reason (e.g. invalid filename).  */
2089         ret = for_dentry_in_tree(root, dentry_calculate_extraction_path, &ctx);
2090         if (ret)
2091                 goto out_dentry_reset_needs_extraction;
2092
2093         /* Build the list of the streams that need to be extracted and
2094          * initialize ctx.progress.extract with stream information.  */
2095         ret = for_dentry_in_tree(ctx.extract_root,
2096                                  dentry_resolve_and_zero_lte_refcnt, &ctx);
2097         if (ret)
2098                 goto out_dentry_reset_needs_extraction;
2099
2100         ret = for_dentry_in_tree(ctx.extract_root,
2101                                  dentry_add_streams_to_extract, &ctx);
2102         if (ret)
2103                 goto out_teardown_stream_list;
2104
2105         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2106                 /* When extracting from a pipe, the number of bytes of data to
2107                  * extract can't be determined in the normal way (examining the
2108                  * lookup table), since at this point all we have is a set of
2109                  * SHA1 message digests of streams that need to be extracted.
2110                  * However, we can get a reasonably accurate estimate by taking
2111                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
2112                  * data.  This does assume that a full image is being extracted,
2113                  * but currently there is no API for doing otherwise.  (Also,
2114                  * subtract <HARDLINKBYTES> from this if hard links are
2115                  * supported by the extraction mode.)  */
2116                 ctx.progress.extract.total_bytes =
2117                         wim_info_get_image_total_bytes(wim->wim_info,
2118                                                        wim->current_image);
2119                 if (ctx.supported_features.hard_links) {
2120                         ctx.progress.extract.total_bytes -=
2121                                 wim_info_get_image_hard_link_bytes(wim->wim_info,
2122                                                                    wim->current_image);
2123                 }
2124         }
2125
2126         /* Handle the special case of extracting a file to standard
2127          * output.  In that case, "root" should be a single file, not a
2128          * directory tree.  (If not, extract_dentry_to_stdout() will
2129          * return an error.)  */
2130         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
2131                 ret = extract_dentry_to_stdout(root);
2132                 goto out_teardown_stream_list;
2133         }
2134
2135         /* If a sequential extraction was specified, sort the streams to be
2136          * extracted by their position in the WIM file so that the WIM file can
2137          * be read sequentially.  */
2138         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SEQUENTIAL |
2139                               WIMLIB_EXTRACT_FLAG_FROM_PIPE))
2140                                         == WIMLIB_EXTRACT_FLAG_SEQUENTIAL)
2141         {
2142                 ret = sort_stream_list_by_sequential_order(
2143                                 &ctx.stream_list,
2144                                 offsetof(struct wim_lookup_table_entry,
2145                                          extraction_list));
2146                 if (ret)
2147                         goto out_teardown_stream_list;
2148         }
2149
2150         if (ctx.ops->realpath_works_on_nonexisting_files &&
2151             ((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) ||
2152              ctx.ops->requires_realtarget_in_paths))
2153         {
2154                 ctx.realtarget = realpath(target, NULL);
2155                 if (!ctx.realtarget) {
2156                         ret = WIMLIB_ERR_NOMEM;
2157                         goto out_teardown_stream_list;
2158                 }
2159                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2160         }
2161
2162         if (progress_func) {
2163                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN :
2164                                                  WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN,
2165                               &ctx.progress);
2166         }
2167
2168         if (!ctx.root_dentry_is_special)
2169         {
2170                 tchar path[ctx.ops->path_max];
2171                 if (build_extraction_path(path, root, &ctx))
2172                 {
2173                         ret = extract_inode(path, &ctx, root->d_inode);
2174                         if (ret)
2175                                 goto out_free_realtarget;
2176                 }
2177         }
2178
2179         /* If we need to fix up the targets of absolute symbolic links
2180          * (WIMLIB_EXTRACT_FLAG_RPFIX) or the extraction mode requires paths to
2181          * be absolute, use realpath() (or its replacement on Windows) to get
2182          * the absolute path to the extraction target.  Note that this requires
2183          * the target directory to exist, unless
2184          * realpath_works_on_nonexisting_files is set in the apply_operations.
2185          * */
2186         if (!ctx.realtarget &&
2187             (((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
2188               required_features.symlink_reparse_points) ||
2189              ctx.ops->requires_realtarget_in_paths))
2190         {
2191                 ctx.realtarget = realpath(target, NULL);
2192                 if (!ctx.realtarget) {
2193                         ret = WIMLIB_ERR_NOMEM;
2194                         goto out_free_realtarget;
2195                 }
2196                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2197         }
2198
2199         /* Finally, the important part: extract the tree of files.  */
2200         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SEQUENTIAL |
2201                              WIMLIB_EXTRACT_FLAG_FROM_PIPE)) {
2202                 /* Sequential extraction requested, so two passes are needed
2203                  * (one for directory structure, one for streams.)  */
2204                 if (progress_func)
2205                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2206                                       &ctx.progress);
2207
2208                 if (!(extract_flags & WIMLIB_EXTRACT_FLAG_RESUME)) {
2209                         ret = for_dentry_in_tree(root, dentry_extract_skeleton, &ctx);
2210                         if (ret)
2211                                 goto out_free_realtarget;
2212                 }
2213                 if (progress_func)
2214                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2215                                       &ctx.progress);
2216                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
2217                         ret = extract_streams_from_pipe(&ctx);
2218                 else
2219                         ret = extract_stream_list(&ctx);
2220                 if (ret)
2221                         goto out_free_realtarget;
2222         } else {
2223                 /* Sequential extraction was not requested, so we can make do
2224                  * with one pass where we both create the files and extract
2225                  * streams.   */
2226                 if (progress_func)
2227                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2228                                       &ctx.progress);
2229                 ret = for_dentry_in_tree(root, dentry_extract, &ctx);
2230                 if (ret)
2231                         goto out_free_realtarget;
2232                 if (progress_func)
2233                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2234                                       &ctx.progress);
2235         }
2236
2237         /* If the total number of bytes to extract was miscalculated, just jump
2238          * to the calculated number in order to avoid confusing the progress
2239          * function.  This should only occur when extracting from a pipe.  */
2240         if (ctx.progress.extract.completed_bytes != ctx.progress.extract.total_bytes)
2241         {
2242                 DEBUG("Calculated %"PRIu64" bytes to extract, but actually "
2243                       "extracted %"PRIu64,
2244                       ctx.progress.extract.total_bytes,
2245                       ctx.progress.extract.completed_bytes);
2246         }
2247         if (progress_func &&
2248             ctx.progress.extract.completed_bytes < ctx.progress.extract.total_bytes)
2249         {
2250                 ctx.progress.extract.completed_bytes = ctx.progress.extract.total_bytes;
2251                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, &ctx.progress);
2252         }
2253
2254         /* Apply security descriptors and timestamps.  This is done at the end,
2255          * and in a depth-first manner, to prevent timestamps from getting
2256          * changed by subsequent extract operations and to minimize the chance
2257          * of the restored security descriptors getting in our way.  */
2258         if (progress_func)
2259                 progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
2260                               &ctx.progress);
2261         ret = for_dentry_in_tree_depth(root, dentry_extract_final, &ctx);
2262         if (ret)
2263                 goto out_free_realtarget;
2264
2265         if (progress_func) {
2266                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END :
2267                               WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END,
2268                               &ctx.progress);
2269         }
2270
2271         ret = 0;
2272 out_free_realtarget:
2273         FREE(ctx.realtarget);
2274 out_teardown_stream_list:
2275         /* Free memory allocated as part of the mapping from each
2276          * wim_lookup_table_entry to the dentries that reference it.  */
2277         if (ctx.extract_flags & WIMLIB_EXTRACT_FLAG_SEQUENTIAL)
2278                 list_for_each_entry(lte, &ctx.stream_list, extraction_list)
2279                         if (lte->out_refcnt > ARRAY_LEN(lte->inline_lte_dentries))
2280                                 FREE(lte->lte_dentries);
2281 out_dentry_reset_needs_extraction:
2282         for_dentry_in_tree(root, dentry_reset_needs_extraction, NULL);
2283 out_finish_or_abort_extract:
2284         if (ret) {
2285                 if (ctx.ops->abort_extract)
2286                         ctx.ops->abort_extract(&ctx);
2287         } else {
2288                 if (ctx.ops->finish_extract)
2289                         ret = ctx.ops->finish_extract(&ctx);
2290         }
2291 out:
2292         return ret;
2293 }
2294
2295 /* Validates a single wimlib_extract_command, mostly checking to make sure the
2296  * extract flags make sense. */
2297 static int
2298 check_extract_command(struct wimlib_extract_command *cmd, int wim_header_flags)
2299 {
2300         int extract_flags;
2301
2302         /* Empty destination path? */
2303         if (cmd->fs_dest_path[0] == T('\0'))
2304                 return WIMLIB_ERR_INVALID_PARAM;
2305
2306         extract_flags = cmd->extract_flags;
2307
2308         /* Check for invalid flag combinations  */
2309         if ((extract_flags &
2310              (WIMLIB_EXTRACT_FLAG_SYMLINK |
2311               WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK |
2312                                                  WIMLIB_EXTRACT_FLAG_HARDLINK))
2313                 return WIMLIB_ERR_INVALID_PARAM;
2314
2315         if ((extract_flags &
2316              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2317               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2318                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
2319                 return WIMLIB_ERR_INVALID_PARAM;
2320
2321         if ((extract_flags &
2322              (WIMLIB_EXTRACT_FLAG_RPFIX |
2323               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
2324                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
2325                 return WIMLIB_ERR_INVALID_PARAM;
2326
2327         if ((extract_flags &
2328              (WIMLIB_EXTRACT_FLAG_RESUME |
2329               WIMLIB_EXTRACT_FLAG_FROM_PIPE)) == WIMLIB_EXTRACT_FLAG_RESUME)
2330                 return WIMLIB_ERR_INVALID_PARAM;
2331
2332         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2333 #ifndef WITH_NTFS_3G
2334                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
2335                       "        we cannot apply a WIM image directly to a NTFS volume.");
2336                 return WIMLIB_ERR_UNSUPPORTED;
2337 #endif
2338         }
2339
2340         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
2341                               WIMLIB_EXTRACT_FLAG_NORPFIX)) == 0)
2342         {
2343                 /* Do reparse point fixups by default if the WIM header says
2344                  * they are enabled and we are extracting a full image. */
2345                 if (wim_header_flags & WIM_HDR_FLAG_RP_FIX)
2346                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
2347         }
2348
2349         /* TODO: Since UNIX data entries are stored in the file resources, in a
2350          * completely sequential extraction they may come up before the
2351          * corresponding file or symbolic link data.  This needs to be handled
2352          * better.  */
2353         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_UNIX_DATA |
2354                               WIMLIB_EXTRACT_FLAG_SEQUENTIAL))
2355                                     == (WIMLIB_EXTRACT_FLAG_UNIX_DATA |
2356                                         WIMLIB_EXTRACT_FLAG_SEQUENTIAL))
2357         {
2358                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2359                         WARNING("Setting UNIX file/owner group may "
2360                                 "be impossible on some\n"
2361                                 "          symbolic links "
2362                                 "when applying from a pipe.");
2363                 } else {
2364                         extract_flags &= ~WIMLIB_EXTRACT_FLAG_SEQUENTIAL;
2365                         WARNING("Disabling sequential extraction for "
2366                                 "UNIX data mode");
2367                 }
2368         }
2369
2370         cmd->extract_flags = extract_flags;
2371         return 0;
2372 }
2373
2374
2375 /* Internal function to execute extraction commands for a WIM image.  The paths
2376  * in the extract commands are expected to be already "canonicalized".  */
2377 static int
2378 do_wimlib_extract_files(WIMStruct *wim,
2379                         int image,
2380                         struct wimlib_extract_command *cmds,
2381                         size_t num_cmds,
2382                         wimlib_progress_func_t progress_func)
2383 {
2384         int ret;
2385         bool found_link_cmd = false;
2386         bool found_nolink_cmd = false;
2387
2388         /* Select the image from which we are extracting files */
2389         ret = select_wim_image(wim, image);
2390         if (ret)
2391                 return ret;
2392
2393         /* Make sure there are no streams in the WIM that have not been
2394          * checksummed yet.  */
2395         ret = wim_checksum_unhashed_streams(wim);
2396         if (ret)
2397                 return ret;
2398
2399         /* Check for problems with the extraction commands */
2400         for (size_t i = 0; i < num_cmds; i++) {
2401                 ret = check_extract_command(&cmds[i], wim->hdr.flags);
2402                 if (ret)
2403                         return ret;
2404                 if (cmds[i].extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2405                                              WIMLIB_EXTRACT_FLAG_HARDLINK)) {
2406                         found_link_cmd = true;
2407                 } else {
2408                         found_nolink_cmd = true;
2409                 }
2410                 if (found_link_cmd && found_nolink_cmd) {
2411                         ERROR("Symlink or hardlink extraction mode must "
2412                               "be set on all extraction commands");
2413                         return WIMLIB_ERR_INVALID_PARAM;
2414                 }
2415         }
2416
2417         /* Execute the extraction commands */
2418         for (size_t i = 0; i < num_cmds; i++) {
2419                 ret = extract_tree(wim,
2420                                    cmds[i].wim_source_path,
2421                                    cmds[i].fs_dest_path,
2422                                    cmds[i].extract_flags,
2423                                    progress_func);
2424                 if (ret)
2425                         return ret;
2426         }
2427         return 0;
2428 }
2429
2430 /* API function documented in wimlib.h  */
2431 WIMLIBAPI int
2432 wimlib_extract_files(WIMStruct *wim,
2433                      int image,
2434                      const struct wimlib_extract_command *cmds,
2435                      size_t num_cmds,
2436                      int default_extract_flags,
2437                      WIMStruct **additional_swms,
2438                      unsigned num_additional_swms,
2439                      wimlib_progress_func_t progress_func)
2440 {
2441         int ret;
2442         struct wimlib_extract_command *cmds_copy;
2443         int all_flags = 0;
2444
2445         default_extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
2446
2447         ret = verify_swm_set(wim, additional_swms, num_additional_swms);
2448         if (ret)
2449                 goto out;
2450
2451         if (num_cmds == 0)
2452                 goto out;
2453
2454         if (num_additional_swms)
2455                 merge_lookup_tables(wim, additional_swms, num_additional_swms);
2456
2457         cmds_copy = CALLOC(num_cmds, sizeof(cmds[0]));
2458         if (!cmds_copy) {
2459                 ret = WIMLIB_ERR_NOMEM;
2460                 goto out_restore_lookup_table;
2461         }
2462
2463         for (size_t i = 0; i < num_cmds; i++) {
2464                 cmds_copy[i].extract_flags = (default_extract_flags |
2465                                                  cmds[i].extract_flags)
2466                                                 & WIMLIB_EXTRACT_MASK_PUBLIC;
2467                 all_flags |= cmds_copy[i].extract_flags;
2468
2469                 cmds_copy[i].wim_source_path = canonicalize_wim_path(cmds[i].wim_source_path);
2470                 if (!cmds_copy[i].wim_source_path) {
2471                         ret = WIMLIB_ERR_NOMEM;
2472                         goto out_free_cmds_copy;
2473                 }
2474
2475                 cmds_copy[i].fs_dest_path = canonicalize_fs_path(cmds[i].fs_dest_path);
2476                 if (!cmds_copy[i].fs_dest_path) {
2477                         ret = WIMLIB_ERR_NOMEM;
2478                         goto out_free_cmds_copy;
2479                 }
2480
2481         }
2482         ret = do_wimlib_extract_files(wim, image,
2483                                       cmds_copy, num_cmds,
2484                                       progress_func);
2485
2486         if (all_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2487                          WIMLIB_EXTRACT_FLAG_HARDLINK))
2488         {
2489                 for_lookup_table_entry(wim->lookup_table,
2490                                        lte_free_extracted_file, NULL);
2491         }
2492 out_free_cmds_copy:
2493         for (size_t i = 0; i < num_cmds; i++) {
2494                 FREE(cmds_copy[i].wim_source_path);
2495                 FREE(cmds_copy[i].fs_dest_path);
2496         }
2497         FREE(cmds_copy);
2498 out_restore_lookup_table:
2499         if (num_additional_swms)
2500                 unmerge_lookup_table(wim);
2501 out:
2502         return ret;
2503 }
2504
2505 /*
2506  * Extracts an image from a WIM file.
2507  *
2508  * @wim:                WIMStruct for the WIM file.
2509  *
2510  * @image:              Number of the single image to extract.
2511  *
2512  * @target:             Directory or NTFS volume to extract the image to.
2513  *
2514  * @extract_flags:      Bitwise or of WIMLIB_EXTRACT_FLAG_*.
2515  *
2516  * @progress_func:      If non-NULL, a progress function to be called
2517  *                      periodically.
2518  *
2519  * Returns 0 on success; nonzero on failure.
2520  */
2521 static int
2522 extract_single_image(WIMStruct *wim, int image,
2523                      const tchar *target, int extract_flags,
2524                      wimlib_progress_func_t progress_func)
2525 {
2526         int ret;
2527         tchar *target_copy = canonicalize_fs_path(target);
2528         if (!target_copy)
2529                 return WIMLIB_ERR_NOMEM;
2530         struct wimlib_extract_command cmd = {
2531                 .wim_source_path = T(""),
2532                 .fs_dest_path = target_copy,
2533                 .extract_flags = extract_flags,
2534         };
2535         ret = do_wimlib_extract_files(wim, image, &cmd, 1, progress_func);
2536         FREE(target_copy);
2537         return ret;
2538 }
2539
2540 static const tchar * const filename_forbidden_chars =
2541 T(
2542 #ifdef __WIN32__
2543 "<>:\"/\\|?*"
2544 #else
2545 "/"
2546 #endif
2547 );
2548
2549 /* This function checks if it is okay to use a WIM image's name as a directory
2550  * name.  */
2551 static bool
2552 image_name_ok_as_dir(const tchar *image_name)
2553 {
2554         return image_name && *image_name &&
2555                 !tstrpbrk(image_name, filename_forbidden_chars) &&
2556                 tstrcmp(image_name, T(".")) &&
2557                 tstrcmp(image_name, T(".."));
2558 }
2559
2560 /* Extracts all images from the WIM to the directory @target, with the images
2561  * placed in subdirectories named by their image names. */
2562 static int
2563 extract_all_images(WIMStruct *wim,
2564                    const tchar *target,
2565                    int extract_flags,
2566                    wimlib_progress_func_t progress_func)
2567 {
2568         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
2569         size_t output_path_len = tstrlen(target);
2570         tchar buf[output_path_len + 1 + image_name_max_len + 1];
2571         int ret;
2572         int image;
2573         const tchar *image_name;
2574         struct stat stbuf;
2575
2576         extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
2577
2578         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2579                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
2580                 return WIMLIB_ERR_INVALID_PARAM;
2581         }
2582
2583         if (tstat(target, &stbuf)) {
2584                 if (errno == ENOENT) {
2585                         if (tmkdir(target, 0755)) {
2586                                 ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
2587                                 return WIMLIB_ERR_MKDIR;
2588                         }
2589                 } else {
2590                         ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
2591                         return WIMLIB_ERR_STAT;
2592                 }
2593         } else if (!S_ISDIR(stbuf.st_mode)) {
2594                 ERROR("\"%"TS"\" is not a directory", target);
2595                 return WIMLIB_ERR_NOTDIR;
2596         }
2597
2598         tmemcpy(buf, target, output_path_len);
2599         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
2600         for (image = 1; image <= wim->hdr.image_count; image++) {
2601                 image_name = wimlib_get_image_name(wim, image);
2602                 if (image_name_ok_as_dir(image_name)) {
2603                         tstrcpy(buf + output_path_len + 1, image_name);
2604                 } else {
2605                         /* Image name is empty or contains forbidden characters.
2606                          * Use image number instead. */
2607                         tsprintf(buf + output_path_len + 1, T("%d"), image);
2608                 }
2609                 ret = extract_single_image(wim, image, buf, extract_flags,
2610                                            progress_func);
2611                 if (ret)
2612                         return ret;
2613         }
2614         return 0;
2615 }
2616
2617 static int
2618 do_wimlib_extract_image(WIMStruct *wim,
2619                         int image,
2620                         const tchar *target,
2621                         int extract_flags,
2622                         WIMStruct **additional_swms,
2623                         unsigned num_additional_swms,
2624                         wimlib_progress_func_t progress_func)
2625 {
2626         int ret;
2627
2628         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2629                 wimlib_assert(wim->hdr.part_number == 1);
2630                 wimlib_assert(num_additional_swms == 0);
2631         } else {
2632                 ret = verify_swm_set(wim, additional_swms, num_additional_swms);
2633                 if (ret)
2634                         return ret;
2635
2636                 if (num_additional_swms)
2637                         merge_lookup_tables(wim, additional_swms, num_additional_swms);
2638         }
2639
2640         if (image == WIMLIB_ALL_IMAGES) {
2641                 ret = extract_all_images(wim, target, extract_flags,
2642                                          progress_func);
2643         } else {
2644                 ret = extract_single_image(wim, image, target, extract_flags,
2645                                            progress_func);
2646         }
2647
2648         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2649                              WIMLIB_EXTRACT_FLAG_HARDLINK))
2650         {
2651                 for_lookup_table_entry(wim->lookup_table,
2652                                        lte_free_extracted_file,
2653                                        NULL);
2654         }
2655         if (num_additional_swms)
2656                 unmerge_lookup_table(wim);
2657         return ret;
2658 }
2659
2660 /* API function documented in wimlib.h  */
2661 WIMLIBAPI int
2662 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
2663                                const tchar *target, int extract_flags,
2664                                wimlib_progress_func_t progress_func)
2665 {
2666         int ret;
2667         WIMStruct *pwm;
2668         struct filedes *in_fd;
2669         int image;
2670         unsigned i;
2671
2672         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
2673
2674         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT)
2675                 return WIMLIB_ERR_INVALID_PARAM;
2676
2677         extract_flags |= WIMLIB_EXTRACT_FLAG_SEQUENTIAL;
2678
2679         /* Read the WIM header from the pipe and get a WIMStruct to represent
2680          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
2681          * wimlib_open_wim(), getting a WIMStruct in this way will result in
2682          * an empty lookup table, no XML data read, and no filename set.  */
2683         ret = open_wim_as_WIMStruct(&pipe_fd,
2684                                     WIMLIB_OPEN_FLAG_FROM_PIPE |
2685                                                 WIMLIB_OPEN_FLAG_SPLIT_OK,
2686                                     &pwm, progress_func);
2687         if (ret)
2688                 return ret;
2689
2690         /* Sanity check to make sure this is a pipable WIM.  */
2691         if (pwm->hdr.magic != PWM_MAGIC) {
2692                 ERROR("The WIM being read from file descriptor %d "
2693                       "is not pipable!", pipe_fd);
2694                 ret = WIMLIB_ERR_NOT_PIPABLE;
2695                 goto out_wimlib_free;
2696         }
2697
2698         /* Sanity check to make sure the first part of a pipable split WIM is
2699          * sent over the pipe first.  */
2700         if (pwm->hdr.part_number != 1) {
2701                 ERROR("The first part of the split WIM must be "
2702                       "sent over the pipe first.");
2703                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
2704                 goto out_wimlib_free;
2705         }
2706
2707         in_fd = &pwm->in_fd;
2708         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
2709
2710         /* As mentioned, the WIMStruct we created from the pipe does not have
2711          * XML data yet.  Fix this by reading the extra copy of the XML data
2712          * that directly follows the header in pipable WIMs.  (Note: see
2713          * write_pipable_wim() for more details about the format of pipable
2714          * WIMs.)  */
2715         {
2716                 struct wim_lookup_table_entry xml_lte;
2717                 ret = read_pwm_stream_header(pwm, &xml_lte, 0, NULL);
2718                 if (ret)
2719                         goto out_wimlib_free;
2720
2721                 if (!(xml_lte.resource_entry.flags & WIM_RESHDR_FLAG_METADATA))
2722                 {
2723                         ERROR("Expected XML data, but found non-metadata "
2724                               "stream.");
2725                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
2726                         goto out_wimlib_free;
2727                 }
2728
2729                 copy_resource_entry(&pwm->hdr.xml_res_entry,
2730                                     &xml_lte.resource_entry);
2731
2732                 ret = read_wim_xml_data(pwm);
2733                 if (ret)
2734                         goto out_wimlib_free;
2735                 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
2736                         ERROR("Image count in XML data is not the same as in WIM header.");
2737                         ret = WIMLIB_ERR_XML;
2738                         goto out_wimlib_free;
2739                 }
2740         }
2741
2742         /* Get image index (this may use the XML data that was just read to
2743          * resolve an image name).  */
2744         if (image_num_or_name) {
2745                 image = wimlib_resolve_image(pwm, image_num_or_name);
2746                 if (image == WIMLIB_NO_IMAGE) {
2747                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
2748                               image_num_or_name);
2749                         ret = WIMLIB_ERR_INVALID_IMAGE;
2750                         goto out_wimlib_free;
2751                 } else if (image == WIMLIB_ALL_IMAGES) {
2752                         ERROR("Applying all images from a pipe is not supported.");
2753                         ret = WIMLIB_ERR_INVALID_IMAGE;
2754                         goto out_wimlib_free;
2755                 }
2756         } else {
2757                 if (pwm->hdr.image_count != 1) {
2758                         ERROR("No image was specified, but the pipable WIM "
2759                               "did not contain exactly 1 image");
2760                         ret = WIMLIB_ERR_INVALID_IMAGE;
2761                         goto out_wimlib_free;
2762                 }
2763                 image = 1;
2764         }
2765
2766         /* Load the needed metadata resource.  */
2767         for (i = 1; i <= pwm->hdr.image_count; i++) {
2768                 struct wim_lookup_table_entry *metadata_lte;
2769                 struct wim_image_metadata *imd;
2770
2771                 metadata_lte = new_lookup_table_entry();
2772                 if (!metadata_lte) {
2773                         ret = WIMLIB_ERR_NOMEM;
2774                         goto out_wimlib_free;
2775                 }
2776
2777                 ret = read_pwm_stream_header(pwm, metadata_lte, 0, NULL);
2778                 imd = pwm->image_metadata[i - 1];
2779                 imd->metadata_lte = metadata_lte;
2780                 if (ret)
2781                         goto out_wimlib_free;
2782
2783                 if (!(metadata_lte->resource_entry.flags &
2784                       WIM_RESHDR_FLAG_METADATA))
2785                 {
2786                         ERROR("Expected metadata resource, but found "
2787                               "non-metadata stream.");
2788                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
2789                         goto out_wimlib_free;
2790                 }
2791
2792                 if (i == image) {
2793                         /* Metadata resource is for the images being extracted.
2794                          * Parse it and save the metadata in memory.  */
2795                         ret = read_metadata_resource(pwm, imd);
2796                         if (ret)
2797                                 goto out_wimlib_free;
2798                         imd->modified = 1;
2799                 } else {
2800                         /* Metadata resource is not for the image being
2801                          * extracted.  Skip over it.  */
2802                         ret = skip_pwm_stream(metadata_lte);
2803                         if (ret)
2804                                 goto out_wimlib_free;
2805                 }
2806         }
2807         /* Extract the image.  */
2808         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
2809         ret = do_wimlib_extract_image(pwm, image, target,
2810                                       extract_flags, NULL, 0, progress_func);
2811         /* Clean up and return.  */
2812 out_wimlib_free:
2813         wimlib_free(pwm);
2814         return ret;
2815 }
2816
2817 /* API function documented in wimlib.h  */
2818 WIMLIBAPI int
2819 wimlib_extract_image(WIMStruct *wim,
2820                      int image,
2821                      const tchar *target,
2822                      int extract_flags,
2823                      WIMStruct **additional_swms,
2824                      unsigned num_additional_swms,
2825                      wimlib_progress_func_t progress_func)
2826 {
2827         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
2828         return do_wimlib_extract_image(wim, image, target, extract_flags,
2829                                        additional_swms, num_additional_swms,
2830                                        progress_func);
2831 }